PLoS Biology

Ni, Jun;Yang, Jiaxin;Ma, Yina

doi: 10.1371/journal.pbio.3002545pmid: 38502637

Introduction Most social groups, from the basic family unit to professional organizations, and societal institutions, are hierarchically structured [1,2]. The hierarchical structure and different status relationships are one of the most fundamental features of social groups and shape interpersonal interactions among group members [3,4] to facilitate group stability [5] and maximize group productivity [6]. However, hierarchical structure comes with challenges and costs for social groups [7]. In hierarchical groups, high-ranking individuals may bully subordinates and usurp a disproportionate share of resources, social influence, and reproductive opportunities [8,9], which may amplify intragroup inequality and competitions [10], undermine the authority and legitimacy of group leaders [7]. Small groups overcome these problems through in-group social bonding [11,12], an adaptive means of forming, strengthening, and maintaining interpersonal connections with in-group members [13–15]. Social bonding exercises, such as grooming behaviors in nonhuman primate, collective rituals, traditions, and team-building activities in human society [16–19], have been widely adopted to facilitate leader influence [20], increase group cohesion [21], and reinforce the hierarchical structure [22]. Yet, despite the significance and impact of in-group social bonding, the neurocognitive mechanisms underlying the effects of social bonding on hierarchical groups remain largely unknown. Specifically, we asked how social bonding facilitated interpersonal interactions within hierarchical groups and examined here at both the behavioral and neural levels. The hierarchical structure places individuals at different positions in the group (i.e., individuals with different social statuses), such as group leader and followers, varying in levels of resources, social influence, or competence [10,23]. Sensitivity to status information and recognizing one’s relative social status in the group are essential for successful social interaction, and interpersonal interaction within a hierarchical group is shaped by different status relationships [3]. Interpersonal interactions within a simply structured hierarchical group are thus classified into 2 fundamental, status-related types: (i) interactions between individuals of different social status, e.g., leader–follower interaction (henceforth inter-status interaction); and (ii) interactions between individuals of the same social status, e.g., follower–follower interaction, the most common representative of intra-status interactions. We further asked whether and if so, how social bonding differentially influenced these 2 types of interpersonal interactions within a hierarchical group. Previous studies have demonstrated distinct functions of inter-status and intra-status interactions within hierarchical groups [24,25]. Inter-status interaction facilitates the exchange of asymmetric information between the group leader and followers [26], as well as leader–follower coordination and alignment [27]. On the other hand, intra-status interaction fosters reciprocal relationships and peer support among individuals with similar social status (e.g., followers) [1,28]. Therefore, we expected that social bonding would exert different effects on inter-status and intra-status interpersonal interactions within a hierarchical group. Specifically, we examined whether social bonding would exhibit stronger or weaker effects or even opposite effects on inter- versus intra-status interactions. At the behavioral level, we tested whether social bonding facilitated inter-status interaction, intra-status interaction, or both, and if so, whether such bonding effect on intra- and/or inter-status interaction was linked to the facilitation of in-group cohesion and leader’s influence. Moreover, in-group social bonding may increase not only in-group cohesion, but also hostility towards out-group members [15,29,30], resulting in increased intergroup discrimination. We thus also examined how in-group social bonding influenced intergroup discrimination, especially for individuals with different social status (i.e., group leader and followers). At the neural level, we employed functional near-infrared spectroscopy (fNIRS) hyper-scanning to measure neural synchronization among group members. Recent neuroscience research has suggested inter-brain neural synchronization (INS) as a reliable indicator of social interactions [31–33]. Neural synchronization emerges in a variety of social encounters, including interactions between peers [34], romantic partners [35], parent and child [36,37], teacher and student [38]. The degree of neural synchronization was predictive of interaction quality [39]. Of particular interest, neural synchronization among group members has been suggested as a candidate mechanism mediating within-group interaction [15,40], and in-group bonding increased within-group neural synchronization [15,41]. However, previous studies merely focused on the egalitarian group, suffering from not being able to differentiate inter- and intra-status interactions and leaving the social bonding effect on hierarchical interactions an open question. Here, we aimed to reveal whether and how social bonding influenced neural synchronization within a hierarchical group and, in particular, the inter-status and intra-status neural synchronization. Recent research has documented stronger neural synchronization during social interactions between individuals with different social roles than those with the same roles (e.g., teacher–student versus student–student, [42,43]; leader–follower versus follower–follower, [44,45]). Therefore, it could be expected that in-group social bonding would have differential effects on inter-status and intra-status neural synchronizations. To address these questions, we applied fNIRS hyper-scanning to 176 three-person groups (the most basic hierarchical group with 1 leader and 2 followers, S1 Table) and simultaneously recorded neural activities of 3 group members of each group during online within-group interactions. Participants democratically elected a group leader and discussed group strategies for potential intergroup contests after in-group social bonding or no-bonding control manipulation (Figs 1A and S1). The in-group social bonding manipulation employed in the current study was adapted from several previously validated procedures [15,46–49]. Specifically, we integrated 3 fundamental procedures to manipulate in-group social bonding: (i) shared preference [46]; (ii) symbolic marker [47]; and (iii) similarity among group members [48]. Taking advantage of fNIRS technology that provides noninvasive measures of neural activity with minimal sensitivity to motion artifacts [50], we measured neural synchronization in the right dorsolateral prefrontal cortex (rDLPFC) and right temporal-parental junction (rTPJ) in inter-status and intra-status dyads (Fig 1B). Brain regions of interest in the current study included rDLPFC and rTPJ. Previous studies have shown that the right (but not left) DLPFC was involved in allocating attention and making strategic decision during social interaction [51,52]. The TPJ, particularly in the right hemisphere, is a core region of the mentalizing network and involved in alignment with in-group members regarding consensus decision and group norms [53]. This ROI choice was also based on earlier studies that have linked neural synchronization in the rDLPFC and rTPJ with social interactive processes [32,36,45,54]. Moreover, neural synchronization in the rDLPFC activity predicted in-group cooperation during intergroup conflict [15], while neural synchronization in rTPJ was associated with leader–follower interaction [45]. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 1. In-group social bonding selectively facilitates inter-status communication and cohesion. (A) Experimental setting. During the group interaction, 3 group members’ rDLPFC and rTPJ signals were simultaneously recorded by the same fNIRS system. Shown are snapshots of a control session and a bonding session. (B) Illustration of fNIRS probe configurations. Two identical 3 × 2 optode probe sets, with each consisting of 3 emitters (light blue) and 3 detectors (dark blue), were placed on rDLPFC and rTPJ, respectively. The integers in between indicate the recording channels. (C) Group members were engaged in turn-taking conversation session. Each turn transition was defined as a discrete pair of utterances from different individuals (depicted by rectangles). The turn-response time was calculated as the time interval of corresponding turn transition. (D) Social bonding increased inter-status turn transitions (control: 5.960 ± 2.971, bonding: 8.854 ± 3.759) but not intra-status ones (control: 5.470 ± 3.560, bonding: 6.320 ± 3.499). (E) Bonding significantly sped up turn transitions (inter-status: control: 0.997 ± 0.188, bonding: 0.898 ± 0.172; intra-status: control: 0.947 ± 0.284, bonding: 0.909 ± 0.228). (F/G) Interpersonal cohesion was positively associated with turn transition frequency, respectively, for inter-status (F) and intra-status dyads (G). Each solid line represents the least squares fit, with shading showing the 95% CI. (H) Bonding selectively increased inter-status (control: 6.310 ± 2.025, bonding: 6.970 ± 1.855) but not intra-status cohesion (control: 6.220 ± 1.870, bonding: 6.240 ± 2.035). Data are plotted as box plots for each condition, with horizontal lines indicating median values, boxes indicating 25% and 75% quartiles, and whiskers indicating the 2.5%–97.5% percentile range. Cross symbols in each box represent the mean values. Data points outside the range are shown separately as circles. *p < 0.05, **p < 0.01, ***p < 0.001. Data used to generate Fig 1D–1H can be found in S1 Data. fNIRS, functional near-infrared spectroscopy; rDLPFC, right dorsolateral prefrontal cortex; rTPJ, right temporal-parental junction. https://doi.org/10.1371/journal.pbio.3002545.g001 Results In-group social bonding facilitated inter-status communication and cohesion We performed conversation analysis [55] on the transcripts of within-group communication for each group. The within-group communications were operationalized on a turn-taking basis (Fig 1C). Therefore, we focused on the number of utterances, the number of turn transition, and turn response time. Turn transition refers to the exchange of utterances among different group members. The number of turn transitions is suggested to reflect the frequency of mutual understanding and engagement [56], with more turn transitions indicating more interactive and engaging communications between group members. Turn response time is measured by the time interval between turns, with faster response times reflecting stronger social connection and more efficient and interactive communication [57]. We first compared the number of utterances of group leader and followers between social bonding and no-bonding control conditions (Methods). There was a significant main effect of bonding (F1, 173 = 46.570, p = 1.429 × 10−10, η2 = 0.212, ANCOVA, controlling for the total length of utterances, S2 Fig), suggesting that group members communicated more often under in-group social bonding. Moreover, social bonding increased leader’s utterance to a greater degree than followers (hierarchy × bonding: F1, 173 = 3.977, p = 0.048, η2 = 0.022; leader: t174 = 5.427, p = 1.897 × 10−7, Cohen’s d = 0.818, 95% CI: 2.651, 5.682; follower: t174 = 4.273, p = 3.172 × 10−5, Cohen’s d = 0.644, 95% CI: 1.389, 3.767, S2 Fig). The number of turn-transition and turn-response time was then calculated separately for inter-status (a discrete pair of utterances between a group leader and a follower) and intra-status (a discrete pair of utterances between 2 followers) communications. We compared these measurements between social bonding and control conditions using hierarchy (leader versus follower) × bonding (bonding versus control) ANCOVAs (controlling for the total length of utterances) and corresponding linear mixed models (LMMs, with hierarchy and bonding as fixed effects and group as a random effect). This analysis revealed that social bonding increased the frequency of inter-status (compared to intra-status), communications to a great extent (increased the number of leader–follower turn transitions, bonding × hierarchy: F1, 172 = 9.951, p = 0.002, η2 = 0.055, inter-status: t174 = 5.658, p = 6.182 × 10−8, Cohen’s d = 0.853, 95% CI: 1.885, 3.904, intra-status: t173 = 1.587, p = 0.114, Cohen’s d = 0.240, 95% CI: −0.206, 1.900; LMM: F1, 347 = 7.673, p = 0.006, Fig 1D) and shortened the turn response time (bonding main effect: F1, 166 = 9.793, p = 0.002, η2 = 0.056, especially for inter-status turns: t172 = −3.406, p = 0.001, Cohen’s d = −0.516, 95% CI: −0.154, −0.040; LMM: F1, 340 = 12.924, p = 3.72 × 10−4, Fig 1E). These results together suggested that social bonding was efficient in increasing group communication, especially promoted more frequent and responsive inter-status interactions and strengthened inter-status social connections. At the end of the experiment, we asked participants to report subjective evaluations on inter- and intra-status cohesion. First, we found that the frequency of group communication predicted group cohesion. In groups with more frequent communications, their group members reported a higher level of group cohesion (r172 = 0.206, p = 0.006). Interestingly, more inter-status turn transitions selectively predicted inter-status cohesion (r173 = 0.165, p = 0.029, Fig 1F), but not intra-status cohesion (r173 = 0.052, p = 0.496). Similarly, more intra-status turn transitions predicted a higher level of intra-status cohesion (r172 = 0.162, p = 0.033, Fig 1G, but not inter-status cohesion, r172 = 0.146, p = 0.055). Second, social bonding selectively facilitated inter-status cohesion (t174 = 2.261, p = 0.025, Cohen’s d = 0.340, 95% CI: 0.082, 1.238) rather than intra-status cohesion (t174 = 0.040, p = 0.968, Cohen’s d = 0.010, 95% CI: −0.562, 0.602), confirmed by a significant interaction between hierarchy (inter- versus intra-status) and bonding (bonding versus control) on in-group cohesion rating (F1, 174 = 4.914, p = 0.028, η2 = 0.027, Fig 1H). In addition, within-group interactions under social bonding were also perceived as more frequent and cohesive by third-party observers (Methods, S3 Fig). In-group social bonding influenced leader behavior and social perception of leader Next, we examined whether social bonding influenced behaviors toward in- and out-group members differently (or not) in individuals of different social statuses (i.e., group leader and followers). Participants completed 2 economic games related to intergroup discrimination: (i) an intergroup dictator game (IDG) where participants donated to in-group and out-group members [15,58]; (ii) an intergroup prisoner’s dilemma-maximizing differences game (IPD-MDG) where participants self-sacrificed separately to benefit in-group members (“ingroup love”) and to derogate out-group members (“outgroup hate”) [59,60]. We found that groups in the bonding (versus control) condition donated more to in-group members than to out-group members in the IDG (F1, 174 = 26.406, p = 7.375 × 10−7, η2 = 0.132, Fig 2A), and such bonding-facilitated intergroup discrimination was stronger in group leaders than followers (hierarchy × bonding: F1, 174 = 6.109, p = 0.014, η2 = 0.034; leader: t174 = 4.631, p = 7.087 × 10−6, Cohen’s d = 0.698, 95% CI: 15.685, 38.983; follower: t174 = 2.239, p = 0.026, Cohen’s d = 0.338, 95% CI: 1.101, 17.479; LMM: F1, 348 = 6.256, p = 0.013, Fig 2A). In the IPD-MDG, participants showed stronger in-group love (paired t test: ingroup love (Mean ± SD): 34.332 ± 16.650, outgroup hate: 26.319 ± 15.764, t175 = 3.966, p = 1.065 × 10−4, Cohen’s d = 0.299, 95% CI: 4.025, 12.001). Interestingly, the interactive effect of bonding and hierarchy was observed in out-group hate (F1, 172 = 4.470, p = 0.036, η2 = 0.025; leader: t172 = 1.995, p = 0.048, Cohen’s d = 0.302, 95% CI: 0.084, 15.830; follower: t174 = −0.605, p = 0.546, Cohen’s d = −0.091, 95% CI: −7.443, 3.950; LMM: F1, 346 = 3.898, p = 0.049, Fig 2B) but not in in-group love (F1, 172 = 0.445, p = 0.506, η2 = 0.003). Taken together, in-group social bonding increased intergroup discrimination and “hate” towards outgroup, especially in group leaders. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 2. The effect of in-group social bonding on leader behavior and the perception of leader. (A) In-group social bonding increased intergroup discrimination to a greater degree in leaders (control: 40.314 ± 39.547, bonding: 67.649 ± 38.752) than followers (control: 47.971 ± 28.122, bonding: 57.261 ± 26.918). (B) Bonding increased out-group hate in leaders (control: 19.858 ± 22.151, bonding: 27.814 ± 29.810) but not in followers (control: 28.557 ± 20.210, bonding: 26.810 ± 18.040). (C) Under social bonding, followers perceived greater social influence of the leader (control: 6.270 ± 2.061, bonding: 6.970 ± 1.963). Data are plotted as box plots for each condition, with horizontal lines indicating median values, boxes indicating 25% and 75% quartiles and whiskers indicating the 2.5%–97.5% percentile range. Cross symbols in each box represent the mean values. Data points outside the range are shown separately as circles. (D) Leader’s social influence was positively associated with inter-status cohesion (Pearson’s correlation analysis). Each solid line represents the least squares fit, with shading showing the 95% CI. (E) Bonding increased perceived social influence of the leader through enhancing inter-status cohesion. *p < 0.05, ***p < 0.001. Data used to generate Fig 2A–2E can be found in S1 Data. https://doi.org/10.1371/journal.pbio.3002545.g002 We concluded our behavioral analysis by examining how in-group social bonding influenced followers’ perception of the leader (i.e., leaders’ social influence and social attraction). We found that followers in groups under social bonding (versus control) perceived their leaders as more influential (t174 = 2.313, p = 0.022, Cohen’s d = 0.348, 95% CI: 0.103, 1.301, Fig 2C) and more attractive (t174 = 2.944, p = 0.004, Cohen’s d = 0.444, 95% CI: 0.265, 1.339, S4A Fig). Moreover, the perceived social influence and social attraction of leaders were positively associated with in-group cohesion, especially evident with inter-status cohesion (social influence: r176 = 0.765, p = 4.183 × 10−35, Fig 2D; social attraction: r176 = 0.702, p = 1.743 × 10−27, S4B Fig; weaker but also with intra-status cohesion, social influence: r176 = 0.435, p = 1.548 × 10−9, S4C Fig; attraction: r176 = 0.287, p = 1.107 × 10−4, S4D Fig; slope test: social influence: z = 5.04, p = 1.164 × 10−7; attraction: z = 5.36, p = 2.081 × 10−8), suggesting that followers perceived their leaders as more influential and attractive in more cohesive groups, especially when they coordinated better with the leaders. Importantly, we established a potential mediation path that the effects of social bonding on perceived social influence (Indirect effect = 0.520, SE = 0.238, 95% bootstrap CI: 0.073, 1.001, Sobel test, Z = 2.224, p = 0.025, Fig 2E) and attraction (Indirect effect = 0.426, SE = 0.196, 95% bootstrap CI: 0.057, 0.0832, Sobel test, Z = 2.224, p = 0.026, S4E Fig) in leaders were fully mediated by inter-status cohesion (Methods). The behavioral results together suggested that social bonding (i) increased contributions of group leader in both within-group communication and intergroup conflict; and (ii) selectively strengthened inter-status (but not intra-status) communication and cohesion, which possibly resulted in a better impression and more social influence of group leader. In-group social bonding selectively increases inter-status neural synchronization in the rDLPFC We applied fNIRS to each hierarchical group and simultaneously recorded all group members’ neural activity, captured by the dynamic hemodynamic signals, from the rDLPFC (7 channels, Fig 1B) and the right temporoparietal junction (rTPJ, 7 channels, Fig 1B), during resting-state and interaction stages. Consistent with previous studies [15,34–37], we operationalized the INS in terms of wavelet transform coherence (WTC). The WTC value indicates the cross-correlation between 2 fNIRS time series of concentration changes in oxygenated hemoglobin (oxy-Hb) in dyads of individuals as a function of frequency and time. Within each three-person group, we calculated the coherence values from the leader–follower dyads to index the inter-status INS, and the coherence value from the follower–follower dyads to index the intra-status INS (Methods, Fig 3A). We were interested in the INS specific to group interaction, thus focused on the INS increases during group interaction relative to the resting-state. We compared coherence values between the resting-state and group interaction to identify the frequency band of interest (FOI, Methods, Fig 3A). Moreover, the INS specific to group interaction was indicated by the FOI-averaged coherence differences (Group interaction—Resting) and then submitted into the following analyses. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 3. In-group social bonding selectively increases inter-status neural synchronization in the rDLPFC. (A) Illustration of inter-status neural synchronization calculation. The concentration changes in oxygenated hemoglobin (oxy-Hb) were simultaneously collected in each channel from each member of the three-person group. The cross-correlations between oxy-Hb time series of leader–follower pairs were generated through WTC analysis, and the 2 pairs were then averaged to indicate INS of inter-status dyads. Comparison of coherence values between group interaction and resting-state identified INS specific to group interaction and the frequency band of interest. (B) Stronger inter-status (vs. intra-status) INS at channel 3 in the rTPJ. Data are plotted as box plots for each condition, with horizontal lines indicating median values, boxes indicating 25% and 75% quartiles and whiskers indicating the 2.5%–97.5% percentile range. Cross symbols in each box represent the mean values. Data points outside the range are shown separately as circles. (C) Positive association between inter-status INS at channel 3 in the rTPJ and group leader’s social influence (Pearson’s correlation analysis). The solid line represents the least squares fit, with shading showing the 95% CI. (D) Bonding increased inter-status INS at channel 9 in the rDLPFC (control: 0.001 ± 0.053, bonding: 0.019 ± 0.055) but decreased that of intra-status dyads (control: 0.019 ± 0.076, bonding: −0.004 ± 0.079). (E) Positive association between inter-status INS at channel 9 in the rDLPFC and perceived group cohesion. (F/G) INS validation by nonparametric permutation tests. We compared the hierarchy main effect in the rTPJ and the interaction effect in the rDLPFC of real group against within-condition permutation distributions (n = 1,000). The observed effects of hierarchy in the rTPJ (F) and of hierarchy × bonding interaction in the rDLPFC (G) exceeded the upper limits of 99% CI of the permutation distributions. (H) The inter-status (but not intra-status) INS at channel 9 in the rDLPFC was associated with intergroup discrimination. *p < 0.05, **p < 0.01. Data used to generate Fig 3B–3H can be found in S1 Data. INS, inter-brain neural synchronization; rDLPFC, right dorsolateral prefrontal cortex; rTPJ, right temporal-parental junction; WTC, wavelet transform coherence. https://doi.org/10.1371/journal.pbio.3002545.g003 To examine whether social bonding differently influenced the inter- and intra-status INS, we submitted INS in each channel of the rDLPFC and rTPJ to 2 (hierarchy: inter- versus intra-status dyads) × 2 (bonding: bonding versus control) mixed-model ANOVAs and LMMs (hierarchy and bonding as fixed effects, group as a random effect). Significant effects were identified after false discovery rate (FDR)-corrected for multiple comparisons for 14 channels. First, the analysis revealed stronger INS in the rTPJ for the inter-status than intra-status dyads (main effect of hierarchy at channel 3, F1, 174 = 10.207, p = 0.002, η2 = 0.055, survived 14-channel-wise FDR-correction; LMM: F1, 348 = 9.684, p = 0.002, Fig 3B and S2 Table). Moreover, stronger inter-status INS in the rTPJ was associated with stronger social influence of the group leader (r176 = 0.188, p = 0.012, Fig 3C), suggesting that followers perceived their leaders more influential when their rTPJ activity synchronized with that of leader to a greater degree. Interestingly, we observed a significant hierarchy × bonding interaction on INS at channel 9 in the rDLPFC (F1, 174 = 9.577, p = 0.002, η2 = 0.052, survived 14-channel-wise FDR-correction; LMM: F1, 348 = 8.912, p = 0.003, Fig 3D and S2 Table). Specifically, social bonding selectively increased the inter-status INS (independent-sample t tests, t174 = 2.357, p = 0.020, Cohen’s d = 0.355, 95% CI: 0.003, 0.035) but decreased intra-status INS (t174 = −1.998, p = 0.047, Cohen’s d = −0.301, 95% CI: −0.047, 0.000). Taking another perspective to interpret the interaction, we observed stronger inter-status (versus intra-status) INS in the bonding condition (paired-sample t tests, t88 = 2.521, p = 0.013, Cohen’s d = 0.267, 95% CI: 0.005, 0.043), which was comparable even in an opposite trend in the control condition (t86 = −1.875, p = 0.064, Cohen’s d = −0.201, 95% CI: −0.039, 0.001). Interestingly, we found that inter-status but not intra-status INS in the rDLPFC was predictive of how the group interaction was perceived. Independent rater perceived the groups with stronger inter-status INS in rDLPFC more cohesive (inter-status INS: r175 = 0.177, p = 0.019, Fig 3E; intra-status INS: r175 = 0.085, p = 0.265). We next conducted 2 sets of validation analyses to exclude the possibility that the observed INS was partially reflected participants sharing the same environment or performing the same task. First, within the bonding and control conditions, we generated 176 within-condition three-person pseudo-groups by randomly grouping a real leader and 2 real followers from different original groups in the same bonding or control condition as 1 three-person pseudo-group (Methods, S5A Fig). We recalculated the inter- and intra-status INS for pseudo groups and repeated these procedures for 1,000 times. We then conducted nonparametric permutation tests on the observed effects of the real interacting groups against the 1,000 permutation samples. This analysis confirmed that both the main effect of hierarchy in the rTPJ (real group: 0.041, permutation: 95% CI: −0.020, 0.031, 99% CI: −0.029, 0.039, p = 0.003, Fig 3F) and the interactive effect of hierarchy × bonding in the rDLPFC (real group: 0.043, permutation: 95% CI: −0.020, 0.031, 99% CI: −0.028, 0.039, p = 0.003, Fig 3G) in the real groups exceeded the upper limit of 99% CI of the permutation distributions. Second, similar analysis conducted on the cross-condition permutation samples (i.e., 1 leader and 2 followers randomly from the bonding or control condition were organized into a pseudo-group, S5B Fig) again confirmed the observed effects in the real interacting groups (S5C and S5D Fig). Taken together, the 2 validation analyses confirmed that the observed bonding and/or hierarchy effects on INS in the real interactive groups were not due to same experimental environment or performing the same task. In addition, we further eliminated potential influence of global physiological noises by (i) using a wavelet-based denoising method [61]; and (ii) controlling the globally co-varying signals in the hierarchy × bonding ANCOVAs (i.e., including the global mean of INS across all channels as a covariant, [62]). These 2 complementary analyses well replicated the aforementioned patterns (S3 Table). In-group social bonding influences group behavior via inter-status INS in the rDLPFC Next, we aimed to reveal whether and how the inter-status or intra-status INS within a group was linked to behaviors towards in-group and out-group members. We found that stronger inter-status (but not intra-status) INS in the rDLPFC was predictive of stronger intergroup discrimination (i.e., donations to in-group versus out-group members; inter-status: r176 = 0.216, p = 0.004; intra-status: r176 = −0.096, p = 0.206, Fig 3H). Further modulation analysis compared the Fisher-transformed correlation coefficients and confirmed selective prediction of inter- (versus intra-) status INS on intergroup discrimination (z = 2.94, p = 0.003). These results suggested that leader and followers synchronized their rDLPFC activity in a way that predicted how they differently treated in-group and out-group members. In-group social bonding facilitates leader-to-follower neural alignment in the rDLPFC Next, we aimed to probe the directionality of the inter-status neural synchronization. We specifically asked whether social bonding influenced the leader-to-follower (i.e., neural activity of leaders preceded that of followers) or follower-to-leader (i.e., neural activity of followers preceded that of leaders) neural alignment, or both. The leader-to-follower neural alignment reflects situations in which the group leader leads and followers follow, while the follower-to-leader neural alignment may indicate instances where followers take the lead and group leader follows. We thus conducted time-lag analysis that has been employed in previous studies to reveal the directional influence between leader and follower’s neural activity [63–65]. The time course of leader’s neural activity was shifted relative to that of the followers from −10 to 10 s (in 1-s increment). Positive time lags indicated leader-to-follower neural alignment and negative time lags reflected follower-to-leader neural alignment (Fig 4A). On each time lag, the coherence values of inter-/intra-status dyads were recomputed for both resting-state and interaction stage. The coherence value increase (i.e., lagged INS during interaction minus that during resting) were used to indicate lagged neural alignment and submitted into subsequent analysis (Methods). It should be noted that the time lag analyses were conducted for the channels that showed increased neural synchronization between leader and followers during the interaction stage, i.e., channel 3 in the rTPJ and channel 9 in the rDLPFC. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 4. In-group social bonding facilitates leader-to-follower neural alignment in the rDLPFC. (A) Illustration of time lag analysis. The neural activity of the leader is shifted forwards (backwards) in relative to that of followers in the positive (negative) time lags, indicating leader-to-follower (follower-to-leader) alignment. (B) Bonding (vs. control) facilitated leader-to-follower neural alignment on +1 to +6 time lags (peaked at +5 s), survived FDR multiple correction. The dashed line indicates the corrected significance threshold. The significant time lags (survived multiple correction) are highlighted with the horizontal line on the x-axis. (C) Significant increases of inter-status neural alignment on +1 to +6 time lags was only found in the bonding condition. Shaded areas represent SE. (D/E) Inter-status neural alignment was positively correlated with intergroup discrimination on +1 to +6 time lags in the bonding (but not control) conditions (correlation coefficients on each time lag of −10 to +10 lags, D; correlations for the averaged inter-status neural alignment, E). Correlations were performed by Pearson’s correlation coefficient analysis. Each solid line represents the least squares fit, with shading showing the 95% CI. Data used to generate Fig 4B–4E can be found in S1 Data. FDR, false discovery rate; rDLPFC, right dorsolateral prefrontal cortex; SE, standard error. https://doi.org/10.1371/journal.pbio.3002545.g004 While the time-lagged neural alignment in the rTPJ was significant in both the leader-to-follower and follower-to-leader directions (with peak centered at 0 second, S6A Fig), regardless of the bonding/control conditions (S6B Fig), the time-lagged neural alignment in the rDLPFC was significant mainly in the leader-to-follower direction and modulated by social bonding. We conducted independent t tests on time-lagged neural alignment between bonding and control conditions on each time lag and revealed a right-skewed bell-shaped curve for the bonding effect (Fig 4B). To be specific, bonding (relative to control condition) significantly facilitated inter-status neural alignment on +1 to +6 time lags (with peak centered at +5 s, Fig 4B and S4 Table), indicating that leaders’ neural activity preceded that of followers for 1 to 6 s (survived multiple corrections for 21 time lags). Separate analyses for bonding and control conditions confirmed significant increase (contrast to zero) of inter-status neural alignment only occurred in the bonding condition (Fig 4C and S5 Table); however, neither the leader-to-follower nor follower-to-leader neural alignment was significant in the control condition (Fig 4C). Moreover, such leader-to-follower neural alignment facilitated intergroup discrimination. Specifically, we correlated leader-to-follower neural alignment with intergroup discrimination and showed that stronger leader-to-follower neural alignment (when leaders’ neural activity preceded that of followers +1 to +6 s) predicted larger intergroup discrimination (Fig 4D, FDR corrected for 21 time lags, S6 Table). Moreover, this relationship was especially true in the bonding but not control condition (averaged neural alignment of +1 to +6 time lag, bonding: r89 = 0.281, p = 0.008, control: r87 = 0.081, p = 0.455, Figs 4E and S7 for each time lag separately). Control analyses were conducted for the intra-status INS for −10 to 10 s (in 1-s increment). Neither the bonding effect (S8A Fig) nor the correlation with intergroup discrimination (S8B Fig) was significant on intra-status neural alignment at any time lags. Taken together, the bonding-elevated inter-status INS selectively emerged when leaders’ neural activity preceded that of followers, indicating that leaders predicted or anticipated followers’ mental states and followers tracked the leader’s mental states to achieve the leader-to-follower neural alignment under in-group social bonding condition. Such leader-to-follower neural alignment may further result in stronger intergroup discrimination under in-group social bonding. Stronger rDLPFC-rTPJ functional connectivity in the leader accounted for leader-to-follower neural alignment The observation that the bonding effect was selectively exhibited on the inter-status dyads, especially in a leader-to-follower manner, led us to further examine the bonding effects respectively in leaders and followers. The functional connectivity between rDLPFC and rTPJ has been shown to play a key role in perspective taking, mental inference, and information integrating [66,67]. As leader-to-follower neural alignment may indicate situations in which group leaders predict followers’ mental states or perspectives [64], as well as when followers strategically attend to and track the group leader [68], we compared rDLPFC-rTPJ functional connectivity between leaders than followers. Furthermore, we tested whether rDLPFC-rTPJ functional connectivity could account for the leader-to-follower neural alignment. To this end, we applied cross-correlation analysis to assess the functional connectivity of rDLPFC and rTPJ in leaders and followers (Methods). Results showed that, leaders (versus followers) showed stronger rDLPFC-rTPJ connectivity (Fig 5A for channel-pairwise rDLPFC-rTPJ connectivity, 28 rDLPFC-rTPJ channel pairs survived FDR correction for 49 channel pairs, S7 Table; Fig 5B for the grand mean rDLPFC-rTPJ connectivity, two-way mixed-model ANOVA, F1, 174 = 12.006, p = 6.679 × 10−4, η2 = 0.065; LMM: F1, 348 = 12.438, p = 4.77 × 10−4). We next correlated the strength of rDLPFC-rTPJ connectivity (averaged connectivity between channel 9 in the rDLPFC and each channel in the rTPJ) in leaders and followers respectively with the leader-to-follower neural alignment at channel 9 in the rDLPFC (Methods). Results endorsed a positive relationship between rDLPFC-rTPJ functional connectivity in leaders (but not followers) and the leader-to-follower neural alignment (leader: r176 = 0.177, p = 0.019, Fig 5C, for average of lagged inter-status neural alignment, S9 Fig for correlations to neural alignment on each time lag; follower: r176 = 0.004, p = 0.955), suggesting that leaders with stronger rDLPFC-rTPJ connectivity predicted their followers’ neural activity to a greater degree. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 5. Stronger rDLPFC-rTPJ functional connectivity in leaders accounted for leader-to-follower neural alignment. (A/B) Leaders (vs. followers) showed stronger rDLPFC-rTPJ functional connectivity at channel-pairwise level (28 rDLPFC-rTPJ channel pairs survived FDR correction for 49 channel pairs, A) and the grand mean level (i.e., averaged coherence value of 49 channel pairs between the rDLPFC and rTPJ, B, displayed as box plots, with horizontal lines indicating median values, boxes indicating 25% and 75% quartiles and whiskers indicating the 2.5%–97.5% percentile range. Cross symbols in each box represent the mean values. Data points outside the range are shown separately as circles). (C) The functional connectivity between channel 9 in the rDLPFC and the rTPJ (averaged across 7 channel pairs) was associated with leader-to-follower neural alignment at channel 9. The solid line represents the least squares fit, with shading showing the 95% CI., **p < 0.01. Data used to generate Fig 5A–5C can be found in S1 Data. FDR, false discovery rate; rDLPFC, right dorsolateral prefrontal cortex; rTPJ, right temporal-parental junction. https://doi.org/10.1371/journal.pbio.3002545.g005 In-group social bonding facilitated inter-status communication and cohesion We performed conversation analysis [55] on the transcripts of within-group communication for each group. The within-group communications were operationalized on a turn-taking basis (Fig 1C). Therefore, we focused on the number of utterances, the number of turn transition, and turn response time. Turn transition refers to the exchange of utterances among different group members. The number of turn transitions is suggested to reflect the frequency of mutual understanding and engagement [56], with more turn transitions indicating more interactive and engaging communications between group members. Turn response time is measured by the time interval between turns, with faster response times reflecting stronger social connection and more efficient and interactive communication [57]. We first compared the number of utterances of group leader and followers between social bonding and no-bonding control conditions (Methods). There was a significant main effect of bonding (F1, 173 = 46.570, p = 1.429 × 10−10, η2 = 0.212, ANCOVA, controlling for the total length of utterances, S2 Fig), suggesting that group members communicated more often under in-group social bonding. Moreover, social bonding increased leader’s utterance to a greater degree than followers (hierarchy × bonding: F1, 173 = 3.977, p = 0.048, η2 = 0.022; leader: t174 = 5.427, p = 1.897 × 10−7, Cohen’s d = 0.818, 95% CI: 2.651, 5.682; follower: t174 = 4.273, p = 3.172 × 10−5, Cohen’s d = 0.644, 95% CI: 1.389, 3.767, S2 Fig). The number of turn-transition and turn-response time was then calculated separately for inter-status (a discrete pair of utterances between a group leader and a follower) and intra-status (a discrete pair of utterances between 2 followers) communications. We compared these measurements between social bonding and control conditions using hierarchy (leader versus follower) × bonding (bonding versus control) ANCOVAs (controlling for the total length of utterances) and corresponding linear mixed models (LMMs, with hierarchy and bonding as fixed effects and group as a random effect). This analysis revealed that social bonding increased the frequency of inter-status (compared to intra-status), communications to a great extent (increased the number of leader–follower turn transitions, bonding × hierarchy: F1, 172 = 9.951, p = 0.002, η2 = 0.055, inter-status: t174 = 5.658, p = 6.182 × 10−8, Cohen’s d = 0.853, 95% CI: 1.885, 3.904, intra-status: t173 = 1.587, p = 0.114, Cohen’s d = 0.240, 95% CI: −0.206, 1.900; LMM: F1, 347 = 7.673, p = 0.006, Fig 1D) and shortened the turn response time (bonding main effect: F1, 166 = 9.793, p = 0.002, η2 = 0.056, especially for inter-status turns: t172 = −3.406, p = 0.001, Cohen’s d = −0.516, 95% CI: −0.154, −0.040; LMM: F1, 340 = 12.924, p = 3.72 × 10−4, Fig 1E). These results together suggested that social bonding was efficient in increasing group communication, especially promoted more frequent and responsive inter-status interactions and strengthened inter-status social connections. At the end of the experiment, we asked participants to report subjective evaluations on inter- and intra-status cohesion. First, we found that the frequency of group communication predicted group cohesion. In groups with more frequent communications, their group members reported a higher level of group cohesion (r172 = 0.206, p = 0.006). Interestingly, more inter-status turn transitions selectively predicted inter-status cohesion (r173 = 0.165, p = 0.029, Fig 1F), but not intra-status cohesion (r173 = 0.052, p = 0.496). Similarly, more intra-status turn transitions predicted a higher level of intra-status cohesion (r172 = 0.162, p = 0.033, Fig 1G, but not inter-status cohesion, r172 = 0.146, p = 0.055). Second, social bonding selectively facilitated inter-status cohesion (t174 = 2.261, p = 0.025, Cohen’s d = 0.340, 95% CI: 0.082, 1.238) rather than intra-status cohesion (t174 = 0.040, p = 0.968, Cohen’s d = 0.010, 95% CI: −0.562, 0.602), confirmed by a significant interaction between hierarchy (inter- versus intra-status) and bonding (bonding versus control) on in-group cohesion rating (F1, 174 = 4.914, p = 0.028, η2 = 0.027, Fig 1H). In addition, within-group interactions under social bonding were also perceived as more frequent and cohesive by third-party observers (Methods, S3 Fig). In-group social bonding influenced leader behavior and social perception of leader Next, we examined whether social bonding influenced behaviors toward in- and out-group members differently (or not) in individuals of different social statuses (i.e., group leader and followers). Participants completed 2 economic games related to intergroup discrimination: (i) an intergroup dictator game (IDG) where participants donated to in-group and out-group members [15,58]; (ii) an intergroup prisoner’s dilemma-maximizing differences game (IPD-MDG) where participants self-sacrificed separately to benefit in-group members (“ingroup love”) and to derogate out-group members (“outgroup hate”) [59,60]. We found that groups in the bonding (versus control) condition donated more to in-group members than to out-group members in the IDG (F1, 174 = 26.406, p = 7.375 × 10−7, η2 = 0.132, Fig 2A), and such bonding-facilitated intergroup discrimination was stronger in group leaders than followers (hierarchy × bonding: F1, 174 = 6.109, p = 0.014, η2 = 0.034; leader: t174 = 4.631, p = 7.087 × 10−6, Cohen’s d = 0.698, 95% CI: 15.685, 38.983; follower: t174 = 2.239, p = 0.026, Cohen’s d = 0.338, 95% CI: 1.101, 17.479; LMM: F1, 348 = 6.256, p = 0.013, Fig 2A). In the IPD-MDG, participants showed stronger in-group love (paired t test: ingroup love (Mean ± SD): 34.332 ± 16.650, outgroup hate: 26.319 ± 15.764, t175 = 3.966, p = 1.065 × 10−4, Cohen’s d = 0.299, 95% CI: 4.025, 12.001). Interestingly, the interactive effect of bonding and hierarchy was observed in out-group hate (F1, 172 = 4.470, p = 0.036, η2 = 0.025; leader: t172 = 1.995, p = 0.048, Cohen’s d = 0.302, 95% CI: 0.084, 15.830; follower: t174 = −0.605, p = 0.546, Cohen’s d = −0.091, 95% CI: −7.443, 3.950; LMM: F1, 346 = 3.898, p = 0.049, Fig 2B) but not in in-group love (F1, 172 = 0.445, p = 0.506, η2 = 0.003). Taken together, in-group social bonding increased intergroup discrimination and “hate” towards outgroup, especially in group leaders. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 2. The effect of in-group social bonding on leader behavior and the perception of leader. (A) In-group social bonding increased intergroup discrimination to a greater degree in leaders (control: 40.314 ± 39.547, bonding: 67.649 ± 38.752) than followers (control: 47.971 ± 28.122, bonding: 57.261 ± 26.918). (B) Bonding increased out-group hate in leaders (control: 19.858 ± 22.151, bonding: 27.814 ± 29.810) but not in followers (control: 28.557 ± 20.210, bonding: 26.810 ± 18.040). (C) Under social bonding, followers perceived greater social influence of the leader (control: 6.270 ± 2.061, bonding: 6.970 ± 1.963). Data are plotted as box plots for each condition, with horizontal lines indicating median values, boxes indicating 25% and 75% quartiles and whiskers indicating the 2.5%–97.5% percentile range. Cross symbols in each box represent the mean values. Data points outside the range are shown separately as circles. (D) Leader’s social influence was positively associated with inter-status cohesion (Pearson’s correlation analysis). Each solid line represents the least squares fit, with shading showing the 95% CI. (E) Bonding increased perceived social influence of the leader through enhancing inter-status cohesion. *p < 0.05, ***p < 0.001. Data used to generate Fig 2A–2E can be found in S1 Data. https://doi.org/10.1371/journal.pbio.3002545.g002 We concluded our behavioral analysis by examining how in-group social bonding influenced followers’ perception of the leader (i.e., leaders’ social influence and social attraction). We found that followers in groups under social bonding (versus control) perceived their leaders as more influential (t174 = 2.313, p = 0.022, Cohen’s d = 0.348, 95% CI: 0.103, 1.301, Fig 2C) and more attractive (t174 = 2.944, p = 0.004, Cohen’s d = 0.444, 95% CI: 0.265, 1.339, S4A Fig). Moreover, the perceived social influence and social attraction of leaders were positively associated with in-group cohesion, especially evident with inter-status cohesion (social influence: r176 = 0.765, p = 4.183 × 10−35, Fig 2D; social attraction: r176 = 0.702, p = 1.743 × 10−27, S4B Fig; weaker but also with intra-status cohesion, social influence: r176 = 0.435, p = 1.548 × 10−9, S4C Fig; attraction: r176 = 0.287, p = 1.107 × 10−4, S4D Fig; slope test: social influence: z = 5.04, p = 1.164 × 10−7; attraction: z = 5.36, p = 2.081 × 10−8), suggesting that followers perceived their leaders as more influential and attractive in more cohesive groups, especially when they coordinated better with the leaders. Importantly, we established a potential mediation path that the effects of social bonding on perceived social influence (Indirect effect = 0.520, SE = 0.238, 95% bootstrap CI: 0.073, 1.001, Sobel test, Z = 2.224, p = 0.025, Fig 2E) and attraction (Indirect effect = 0.426, SE = 0.196, 95% bootstrap CI: 0.057, 0.0832, Sobel test, Z = 2.224, p = 0.026, S4E Fig) in leaders were fully mediated by inter-status cohesion (Methods). The behavioral results together suggested that social bonding (i) increased contributions of group leader in both within-group communication and intergroup conflict; and (ii) selectively strengthened inter-status (but not intra-status) communication and cohesion, which possibly resulted in a better impression and more social influence of group leader. In-group social bonding selectively increases inter-status neural synchronization in the rDLPFC We applied fNIRS to each hierarchical group and simultaneously recorded all group members’ neural activity, captured by the dynamic hemodynamic signals, from the rDLPFC (7 channels, Fig 1B) and the right temporoparietal junction (rTPJ, 7 channels, Fig 1B), during resting-state and interaction stages. Consistent with previous studies [15,34–37], we operationalized the INS in terms of wavelet transform coherence (WTC). The WTC value indicates the cross-correlation between 2 fNIRS time series of concentration changes in oxygenated hemoglobin (oxy-Hb) in dyads of individuals as a function of frequency and time. Within each three-person group, we calculated the coherence values from the leader–follower dyads to index the inter-status INS, and the coherence value from the follower–follower dyads to index the intra-status INS (Methods, Fig 3A). We were interested in the INS specific to group interaction, thus focused on the INS increases during group interaction relative to the resting-state. We compared coherence values between the resting-state and group interaction to identify the frequency band of interest (FOI, Methods, Fig 3A). Moreover, the INS specific to group interaction was indicated by the FOI-averaged coherence differences (Group interaction—Resting) and then submitted into the following analyses. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 3. In-group social bonding selectively increases inter-status neural synchronization in the rDLPFC. (A) Illustration of inter-status neural synchronization calculation. The concentration changes in oxygenated hemoglobin (oxy-Hb) were simultaneously collected in each channel from each member of the three-person group. The cross-correlations between oxy-Hb time series of leader–follower pairs were generated through WTC analysis, and the 2 pairs were then averaged to indicate INS of inter-status dyads. Comparison of coherence values between group interaction and resting-state identified INS specific to group interaction and the frequency band of interest. (B) Stronger inter-status (vs. intra-status) INS at channel 3 in the rTPJ. Data are plotted as box plots for each condition, with horizontal lines indicating median values, boxes indicating 25% and 75% quartiles and whiskers indicating the 2.5%–97.5% percentile range. Cross symbols in each box represent the mean values. Data points outside the range are shown separately as circles. (C) Positive association between inter-status INS at channel 3 in the rTPJ and group leader’s social influence (Pearson’s correlation analysis). The solid line represents the least squares fit, with shading showing the 95% CI. (D) Bonding increased inter-status INS at channel 9 in the rDLPFC (control: 0.001 ± 0.053, bonding: 0.019 ± 0.055) but decreased that of intra-status dyads (control: 0.019 ± 0.076, bonding: −0.004 ± 0.079). (E) Positive association between inter-status INS at channel 9 in the rDLPFC and perceived group cohesion. (F/G) INS validation by nonparametric permutation tests. We compared the hierarchy main effect in the rTPJ and the interaction effect in the rDLPFC of real group against within-condition permutation distributions (n = 1,000). The observed effects of hierarchy in the rTPJ (F) and of hierarchy × bonding interaction in the rDLPFC (G) exceeded the upper limits of 99% CI of the permutation distributions. (H) The inter-status (but not intra-status) INS at channel 9 in the rDLPFC was associated with intergroup discrimination. *p < 0.05, **p < 0.01. Data used to generate Fig 3B–3H can be found in S1 Data. INS, inter-brain neural synchronization; rDLPFC, right dorsolateral prefrontal cortex; rTPJ, right temporal-parental junction; WTC, wavelet transform coherence. https://doi.org/10.1371/journal.pbio.3002545.g003 To examine whether social bonding differently influenced the inter- and intra-status INS, we submitted INS in each channel of the rDLPFC and rTPJ to 2 (hierarchy: inter- versus intra-status dyads) × 2 (bonding: bonding versus control) mixed-model ANOVAs and LMMs (hierarchy and bonding as fixed effects, group as a random effect). Significant effects were identified after false discovery rate (FDR)-corrected for multiple comparisons for 14 channels. First, the analysis revealed stronger INS in the rTPJ for the inter-status than intra-status dyads (main effect of hierarchy at channel 3, F1, 174 = 10.207, p = 0.002, η2 = 0.055, survived 14-channel-wise FDR-correction; LMM: F1, 348 = 9.684, p = 0.002, Fig 3B and S2 Table). Moreover, stronger inter-status INS in the rTPJ was associated with stronger social influence of the group leader (r176 = 0.188, p = 0.012, Fig 3C), suggesting that followers perceived their leaders more influential when their rTPJ activity synchronized with that of leader to a greater degree. Interestingly, we observed a significant hierarchy × bonding interaction on INS at channel 9 in the rDLPFC (F1, 174 = 9.577, p = 0.002, η2 = 0.052, survived 14-channel-wise FDR-correction; LMM: F1, 348 = 8.912, p = 0.003, Fig 3D and S2 Table). Specifically, social bonding selectively increased the inter-status INS (independent-sample t tests, t174 = 2.357, p = 0.020, Cohen’s d = 0.355, 95% CI: 0.003, 0.035) but decreased intra-status INS (t174 = −1.998, p = 0.047, Cohen’s d = −0.301, 95% CI: −0.047, 0.000). Taking another perspective to interpret the interaction, we observed stronger inter-status (versus intra-status) INS in the bonding condition (paired-sample t tests, t88 = 2.521, p = 0.013, Cohen’s d = 0.267, 95% CI: 0.005, 0.043), which was comparable even in an opposite trend in the control condition (t86 = −1.875, p = 0.064, Cohen’s d = −0.201, 95% CI: −0.039, 0.001). Interestingly, we found that inter-status but not intra-status INS in the rDLPFC was predictive of how the group interaction was perceived. Independent rater perceived the groups with stronger inter-status INS in rDLPFC more cohesive (inter-status INS: r175 = 0.177, p = 0.019, Fig 3E; intra-status INS: r175 = 0.085, p = 0.265). We next conducted 2 sets of validation analyses to exclude the possibility that the observed INS was partially reflected participants sharing the same environment or performing the same task. First, within the bonding and control conditions, we generated 176 within-condition three-person pseudo-groups by randomly grouping a real leader and 2 real followers from different original groups in the same bonding or control condition as 1 three-person pseudo-group (Methods, S5A Fig). We recalculated the inter- and intra-status INS for pseudo groups and repeated these procedures for 1,000 times. We then conducted nonparametric permutation tests on the observed effects of the real interacting groups against the 1,000 permutation samples. This analysis confirmed that both the main effect of hierarchy in the rTPJ (real group: 0.041, permutation: 95% CI: −0.020, 0.031, 99% CI: −0.029, 0.039, p = 0.003, Fig 3F) and the interactive effect of hierarchy × bonding in the rDLPFC (real group: 0.043, permutation: 95% CI: −0.020, 0.031, 99% CI: −0.028, 0.039, p = 0.003, Fig 3G) in the real groups exceeded the upper limit of 99% CI of the permutation distributions. Second, similar analysis conducted on the cross-condition permutation samples (i.e., 1 leader and 2 followers randomly from the bonding or control condition were organized into a pseudo-group, S5B Fig) again confirmed the observed effects in the real interacting groups (S5C and S5D Fig). Taken together, the 2 validation analyses confirmed that the observed bonding and/or hierarchy effects on INS in the real interactive groups were not due to same experimental environment or performing the same task. In addition, we further eliminated potential influence of global physiological noises by (i) using a wavelet-based denoising method [61]; and (ii) controlling the globally co-varying signals in the hierarchy × bonding ANCOVAs (i.e., including the global mean of INS across all channels as a covariant, [62]). These 2 complementary analyses well replicated the aforementioned patterns (S3 Table). In-group social bonding influences group behavior via inter-status INS in the rDLPFC Next, we aimed to reveal whether and how the inter-status or intra-status INS within a group was linked to behaviors towards in-group and out-group members. We found that stronger inter-status (but not intra-status) INS in the rDLPFC was predictive of stronger intergroup discrimination (i.e., donations to in-group versus out-group members; inter-status: r176 = 0.216, p = 0.004; intra-status: r176 = −0.096, p = 0.206, Fig 3H). Further modulation analysis compared the Fisher-transformed correlation coefficients and confirmed selective prediction of inter- (versus intra-) status INS on intergroup discrimination (z = 2.94, p = 0.003). These results suggested that leader and followers synchronized their rDLPFC activity in a way that predicted how they differently treated in-group and out-group members. In-group social bonding facilitates leader-to-follower neural alignment in the rDLPFC Next, we aimed to probe the directionality of the inter-status neural synchronization. We specifically asked whether social bonding influenced the leader-to-follower (i.e., neural activity of leaders preceded that of followers) or follower-to-leader (i.e., neural activity of followers preceded that of leaders) neural alignment, or both. The leader-to-follower neural alignment reflects situations in which the group leader leads and followers follow, while the follower-to-leader neural alignment may indicate instances where followers take the lead and group leader follows. We thus conducted time-lag analysis that has been employed in previous studies to reveal the directional influence between leader and follower’s neural activity [63–65]. The time course of leader’s neural activity was shifted relative to that of the followers from −10 to 10 s (in 1-s increment). Positive time lags indicated leader-to-follower neural alignment and negative time lags reflected follower-to-leader neural alignment (Fig 4A). On each time lag, the coherence values of inter-/intra-status dyads were recomputed for both resting-state and interaction stage. The coherence value increase (i.e., lagged INS during interaction minus that during resting) were used to indicate lagged neural alignment and submitted into subsequent analysis (Methods). It should be noted that the time lag analyses were conducted for the channels that showed increased neural synchronization between leader and followers during the interaction stage, i.e., channel 3 in the rTPJ and channel 9 in the rDLPFC. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 4. In-group social bonding facilitates leader-to-follower neural alignment in the rDLPFC. (A) Illustration of time lag analysis. The neural activity of the leader is shifted forwards (backwards) in relative to that of followers in the positive (negative) time lags, indicating leader-to-follower (follower-to-leader) alignment. (B) Bonding (vs. control) facilitated leader-to-follower neural alignment on +1 to +6 time lags (peaked at +5 s), survived FDR multiple correction. The dashed line indicates the corrected significance threshold. The significant time lags (survived multiple correction) are highlighted with the horizontal line on the x-axis. (C) Significant increases of inter-status neural alignment on +1 to +6 time lags was only found in the bonding condition. Shaded areas represent SE. (D/E) Inter-status neural alignment was positively correlated with intergroup discrimination on +1 to +6 time lags in the bonding (but not control) conditions (correlation coefficients on each time lag of −10 to +10 lags, D; correlations for the averaged inter-status neural alignment, E). Correlations were performed by Pearson’s correlation coefficient analysis. Each solid line represents the least squares fit, with shading showing the 95% CI. Data used to generate Fig 4B–4E can be found in S1 Data. FDR, false discovery rate; rDLPFC, right dorsolateral prefrontal cortex; SE, standard error. https://doi.org/10.1371/journal.pbio.3002545.g004 While the time-lagged neural alignment in the rTPJ was significant in both the leader-to-follower and follower-to-leader directions (with peak centered at 0 second, S6A Fig), regardless of the bonding/control conditions (S6B Fig), the time-lagged neural alignment in the rDLPFC was significant mainly in the leader-to-follower direction and modulated by social bonding. We conducted independent t tests on time-lagged neural alignment between bonding and control conditions on each time lag and revealed a right-skewed bell-shaped curve for the bonding effect (Fig 4B). To be specific, bonding (relative to control condition) significantly facilitated inter-status neural alignment on +1 to +6 time lags (with peak centered at +5 s, Fig 4B and S4 Table), indicating that leaders’ neural activity preceded that of followers for 1 to 6 s (survived multiple corrections for 21 time lags). Separate analyses for bonding and control conditions confirmed significant increase (contrast to zero) of inter-status neural alignment only occurred in the bonding condition (Fig 4C and S5 Table); however, neither the leader-to-follower nor follower-to-leader neural alignment was significant in the control condition (Fig 4C). Moreover, such leader-to-follower neural alignment facilitated intergroup discrimination. Specifically, we correlated leader-to-follower neural alignment with intergroup discrimination and showed that stronger leader-to-follower neural alignment (when leaders’ neural activity preceded that of followers +1 to +6 s) predicted larger intergroup discrimination (Fig 4D, FDR corrected for 21 time lags, S6 Table). Moreover, this relationship was especially true in the bonding but not control condition (averaged neural alignment of +1 to +6 time lag, bonding: r89 = 0.281, p = 0.008, control: r87 = 0.081, p = 0.455, Figs 4E and S7 for each time lag separately). Control analyses were conducted for the intra-status INS for −10 to 10 s (in 1-s increment). Neither the bonding effect (S8A Fig) nor the correlation with intergroup discrimination (S8B Fig) was significant on intra-status neural alignment at any time lags. Taken together, the bonding-elevated inter-status INS selectively emerged when leaders’ neural activity preceded that of followers, indicating that leaders predicted or anticipated followers’ mental states and followers tracked the leader’s mental states to achieve the leader-to-follower neural alignment under in-group social bonding condition. Such leader-to-follower neural alignment may further result in stronger intergroup discrimination under in-group social bonding. Stronger rDLPFC-rTPJ functional connectivity in the leader accounted for leader-to-follower neural alignment The observation that the bonding effect was selectively exhibited on the inter-status dyads, especially in a leader-to-follower manner, led us to further examine the bonding effects respectively in leaders and followers. The functional connectivity between rDLPFC and rTPJ has been shown to play a key role in perspective taking, mental inference, and information integrating [66,67]. As leader-to-follower neural alignment may indicate situations in which group leaders predict followers’ mental states or perspectives [64], as well as when followers strategically attend to and track the group leader [68], we compared rDLPFC-rTPJ functional connectivity between leaders than followers. Furthermore, we tested whether rDLPFC-rTPJ functional connectivity could account for the leader-to-follower neural alignment. To this end, we applied cross-correlation analysis to assess the functional connectivity of rDLPFC and rTPJ in leaders and followers (Methods). Results showed that, leaders (versus followers) showed stronger rDLPFC-rTPJ connectivity (Fig 5A for channel-pairwise rDLPFC-rTPJ connectivity, 28 rDLPFC-rTPJ channel pairs survived FDR correction for 49 channel pairs, S7 Table; Fig 5B for the grand mean rDLPFC-rTPJ connectivity, two-way mixed-model ANOVA, F1, 174 = 12.006, p = 6.679 × 10−4, η2 = 0.065; LMM: F1, 348 = 12.438, p = 4.77 × 10−4). We next correlated the strength of rDLPFC-rTPJ connectivity (averaged connectivity between channel 9 in the rDLPFC and each channel in the rTPJ) in leaders and followers respectively with the leader-to-follower neural alignment at channel 9 in the rDLPFC (Methods). Results endorsed a positive relationship between rDLPFC-rTPJ functional connectivity in leaders (but not followers) and the leader-to-follower neural alignment (leader: r176 = 0.177, p = 0.019, Fig 5C, for average of lagged inter-status neural alignment, S9 Fig for correlations to neural alignment on each time lag; follower: r176 = 0.004, p = 0.955), suggesting that leaders with stronger rDLPFC-rTPJ connectivity predicted their followers’ neural activity to a greater degree. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 5. Stronger rDLPFC-rTPJ functional connectivity in leaders accounted for leader-to-follower neural alignment. (A/B) Leaders (vs. followers) showed stronger rDLPFC-rTPJ functional connectivity at channel-pairwise level (28 rDLPFC-rTPJ channel pairs survived FDR correction for 49 channel pairs, A) and the grand mean level (i.e., averaged coherence value of 49 channel pairs between the rDLPFC and rTPJ, B, displayed as box plots, with horizontal lines indicating median values, boxes indicating 25% and 75% quartiles and whiskers indicating the 2.5%–97.5% percentile range. Cross symbols in each box represent the mean values. Data points outside the range are shown separately as circles). (C) The functional connectivity between channel 9 in the rDLPFC and the rTPJ (averaged across 7 channel pairs) was associated with leader-to-follower neural alignment at channel 9. The solid line represents the least squares fit, with shading showing the 95% CI., **p < 0.01. Data used to generate Fig 5A–5C can be found in S1 Data. FDR, false discovery rate; rDLPFC, right dorsolateral prefrontal cortex; rTPJ, right temporal-parental junction. https://doi.org/10.1371/journal.pbio.3002545.g005 Discussion Social bonding has been recognized as a potent strategy to enhance in-group interaction and cohesion across species [16–21] and to reinforce hierarchical structures [22]. However, to date, the neurocognitive mechanisms underlying the effects of social bonding on hierarchical interactions remain largely elusive. We applied a multi-brain hyper-scanning approach to real-time within-group communication and differentiated interpersonal interactions within a hierarchical group into 2 status-related types: inter-status and intra-status interactions. By doing so, we provide evidence that in-group social bonding exerts distinct effects on individuals of different statuses and on dyads of different status-related relationships. Specifically, social bonding selectively facilitates communication frequency and responsiveness, as well as synchronized and leader-proceeding neural alignment for inter-status dyads but not for dyads with the same social status. These findings distinguish between inter-status and intra-status interactions and shed lights on the distinct neurocognitive mechanisms through which social bonding shapes group dynamics in hierarchical groups. Hierarchical groups are characterized by complex relationships among group members [3], where their behaviors and neural systems dynamically coordinate and interact [4]. However, most neuroscience studies on leadership and social hierarchy have examined the brains of leaders or followers in isolation [69,70], focusing on leading behaviors or the distinct roles leaders and followers respectively played in a group [71]. This approach has overlooked the interactive nature of hierarchical relationships, limiting our understanding of how different individuals dynamically interact within a hierarchical group. Recently, the emergence of second-person, interactive neuroscience has underscored the significance of studying how the brains of socially interacting individuals entrain to support social interaction and relationships [31–33]. Through the lens of interactive neuroscience, we were able to characterize distinct behavioral and neural profiles of 2 types of interpersonal interactions based on the status relationship of interactive dyads. Within a hierarchical group, group members of different, rather than same, status engaged in more frequent and rapid information exchange, and frequent inter-status interaction was closely related to in-group cohesion for the aggregated group. At the neural level, inter-status interaction was featured with stronger neural synchronization in the rTPJ, a key brain region involved in mentalizing and taking perspectives of others, especially dissimilar others [72]. This finding suggests an important role of perspective-taking and social mentalizing in inter-status interaction, while interactions among individuals with the same status may require less mentalizing efforts and gain social support through “shared mind.” Together, the distinct behavioral and neural profiles for inter-status and intra-status interactions highlight the importance of differentiating status-related interactions and relationships in future work to understand leadership and group dynamics in hierarchical groups. By differentiating between inter-status and intra-status interactions in hierarchical groups, we are able to arbitrate between 2 possible mechanisms: whether social bonding enhances group dynamics in hierarchical groups regardless of the interaction types (i.e., enhancing both inter-status and intra-status interactions in a similar way), or if it is modulated by the interaction type, selectively promoting either inter- or intra-status interaction. Our results, which revealed distinct behavioral and neural effects of social bonding on inter- and intra-status dyads, provide evidence in support of the latter. Specifically, social bonding mainly facilitates inter-status (rather than intra-status) communication and cohesion, and selectively enhances neural alignment between the rDLPFC activity of leaders and followers. Synthesizing and extending findings from previous studies [13,15], we suggest that the mechanisms by which social bonding operates within a group depends on the type of group structure. Unlike non-hierarchical groups in which social bonding fosters interaction and cohesion among group members indiscriminately, social bonding in hierarchical groups primarily serves to foster and reinforce connections among individuals with different social statuses. It selectively enhances mutual understanding and information exchange between leaders and followers rather than among peers, highlighting the importance of status-related interactions in shaping the effects of social bonding within a hierarchical group. We further investigated how social bonding facilitated neural entrainment between individuals of high and low status, specifically addressing whether this entrainment occurred in a high-to-low status or low-to-high status direction or both (i.e., a bidirectional manner). By employing time-lagged analysis, we revealed the temporal dynamics of the inter-status synchronization. The prefrontal activity of the group leader preceded that of followers by 1 to 6 s, indicating a unidirectional neural alignment from group leader to followers after social bonding. Previous studies have associated such sequential neural alignment with anticipatory processing, reflecting active engagement and predictive processing of others’ behaviors and intentions during social interactions [63–65]. For example, in dyadic communication, the listener’s prefrontal activity often precedes that of the speaker, and such anticipatory neural response facilitated mutual understanding and successful communication [65]. Therefore, our findings of leader-to-follower neural alignment suggested a potential underlying process through which social bonding influenced inter-status interaction: the group leader actively engaged in anticipating and predicting the mental states of followers, enabling more frequent communication and coordination with them. Moreover, the effects of social bonding on synchronized and leader-proceeding neural entrainment were only observed in the rDLPFC rather than the rTPJ. Previous work has evidenced a crucial role of the prefrontal cortex, particularly the DLFPC, in predicting forthcoming sensory inputs (e.g., short sound, [73]), social cues (e.g., eye gaze, [63]), and social dominance and future social interactions [74], suggesting DLPFC as a hub region for monitoring errors and updating predictions of upcoming inputs to prepare for appropriate responses [73]. Therefore, the impact of social bonding on rDLPFC synchronization and leader-to-follower alignment may provide a neurocognitive account for increased leader initiation and more frequent, efficient leader-follower information exchange. Interestingly, we showed that group leaders with stronger DLPFC-TPJ functional connectivity exhibited a greater degree of neural alignment with followers. This link suggested that the exchange and integration of information between DLPFC and TPJ may support predictive neural alignment from leader to follower. Taken together, in-group social bonding may enable the group leader to actively adopt followers’ perspective, consider their potential behaviors and intentions, and align predictively with them. In-group social bonding prioritizes the allocation of cognitive resources, emotional attachments, and neural entrainment to inter-status interactions within hierarchical groups, and fails to yield comparable effects on intra-status interactions. This discrepancy cannot be attributed to a ceiling effect resulting from potentially preexisting close bonding among individuals of the same-status due to shared similarities [75], which may limit the extent to which external bonding manipulation can further enhance their interactions. This interpretation is supported by the absence of differences in communication frequency and responsiveness, and neural synchronization between intra-status and inter-status dyads in the control condition without external bonding manipulation. Moreover, the bonding effects on neural synchronization exhibited a distinct pattern, with social bonding decreasing neural synchronization in the rDLPFC for intra-status dyads while increasing it for inter-status dyads. Given the critical role of the rDLPFC in top-down regulation of social attention [76], we propose that the observed decreases in neural synchronization in the rDLPFC of intra-status dyads may reflect a disengagement of attention from fellow members, potentially accompanied by a reallocation of attention towards the group leader [77]. Consistently, followers under social bonding engaged in more frequent and responsive communications with group leaders (compared to other fellow members, as shown in Fig 1E) and perceived leaders as more influential and socially attractive. Such an upward attention shift may contribute to the maintenance of structural stability by facilitating follower’s understanding of the intentions and/or preferences of the group leader while minimizing potential competition and violations [74,77,78]. Together, these findings put forward the hypothesis that the effects of social bonding on modulating neural couplings and redirecting attentional engagement from intra-status to inter-status interaction may serve to reinforce the hierarchical structure. Extending previous findings of bonding effects on egalitarian group, our work reveals the critical roles of social status in shaping the strength and nature of the social bonding experience in hierarchical groups, which operates at both the individual and dyadic levels. At the individual level, social bonding facilitates the initiation and engagement of high-status individuals in group communications, while increases the responsiveness of low-status individuals to group leader and their perceptions of high-status individuals. At the dyadic level, social bonding exerts distinct effects on the inter-status and intra-status dynamics, potentially through the mechanisms of top-down predictive alignment and bottom-up attentional shift. Specifically, social bonding increases a leader’s forward-prediction of follower’s neural activity, while suppresses follower’s neural and attentional entrainment with same-status fellows. These plausible neurocognitive pathways helped us to synthesize social bonding effects, providing neural accounts for the effect of social bonding on group dynamics within hierarchical social contexts. The establishment of social bonds between leader and followers may serve to alleviate inter-status inequality and competition, foster inter-status coalitions, and maintain social hierarchy [71]. Our findings may be limited to the current experimental settings and could raise a number of exciting research questions for future studies. First, nonverbal communication, such as gestures, facial expressions, and eye contact, plays a crucial role in real-life social interactions. However, since our participants were restricted to online communication through typing, the effects of social bonding on nonverbal hierarchical interaction remain unexplored. Second, the current study was conducted within simple three-person hierarchical groups with leaders who represented symbolic, perhaps prestige-style leaders, democratically elected by group members. These leaders made more contributions in intergroup economic games and established positive connections with followers, but lacked the authority to sanction fellow members or allocate resources. These settings deviated from those commonly encountered in complex real-life scenarios. Therefore, caution should be exercised when attempting to generalize our findings and future work is encouraged to explore the generality and specificity of social bonding effects across diverse leadership styles. Third, our study examined the three-person group, which is the minimal unit of a hierarchical group with only 2 levels of social hierarchy. While group dynamics and leadership in such a small-scale society are arguably representative, it lacks the complexity of group structure institutions. It will be interesting for future research to test whether the bonding effect would be weakened or amplified by the despotic power of nonhuman animal groups or the complexity of large-scale social networks. Finally, it should be noted that the current dataset was obtained from participants of a specific cultural background, i.e., East Asian Chinese individuals. This raises the question of whether the observed effects in the current sample can be generalized to individuals from other cultures. Individuals from East Asian cultures place emphasis on group cohesion, interpersonal connection, and social hierarchy [79–81]. In comparison to Western cultures, followers in East Asian cultures tend to display higher levels of obedience and commitment towards their group leader while also encouraging more supportive leadership [82]. Therefore, one may expect cultural differences in how leaders and followers interact, especially after in-group social bonding, in hierarchical groups. To explore this possibility further, we conducted a preliminary examination by assessing individual differences in cultural values within our sample. Previous studies have suggested that cultural group differences in the neural activity underlying social cognition may be mediated by cultural values, such as interdependence of self-construal [83]. In the current study, we employed the Self-Construal Scale [84] to assess individual variations in cultural value of interdependence. We found that individual differences in independence did not influence behavioral and neural indices related to inter- or intra-status interactions nor did they affect observed bonding effects on behavioral and neural indices. These results suggested that our findings may be insensitive to culture-specific values. However, we acknowledge that the lack of modulation by cultural values could potentially be attributed to minimal variability in culture values within a single cultural context. It is important for future cross-cultural research to directly test whether and how our current findings can be generalized to other cultural populations. Materials and methods Participants The current study reported behavioral and neural data from 176 three-person, same-gender groups (528 healthy volunteers). Eighty-nine three-person groups were randomly assigned to the bonding condition, and 87 groups underwent a matched no-bonding control condition. Participants in the bonding/control conditions, regardless of their role of leader or follower, did not differ significantly in age, gender, education, empathic capacity, cooperative personality traits, preference for social hierarchy, cultural value, or baseline intergroup discrimination (all ps > 0.05; S1 Table). The current study has not been preregistered. We conducted the study with a relatively large sample which was chose to exceed comparable behavioral and neuroimaging studies [85,86] to enhance power and to draw robust conclusions. Moreover, we conducted post hoc power analysis and confirmed that the bonding effects on both behavioral and neural indices yielded a statistical power larger than 99% (i.e., 99.8% for 1 main behavioral index of intergroup discrimination and 99.5% for 1 main neural index of the INS in the rDLPFC), based on the current sample (n = 176) with α = 0.05, repeated measures F-tests, and effect size f = 0.188 (0.234). In addition, an independent group of participants (n = 14, 5 males, age: 18 to 23 years old, mean ± SD = 19.85 ± 1.61 years, education: 13 to 15 years, mean ± SD = 14.50 ± 1.09 years) were recruited to evaluate the group interaction quality for the 176 groups. All participants had a normal or corrected-to-normal vision and were free of psychiatric disorders and neurological conditions. Ethics statement All participants provided written informed consent prior to the study and were paid for their participation. The experimental protocol was approved by the local research ethics committee at the State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University (protocol number: IORG0004944) and was conducted in accordance with the Declaration of Helsinki. Experimental procedure and tasks Prior to the formal experiment, participants completed an online survey consisting of a set of questionnaires and color preference for black versus white (for the bonding manipulation). One to 4 days later, groups of 3 same-gender strangers came to the laboratory. During the fNIRS-based hyper-scanning, participants sat face-to-face in a triangle and completed 3 sessions (Figs 1A and S1): (i) a baseline session of 4-min resting-state where participants remained motionless, eyes closed, and mind relaxed [44]; (ii) a 4-min in-group social bonding (or no-bonding control) manipulation session; and (iii) a 4-min online within-group interaction session where participants discussed a given topic with in-group members and democratically elected the leader of their group. We employed online interaction in the current experiment as a growing number of social interactions are taking place online (e.g., video conference, [87]). In addition, the online setting ensures the anonymity of each group member, avoiding potential confounding. After completing the 3 sessions, participants were asked to report subjective evaluations on inter- and intra-status cohesion. Followers also reported perceived social influence and social attraction of their group leader. Moreover, participants completed an IDG and an IPD-MDG before the experiment and at the end of the experiment. In-group social bonding manipulation session. Each three-person group was randomly assigned to the social bonding or no-bonding control session. The in-group bonding manipulation in the current study was adapted from several previously validated procedures [15,46–49]. Specifically, we merged 3 essential procedures used in previous studies to manipulate in-group bonding: shared preference [46], symbolic marker [47], and similarity among group members [48]. First, for each in-group bonding session, we invited 3 participants preferring the same color (black or white, as indicated in the preexperiment survey) and used this color preference to create group identities, with the individuals referring black over white being labeled as “Group Black” and those preferring white over lack being labeled as “Group White.” Second, each participant was then given either a black vest (for “Group Black”) or a white vest (for “Group White”) to wear during the entire experimental session. The same-colored uniforms serve as an arbitrary symbolic marker and have been proven as a strong approach to enhance group identification, group coordination, and in-group favoritism [49]. Third, each group participated in a 4-min online chat session to introduce themselves and identify common features among group members. Brief conversations have been recognized as an effective way to establish and maintain social bond [88] and to strengthen interconnections among group members, particularly when the topic involved similarities or interdependence among them [89]. In contrast, participants in the no-bonding control session did not share color preference nor wear group uniform [15] and talked about their main courses without being explicitly asked to find shared features during the online communications. No verbal communication was allowed in both bonding and control sessions. Leader election and within-group interaction session. Participants engaged in a 4-min within-group online chat session. Similar to previous studies [44,45], participants were asked to: (i) democratically elect a leader for their group; and (ii) discuss the strategies that their group would employ in potential intergroup contests. Participants had the autonomy to decide whether they preferred initiating discussions on strategies, electing a group leader first, or discussing both 2 topics in parallel. We made it clear to participants that leaders in the current experiment were elected rather than appointed and held symbolic roles without formal power or responsibility over other group members. During the online chatting, participants were labeled by different shapes and communicated via sending messages to remain anonymous (verbal communication was not allowed). For each three-person group, any group member could self-nominate or nominate their peers as the leader. Among all groups, 85% of groups (N = 149) nominated 1 candidate as the leader, who was then approved, while the remaining 15% of groups (N = 27) initially nominated multiple candidates but later identified one of them as the final leader later. Of all leaders, 37.5% (N = 66) self-nominated while 62.5% (N = 110) were nominated by their peers. Leader election validation. All 176 three-person groups elected their leaders within the 4-min session and successfully recognized group leader in the post-survey. We conducted conversation analysis and compared the total number and length of the utterance given by leader and followers. We found that the group leader was more talkative than followers during group communications (total number of utterances: t175 = 5.418, p = 1.970 × 10−7, Cohen’s d = 0.408, 95% CI: 1.315, 2.822; total length of utterances: t175 = 8.316, p = 2.468 × 10−14, Cohen’s d = 0.627, 95% CI: 26.932, 43.693). Moreover, the group leader was more likely to initiate group interaction (leader: 40.3% versus three-person-average: 33.3%, p = 0.039). In addition, we asked participants to report their willingness to be the group leader (0 = not willing at all, 10 = extremely willing). We found that, both in bonding and control conditions, the group leader showed higher willingness and motivation to be the group leader (main effect of leader: F1, 172 = 64.615, p = 1.40 × 10−13, η2 = 0.273). Moreover, leaders who underwent social in-group bonding exhibited a higher level of willingness to be the group leader (leader × bonding interaction: F1, 172 = 0.067, p = 0.812, η2 = 3.31 × 10−4). Finally, the chat transcriptions of all 176 groups were evaluated by independent raters, who were blind to experimental hypotheses and conditions, along domains of (i) identifying the group leader of each group (identification accuracy = 97.66%); and (ii) reporting perceived interaction quality and leadership. These results suggested that leaders were not randomly elected or coerced into their positions and validated our experimental setup for establishing the roles of the group leader and followers in three-person groups. Economic games related to intergroup discrimination. In the IDG, participants received an endowment of 100 monetary units (MUs) and decided how to split the 100 MUs between hypothetical in-group and out-group members. The different MUs to in-group and out-group members indicated the level of intergroup discrimination [15,58]. In the IPD-MDG, participants received an endowment of 100 MUs and decided how to split the 100 MUs among a self pool, a within-group pool, and a between-group pool [59,60]. Each MU to the self pool was kept as 1 MU to the participants themselves; each MU contributed to the within-group pool added 0.5 MU to each in-group member, including the contributor, and each MU contributed to the between-group pool added 0.5 MU to each in-group member and in addition, subtracted 0.5 MU from each out-group member. Contributions to the within-group pool reflects “ingroup love” without hurting out-group members. Contributions to the between-group pool, in contrast, reflects spiteful “out-group hate.” fNIRS data acquisition Brain activity of 3 individuals of the same group was simultaneously recorded by 1 LABNIRS optical topography system (52-channel high-speed LABNIRS, Shimadzu Corporation, Japan). For each participant, 2 sets of homologous optode probes were used, with each measuring 7 channels (3 light emitters and 3 detectors, inter-optode distance of 30 mm, Fig 1B). The 2 probe sets separately covered the rDLPFC and the rTPJ, the position of which was based on the relevant standard positions of F4 and P6 in the 10–10 international system for electroencephalogram electrode placement (S8 Table, [90]). The anatomical localization of rDLPFC and rTPJ was confirmed in our previous study using high-resolution T1-weighted structural images [15]. The absorption of near-infrared light at 3 wavelengths (780 nm, 805 nm, and 830 nm) was measured at a sampling rate of 47.62 Hz and later down-sampled to 9.52 Hz by averaging 5 consecutive data points for all the analyses to decrease temporal autocorrelation [91]. The absorption changes were transformed into the relative concentration changes of oxygenated hemoglobin (oxy-Hb), deoxygenated hemoglobin (deoxy-Hb), and total hemoglobin based on the modified Beer–Lambert law [92]. It has been shown that the concentration changes of oxy-Hb are the most sensitive indicator of the regional cerebral blood flow in fNIRS measures and increases in oxy-Hb reflects the consequence of neural activity and corresponding to the blood oxygenation level dependent (BOLD) signal measured by fMRI [93]. Moreover, the oxy-Hb concentration has a better signal-to-noise ratio than deoxy-Hb [34]. Thus, similar to most previous fNRIS studies [15,34–36,40,44,45], the current study focused on the concentration changes of oxy-Hb. Behavioral data analysis Conversation analysis. We analyzed the transcripts of turn-taking conversations for each three-person group. For the within-group communication of each group, we calculated: (i) the total number of utterances of group leader and followers, respectively; (ii) the number of turn-transition; and (iii) the turn response time (i.e., the time gap between turns) separately for inter-status and intra-status transitions. The inter-status turn-transition was defined as a discrete pair of a leader utterance followed by a follower utterance, or vice versa, whereas the intra-status turn-transition was defined as a follower utterance followed by another follower. The turn response time was then log transformed to avoiding skewness. To control for the total length of utterances of all 3 group members, we include the total word number as covariate in the further analysis. These conversation-related indices were submitted to two-way mixed-model ANCOVAs with a within-subjects factor hierarchy (inter-status versus intra-status or leader versus follower), a between-subjects factor bonding (bonding versus control), and total word number as covariate. It should be noted that, the mean turn response time in the current study deviated largely away from the verbal conversation turn response time of about 200 ms, possibly due to that group members communicated via online messaging rather than verbal communication. In addition, to mitigate the potential impact of displaying “typing” in the group chat window on response latency measurement, we have opted not to show status information (i.e., whether someone is typing or not) in the chat window. Only after participants completed and submitted their messages into the group chat window, other members were then able to read others’ messages. Participants’ subjective ratings. Participant’s self-report of inter-status cohesion (i.e., cohesion between leader and follower) and intra-status cohesion (i.e., cohesion between followers) were submitted to a two-way mixed-model ANOVA with a within-subjects factor hierarchy (inter-status versus intra-status) and a between-subjects factor bonding (bonding versus control). Follower’s ratings of perceived social influence and attraction of the group leader was compared between bonding and control conditions with two-tailed independent-sample t tests. Moreover, we performed partial correlation analyses to test the relationship between the perceived social influence (attraction) and the turn transition (response time) separately for inter- and intra-status, controlling the utterance length of leaders and followers, respectively. We next conducted mediation analysis to examine whether the effect of social bonding (the independent variable, X) on the leader’s social influence or attraction (the dependent variable, Y) was mediated by social cohesion (the mediator, M). The mediation analyses were performed using PROCESS (model 4) in SPSS. For each mediation analysis, we computed both the indirect effect (c) and direct effect (c’), where c measures the extent to which X influences Y through M, while c’ represents the effect of X on Y that cannot be explained by c. If c is nonzero and c’ is zero, it indicates full mediation; otherwise, it suggests partial mediation. The significance of mediation was assessed using both Sobel test and Preacher–Hayes bootstrapping. The Sobel test was employed to determine if introducing M significantly reduced the relationship between X and Y. A two-tailed z-test was used to establish statistical significance level. On the other hand, bootstrapping method involved a nonparametric examination of whether zero fell within the bootstrapped 95% confidence intervals (5,000 times). If zero did not fall within the confidence intervals, we could conclude that there existed a significant indirect effect. It showed be noted that causal direction in the mediation analysis was not directly examined. Further research employing experimental and longitudinal approaches is warranted to validate our current findings from the mediation analysis. Intergroup discrimination. Participants completed the IDG and IPD-MD before the experiment (as baseline) and at the end of the experiment. We were interested in the intergroup discrimination, i.e., differences in monetary allocation to in-group members and out-group members. Moreover, in IPD-MD, we were also interested in the amount of MU allocated to the within-group pool (i.e., “ingroup love”) and the between-group pool (i.e., “outgroup hate”), respectively. We first compared the baseline intergroup discrimination measured before the experiment between bonding and control sessions and between leaders and followers. We found comparable baseline intergroup discrimination (main effect of bonding: F1, 174 = 0.541, p = 0.463, η2 = 0.003; main effect of hierarchy: F1, 174 = 0.017, p = 0.897, η2 = 9.712 × 10−5; bonding × hierarchy interaction: F1, 174 = 0.007, p = 0.935, η2 = 3.855 × 10−5), ingroup love (main effect of bonding: F1, 174 = 0.149, p = 0.700, η2 = 0.001; main effect of hierarchy: F1, 174 = 0.257, p = 0.613, η2 = 0.001; bonding × hierarchy interaction: F1, 174 = 1.484, p = 0.225, η2 = 0.008), and outgroup hate (main effect of bonding: F1, 174 = 1.140, p = 0.287, η2 = 0.007; main effect of hierarchy: F1, 174 = 0.447, p = 0.505, η2 = 0.003; bonding × hierarchy interaction: F1, 174 = 0.320, p = 0.572, η2 = 0.002) among the 4 conditions. The intergroup discrimination in the IDG, the ingroup love and outgroup hate in the IPD-MDG, measured at the end of the experiment, were submitted to bonding × hierarchy (leader versus follower) ANOVAs to examine bonding effects on intergroup discrimination in leader and followers. Independent ratings on within-group interaction. To objectively quantify the within-group interactions perceived by third-party observers, independent raters were asked to identify the leader of each group and leader emergence time and to evaluate the chatting messages of all the 176 groups along the following dimensions on a 9-point Likert scale: (i) interaction frequency (1 = least frequent, 9 = most frequent); (ii) interaction intensity (1 = least intense, 9 = most intense); (iii) group cohesion (1 = least cohesive, 9 = most cohesive); and (iv) prominence of the group leader (1 = least prominent leader, 9 = most prominent leader). The independent raters correctly identified the group leader (accuracy: 97.66%). The ratings were reliable among the 14 raters (interaction frequency: ICC consistency: 0.940, agreement: 0.885; interaction intensity: ICC consistency: 0.814, agreement: 0.710; group cohesion: ICC consistency: 0.855, agreement: 0.757; leader prominence: ICC consistency: 0.813, agreement: 0.746). For each rating dimension, rating scores were first z-score transformed for each rater. The normalized ratings were then averaged across raters and compared between bonding and control conditions using two-tailed independent-sample t tests. The rating data of one group was missing. We showed that social bonding indeed facilitated the outsider-perceived group interaction frequency (t173 = 4.426, p = 1.696 × 10−5, Cohen’s d = 0.669, 95% CI: 0.264, 0.688, S3A Fig) and intensity (t173 = 3.943, p = 1.166 × 10−4, Cohen’s d = 0.596, 95% CI: 0.157, 0.472, S3B Fig). Moreover, independent raters identified the leader of groups under bonding (versus control) condition faster (t173 = −2.547, p = 0.012, Cohen’s d = −0.385, 95% CI: −40.522, −5.140, S3C Fig) and also recognized the prominence of group leader in the bonding condition to a greater degree than that in the control condition (t173 = 2.001, p = 0.047, Cohen’s d = 0.303, 95% CI: 0.002, 0.339, S3D Fig). FNIRS data analysis Data quality check. The following steps were performed to check and ensure the data quality of fNIRS signals. First, prior to data recording, the resistance for each optode probe was monitored and adjusted to meet the minimum criteria defined in the LABNIRS recording software, aiming to achieve a good signal-to-noise ratio [94]. Second, during data recording, the raw optical density signal in each channel was transformed into concentration changes in HbO and HbR in real time, which were visually inspected to assess signal quality [95]. Third, during preprocessing of the fNIRS signals, we applied an automatic sliding-window detection to further evaluate data quality. Specifically, for each HbO and HbR time series, extreme values exceeding mean ± 3 standard deviations (SD) within a 10-s time-window were identified as outliers [96]. Channels with severe motion artifacts or containing more than 5% extreme values were labeled as bad. In order to maintain the same number of channels across all groups, any group with one or more bad channels was excluded from further analyses. Based on these criteria, a total of 8 groups were excluded from the final dataset due to poor data quality. Inter-brain neural synchronization (INS). We performed INS analysis on the neural data collected during the resting-state session (4 min, served as a baseline) and within-group interaction session (4 min). Similar to previous studies [15,34–37,44,45], we employed the WTC analysis to assess the cross-correlations between 2 oxy-Hb time series of dyads of participants as a function of frequency and time [97]. Specifically, within each three-person group, the 3 same-length oxy-Hb time series for each channel (i.e., oxy-Hbleader, oxy-Hbfollower1, and oxy-Hbfollower2) were simultaneously acquired by the same fNIRS system. We applied WTC analysis to each pair of 3 oxy-Hb time series and generated 3 time-frequency matrices of the coherence values for each group (i.e., Coherenceleader-follower1, Coherenceleader-follower2, and Coherencefollower1-follower2). The coherence values from the leader-follower dyads (averaged value of Coherenceleader-follower1 and Coherenceleader-follower2) indicated the inter-status INS, and the coherence value from the follower–follower dyads (i.e., Coherencefollower1-follower2) indicated the intra-status INS. In each time-frequency matrix (Fig 3A), each row corresponded to a specific frequency point, each column corresponded to a specific time point and the color bar corresponded to the coherence value. To ensure consistent data size for inter- and intra-status INS, we conducted a set of control analyses by calculating coherence values between the group leader and a randomly selected follower to index inter-status INS. With this approach, the data size and calculation of inter- and intra-status INS were matched. The results obtained from these control analyses replicated our main findings (S10 Fig). fNIRS signals not only reflected task-evoked brain activity but also systemic physiological interference arising from cardiac pulsation (approximately 1 Hz), breathing rate (approximately 0.3 Hz), and other homeostatic processes [95]. Similar to previous studies [4,98], we employed the baseline subtraction approach to mitigate the impact of physiological noise. Specifically, we recorded a resting state session with an identical duration as the task session. As the resting-state predominantly reflects spontaneous hemodynamic oscillations [98], it served as the baseline for comparison. We performed WTC analysis on the neural data collected during both the resting-state session (4 min, serving as a baseline) and within-group interaction session (4 min). To reveal the effects of bonding and hierarchy on INS specific to group interaction, we focused on the increased INS during group interaction relative to the baseline (i.e., the resting-state). First, we compared coherence values (averaged across all channels and for each channel) between the within-group interaction session and resting-state session by performing paired-sample t tests for each frequency (frequency range 0.01 to 1 Hz) [99] to identify the FOI (Fig 3A). This analysis identified increased coherence values in 2 frequency bands: between 0.136 Hz and 0.192 Hz (corresponding to the period between 5.20 and 7.35 s) and between 0.407 and 0.432 Hz (corresponding to the period between 2.31 and 2.46 s). These 2 frequency bands were chosen as the frequency of interest for the subsequent analyses (FOI, Bonferroni family-wise error (FWE) corrected for multiple comparisons). It is worth noting that no significant results were found in the frequency band of 0.407 to 0.432 Hz (full statistics were reported in S9 Table); therefore, only results based on 0.136 to 0.192 Hz were reported in the main text. This chosen period also effectively captures the temporal structure of the within-group interaction task since one-round within-group messaging typically took an average time of 5 to 7 s. In addition, this frequency band also excluded high- and low-frequency physiological noises, such as those related to respiration (about 0.2 to 0.3 Hz), cardiac pulsation (0.7 to 4 Hz), and high-frequency head movements (>1 Hz) [98]. We then calculated the session- and FOI-averaged coherence values and converted into Fisher z-scores. The increased INS (coherence differences between within-group interaction and resting-state) was submitted to bonding × hierarchy (inter-status versus intra-status) ANOVAs. Note that, the WTC algorithm normalized the amplitude of the signal within each time-window defined by the wavelet to make the data less vulnerable to transient spikes or motor artifacts [34–36]. Moreover, we conducted 2 sets of complementary analyses to further control the potential impact of physiological noises. First, we applied a wavelet-based denoising method to identify global physiological components per channel and extracted them out of the hemodynamic signals [15,61]. After the denoising process, the same WTC calculation and statistical analyses were applied, and the observed results were reserved (main effect of hierarchy at channel 3, F1, 174 = 6.296, p = 0.013, η2 = 0.035, hierarchy × bonding interaction effect at channel 9, F1, 174 = 5.311, p = 0.022, η2 = 0.030, S3A Table). Second, we controlled the globally co-varying signal using ANCOVA analyses [62], with the global mean INS (averaged coherence values across all channels) as the covariate. Significant results were fully replicated (main effect of hierarchy at channel 3, F1, 174 = 9.220, p = 0.003, η2 = 0.051, hierarchy × bonding interaction effect at channel 9, F1, 174 = 8.373, p = 0.004, η2 = 0.046, S3B Table). Time-lagged analysis. To investigate the directionality of the inter- and intra-status INS, we conducted time-lag analysis [63–65] for the channels that showed increased INS during the within-group interaction stage, i.e., channel 3 in the rTPJ and channel 9 in the rDLPFC. For each leader–follower (or follower–follower) dyad, the time courses of neural activity of the leader (1 follower) were shifted relative to that of the follower (the other follower) from −10 to 10 s (in 1-s increment). We then recalculated the inter- and intra- status INS on each time lag for both resting-state and within-group interaction. The time-lagged inter- and intra-status neural alignment increases (i.e., lagged INS during within-group interaction minus that during resting) on each time lag were compared with 0 using one-sample t tests and compared between bonding and control conditions using two-tailed independent-sample t tests. Significant effects were thresholded at p < 0.05, FDR corrected for multiple comparisons of the 21 time lags. Next, we performed correlation analysis between the neural alignment on each time lag and intergroup discrimination, for inter-status and intra-status dyads, respectively, the correlation coefficients were FDR corrected for multiple comparisons of the 21 time lags. We also performed Pearson’s correlation coefficient analysis separately for bonding and control conditions. Permutation test. First, we aimed to validate the INS increases (i.e., group interaction versus resting-state) in real groups. To this end, we examined which conditions showed increased INS (i.e., significant INS increases during group interaction compared to resting-state). Specifically, we compared INS differences (group interaction minus resting-state) against zero for the channels that exhibited significant effects of interest (i.e., channel 3 in the rTPJ and channel 9 in the rDLPFC). We found increased INS for the inter-status dyads at channel 3 in the rTPJ under control condition (t1,86 = 3.943, p = 1.64 × 10−4) and bonding condition (t1,88 = 3.394, p = 0.001; survived multiple correction) and at channel 9 in the rDLPFC (bonding: t1,88 = 3.378, p = 0.001; survived multiple correction). We then performed permutation test to examine whether these conditions showed increased INS in real group than pseudo groups. Specifically, within each condition, we generated three-person pseudo-groups by randomly grouping 3 participants from different original real groups. For each pseudo-group, the INS of each dyad was recalculated. This procedure was repeated for 1,000 times to generate a pseudo-group INS distribution. The increased INS in the aforementioned conditions of real groups were compared with each condition-specific permutation distributions. The significance level, p-value was indicated as: p = j/1,000, where j is the number of samples out of the 1,000 permutation samples, of which the examined value was larger than the observed value of real groups. The results indicated that the increases in inter-status INS in the rTPJ (control: p = 0.022; bonding: p = 0.047) and rDLPFC (bonding: p = 0.032) of real groups all exceeded the upper 95% CI of the permutation distribution. These findings further confirmed an increased INS in these specific conditions within real (rather than random) groups. Next, to validate the observed bonding and/or hierarchy effects on INS, we performed another 2 sets of permutation tests: (i) the within-condition permutation test; and (ii) cross-condition permutation test. First, within the bonding and control conditions, leader and followers of the real groups were randomly reassigned into new groups to form three-person pseudo groups. We then recalculated the inter- and intra-status INS for the 176 pseudo groups. This shuffling and recalculation procedures were repeated 1,000 times to generate permutation distributions for the observed hierarchy effect in rTPJ and bonding × hierarchy interaction effect in the rDLPFC. We then compared the observed effects of real interacting groups against 1,000 permutation samples and examined whether real effect exceeded the upper limits 95% or 99% CI of the permutation distribution. Second, similar procedures and statistical analyses were conducted for the cross-condition permutation test except that we generated cross-condition, three-person pseudo groups by randomly grouping 1 leader and 2 followers across bonding and control conditions as a pseudo-group. rDLPFC-rTPJ functional connectivity. We applied cross-correlation analysis using the Functional Connectivity Toolbox [100] implemented in MATLAB to assess the functional connectivity of each rDLPFC-rTPJ channel pair (49 channel pairs, 7 channels in TPJ, and 7 channels in DLPFC) for each participant, which was then Fisher z transformed [101]. We also averaged the 49 channel pairs to index the grand mean of functional connectivity. The channel-pairwise connectivity and grand mean connectivity were separately submitted to bonding × hierarchy (leader versus follower) ANOVAs (FDR correction for multiple comparisons of 49 channel pairs was applied to channel-pairwise analysis). Next, the channel showing significant bonding effect on leader-to-follower neural alignment (i.e., CH9 in the rDLPFC) was used as the seed channel and the averaged functional connectivity across the 7 CH9-rTPJ channel pairs was used to index the channel-based functional connectivity. We then conducted correlation analysis between the channel-based functional connectivity and the leader-to-follower neural alignment in rDLPFC. Additional analyses and results. To test whether the effects of social in-group bonding on hierarchical interaction were modulated by different interaction stages, we conducted additional analyses. First, we identified the specific time point at which the group leader explicitly emerged in each group. This analysis revealed that the establishment of a group leader occurred approximately halfway through the within-group interaction (with a mean value of 145.60 s, SE = 5.76). Second, based on this identified time point for leader emergence, we divided the within-group interaction session into 2 stages: pre- and post-leader emergence stages. Third, we conducted 3-way ANOVAs to examine whether these interaction stages influenced our main findings regarding hierarchy and/or bonding effects on both behavioral and neural indices. The results showed no significant impact of the interaction-stage on our main findings (S10A Table). Furthermore, despite observing an average occurrence of leader emergence around the middle of the within-group interaction period, we further balanced the duration between pre- and post-leader emergence stages by dividing it equally into 2 parts for subsequent analyses. This mid-split approach replicated our observation of no significant impact of different interaction stages (S10B Table). These results suggested that the effects of social bonding on hierarchical interaction remained consistent across different stages of within-group interaction. Alternatively, it is possible that in the current experimental setting, the group leader implicitly emerged prior to the explicit emergence time point (e.g., during the bonding section). Supporting this possibility, we observed that the group leader was more likely to initiate group interaction at the beginning of within-group interaction and was already more talkative in the pre-emergence stage (total number of utterances: t175 = 4.877, p = 2.404 × 10−6; total length of utterances: t175 = 7.451, p = 4.076 × 10−12). Furthermore, a majority of groups (85%, N = 149) early on nominated the individual who later emerged as their leader. We encourage future studies to directly investigate these possibilities. In addition, it is important to note that this examination of time effect was merely an initial exploration conducted at a coarse time scale. Thus, the neural dynamics at second or millisecond resolution needs to be directly and systematically examined in future studies. To present the measured brain activity from different perspectives, we repeated our analysis on the deoxygenated hemoglobin signals (HbR). Specifically, we calculated the INS and intra-brain rDLPFC-rTPJ functional connectivity on HbR signals. We then performed the bonding × hierarchy ANOVAs on HbR-INS and HbR-FC to examine whether the main findings obtained with HbO signals would be similarly observed with HbR signals. Similar to the pattern of the HbO signals, we observed a significant, although weaker (did not survive multiple corrections), hierarchy effect in rTPJ (channel 3, F1, 174 = 4.736, uncorrected p = 0.031, η2 = 0.026, S11 Table and S11A Fig), with stronger inter-status INS than intra-status dyads. However, no bonding × hierarchy interaction effect was observed on HbR signals (channel 9, F1, 174 = 0.420, p = 0.518, η2 = 0.002). The difference observed on INS based on HbO and HbR signals may be caused by different sensitivities of these 2 types of signals in reflecting task-induced changes in neural signals. Regarding the intra-brain FC index, the leader effect on the rDLPFC-rTPJ functional connectivity (stronger in leader than followers) based on HbO signals was similarly observed in the HbR-FC analysis (48 rDLPFC-rTPJ channel pairs survived FDR correction for 49 channel pairs, S12 Table; S11B Fig for the grand mean rDLPFC-rTPJ connectivity, F1, 174 = 27.345, p = 4.842 × 10−7, η2 = 0.136). Statistical analysis Similar to previous studies [4,15,102], data were aggregated at the three-person group level and hierarchy within each group (i.e., leader versus follower, or inter-status versus intra-status) were treated as a within-subjects factor. For both the behavioral and neural data, we averaged the 2 followers or the 2 leader–follower dyads to index the follower or inter-status level. The experimental condition (bonding versus control) was randomly introduced and blinded to the participants during data collection. For each dependent variable, the three-person groups whose value was larger or smaller than 5 SDs from the mean value were excluded. This data cleaning procedure led to exclusion of data in the following variables: intra-status turn transition (n = 1), intra-status turn response time (n = 1), and inter-status turn response time (n = 2). Two-way mixed-model ANOVAs were conducted on final behavioral and neural datasets with bonding (bonding versus no-bonding control) as a between-subjects factor and hierarchy (inter-status versus intra-status, or leader versus follower) as a within-subjects factor. Furthermore, the LMM is another optimal method for analyzing such structural data. Therefore, we performed a series of LMM analyses on behavioral and neural indices, considering hierarchy and bonding as fixed effects while treating each group as a random effect. This set of analyses yielded similar results to ANOVA. ANOVA with significant interaction were followed by planned two-tailed t tests to examine: (i) bonding effects (two-tailed independent-sample t test) separately on inter-status and intra-status dyads or leaders and followers; and (ii) hierarchy effects (two-tailed paired-sample t test) separately in bonding and control condition. Statistical significance was thresholded at p < 0.05. Data distributions were assumed to be normal, but this was not formally tested. For ratings from independent sample, data were first normalized across items for each rater and then averaged across all raters. Correlation analyses were conducted using Pearson’s correlation coefficient analysis. It should be noted that the reported behavior-neural correlations, although statistically significant, should be interpreted and applied with caution due to their small to medium effect sizes [103]. All statistical analyses were performed with SPSS (IBM SPSS Statistics 25) and custom scripts in MATLAB (R2017b & R2020b, The MathWorks, United States of America). The wavelet coherence analysis was performed by Wavelet Coherence Package [104] implemented in MATLAB (which is available in https://noc.ac.uk/business/marine-data-products/cross-wavelet-wavelet-coherence-toolbox-matlab). Participants The current study reported behavioral and neural data from 176 three-person, same-gender groups (528 healthy volunteers). Eighty-nine three-person groups were randomly assigned to the bonding condition, and 87 groups underwent a matched no-bonding control condition. Participants in the bonding/control conditions, regardless of their role of leader or follower, did not differ significantly in age, gender, education, empathic capacity, cooperative personality traits, preference for social hierarchy, cultural value, or baseline intergroup discrimination (all ps > 0.05; S1 Table). The current study has not been preregistered. We conducted the study with a relatively large sample which was chose to exceed comparable behavioral and neuroimaging studies [85,86] to enhance power and to draw robust conclusions. Moreover, we conducted post hoc power analysis and confirmed that the bonding effects on both behavioral and neural indices yielded a statistical power larger than 99% (i.e., 99.8% for 1 main behavioral index of intergroup discrimination and 99.5% for 1 main neural index of the INS in the rDLPFC), based on the current sample (n = 176) with α = 0.05, repeated measures F-tests, and effect size f = 0.188 (0.234). In addition, an independent group of participants (n = 14, 5 males, age: 18 to 23 years old, mean ± SD = 19.85 ± 1.61 years, education: 13 to 15 years, mean ± SD = 14.50 ± 1.09 years) were recruited to evaluate the group interaction quality for the 176 groups. All participants had a normal or corrected-to-normal vision and were free of psychiatric disorders and neurological conditions. Ethics statement All participants provided written informed consent prior to the study and were paid for their participation. The experimental protocol was approved by the local research ethics committee at the State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University (protocol number: IORG0004944) and was conducted in accordance with the Declaration of Helsinki. Experimental procedure and tasks Prior to the formal experiment, participants completed an online survey consisting of a set of questionnaires and color preference for black versus white (for the bonding manipulation). One to 4 days later, groups of 3 same-gender strangers came to the laboratory. During the fNIRS-based hyper-scanning, participants sat face-to-face in a triangle and completed 3 sessions (Figs 1A and S1): (i) a baseline session of 4-min resting-state where participants remained motionless, eyes closed, and mind relaxed [44]; (ii) a 4-min in-group social bonding (or no-bonding control) manipulation session; and (iii) a 4-min online within-group interaction session where participants discussed a given topic with in-group members and democratically elected the leader of their group. We employed online interaction in the current experiment as a growing number of social interactions are taking place online (e.g., video conference, [87]). In addition, the online setting ensures the anonymity of each group member, avoiding potential confounding. After completing the 3 sessions, participants were asked to report subjective evaluations on inter- and intra-status cohesion. Followers also reported perceived social influence and social attraction of their group leader. Moreover, participants completed an IDG and an IPD-MDG before the experiment and at the end of the experiment. In-group social bonding manipulation session. Each three-person group was randomly assigned to the social bonding or no-bonding control session. The in-group bonding manipulation in the current study was adapted from several previously validated procedures [15,46–49]. Specifically, we merged 3 essential procedures used in previous studies to manipulate in-group bonding: shared preference [46], symbolic marker [47], and similarity among group members [48]. First, for each in-group bonding session, we invited 3 participants preferring the same color (black or white, as indicated in the preexperiment survey) and used this color preference to create group identities, with the individuals referring black over white being labeled as “Group Black” and those preferring white over lack being labeled as “Group White.” Second, each participant was then given either a black vest (for “Group Black”) or a white vest (for “Group White”) to wear during the entire experimental session. The same-colored uniforms serve as an arbitrary symbolic marker and have been proven as a strong approach to enhance group identification, group coordination, and in-group favoritism [49]. Third, each group participated in a 4-min online chat session to introduce themselves and identify common features among group members. Brief conversations have been recognized as an effective way to establish and maintain social bond [88] and to strengthen interconnections among group members, particularly when the topic involved similarities or interdependence among them [89]. In contrast, participants in the no-bonding control session did not share color preference nor wear group uniform [15] and talked about their main courses without being explicitly asked to find shared features during the online communications. No verbal communication was allowed in both bonding and control sessions. Leader election and within-group interaction session. Participants engaged in a 4-min within-group online chat session. Similar to previous studies [44,45], participants were asked to: (i) democratically elect a leader for their group; and (ii) discuss the strategies that their group would employ in potential intergroup contests. Participants had the autonomy to decide whether they preferred initiating discussions on strategies, electing a group leader first, or discussing both 2 topics in parallel. We made it clear to participants that leaders in the current experiment were elected rather than appointed and held symbolic roles without formal power or responsibility over other group members. During the online chatting, participants were labeled by different shapes and communicated via sending messages to remain anonymous (verbal communication was not allowed). For each three-person group, any group member could self-nominate or nominate their peers as the leader. Among all groups, 85% of groups (N = 149) nominated 1 candidate as the leader, who was then approved, while the remaining 15% of groups (N = 27) initially nominated multiple candidates but later identified one of them as the final leader later. Of all leaders, 37.5% (N = 66) self-nominated while 62.5% (N = 110) were nominated by their peers. Leader election validation. All 176 three-person groups elected their leaders within the 4-min session and successfully recognized group leader in the post-survey. We conducted conversation analysis and compared the total number and length of the utterance given by leader and followers. We found that the group leader was more talkative than followers during group communications (total number of utterances: t175 = 5.418, p = 1.970 × 10−7, Cohen’s d = 0.408, 95% CI: 1.315, 2.822; total length of utterances: t175 = 8.316, p = 2.468 × 10−14, Cohen’s d = 0.627, 95% CI: 26.932, 43.693). Moreover, the group leader was more likely to initiate group interaction (leader: 40.3% versus three-person-average: 33.3%, p = 0.039). In addition, we asked participants to report their willingness to be the group leader (0 = not willing at all, 10 = extremely willing). We found that, both in bonding and control conditions, the group leader showed higher willingness and motivation to be the group leader (main effect of leader: F1, 172 = 64.615, p = 1.40 × 10−13, η2 = 0.273). Moreover, leaders who underwent social in-group bonding exhibited a higher level of willingness to be the group leader (leader × bonding interaction: F1, 172 = 0.067, p = 0.812, η2 = 3.31 × 10−4). Finally, the chat transcriptions of all 176 groups were evaluated by independent raters, who were blind to experimental hypotheses and conditions, along domains of (i) identifying the group leader of each group (identification accuracy = 97.66%); and (ii) reporting perceived interaction quality and leadership. These results suggested that leaders were not randomly elected or coerced into their positions and validated our experimental setup for establishing the roles of the group leader and followers in three-person groups. Economic games related to intergroup discrimination. In the IDG, participants received an endowment of 100 monetary units (MUs) and decided how to split the 100 MUs between hypothetical in-group and out-group members. The different MUs to in-group and out-group members indicated the level of intergroup discrimination [15,58]. In the IPD-MDG, participants received an endowment of 100 MUs and decided how to split the 100 MUs among a self pool, a within-group pool, and a between-group pool [59,60]. Each MU to the self pool was kept as 1 MU to the participants themselves; each MU contributed to the within-group pool added 0.5 MU to each in-group member, including the contributor, and each MU contributed to the between-group pool added 0.5 MU to each in-group member and in addition, subtracted 0.5 MU from each out-group member. Contributions to the within-group pool reflects “ingroup love” without hurting out-group members. Contributions to the between-group pool, in contrast, reflects spiteful “out-group hate.” In-group social bonding manipulation session. Each three-person group was randomly assigned to the social bonding or no-bonding control session. The in-group bonding manipulation in the current study was adapted from several previously validated procedures [15,46–49]. Specifically, we merged 3 essential procedures used in previous studies to manipulate in-group bonding: shared preference [46], symbolic marker [47], and similarity among group members [48]. First, for each in-group bonding session, we invited 3 participants preferring the same color (black or white, as indicated in the preexperiment survey) and used this color preference to create group identities, with the individuals referring black over white being labeled as “Group Black” and those preferring white over lack being labeled as “Group White.” Second, each participant was then given either a black vest (for “Group Black”) or a white vest (for “Group White”) to wear during the entire experimental session. The same-colored uniforms serve as an arbitrary symbolic marker and have been proven as a strong approach to enhance group identification, group coordination, and in-group favoritism [49]. Third, each group participated in a 4-min online chat session to introduce themselves and identify common features among group members. Brief conversations have been recognized as an effective way to establish and maintain social bond [88] and to strengthen interconnections among group members, particularly when the topic involved similarities or interdependence among them [89]. In contrast, participants in the no-bonding control session did not share color preference nor wear group uniform [15] and talked about their main courses without being explicitly asked to find shared features during the online communications. No verbal communication was allowed in both bonding and control sessions. Leader election and within-group interaction session. Participants engaged in a 4-min within-group online chat session. Similar to previous studies [44,45], participants were asked to: (i) democratically elect a leader for their group; and (ii) discuss the strategies that their group would employ in potential intergroup contests. Participants had the autonomy to decide whether they preferred initiating discussions on strategies, electing a group leader first, or discussing both 2 topics in parallel. We made it clear to participants that leaders in the current experiment were elected rather than appointed and held symbolic roles without formal power or responsibility over other group members. During the online chatting, participants were labeled by different shapes and communicated via sending messages to remain anonymous (verbal communication was not allowed). For each three-person group, any group member could self-nominate or nominate their peers as the leader. Among all groups, 85% of groups (N = 149) nominated 1 candidate as the leader, who was then approved, while the remaining 15% of groups (N = 27) initially nominated multiple candidates but later identified one of them as the final leader later. Of all leaders, 37.5% (N = 66) self-nominated while 62.5% (N = 110) were nominated by their peers. Leader election validation. All 176 three-person groups elected their leaders within the 4-min session and successfully recognized group leader in the post-survey. We conducted conversation analysis and compared the total number and length of the utterance given by leader and followers. We found that the group leader was more talkative than followers during group communications (total number of utterances: t175 = 5.418, p = 1.970 × 10−7, Cohen’s d = 0.408, 95% CI: 1.315, 2.822; total length of utterances: t175 = 8.316, p = 2.468 × 10−14, Cohen’s d = 0.627, 95% CI: 26.932, 43.693). Moreover, the group leader was more likely to initiate group interaction (leader: 40.3% versus three-person-average: 33.3%, p = 0.039). In addition, we asked participants to report their willingness to be the group leader (0 = not willing at all, 10 = extremely willing). We found that, both in bonding and control conditions, the group leader showed higher willingness and motivation to be the group leader (main effect of leader: F1, 172 = 64.615, p = 1.40 × 10−13, η2 = 0.273). Moreover, leaders who underwent social in-group bonding exhibited a higher level of willingness to be the group leader (leader × bonding interaction: F1, 172 = 0.067, p = 0.812, η2 = 3.31 × 10−4). Finally, the chat transcriptions of all 176 groups were evaluated by independent raters, who were blind to experimental hypotheses and conditions, along domains of (i) identifying the group leader of each group (identification accuracy = 97.66%); and (ii) reporting perceived interaction quality and leadership. These results suggested that leaders were not randomly elected or coerced into their positions and validated our experimental setup for establishing the roles of the group leader and followers in three-person groups. Economic games related to intergroup discrimination. In the IDG, participants received an endowment of 100 monetary units (MUs) and decided how to split the 100 MUs between hypothetical in-group and out-group members. The different MUs to in-group and out-group members indicated the level of intergroup discrimination [15,58]. In the IPD-MDG, participants received an endowment of 100 MUs and decided how to split the 100 MUs among a self pool, a within-group pool, and a between-group pool [59,60]. Each MU to the self pool was kept as 1 MU to the participants themselves; each MU contributed to the within-group pool added 0.5 MU to each in-group member, including the contributor, and each MU contributed to the between-group pool added 0.5 MU to each in-group member and in addition, subtracted 0.5 MU from each out-group member. Contributions to the within-group pool reflects “ingroup love” without hurting out-group members. Contributions to the between-group pool, in contrast, reflects spiteful “out-group hate.” fNIRS data acquisition Brain activity of 3 individuals of the same group was simultaneously recorded by 1 LABNIRS optical topography system (52-channel high-speed LABNIRS, Shimadzu Corporation, Japan). For each participant, 2 sets of homologous optode probes were used, with each measuring 7 channels (3 light emitters and 3 detectors, inter-optode distance of 30 mm, Fig 1B). The 2 probe sets separately covered the rDLPFC and the rTPJ, the position of which was based on the relevant standard positions of F4 and P6 in the 10–10 international system for electroencephalogram electrode placement (S8 Table, [90]). The anatomical localization of rDLPFC and rTPJ was confirmed in our previous study using high-resolution T1-weighted structural images [15]. The absorption of near-infrared light at 3 wavelengths (780 nm, 805 nm, and 830 nm) was measured at a sampling rate of 47.62 Hz and later down-sampled to 9.52 Hz by averaging 5 consecutive data points for all the analyses to decrease temporal autocorrelation [91]. The absorption changes were transformed into the relative concentration changes of oxygenated hemoglobin (oxy-Hb), deoxygenated hemoglobin (deoxy-Hb), and total hemoglobin based on the modified Beer–Lambert law [92]. It has been shown that the concentration changes of oxy-Hb are the most sensitive indicator of the regional cerebral blood flow in fNIRS measures and increases in oxy-Hb reflects the consequence of neural activity and corresponding to the blood oxygenation level dependent (BOLD) signal measured by fMRI [93]. Moreover, the oxy-Hb concentration has a better signal-to-noise ratio than deoxy-Hb [34]. Thus, similar to most previous fNRIS studies [15,34–36,40,44,45], the current study focused on the concentration changes of oxy-Hb. Behavioral data analysis Conversation analysis. We analyzed the transcripts of turn-taking conversations for each three-person group. For the within-group communication of each group, we calculated: (i) the total number of utterances of group leader and followers, respectively; (ii) the number of turn-transition; and (iii) the turn response time (i.e., the time gap between turns) separately for inter-status and intra-status transitions. The inter-status turn-transition was defined as a discrete pair of a leader utterance followed by a follower utterance, or vice versa, whereas the intra-status turn-transition was defined as a follower utterance followed by another follower. The turn response time was then log transformed to avoiding skewness. To control for the total length of utterances of all 3 group members, we include the total word number as covariate in the further analysis. These conversation-related indices were submitted to two-way mixed-model ANCOVAs with a within-subjects factor hierarchy (inter-status versus intra-status or leader versus follower), a between-subjects factor bonding (bonding versus control), and total word number as covariate. It should be noted that, the mean turn response time in the current study deviated largely away from the verbal conversation turn response time of about 200 ms, possibly due to that group members communicated via online messaging rather than verbal communication. In addition, to mitigate the potential impact of displaying “typing” in the group chat window on response latency measurement, we have opted not to show status information (i.e., whether someone is typing or not) in the chat window. Only after participants completed and submitted their messages into the group chat window, other members were then able to read others’ messages. Participants’ subjective ratings. Participant’s self-report of inter-status cohesion (i.e., cohesion between leader and follower) and intra-status cohesion (i.e., cohesion between followers) were submitted to a two-way mixed-model ANOVA with a within-subjects factor hierarchy (inter-status versus intra-status) and a between-subjects factor bonding (bonding versus control). Follower’s ratings of perceived social influence and attraction of the group leader was compared between bonding and control conditions with two-tailed independent-sample t tests. Moreover, we performed partial correlation analyses to test the relationship between the perceived social influence (attraction) and the turn transition (response time) separately for inter- and intra-status, controlling the utterance length of leaders and followers, respectively. We next conducted mediation analysis to examine whether the effect of social bonding (the independent variable, X) on the leader’s social influence or attraction (the dependent variable, Y) was mediated by social cohesion (the mediator, M). The mediation analyses were performed using PROCESS (model 4) in SPSS. For each mediation analysis, we computed both the indirect effect (c) and direct effect (c’), where c measures the extent to which X influences Y through M, while c’ represents the effect of X on Y that cannot be explained by c. If c is nonzero and c’ is zero, it indicates full mediation; otherwise, it suggests partial mediation. The significance of mediation was assessed using both Sobel test and Preacher–Hayes bootstrapping. The Sobel test was employed to determine if introducing M significantly reduced the relationship between X and Y. A two-tailed z-test was used to establish statistical significance level. On the other hand, bootstrapping method involved a nonparametric examination of whether zero fell within the bootstrapped 95% confidence intervals (5,000 times). If zero did not fall within the confidence intervals, we could conclude that there existed a significant indirect effect. It showed be noted that causal direction in the mediation analysis was not directly examined. Further research employing experimental and longitudinal approaches is warranted to validate our current findings from the mediation analysis. Intergroup discrimination. Participants completed the IDG and IPD-MD before the experiment (as baseline) and at the end of the experiment. We were interested in the intergroup discrimination, i.e., differences in monetary allocation to in-group members and out-group members. Moreover, in IPD-MD, we were also interested in the amount of MU allocated to the within-group pool (i.e., “ingroup love”) and the between-group pool (i.e., “outgroup hate”), respectively. We first compared the baseline intergroup discrimination measured before the experiment between bonding and control sessions and between leaders and followers. We found comparable baseline intergroup discrimination (main effect of bonding: F1, 174 = 0.541, p = 0.463, η2 = 0.003; main effect of hierarchy: F1, 174 = 0.017, p = 0.897, η2 = 9.712 × 10−5; bonding × hierarchy interaction: F1, 174 = 0.007, p = 0.935, η2 = 3.855 × 10−5), ingroup love (main effect of bonding: F1, 174 = 0.149, p = 0.700, η2 = 0.001; main effect of hierarchy: F1, 174 = 0.257, p = 0.613, η2 = 0.001; bonding × hierarchy interaction: F1, 174 = 1.484, p = 0.225, η2 = 0.008), and outgroup hate (main effect of bonding: F1, 174 = 1.140, p = 0.287, η2 = 0.007; main effect of hierarchy: F1, 174 = 0.447, p = 0.505, η2 = 0.003; bonding × hierarchy interaction: F1, 174 = 0.320, p = 0.572, η2 = 0.002) among the 4 conditions. The intergroup discrimination in the IDG, the ingroup love and outgroup hate in the IPD-MDG, measured at the end of the experiment, were submitted to bonding × hierarchy (leader versus follower) ANOVAs to examine bonding effects on intergroup discrimination in leader and followers. Independent ratings on within-group interaction. To objectively quantify the within-group interactions perceived by third-party observers, independent raters were asked to identify the leader of each group and leader emergence time and to evaluate the chatting messages of all the 176 groups along the following dimensions on a 9-point Likert scale: (i) interaction frequency (1 = least frequent, 9 = most frequent); (ii) interaction intensity (1 = least intense, 9 = most intense); (iii) group cohesion (1 = least cohesive, 9 = most cohesive); and (iv) prominence of the group leader (1 = least prominent leader, 9 = most prominent leader). The independent raters correctly identified the group leader (accuracy: 97.66%). The ratings were reliable among the 14 raters (interaction frequency: ICC consistency: 0.940, agreement: 0.885; interaction intensity: ICC consistency: 0.814, agreement: 0.710; group cohesion: ICC consistency: 0.855, agreement: 0.757; leader prominence: ICC consistency: 0.813, agreement: 0.746). For each rating dimension, rating scores were first z-score transformed for each rater. The normalized ratings were then averaged across raters and compared between bonding and control conditions using two-tailed independent-sample t tests. The rating data of one group was missing. We showed that social bonding indeed facilitated the outsider-perceived group interaction frequency (t173 = 4.426, p = 1.696 × 10−5, Cohen’s d = 0.669, 95% CI: 0.264, 0.688, S3A Fig) and intensity (t173 = 3.943, p = 1.166 × 10−4, Cohen’s d = 0.596, 95% CI: 0.157, 0.472, S3B Fig). Moreover, independent raters identified the leader of groups under bonding (versus control) condition faster (t173 = −2.547, p = 0.012, Cohen’s d = −0.385, 95% CI: −40.522, −5.140, S3C Fig) and also recognized the prominence of group leader in the bonding condition to a greater degree than that in the control condition (t173 = 2.001, p = 0.047, Cohen’s d = 0.303, 95% CI: 0.002, 0.339, S3D Fig). Conversation analysis. We analyzed the transcripts of turn-taking conversations for each three-person group. For the within-group communication of each group, we calculated: (i) the total number of utterances of group leader and followers, respectively; (ii) the number of turn-transition; and (iii) the turn response time (i.e., the time gap between turns) separately for inter-status and intra-status transitions. The inter-status turn-transition was defined as a discrete pair of a leader utterance followed by a follower utterance, or vice versa, whereas the intra-status turn-transition was defined as a follower utterance followed by another follower. The turn response time was then log transformed to avoiding skewness. To control for the total length of utterances of all 3 group members, we include the total word number as covariate in the further analysis. These conversation-related indices were submitted to two-way mixed-model ANCOVAs with a within-subjects factor hierarchy (inter-status versus intra-status or leader versus follower), a between-subjects factor bonding (bonding versus control), and total word number as covariate. It should be noted that, the mean turn response time in the current study deviated largely away from the verbal conversation turn response time of about 200 ms, possibly due to that group members communicated via online messaging rather than verbal communication. In addition, to mitigate the potential impact of displaying “typing” in the group chat window on response latency measurement, we have opted not to show status information (i.e., whether someone is typing or not) in the chat window. Only after participants completed and submitted their messages into the group chat window, other members were then able to read others’ messages. Participants’ subjective ratings. Participant’s self-report of inter-status cohesion (i.e., cohesion between leader and follower) and intra-status cohesion (i.e., cohesion between followers) were submitted to a two-way mixed-model ANOVA with a within-subjects factor hierarchy (inter-status versus intra-status) and a between-subjects factor bonding (bonding versus control). Follower’s ratings of perceived social influence and attraction of the group leader was compared between bonding and control conditions with two-tailed independent-sample t tests. Moreover, we performed partial correlation analyses to test the relationship between the perceived social influence (attraction) and the turn transition (response time) separately for inter- and intra-status, controlling the utterance length of leaders and followers, respectively. We next conducted mediation analysis to examine whether the effect of social bonding (the independent variable, X) on the leader’s social influence or attraction (the dependent variable, Y) was mediated by social cohesion (the mediator, M). The mediation analyses were performed using PROCESS (model 4) in SPSS. For each mediation analysis, we computed both the indirect effect (c) and direct effect (c’), where c measures the extent to which X influences Y through M, while c’ represents the effect of X on Y that cannot be explained by c. If c is nonzero and c’ is zero, it indicates full mediation; otherwise, it suggests partial mediation. The significance of mediation was assessed using both Sobel test and Preacher–Hayes bootstrapping. The Sobel test was employed to determine if introducing M significantly reduced the relationship between X and Y. A two-tailed z-test was used to establish statistical significance level. On the other hand, bootstrapping method involved a nonparametric examination of whether zero fell within the bootstrapped 95% confidence intervals (5,000 times). If zero did not fall within the confidence intervals, we could conclude that there existed a significant indirect effect. It showed be noted that causal direction in the mediation analysis was not directly examined. Further research employing experimental and longitudinal approaches is warranted to validate our current findings from the mediation analysis. Intergroup discrimination. Participants completed the IDG and IPD-MD before the experiment (as baseline) and at the end of the experiment. We were interested in the intergroup discrimination, i.e., differences in monetary allocation to in-group members and out-group members. Moreover, in IPD-MD, we were also interested in the amount of MU allocated to the within-group pool (i.e., “ingroup love”) and the between-group pool (i.e., “outgroup hate”), respectively. We first compared the baseline intergroup discrimination measured before the experiment between bonding and control sessions and between leaders and followers. We found comparable baseline intergroup discrimination (main effect of bonding: F1, 174 = 0.541, p = 0.463, η2 = 0.003; main effect of hierarchy: F1, 174 = 0.017, p = 0.897, η2 = 9.712 × 10−5; bonding × hierarchy interaction: F1, 174 = 0.007, p = 0.935, η2 = 3.855 × 10−5), ingroup love (main effect of bonding: F1, 174 = 0.149, p = 0.700, η2 = 0.001; main effect of hierarchy: F1, 174 = 0.257, p = 0.613, η2 = 0.001; bonding × hierarchy interaction: F1, 174 = 1.484, p = 0.225, η2 = 0.008), and outgroup hate (main effect of bonding: F1, 174 = 1.140, p = 0.287, η2 = 0.007; main effect of hierarchy: F1, 174 = 0.447, p = 0.505, η2 = 0.003; bonding × hierarchy interaction: F1, 174 = 0.320, p = 0.572, η2 = 0.002) among the 4 conditions. The intergroup discrimination in the IDG, the ingroup love and outgroup hate in the IPD-MDG, measured at the end of the experiment, were submitted to bonding × hierarchy (leader versus follower) ANOVAs to examine bonding effects on intergroup discrimination in leader and followers. Independent ratings on within-group interaction. To objectively quantify the within-group interactions perceived by third-party observers, independent raters were asked to identify the leader of each group and leader emergence time and to evaluate the chatting messages of all the 176 groups along the following dimensions on a 9-point Likert scale: (i) interaction frequency (1 = least frequent, 9 = most frequent); (ii) interaction intensity (1 = least intense, 9 = most intense); (iii) group cohesion (1 = least cohesive, 9 = most cohesive); and (iv) prominence of the group leader (1 = least prominent leader, 9 = most prominent leader). The independent raters correctly identified the group leader (accuracy: 97.66%). The ratings were reliable among the 14 raters (interaction frequency: ICC consistency: 0.940, agreement: 0.885; interaction intensity: ICC consistency: 0.814, agreement: 0.710; group cohesion: ICC consistency: 0.855, agreement: 0.757; leader prominence: ICC consistency: 0.813, agreement: 0.746). For each rating dimension, rating scores were first z-score transformed for each rater. The normalized ratings were then averaged across raters and compared between bonding and control conditions using two-tailed independent-sample t tests. The rating data of one group was missing. We showed that social bonding indeed facilitated the outsider-perceived group interaction frequency (t173 = 4.426, p = 1.696 × 10−5, Cohen’s d = 0.669, 95% CI: 0.264, 0.688, S3A Fig) and intensity (t173 = 3.943, p = 1.166 × 10−4, Cohen’s d = 0.596, 95% CI: 0.157, 0.472, S3B Fig). Moreover, independent raters identified the leader of groups under bonding (versus control) condition faster (t173 = −2.547, p = 0.012, Cohen’s d = −0.385, 95% CI: −40.522, −5.140, S3C Fig) and also recognized the prominence of group leader in the bonding condition to a greater degree than that in the control condition (t173 = 2.001, p = 0.047, Cohen’s d = 0.303, 95% CI: 0.002, 0.339, S3D Fig). FNIRS data analysis Data quality check. The following steps were performed to check and ensure the data quality of fNIRS signals. First, prior to data recording, the resistance for each optode probe was monitored and adjusted to meet the minimum criteria defined in the LABNIRS recording software, aiming to achieve a good signal-to-noise ratio [94]. Second, during data recording, the raw optical density signal in each channel was transformed into concentration changes in HbO and HbR in real time, which were visually inspected to assess signal quality [95]. Third, during preprocessing of the fNIRS signals, we applied an automatic sliding-window detection to further evaluate data quality. Specifically, for each HbO and HbR time series, extreme values exceeding mean ± 3 standard deviations (SD) within a 10-s time-window were identified as outliers [96]. Channels with severe motion artifacts or containing more than 5% extreme values were labeled as bad. In order to maintain the same number of channels across all groups, any group with one or more bad channels was excluded from further analyses. Based on these criteria, a total of 8 groups were excluded from the final dataset due to poor data quality. Inter-brain neural synchronization (INS). We performed INS analysis on the neural data collected during the resting-state session (4 min, served as a baseline) and within-group interaction session (4 min). Similar to previous studies [15,34–37,44,45], we employed the WTC analysis to assess the cross-correlations between 2 oxy-Hb time series of dyads of participants as a function of frequency and time [97]. Specifically, within each three-person group, the 3 same-length oxy-Hb time series for each channel (i.e., oxy-Hbleader, oxy-Hbfollower1, and oxy-Hbfollower2) were simultaneously acquired by the same fNIRS system. We applied WTC analysis to each pair of 3 oxy-Hb time series and generated 3 time-frequency matrices of the coherence values for each group (i.e., Coherenceleader-follower1, Coherenceleader-follower2, and Coherencefollower1-follower2). The coherence values from the leader-follower dyads (averaged value of Coherenceleader-follower1 and Coherenceleader-follower2) indicated the inter-status INS, and the coherence value from the follower–follower dyads (i.e., Coherencefollower1-follower2) indicated the intra-status INS. In each time-frequency matrix (Fig 3A), each row corresponded to a specific frequency point, each column corresponded to a specific time point and the color bar corresponded to the coherence value. To ensure consistent data size for inter- and intra-status INS, we conducted a set of control analyses by calculating coherence values between the group leader and a randomly selected follower to index inter-status INS. With this approach, the data size and calculation of inter- and intra-status INS were matched. The results obtained from these control analyses replicated our main findings (S10 Fig). fNIRS signals not only reflected task-evoked brain activity but also systemic physiological interference arising from cardiac pulsation (approximately 1 Hz), breathing rate (approximately 0.3 Hz), and other homeostatic processes [95]. Similar to previous studies [4,98], we employed the baseline subtraction approach to mitigate the impact of physiological noise. Specifically, we recorded a resting state session with an identical duration as the task session. As the resting-state predominantly reflects spontaneous hemodynamic oscillations [98], it served as the baseline for comparison. We performed WTC analysis on the neural data collected during both the resting-state session (4 min, serving as a baseline) and within-group interaction session (4 min). To reveal the effects of bonding and hierarchy on INS specific to group interaction, we focused on the increased INS during group interaction relative to the baseline (i.e., the resting-state). First, we compared coherence values (averaged across all channels and for each channel) between the within-group interaction session and resting-state session by performing paired-sample t tests for each frequency (frequency range 0.01 to 1 Hz) [99] to identify the FOI (Fig 3A). This analysis identified increased coherence values in 2 frequency bands: between 0.136 Hz and 0.192 Hz (corresponding to the period between 5.20 and 7.35 s) and between 0.407 and 0.432 Hz (corresponding to the period between 2.31 and 2.46 s). These 2 frequency bands were chosen as the frequency of interest for the subsequent analyses (FOI, Bonferroni family-wise error (FWE) corrected for multiple comparisons). It is worth noting that no significant results were found in the frequency band of 0.407 to 0.432 Hz (full statistics were reported in S9 Table); therefore, only results based on 0.136 to 0.192 Hz were reported in the main text. This chosen period also effectively captures the temporal structure of the within-group interaction task since one-round within-group messaging typically took an average time of 5 to 7 s. In addition, this frequency band also excluded high- and low-frequency physiological noises, such as those related to respiration (about 0.2 to 0.3 Hz), cardiac pulsation (0.7 to 4 Hz), and high-frequency head movements (>1 Hz) [98]. We then calculated the session- and FOI-averaged coherence values and converted into Fisher z-scores. The increased INS (coherence differences between within-group interaction and resting-state) was submitted to bonding × hierarchy (inter-status versus intra-status) ANOVAs. Note that, the WTC algorithm normalized the amplitude of the signal within each time-window defined by the wavelet to make the data less vulnerable to transient spikes or motor artifacts [34–36]. Moreover, we conducted 2 sets of complementary analyses to further control the potential impact of physiological noises. First, we applied a wavelet-based denoising method to identify global physiological components per channel and extracted them out of the hemodynamic signals [15,61]. After the denoising process, the same WTC calculation and statistical analyses were applied, and the observed results were reserved (main effect of hierarchy at channel 3, F1, 174 = 6.296, p = 0.013, η2 = 0.035, hierarchy × bonding interaction effect at channel 9, F1, 174 = 5.311, p = 0.022, η2 = 0.030, S3A Table). Second, we controlled the globally co-varying signal using ANCOVA analyses [62], with the global mean INS (averaged coherence values across all channels) as the covariate. Significant results were fully replicated (main effect of hierarchy at channel 3, F1, 174 = 9.220, p = 0.003, η2 = 0.051, hierarchy × bonding interaction effect at channel 9, F1, 174 = 8.373, p = 0.004, η2 = 0.046, S3B Table). Time-lagged analysis. To investigate the directionality of the inter- and intra-status INS, we conducted time-lag analysis [63–65] for the channels that showed increased INS during the within-group interaction stage, i.e., channel 3 in the rTPJ and channel 9 in the rDLPFC. For each leader–follower (or follower–follower) dyad, the time courses of neural activity of the leader (1 follower) were shifted relative to that of the follower (the other follower) from −10 to 10 s (in 1-s increment). We then recalculated the inter- and intra- status INS on each time lag for both resting-state and within-group interaction. The time-lagged inter- and intra-status neural alignment increases (i.e., lagged INS during within-group interaction minus that during resting) on each time lag were compared with 0 using one-sample t tests and compared between bonding and control conditions using two-tailed independent-sample t tests. Significant effects were thresholded at p < 0.05, FDR corrected for multiple comparisons of the 21 time lags. Next, we performed correlation analysis between the neural alignment on each time lag and intergroup discrimination, for inter-status and intra-status dyads, respectively, the correlation coefficients were FDR corrected for multiple comparisons of the 21 time lags. We also performed Pearson’s correlation coefficient analysis separately for bonding and control conditions. Permutation test. First, we aimed to validate the INS increases (i.e., group interaction versus resting-state) in real groups. To this end, we examined which conditions showed increased INS (i.e., significant INS increases during group interaction compared to resting-state). Specifically, we compared INS differences (group interaction minus resting-state) against zero for the channels that exhibited significant effects of interest (i.e., channel 3 in the rTPJ and channel 9 in the rDLPFC). We found increased INS for the inter-status dyads at channel 3 in the rTPJ under control condition (t1,86 = 3.943, p = 1.64 × 10−4) and bonding condition (t1,88 = 3.394, p = 0.001; survived multiple correction) and at channel 9 in the rDLPFC (bonding: t1,88 = 3.378, p = 0.001; survived multiple correction). We then performed permutation test to examine whether these conditions showed increased INS in real group than pseudo groups. Specifically, within each condition, we generated three-person pseudo-groups by randomly grouping 3 participants from different original real groups. For each pseudo-group, the INS of each dyad was recalculated. This procedure was repeated for 1,000 times to generate a pseudo-group INS distribution. The increased INS in the aforementioned conditions of real groups were compared with each condition-specific permutation distributions. The significance level, p-value was indicated as: p = j/1,000, where j is the number of samples out of the 1,000 permutation samples, of which the examined value was larger than the observed value of real groups. The results indicated that the increases in inter-status INS in the rTPJ (control: p = 0.022; bonding: p = 0.047) and rDLPFC (bonding: p = 0.032) of real groups all exceeded the upper 95% CI of the permutation distribution. These findings further confirmed an increased INS in these specific conditions within real (rather than random) groups. Next, to validate the observed bonding and/or hierarchy effects on INS, we performed another 2 sets of permutation tests: (i) the within-condition permutation test; and (ii) cross-condition permutation test. First, within the bonding and control conditions, leader and followers of the real groups were randomly reassigned into new groups to form three-person pseudo groups. We then recalculated the inter- and intra-status INS for the 176 pseudo groups. This shuffling and recalculation procedures were repeated 1,000 times to generate permutation distributions for the observed hierarchy effect in rTPJ and bonding × hierarchy interaction effect in the rDLPFC. We then compared the observed effects of real interacting groups against 1,000 permutation samples and examined whether real effect exceeded the upper limits 95% or 99% CI of the permutation distribution. Second, similar procedures and statistical analyses were conducted for the cross-condition permutation test except that we generated cross-condition, three-person pseudo groups by randomly grouping 1 leader and 2 followers across bonding and control conditions as a pseudo-group. rDLPFC-rTPJ functional connectivity. We applied cross-correlation analysis using the Functional Connectivity Toolbox [100] implemented in MATLAB to assess the functional connectivity of each rDLPFC-rTPJ channel pair (49 channel pairs, 7 channels in TPJ, and 7 channels in DLPFC) for each participant, which was then Fisher z transformed [101]. We also averaged the 49 channel pairs to index the grand mean of functional connectivity. The channel-pairwise connectivity and grand mean connectivity were separately submitted to bonding × hierarchy (leader versus follower) ANOVAs (FDR correction for multiple comparisons of 49 channel pairs was applied to channel-pairwise analysis). Next, the channel showing significant bonding effect on leader-to-follower neural alignment (i.e., CH9 in the rDLPFC) was used as the seed channel and the averaged functional connectivity across the 7 CH9-rTPJ channel pairs was used to index the channel-based functional connectivity. We then conducted correlation analysis between the channel-based functional connectivity and the leader-to-follower neural alignment in rDLPFC. Additional analyses and results. To test whether the effects of social in-group bonding on hierarchical interaction were modulated by different interaction stages, we conducted additional analyses. First, we identified the specific time point at which the group leader explicitly emerged in each group. This analysis revealed that the establishment of a group leader occurred approximately halfway through the within-group interaction (with a mean value of 145.60 s, SE = 5.76). Second, based on this identified time point for leader emergence, we divided the within-group interaction session into 2 stages: pre- and post-leader emergence stages. Third, we conducted 3-way ANOVAs to examine whether these interaction stages influenced our main findings regarding hierarchy and/or bonding effects on both behavioral and neural indices. The results showed no significant impact of the interaction-stage on our main findings (S10A Table). Furthermore, despite observing an average occurrence of leader emergence around the middle of the within-group interaction period, we further balanced the duration between pre- and post-leader emergence stages by dividing it equally into 2 parts for subsequent analyses. This mid-split approach replicated our observation of no significant impact of different interaction stages (S10B Table). These results suggested that the effects of social bonding on hierarchical interaction remained consistent across different stages of within-group interaction. Alternatively, it is possible that in the current experimental setting, the group leader implicitly emerged prior to the explicit emergence time point (e.g., during the bonding section). Supporting this possibility, we observed that the group leader was more likely to initiate group interaction at the beginning of within-group interaction and was already more talkative in the pre-emergence stage (total number of utterances: t175 = 4.877, p = 2.404 × 10−6; total length of utterances: t175 = 7.451, p = 4.076 × 10−12). Furthermore, a majority of groups (85%, N = 149) early on nominated the individual who later emerged as their leader. We encourage future studies to directly investigate these possibilities. In addition, it is important to note that this examination of time effect was merely an initial exploration conducted at a coarse time scale. Thus, the neural dynamics at second or millisecond resolution needs to be directly and systematically examined in future studies. To present the measured brain activity from different perspectives, we repeated our analysis on the deoxygenated hemoglobin signals (HbR). Specifically, we calculated the INS and intra-brain rDLPFC-rTPJ functional connectivity on HbR signals. We then performed the bonding × hierarchy ANOVAs on HbR-INS and HbR-FC to examine whether the main findings obtained with HbO signals would be similarly observed with HbR signals. Similar to the pattern of the HbO signals, we observed a significant, although weaker (did not survive multiple corrections), hierarchy effect in rTPJ (channel 3, F1, 174 = 4.736, uncorrected p = 0.031, η2 = 0.026, S11 Table and S11A Fig), with stronger inter-status INS than intra-status dyads. However, no bonding × hierarchy interaction effect was observed on HbR signals (channel 9, F1, 174 = 0.420, p = 0.518, η2 = 0.002). The difference observed on INS based on HbO and HbR signals may be caused by different sensitivities of these 2 types of signals in reflecting task-induced changes in neural signals. Regarding the intra-brain FC index, the leader effect on the rDLPFC-rTPJ functional connectivity (stronger in leader than followers) based on HbO signals was similarly observed in the HbR-FC analysis (48 rDLPFC-rTPJ channel pairs survived FDR correction for 49 channel pairs, S12 Table; S11B Fig for the grand mean rDLPFC-rTPJ connectivity, F1, 174 = 27.345, p = 4.842 × 10−7, η2 = 0.136). Data quality check. The following steps were performed to check and ensure the data quality of fNIRS signals. First, prior to data recording, the resistance for each optode probe was monitored and adjusted to meet the minimum criteria defined in the LABNIRS recording software, aiming to achieve a good signal-to-noise ratio [94]. Second, during data recording, the raw optical density signal in each channel was transformed into concentration changes in HbO and HbR in real time, which were visually inspected to assess signal quality [95]. Third, during preprocessing of the fNIRS signals, we applied an automatic sliding-window detection to further evaluate data quality. Specifically, for each HbO and HbR time series, extreme values exceeding mean ± 3 standard deviations (SD) within a 10-s time-window were identified as outliers [96]. Channels with severe motion artifacts or containing more than 5% extreme values were labeled as bad. In order to maintain the same number of channels across all groups, any group with one or more bad channels was excluded from further analyses. Based on these criteria, a total of 8 groups were excluded from the final dataset due to poor data quality. Inter-brain neural synchronization (INS). We performed INS analysis on the neural data collected during the resting-state session (4 min, served as a baseline) and within-group interaction session (4 min). Similar to previous studies [15,34–37,44,45], we employed the WTC analysis to assess the cross-correlations between 2 oxy-Hb time series of dyads of participants as a function of frequency and time [97]. Specifically, within each three-person group, the 3 same-length oxy-Hb time series for each channel (i.e., oxy-Hbleader, oxy-Hbfollower1, and oxy-Hbfollower2) were simultaneously acquired by the same fNIRS system. We applied WTC analysis to each pair of 3 oxy-Hb time series and generated 3 time-frequency matrices of the coherence values for each group (i.e., Coherenceleader-follower1, Coherenceleader-follower2, and Coherencefollower1-follower2). The coherence values from the leader-follower dyads (averaged value of Coherenceleader-follower1 and Coherenceleader-follower2) indicated the inter-status INS, and the coherence value from the follower–follower dyads (i.e., Coherencefollower1-follower2) indicated the intra-status INS. In each time-frequency matrix (Fig 3A), each row corresponded to a specific frequency point, each column corresponded to a specific time point and the color bar corresponded to the coherence value. To ensure consistent data size for inter- and intra-status INS, we conducted a set of control analyses by calculating coherence values between the group leader and a randomly selected follower to index inter-status INS. With this approach, the data size and calculation of inter- and intra-status INS were matched. The results obtained from these control analyses replicated our main findings (S10 Fig). fNIRS signals not only reflected task-evoked brain activity but also systemic physiological interference arising from cardiac pulsation (approximately 1 Hz), breathing rate (approximately 0.3 Hz), and other homeostatic processes [95]. Similar to previous studies [4,98], we employed the baseline subtraction approach to mitigate the impact of physiological noise. Specifically, we recorded a resting state session with an identical duration as the task session. As the resting-state predominantly reflects spontaneous hemodynamic oscillations [98], it served as the baseline for comparison. We performed WTC analysis on the neural data collected during both the resting-state session (4 min, serving as a baseline) and within-group interaction session (4 min). To reveal the effects of bonding and hierarchy on INS specific to group interaction, we focused on the increased INS during group interaction relative to the baseline (i.e., the resting-state). First, we compared coherence values (averaged across all channels and for each channel) between the within-group interaction session and resting-state session by performing paired-sample t tests for each frequency (frequency range 0.01 to 1 Hz) [99] to identify the FOI (Fig 3A). This analysis identified increased coherence values in 2 frequency bands: between 0.136 Hz and 0.192 Hz (corresponding to the period between 5.20 and 7.35 s) and between 0.407 and 0.432 Hz (corresponding to the period between 2.31 and 2.46 s). These 2 frequency bands were chosen as the frequency of interest for the subsequent analyses (FOI, Bonferroni family-wise error (FWE) corrected for multiple comparisons). It is worth noting that no significant results were found in the frequency band of 0.407 to 0.432 Hz (full statistics were reported in S9 Table); therefore, only results based on 0.136 to 0.192 Hz were reported in the main text. This chosen period also effectively captures the temporal structure of the within-group interaction task since one-round within-group messaging typically took an average time of 5 to 7 s. In addition, this frequency band also excluded high- and low-frequency physiological noises, such as those related to respiration (about 0.2 to 0.3 Hz), cardiac pulsation (0.7 to 4 Hz), and high-frequency head movements (>1 Hz) [98]. We then calculated the session- and FOI-averaged coherence values and converted into Fisher z-scores. The increased INS (coherence differences between within-group interaction and resting-state) was submitted to bonding × hierarchy (inter-status versus intra-status) ANOVAs. Note that, the WTC algorithm normalized the amplitude of the signal within each time-window defined by the wavelet to make the data less vulnerable to transient spikes or motor artifacts [34–36]. Moreover, we conducted 2 sets of complementary analyses to further control the potential impact of physiological noises. First, we applied a wavelet-based denoising method to identify global physiological components per channel and extracted them out of the hemodynamic signals [15,61]. After the denoising process, the same WTC calculation and statistical analyses were applied, and the observed results were reserved (main effect of hierarchy at channel 3, F1, 174 = 6.296, p = 0.013, η2 = 0.035, hierarchy × bonding interaction effect at channel 9, F1, 174 = 5.311, p = 0.022, η2 = 0.030, S3A Table). Second, we controlled the globally co-varying signal using ANCOVA analyses [62], with the global mean INS (averaged coherence values across all channels) as the covariate. Significant results were fully replicated (main effect of hierarchy at channel 3, F1, 174 = 9.220, p = 0.003, η2 = 0.051, hierarchy × bonding interaction effect at channel 9, F1, 174 = 8.373, p = 0.004, η2 = 0.046, S3B Table). Time-lagged analysis. To investigate the directionality of the inter- and intra-status INS, we conducted time-lag analysis [63–65] for the channels that showed increased INS during the within-group interaction stage, i.e., channel 3 in the rTPJ and channel 9 in the rDLPFC. For each leader–follower (or follower–follower) dyad, the time courses of neural activity of the leader (1 follower) were shifted relative to that of the follower (the other follower) from −10 to 10 s (in 1-s increment). We then recalculated the inter- and intra- status INS on each time lag for both resting-state and within-group interaction. The time-lagged inter- and intra-status neural alignment increases (i.e., lagged INS during within-group interaction minus that during resting) on each time lag were compared with 0 using one-sample t tests and compared between bonding and control conditions using two-tailed independent-sample t tests. Significant effects were thresholded at p < 0.05, FDR corrected for multiple comparisons of the 21 time lags. Next, we performed correlation analysis between the neural alignment on each time lag and intergroup discrimination, for inter-status and intra-status dyads, respectively, the correlation coefficients were FDR corrected for multiple comparisons of the 21 time lags. We also performed Pearson’s correlation coefficient analysis separately for bonding and control conditions. Permutation test. First, we aimed to validate the INS increases (i.e., group interaction versus resting-state) in real groups. To this end, we examined which conditions showed increased INS (i.e., significant INS increases during group interaction compared to resting-state). Specifically, we compared INS differences (group interaction minus resting-state) against zero for the channels that exhibited significant effects of interest (i.e., channel 3 in the rTPJ and channel 9 in the rDLPFC). We found increased INS for the inter-status dyads at channel 3 in the rTPJ under control condition (t1,86 = 3.943, p = 1.64 × 10−4) and bonding condition (t1,88 = 3.394, p = 0.001; survived multiple correction) and at channel 9 in the rDLPFC (bonding: t1,88 = 3.378, p = 0.001; survived multiple correction). We then performed permutation test to examine whether these conditions showed increased INS in real group than pseudo groups. Specifically, within each condition, we generated three-person pseudo-groups by randomly grouping 3 participants from different original real groups. For each pseudo-group, the INS of each dyad was recalculated. This procedure was repeated for 1,000 times to generate a pseudo-group INS distribution. The increased INS in the aforementioned conditions of real groups were compared with each condition-specific permutation distributions. The significance level, p-value was indicated as: p = j/1,000, where j is the number of samples out of the 1,000 permutation samples, of which the examined value was larger than the observed value of real groups. The results indicated that the increases in inter-status INS in the rTPJ (control: p = 0.022; bonding: p = 0.047) and rDLPFC (bonding: p = 0.032) of real groups all exceeded the upper 95% CI of the permutation distribution. These findings further confirmed an increased INS in these specific conditions within real (rather than random) groups. Next, to validate the observed bonding and/or hierarchy effects on INS, we performed another 2 sets of permutation tests: (i) the within-condition permutation test; and (ii) cross-condition permutation test. First, within the bonding and control conditions, leader and followers of the real groups were randomly reassigned into new groups to form three-person pseudo groups. We then recalculated the inter- and intra-status INS for the 176 pseudo groups. This shuffling and recalculation procedures were repeated 1,000 times to generate permutation distributions for the observed hierarchy effect in rTPJ and bonding × hierarchy interaction effect in the rDLPFC. We then compared the observed effects of real interacting groups against 1,000 permutation samples and examined whether real effect exceeded the upper limits 95% or 99% CI of the permutation distribution. Second, similar procedures and statistical analyses were conducted for the cross-condition permutation test except that we generated cross-condition, three-person pseudo groups by randomly grouping 1 leader and 2 followers across bonding and control conditions as a pseudo-group. rDLPFC-rTPJ functional connectivity. We applied cross-correlation analysis using the Functional Connectivity Toolbox [100] implemented in MATLAB to assess the functional connectivity of each rDLPFC-rTPJ channel pair (49 channel pairs, 7 channels in TPJ, and 7 channels in DLPFC) for each participant, which was then Fisher z transformed [101]. We also averaged the 49 channel pairs to index the grand mean of functional connectivity. The channel-pairwise connectivity and grand mean connectivity were separately submitted to bonding × hierarchy (leader versus follower) ANOVAs (FDR correction for multiple comparisons of 49 channel pairs was applied to channel-pairwise analysis). Next, the channel showing significant bonding effect on leader-to-follower neural alignment (i.e., CH9 in the rDLPFC) was used as the seed channel and the averaged functional connectivity across the 7 CH9-rTPJ channel pairs was used to index the channel-based functional connectivity. We then conducted correlation analysis between the channel-based functional connectivity and the leader-to-follower neural alignment in rDLPFC. Additional analyses and results. To test whether the effects of social in-group bonding on hierarchical interaction were modulated by different interaction stages, we conducted additional analyses. First, we identified the specific time point at which the group leader explicitly emerged in each group. This analysis revealed that the establishment of a group leader occurred approximately halfway through the within-group interaction (with a mean value of 145.60 s, SE = 5.76). Second, based on this identified time point for leader emergence, we divided the within-group interaction session into 2 stages: pre- and post-leader emergence stages. Third, we conducted 3-way ANOVAs to examine whether these interaction stages influenced our main findings regarding hierarchy and/or bonding effects on both behavioral and neural indices. The results showed no significant impact of the interaction-stage on our main findings (S10A Table). Furthermore, despite observing an average occurrence of leader emergence around the middle of the within-group interaction period, we further balanced the duration between pre- and post-leader emergence stages by dividing it equally into 2 parts for subsequent analyses. This mid-split approach replicated our observation of no significant impact of different interaction stages (S10B Table). These results suggested that the effects of social bonding on hierarchical interaction remained consistent across different stages of within-group interaction. Alternatively, it is possible that in the current experimental setting, the group leader implicitly emerged prior to the explicit emergence time point (e.g., during the bonding section). Supporting this possibility, we observed that the group leader was more likely to initiate group interaction at the beginning of within-group interaction and was already more talkative in the pre-emergence stage (total number of utterances: t175 = 4.877, p = 2.404 × 10−6; total length of utterances: t175 = 7.451, p = 4.076 × 10−12). Furthermore, a majority of groups (85%, N = 149) early on nominated the individual who later emerged as their leader. We encourage future studies to directly investigate these possibilities. In addition, it is important to note that this examination of time effect was merely an initial exploration conducted at a coarse time scale. Thus, the neural dynamics at second or millisecond resolution needs to be directly and systematically examined in future studies. To present the measured brain activity from different perspectives, we repeated our analysis on the deoxygenated hemoglobin signals (HbR). Specifically, we calculated the INS and intra-brain rDLPFC-rTPJ functional connectivity on HbR signals. We then performed the bonding × hierarchy ANOVAs on HbR-INS and HbR-FC to examine whether the main findings obtained with HbO signals would be similarly observed with HbR signals. Similar to the pattern of the HbO signals, we observed a significant, although weaker (did not survive multiple corrections), hierarchy effect in rTPJ (channel 3, F1, 174 = 4.736, uncorrected p = 0.031, η2 = 0.026, S11 Table and S11A Fig), with stronger inter-status INS than intra-status dyads. However, no bonding × hierarchy interaction effect was observed on HbR signals (channel 9, F1, 174 = 0.420, p = 0.518, η2 = 0.002). The difference observed on INS based on HbO and HbR signals may be caused by different sensitivities of these 2 types of signals in reflecting task-induced changes in neural signals. Regarding the intra-brain FC index, the leader effect on the rDLPFC-rTPJ functional connectivity (stronger in leader than followers) based on HbO signals was similarly observed in the HbR-FC analysis (48 rDLPFC-rTPJ channel pairs survived FDR correction for 49 channel pairs, S12 Table; S11B Fig for the grand mean rDLPFC-rTPJ connectivity, F1, 174 = 27.345, p = 4.842 × 10−7, η2 = 0.136). Statistical analysis Similar to previous studies [4,15,102], data were aggregated at the three-person group level and hierarchy within each group (i.e., leader versus follower, or inter-status versus intra-status) were treated as a within-subjects factor. For both the behavioral and neural data, we averaged the 2 followers or the 2 leader–follower dyads to index the follower or inter-status level. The experimental condition (bonding versus control) was randomly introduced and blinded to the participants during data collection. For each dependent variable, the three-person groups whose value was larger or smaller than 5 SDs from the mean value were excluded. This data cleaning procedure led to exclusion of data in the following variables: intra-status turn transition (n = 1), intra-status turn response time (n = 1), and inter-status turn response time (n = 2). Two-way mixed-model ANOVAs were conducted on final behavioral and neural datasets with bonding (bonding versus no-bonding control) as a between-subjects factor and hierarchy (inter-status versus intra-status, or leader versus follower) as a within-subjects factor. Furthermore, the LMM is another optimal method for analyzing such structural data. Therefore, we performed a series of LMM analyses on behavioral and neural indices, considering hierarchy and bonding as fixed effects while treating each group as a random effect. This set of analyses yielded similar results to ANOVA. ANOVA with significant interaction were followed by planned two-tailed t tests to examine: (i) bonding effects (two-tailed independent-sample t test) separately on inter-status and intra-status dyads or leaders and followers; and (ii) hierarchy effects (two-tailed paired-sample t test) separately in bonding and control condition. Statistical significance was thresholded at p < 0.05. Data distributions were assumed to be normal, but this was not formally tested. For ratings from independent sample, data were first normalized across items for each rater and then averaged across all raters. Correlation analyses were conducted using Pearson’s correlation coefficient analysis. It should be noted that the reported behavior-neural correlations, although statistically significant, should be interpreted and applied with caution due to their small to medium effect sizes [103]. All statistical analyses were performed with SPSS (IBM SPSS Statistics 25) and custom scripts in MATLAB (R2017b & R2020b, The MathWorks, United States of America). The wavelet coherence analysis was performed by Wavelet Coherence Package [104] implemented in MATLAB (which is available in https://noc.ac.uk/business/marine-data-products/cross-wavelet-wavelet-coherence-toolbox-matlab). Supporting information S1 Data. Numerical data underlying graphs. Names of individual sheets correspond to figure panels for which the numerical data is used: Figs 1D–1H, 2A–2E, 3B–3H, 4B–4E and 5A–5C. https://doi.org/10.1371/journal.pbio.3002545.s001 (XLSX) S1 Fig. Experimental procedure. Before coming to the laboratory, participants completed an online survey that included demographic and psychological information as well as color preference (white versus black, for bonding manipulation). One to 4 days later, participants were invited to the laboratory in groups of 3 same-gender strangers and randomly assigned into either the bonding or control condition. They were instructed to sit face-to-face in a triangle and completed 3 sessions during fNIRS-based hyper-scanning: (i) a 4-min resting-state session; (ii) a 4-min in-group social bonding (or no-bonding control) manipulation session; and (iii) a 4-min online within-group interaction session. At the end of the experiment, participants were asked to complete a series of intergroup-related economic games (including intergroup dictator game and intergroup prisoner’s dilemma-maximizing differences game), as well as report subjective evaluations on group cohesion, leader influence and attraction, positive attitudes, willingness to become the leader, etc. (details in Methods). https://doi.org/10.1371/journal.pbio.3002545.s002 (TIF) S2 Fig. In-group social bonding facilitates the within-group communication. Social bonding particularly increased the utterance numbers given by leaders (control: 8.590 ± 4.088, bonding: 12.750 ± 5.911) than followers (control: 7.322 ± 3.655, bonding: 9.899 ± 4.311). Data are plotted as box plots for each condition, with horizontal lines indicating median values, boxes indicating 25% and 75% quartiles and whiskers indicating the 2.5%–97.5% percentile range. Cross symbols in each box represent the mean values. Data points outside the range are shown separately as circles. *p < 0.05, ***p < 0.001. https://doi.org/10.1371/journal.pbio.3002545.s003 (TIF) S3 Fig. Bonding effect on within-group communication was perceived by third-party observers. Social bonding increased the perceived group interaction frequency (control: −0.237 ± 0.787, bonding: 0.240 ± 0.629, A) and intensity (control: −0.158 ± 0.542, bonding: 0.156 ± 0.512, B). Third-party observers identified the group leader faster (control: 189.988 ± 49.554, bonding: 167.157 ± 67.349 C), and perceived the leader as more prominent (control: −0.0095 ± 0.582, bonding: 0.076 ± 0.547, D) in the bonding condition. Data are plotted as box plots for each condition, with horizontal lines indicating median values, boxes indicating 25% and 75% quartiles and whiskers indicating the 2.5%–97.5% percentile range. Cross symbols in each box represent the mean values. Data points outside the range are shown separately as circles. *p < 0.05, ***p < 0.001. https://doi.org/10.1371/journal.pbio.3002545.s004 (TIF) S4 Fig. The effect of in-group social bonding on leader behavior and the perception of leader. (A) Under social bonding, followers perceived greater social attraction of the leader (control: 6.333 ± 2.036, bonding: 7.135 ± 1.548). (B) Leader’s social attraction was positively associated with inter-status cohesion (Pearson’s correlation analysis). Each solid line represents the least squares fit, with shading showing the 95% CI. (C/D) Leader’s social influence (C) and attraction (D) were positively associated with intra-status cohesion. (E) Bonding increased perceived social attraction of the leader through enhancing inter-status cohesion. *p < 0.05, **p < 0.01, ***p < 0.001. https://doi.org/10.1371/journal.pbio.3002545.s005 (TIF) S5 Fig. Validation of INS results by nonparametric permutation tests. (A/B) We generated within-condition pseudo-groups by randomly grouping a real leader and 2 real followers from different original groups in the same bonding or control condition to 1 pseudo-group (A), or generate across-condition pseudo-groups by randomly grouping 1 leader and 2 followers across bonding and control conditions as one pseudo-group (B). The inter- and intra-status INS for each pseudo group were recalculated. These procedures were repeated for 1,000 times to generate permutation distributions. (C/D) We compared the hierarchy main effect in the rTPJ and the interaction effect in the rDLPFC of real group against cross-condition permutation distributions (n = 1,000). The observed effects of Hierarchy in the rTPJ (C) and of Hierarchy × Bonding interaction in the rDLPFC (D) exceeded the upper limits of 99% CI of the permutation distributions. https://doi.org/10.1371/journal.pbio.3002545.s006 (TIF) S6 Fig. Inter-status neural alignment in the rTPJ was significant in both directions. (A) Inter-status neural alignment in rTPJ is significant from −10 to +10 time lags (peaked at 0 s), survived FDR multiple correction. The significant time lags (survived multiple correction) are highlighted with the horizontal line on the x-axis. (B) The inter-status neural alignment in rTPJ is significant at all time lags in bonding and control conditions separately. Shaded areas represent standard error (SE). https://doi.org/10.1371/journal.pbio.3002545.s007 (TIF) S7 Fig. Leader-to-follower neural alignment in rDLPFC was positively associated with intergroup discrimination. (A–F) Correlation analyses between leader-to-follower neural alignment at each time lag (+1 to +6) with intergroup discrimination. Correlations were performed by Pearson’s correlation coefficient analysis. Each solid line represents the least squares fit, with shading showing the 95% CI. † p < 0.06, *p < 0.05, **p < 0.01. https://doi.org/10.1371/journal.pbio.3002545.s008 (TIF) S8 Fig. No significant bonding effect or correlation on intra-status neural alignment. (A) In-group social bonding showed no significant effect on intra-status neural alignment in rDLFPC at any time lags. (B) Intra-status neural alignment in rDLFPC did not correlate with intergroup discrimination at any time lags. https://doi.org/10.1371/journal.pbio.3002545.s009 (TIF) S9 Fig. Stronger rDLPFC-rTPJ functional connectivity in leaders accounted for leader-to-follower neural alignment. (A–F) Correlation analyses between leaders’ rDLPFC-rTPJ connectivity with leader-to-follower neural alignment at each time lag (+1 to +6). Correlations were performed by Pearson’s correlation coefficient analysis. Each solid line represents the least squares fit, with shading showing the 95% CI. † p < 0.07, *p < 0.05. https://doi.org/10.1371/journal.pbio.3002545.s010 (TIF) S10 Fig. Neural synchronization between the group leader and a randomly selected follower. The significant hierarchy main effect in the rTPJ (A) and bonding × hierarchy interaction effect in the rDLPFC (B) were fully replicated when considering inter-status INS for the leader and a randomly selected follower (channel 3, F1, 174 = 8.330, p = 0.004, η2 = 0.046, A; channel 9, F1, 174 = 11.133, p = 0.001, η2 = 0.060, B). Data are plotted as box plots for each condition, with horizontal lines indicating median values, boxes indicating 25% and 75% quartiles and whiskers indicating the 2.5%–97.5% percentile range. Cross symbols in each box represent the mean values. Data points outside the range are shown separately as circles. **p < 0.01. https://doi.org/10.1371/journal.pbio.3002545.s011 (TIF) S11 Fig. Inter-brain neural synchronization and intra-brain functional connectivity in HbR signals. (A) HbR-INS in rTPJ showed a significant, but weaker (did not survive multiple corrections), hierarchy main effect (channel 3, F1, 174 = 4.736, uncorrected p = 0.031, η2 = 0.026), with stronger inter-status INS than intra-status one. (B) The grand mean of rDLPFC-rTPJ connectivity in HbR exhibited a significant hierarchy main effect (F1, 174 = 27.345, p = 4.842 × 10−7, η2 = 0.136), with stronger rDLPFC-rTPJ functional connectivity in leaders (vs. followers). Data are plotted as box plots for each condition, with horizontal lines indicating median values, boxes indicating 25% and 75% quartiles and whiskers indicating the 2.5%–97.5% percentile range. Cross symbols in each box represent the mean values. Data points outside the range are shown separately as circles. ***p < 0.001. https://doi.org/10.1371/journal.pbio.3002545.s012 (TIF) S1 Table. Demographic and psychological information of participants. https://doi.org/10.1371/journal.pbio.3002545.s013 (DOCX) S2 Table. Full statistical reports of the results of Hierarchy × Bonding mixed-model ANOVAs on inter-brain neural synchronization. https://doi.org/10.1371/journal.pbio.3002545.s014 (DOCX) S3 Table. Statistical reports of inter-brain neural synchronization in 2 complementary analyses. (A) ANOVA analysis after the wavelet-based denoising. (B) ANCOVA analysis controlling global mean INS. https://doi.org/10.1371/journal.pbio.3002545.s015 (DOCX) S4 Table. Full statistical reports of the results of bonding effect (Bonding vs. Control) on inter-status neural alignment (CH9) for each time lag. https://doi.org/10.1371/journal.pbio.3002545.s016 (DOCX) S5 Table. Full statistical reports of the results of inter-status INS increase (CH9, one-sample t tests) of each time lag under bonding and control conditions, respectively. https://doi.org/10.1371/journal.pbio.3002545.s017 (DOCX) S6 Table. Full statistical reports of the results of the Pearson’s correlations between inter-status neural alignment (CH9) and intergroup discrimination on each time lag. https://doi.org/10.1371/journal.pbio.3002545.s018 (DOCX) S7 Table. Full statistical reports of hierarchy main effect on rDLPFC-rTPJ functional connectivity for each channel pair. https://doi.org/10.1371/journal.pbio.3002545.s019 (DOCX) S8 Table. The anatomical position for each recording channel. https://doi.org/10.1371/journal.pbio.3002545.s020 (DOCX) S9 Table. Full statistical reports of the results of Hierarchy × Bonding mixed-model ANOVAs on inter-brain neural synchronization in frequency band 0.407–0.432 Hz. https://doi.org/10.1371/journal.pbio.3002545.s021 (DOCX) S10 Table. Statistical reports of interaction-stage/time-bin modulation effects on main findings. (A) Analysis on pre- and post-leader emergence stages (based on leader emergence time). (B) Analysis on early and late time-bin (equally split by mid-point). https://doi.org/10.1371/journal.pbio.3002545.s022 (DOCX) S11 Table. Full statistical reports of the results of Hierarchy × Bonding mixed-model ANOVAs on HbR-INS. https://doi.org/10.1371/journal.pbio.3002545.s023 (DOCX) S12 Table. Full statistical reports of hierarchy main effect on rDLPFC-rTPJ HbR-FC for each channel pair. https://doi.org/10.1371/journal.pbio.3002545.s024 (DOCX) Acknowledgments We thank H. Zhang, C. Yang, and X. Zou for their assistance in data collection.

Lucas, Robert J.;Allen, Annette E.;Brainard, George C.;Brown, Timothy M.;Dauchy, Robert T.;Didikoglu, Altug;Do, Michael Tri H.;Gaskill, Brianna N.;Hattar, Samer;Hawkins, Penny;Hut, Roelof A.;McDowell, Richard J.;Nelson, Randy J.;Prins, Jan-Bas;Schmidt, Tiffany M.;Takahashi, Joseph S.;Verma, Vandana;Voikar, Vootele;Wells, Sara;

Wu, Zhimao;Chen, Huicheng;Zhang, Yingying;Wang, Yaru;Wang, Qiaoling;Augière, Céline;Hou, Yanan;Fu, Yuejun;Peng, Ying;Durand, Bénédicte;Wei, Qing

doi: 10.1371/journal.pbio.3002330pmid: 38442096

Introduction Cilia are microtubule-based organelles that extend from the surface of many cell types and are widely present in eukaryotes. They play crucial roles in the development and maintenance of various organs in humans [1–4], and their dysfunction has been linked to a wide range of human genetics diseases known as ciliopathies [5–7]. The structure of cilia remains highly conserved across evolution [8]. A cilium comprises a basal body (BB), a transition zone (TZ), an axoneme and its overlying membrane. The BB originates from the mother centriole and attaches to the membrane through transition fibers that come from the distal appendages of the mother centriole [9–11]. The TZ is situated just above transition fibers [12] and is characterized by Y-linker structures that connect the axoneme and ciliary membrane, serving as a diffusion barrier to control ciliary protein entry [9]. The axoneme forms the cilium skeleton and consists of a 9-fold array of doublet microtubules, which are templated from the BB microtubules and surrounded by the ciliary membrane [5,8]. The formation of cilia involves 2 main processes: ciliogenesis initiation and axoneme elongation. The elongation of axoneme relies on the intraflagellar transport (IFT), a evolutionarily conserved transport machinery within cilia [13,14]. The initiation of ciliogenesis involves the membrane docking of BB and the formation of ciliary bud [10,15,16]. In mammals, the membrane docking of BB is mediated by centriole distal appendages/ciliary transition fibers [17–21]. However in invertebrate model organism Drosophila, transition fiber proteins are dispensable for BB membrane docking [22], suggesting the presence of alternative mechanisms. The ciliary bud consists of the TZ and its surrounding membrane [10,23,24]. Dozens of proteins have been identified as components of TZ, and mutations in most of them lead to ciliopathy [5]. Studies have categorized TZ proteins into 3 functional modules: Meckel–Gruber syndrome (MKS) module, Nephronophthisis (NPHP) module, and Cep290 [25]. The formation of ciliary membrane relies on the membrane transport regulator small GTPase Rab8 related signaling cascade [26–28]. Dzip1 (DAZ interacting zinc finger protein 1) and its close paralog Dzip1L have recently been found to play a role in ciliary bud formation in both Drosophila and mammals [24,29,30]. Downstream of Dzip1/1L, Rab8, and Cby (Chibby)-Fam92 (family with sequence similarity 92) module work together to regulate ciliary membrane formation [24,29,31]. Although speculations have accumulated that ciliary bud formation must be a coordinated event matching the TZ assembly and the ciliary membrane formation [27,29], a detailed molecular linkage between these 2 seemingly independent processes remains largely elusive. Using Drosophila model, we provide evidence that the core TZ protein Cep290 play a pivotal role in coordinating ciliary membrane formation and TZ assembly [29]. We demonstrated that, in addition to its classical role in TZ assembly, Cep290 acts upstream of Dzip1, playing an essential role in ciliogenesis initiation and ciliary bud formation [29]. Cep290 is an intriguing cilia gene associated with several ciliopathies [32], including Leber congenital amaurosis (LCA), Senior–Loken syndrome (SLS), Joubert syndrome (JBTS), MKS, and Bardet–Biedl syndrome (BBS). More than 100 ciliopathy mutations have been identified in Cep290 [32]. The broad spectrum of diseases highlights the critical roles of Cep290 in cilia. We and others have demonstrated that the N-terminus of Cep290 associates with the ciliary membrane, while its C-terminus connects with the ciliary axoneme, both being critical for ciliogenesis [29,33,34]. Nonetheless, how Cep290 is targeted to the TZ and whether axoneme derived signal and ciliary membrane localized cues converge to determine the localization and stability of Cep290 remain unknown. Centrosome protein 131 (Cep131) localizes to both centrosome centriolar satellites and ciliary TZ in mammals [35–37] and has been demonstrated to be required for ciliogenesis, centriole amplification, genome stability, and cancer [35,36,38–40]. Studies on the model organisms Drosophila and zebrafish revealed that Cep131 is an evolutionarily conserved BB protein [41,42], and that deletion of Cep131 results in abnormal cilia formation, suggesting that Cep131 has an evolutionarily conserved role in ciliogenesis. But the precise mechanism by which Cep131 regulates ciliogenesis remains largely unknown. In Drosophila, Cep131 (also called dilatory in flies) has been shown to localize to the lumen of the distal BB and TZ, playing a role in the initiation of ciliogenesis [42,43]. Interestingly, although both Cep131 and Cby single mutants display mild defects in cilium assembly, the initiation of ciliogenesis is completely blocked and Cep290 is totally absent from basal bodies in cep131; cby double mutant [43]. However, the molecular function of Cep131 in the initiation of ciliogenesis is still largely unknown, and the underlying mechanism by which Cep131 genetically interacts with Cby to regulate Cep290 localization remains unclear. Here, we report that Cep131 recruits Cep162 to regulate the TZ localization of Cep290 C-terminus and promote ciliogenesis. We show that Cep162 is a Cep131-interacting protein and acts downstream of Cep131 to mediate the association of Cep290 C-terminus with the axonemal microtubules. In addition, we demonstrate that Cby-Fam92 module regulates the TZ localization of the N-terminus of Cep290. Both modules cooperate to recruit and stabilize Cep290 at the TZ, as combined loss of either module (Cep131-Cep162 or Cby-Fam92) results in complete failure of Cep290 localization to the TZ and ultimately prevents the initiation of ciliogenesis. Our results reveal the crucial molecular function of Cep131 in ciliogenesis and unveil a cooperative and orderly assembly of Cep290 facilitated by Cep131-Cep162 and Cby-Fam92 modules during the initiation of ciliogenesis. Thus, our work defines a central molecular pathway composed of 3 modules: Cep131-Cep162, Cep290, and Dzip1-Cby-Fam92, which cooperatively regulate the initiation of cilium assembly. Results Cep131 is required for the basal body localization of Cep290 C-terminus To understand the function of the Cep131 in ciliogenesis, we generated the cep1311 (C terminal deletion) mutant flies using the CRISPR-Cas9 system (S1A–S1C Fig). Similar to reported cep131 mutant [42], our cep1311 flies exhibited typical symptoms related to ciliary defects, and were severely uncoordinated during walking and flying. Consistent with previous reports that cep131 single mutant fly has mild defects in the initiation of ciliogenesis [42], we observed that approximately 33.6% of cep1311 spermatocyte centrioles showed abnormal elongation of the microtubules labeled by CG6652::GFP (S1D Fig), a Drosophila spermatocyte-specific phenotype associated with defects in BB docking and TZ membrane cap formation. Consistent with this observation, the signal of the ciliogenesis initiation regulators Cep290, Dzip1, and Cby, as well as the TZ marker Mks1 were notably reduced at the basal bodies of spermatocytes of cep1311 mutants, although they were still detectable (Fig 1A). Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 1. Cep131 is required for the BB localization of Cep290-C terminus. (A) Localization of various TZ proteins in spermatocyte cilia of WT flies and cep131 mutants and quantifications of corresponding relative fluorescence intensities. In cep131 mutants, the signals of Cep290, Dzip1, Cby::GFP, and Mks1::GFP are significantly reduced compared to WT. Importantly, Cep290-C::GFP signal is almost completely lost in cep131 mutants. The BB is labeled with γ-Tubulin (red). The error bars represent the mean ± SD, n = 30. (B) The localization of various TZ proteins in auditory cilia of WT flies and cep131 mutants. Similar to what we observed in spermatocyte cilia, the signals of Cep290, Dzip1, and Cby::GFP are significantly decreased, and Cep290-C::GFP is completely lost at the base of the sensory cilia. 21A6 (blue) marks the cilia base, Actin (red) marks the ciliated region. The error bars represent the mean ± SD, n = 30. Scale bars: 2 μm (A), 5 μm (B, full-scale images on the left), 1 μm (B, insets or zoomed in areas on the right). The data underlying this figure can be found in S1 Data. BB, basal body; TZ, transition zone; WT, wild type. https://doi.org/10.1371/journal.pbio.3002330.g001 Cep290 is the most upstream protein known in the initiation of ciliogenesis in Drosophila [29,30]. Works in fly and mammalian cells have suggested that Cep290 bridges the ciliary axoneme and the membrane, with its C-terminus associated with microtubule doublets and N-terminus associated with the membrane [29,33,34]. Previously, we have demonstrated that both N-terminal truncation (Cep290-N, aa 1–650) and C-terminal truncation (Cep290-C, aa 1385 to the end at 1978) of Drosophila Cep290 are capable of localizing to the TZ, independently of each other [29]. 3D-SIM microscopy revealed that Cep290-C::GFP localizes close to the axoneme, whereas Cep290-N::GFP localizes close to the membrane, displaying a noticeably larger diameter [29]. Given that Cep131 localizes to the lumen of TZ [43], we hypothesized that Cep131 might have a role in regulating the localization of Cep290 C-terminus. As expected, we found that the signal of Cep290-C::GFP signal was nearly completely lost in spermatocyte cilia of cep131 mutants, whereas the signal intensity of Cep290-N::GFP was similar to that in wild-type (WT) (Fig 1A). Interestingly, we noticed that both the diameter of endogenous Cep290 N-terminus (labeled by anti-Cep290 antibody against aa 292–541 in its N-terminus) and Cep290-N::GFP in cep131 mutants were significantly smaller than that in WT, suggesting a potential alternation in the conformation of Cep290 in cep131 mutants (Figs 1A and S1E). Collectively, our results indicate that Cep131 specifically regulates the localization of Cep290 C-terminus in spermatocyte cilia. To determine whether our observation is specific to spermatocyte cilia, we focused on sensory cilia, another type of cilia in Drosophila. Similar to our observation in spermatocytes, the signal intensities of Cep290, Cep290-N, Dzip1, Cby, and Mks1 were mildly affected, whereas Cep290-C::GFP was almost completely lost from the basal bodies in auditory cilia of cep131 mutants (Fig 1B). Hence, Cep131 also promotes the binding of Cep290 C-terminus to the axoneme in sensory cilia. Cep162 bridges Cep131 and Cep290 Next, we wondered whether Cep131 directly interacts with Cep290. However, no interaction between Cep131 and Cep290 was observed in our yeast two-hybrid (Y2H) assay (S2A Fig), suggesting that additional proteins might mediate the functional interaction between Cep131 and Cep290. In mammalian cells, it has been reported that Cep162 localizes to the centriole distal end and interacts with Cep290 to promote its association with microtubules [44]. Protein homology search using NCBI protein–protein BLAST identified CG42699 as the sole homolog of Cep162 in Drosophila (S3 Fig). Consistent with reports in mammalian cells, Y2H assay showed that CG42669 interacts with Drosophila Cep290 (S2A Fig). Interestingly, our Y2H assay also revealed that CG42699 interacts with Drosophila Cep131 (Fig 2A and 2B). Specifically, the C-terminal half of CG42699 (aa 448 to the end at 897) interacts with Cep131, while the N-terminal half does not. GST pull-down assay further confirmed the direct interaction between Cep131 and Cep162 C-terminus, and showed that Cep162 C-terminus interacts with both the N-terminal half (aa 1–549) and C-terminal half (aa 550 to the end at 1114) of Cep131 (S2B Fig). Subsequent analysis using Y2H showed that Cep162 C-terminus does not interact with the middle region of Cep131 (aa 481–781), but it does interacts with both the N-terminus (aa 1–480) and C-terminus (aa 782 to the end at 1114) (S2C Fig), indicating the presence of 2 binding sites for Cep162 in Cep131. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 2. Cep131 interacts with and recruits Cep162 to centriole tips and the TZ. (A, B) Cep131 directly interacts with Cep162. (A) Schematic representation of full-length (Cep162-FL) or truncated (Cep162-N, Cep162-C) Cep162 proteins used for interaction assays in B. (B) Cep131 interacts with Cep162-FL and Cep162-C, but not Cep162-N in the Y2H assay. The upper panel shows the presence of Y2H plasmids as evidenced by colony growth on SD-Leu-Trp plates. The lower panel shows the positive interaction between Cep131 and Cep162 as evidenced by colony growth on SD-Ade-Leu-Trp-His plates. (C) Immunostaining of Cep162::GFP (green) in spermatocyte cilia in cep131 mutants. Quantification of Cep162 signal intensity at the ciliary base is shown on the right. γ-Tubulin (red) labels the centriole/basal body. The error bars represent the mean ± SD, n = 30. (D) Immunostaining of Cep162::GFP (green), Actin (red), and 21A6 (blue) in auditory cilia or olfactory cilia in WT or cep131 mutants. Quantification of Cep162 signal intensity at the ciliary base is shown on the right. The error bars represent the mean ± SD, n = 30. (E) The subcellular localization of exogenous Cep162::GFP during spermatogenesis. From spermatogonia to late spermatocytes, Cep162 is localized at the tip of centriole/basal body. In round spermatids, as flagella elongates and the ciliary cap moves away from the BBs, Cep162 migrates with the ring centriole labeled by γ-Tubulin (arrowhead) and no signal is maintained at the BBs (arrow). During subsequent spermatid flagellar elongation, Cep162 signal disappears from the ciliary cap base labeled with Fbf1 (a transition fiber protein, red). γ-Tubulin (red) labels the centriole/basal body, axoneme is marked with Ac-Tub (magenta) and nuclei are marked with DAPI (blue). (F) The localization of GFP-tagged Cep162 N-terminus (1–447 aa) and C-terminus (448–897 aa) in spermatocytes. γ-Tubulin was used to label the BBs (red). (G) 3D-SIM images of Cep131::GFP, Cep162::GFP, Cep290-C::GFP, or Cep290-N::GFP co-immunostained with antibody against Dzip1 (red). The plots of the signal intensity are shown on the right, respectively. (H) Graph showing the radial diameter of Cep131, Cep162, Cep290-N, and Cep290-C. (I) Schematic diagrams of localization pattern of Cep131 and Cep162 in the cross section of TZ. Scale bars: 5 μm (C, D), 2 μm (F), 500 nm (G), Zoom, 1 μm (D, E). The data underlying this figure can be found in S1 Data. BB, basal body; TZ, transition zone; WT, wild type. https://doi.org/10.1371/journal.pbio.3002330.g002 The function of CG42699/Cep162 in fly is unknown yet. We constructed a transgenic fly strain expressing Cep162::GFP under the control of its endogenous promoter to examine its subcellular localization in testis and ciliated sensory neurons. We observed that Cep162 was localized to the BB in all types of cilia (Fig 2C–2E), indicating that the subcellular localization of Cep162 in Drosophila is conserved. Interestingly, we found that the BB signal of Cep162 was completely lost in cep131 mutants (Fig 2C and 2D), indicating that Cep131 plays a critical role in recruiting Cep162. Notably, using transgenic flies expressing a GFP-tagged C-terminal fragment of Cep162 (Cep162-C::GFP) encompassing amino acids 448–897, we observed that Cep162 C-terminus alone was able to target to the BB (Fig 2F). Conversely, the GFP-tagged N-terminal fragment Cep162-N::GFP comprising amino acids 1–447 failed to localize to the BB. These results indicate that the C-terminus of Cep162 is critical for BB targeting. This characteristic is conserved in mammals, as it has been reported previously that the centrosome localization of mammalian Cep162 also depends on its C-terminus [44]. In spermatocyte cilia, Cep162::GFP was localized at the tips of the basal bodies (Fig 2E). To more accurately determine the localization of Cep162, we performed 3D-SIM and examined its spatial distribution with respect to other BB proteins. As shown in Fig 2G, both Cep131::GFP and Cep162::GFP were surrounded by the TZ protein Dzip1. Notably, the plot profile of fluorescence showed that the distribution of Cep162 was bimodal, while the distribution of Cep131 has a single peak. Consistent with this observation, the average radial diameter of Cep162 signal was larger than that formed by Cep131 (Fig 2G and 2H), suggesting that Cep162 surrounds Cep131 (Fig 2I). Furthermore, we calculated the average radial diameters of Cep290-C::GFP and Cep290-N::GFP signals, and found that Cep290-C::GFP was close to Cep162, and formed a smaller diameter domain than that formed by Cep290-N (Fig 2G–2I). Taken together, our 3D-SIM spatial distribution data support the role of Cep162 in mediating the connection between Cep131 and Cep290 C-terminus at the TZ. In round spermatids, the TZ starts migrating along the growing axoneme. Cep162::GFP also migrated away from the BB with the ring centriole labeled by γ-tubulin, but its signal was gradually decreased and eventually completely disappeared from the ciliary cap base (labeled by Fbf1) in elongating spermatids (Fig 2E). This behavior is similar to that of Cep131 but different from other TZ proteins [43]. Such unique temporal localization pattern of Cep131 and Cep162 suggests a specific role in the initiation stage of ciliogenesis, but not as a constitutive component of the mature TZ. In addition, we noticed that unlike other TZ proteins, Cep162 was localized to the distal end of centrioles in spermatogonium (Fig 2E), suggesting that it was recruited to the centriole before cilia formation. Cep162 acts downstream of Cep131 to regulate ciliogenesis To elucidate the role of Cep162 in fly ciliogenesis, we designed 2 gRNA to knockout Cep162 using the CRISPR-Cas9 system. We obtained a deletion mutant line, cep1621 (c.981-1306Del), in which the C-terminus of Cep162 was lost due to reading frame shift caused by the deletion (Figs 3A, S4A and S4B). cep1621 mutants were viable, but showed defects in cilia-related behaviors such as movement and hearing, which could be effectively rescued by expression of Cep162 (Fig 3B). Examination of the cilia morphology in auditory organ showed that about 20.2% of cilia were missing or very short in cep162 mutants (Fig 3C). In addition, TEM analysis showed that missing spermatids were frequently observed in the cysts of cep162 testes (Fig 3D). Therefore, Cep162 is indeed a key component for ciliogenesis in Drosophila. Notably, our TEM images revealed evident mitochondrial abnormalities in cep162 spermatids, with some axonemes displaying very small or entirely lost mitochondrial derivatives (highlighted in Fig 3D). Additionally, the sizes of mitochondrial derivatives in the cep162 mutant vary, contrasting with the overall even size observed in WT. How Cep162 affects mitochondria dynamics remain unclear. Interestingly, a recent study by Bauerly and colleagues reported the impact of cilia-related gene on mitochondria dynamics during Drosophila spermatogenesis [45]. Mutants in dynein-related ciliary genes exhibited the absence or reduced size of minor mitochondrial derivatives. Therefore, the influence of cilia-related genes on mitochondria is not a singular occurrence. While the underlying mechanism is presently unclear, further investigating is needed. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 3. cep162 mutant mimics the phenotype of cep131 mutant. (A) Generation of cep162 deletion mutants. Schematics show the genomic (upper panel) and protein (lower panel) structures of Cep162, along with the predicted protein products of cep1621 mutant (Cep1621_p.(Glu327_Met436delfsTer24)). Arrows point to 2 gRNA targeting sites. cep1621 mutant has a deletion in cDNA from nt 981 to 1306, resulting in a reading frame shift and C-terminus loss. (B) Analysis of hearing and negative geotaxis of cep162 mutants. cep1621 flies show mild hearing defects. The retraction index indicates the larval response to a 1k Hz tone. The box shows the median and interquartile range; n = 25. The percentage of cep1621 flies passing the 8 cm high scale was significantly lower than that of WT flies. The error bars represent the mean ± SD, n = 50. (C) Living images of cilia morphology in antennal auditory organ of WT fly and cep162 mutant pupae. Sensory neurons were labeled by nompC-Gal4/UAS::GFP (green), cilia are localized at the tip of dendrites. Sensory cilia are lost in partial sensory neurons (white asterisks) of cep1621 mutant. The graph on the right shows the percentage of sensory neurons with ciliary defects. (D) Representative TEM images of elongating spermatid cysts in WT and cep162 mutants. There are 64 spermatids per cyst in WT, whereas the number of spermatids per cyst is reduced in cep162 mutant. (E) Immunostaining of Cep131, Cep290, Dzip1, Cby, Mks1, and Mks6 in WT or cep1621 testis. The quantification of the TZ protein intensities is shown on the lower panel. Unlike other TZ proteins, the localization of Cep131 in the TZ is normal. The error bars represent the mean ± SD, n = 30. Scale bars, 5 μm (C), 1 μm (D), 2 μm (E). The data underlying this figure can be found in S1 Data. TZ, transition zone; WT, wild type. https://doi.org/10.1371/journal.pbio.3002330.g003 In spermatocyte cilia of cep162 mutants, similar to cep131 mutants, there was a significantly reduction in the signal intensities of Cep290, Dzip1, Cby, and TZ proteins Mks1 and Mks6 (Fig 3E). Additionally, we observed abnormally extended CG6652::GFP signals in 12.6% of spermatocyte centrioles (S4C Fig), indicating impaired BB docking in cep162 mutants. Live imaging of the connection between the BB and the plasma membrane further confirmed the defective BB docking in some round spermatids (S4D Fig). Importantly, we found that the BB localization of Cep131 was normal in cep162 mutants (Fig 3E). Collectively, our data indicate that cep162 and cep131 mutants exhibit similar phenotypes, with Cep131 being necessary for recruiting Cep162, whereas the reverse is not true. Cep162 is required for the correct localization of C-terminus of Cep290 We then asked whether Cep162 is the downstream protein of Cep131 responsible for regulating the localization of Cep290 C-terminus to the TZ. In fact, the signal of overexpressed Cep290-C::GFP in cep162 mutant spermatocytes was significantly reduced compared to WT (Fig 4A), but a certain amount of signal could still be observed. Notably, such residual Cep290-C::GFP signal was much stronger than that observed in cep131 mutants (Figs 1A and 3E). Mammalian Cep290 C-terminal fragment was previously shown to interact with itself or to Cep290 N-terminal fragment, forming homodimers or heterodimers [46]. As part of endogenous Cep290 was still able to localize to the TZ in cep162 mutants, we therefore speculated that overexpressed Cep290-C::GFP might bind to remaining endogenous Cep290. To exclude this possibility, we generated the cep162 and cep2901 double mutant. Previously, we have shown that the TZ assembly is completely blocked and that centriole/basal body microtubules extend abnormally in cep2901 mutant [29]. Interestingly, we observed that both Cep162-FL::GFP and Cep162-C::GFP were localized along the abnormally extended microtubules in cep2901 single mutant (Fig 4B), suggesting that Cep162 can recognize microtubules independently of Cep290. Intriguingly, in cep2901 single mutants, Cep290-C::GFP showed a similar localization pattern as Cep162::GFP along the abnormally extended microtubules (Fig 4A), demonstrating that Cep290-C::GFP does not need full length Cep290 to be targeted to the axoneme. Furthermore, the signal of Cep290-C::GFP was significantly reduced in spermatocytes of cep162; cep2901 double mutants (Fig 4A), despite the persistence of abnormal microtubule extensions indicated by Ac-tub (S4E Fig) in these double mutants. This observation suggests that Cep162 regulates the association between Cep290 C-terminus with microtubules in spermatocyte cilia. Collectively, our results indicate that the localization of Cep290-C to the axoneme is mediated by Cep162, while it can still be retained by endogenous Cep290 in cep162 mutants. Similar results were also observed in sensory cilia, where the deletion of Cep290 did not affect the targeting of Cep162-FL or Cep162-C to the ciliary base in sensory neurons (Fig 4C), but Cep290-C::GFP was missing from the TZ in cep162; cep2901 sensory neurons (Fig 4D). Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 4. Cep162 is required for the BB localization of Cep290 C-terminus. (A) Immunostaining of Cep290-C::GFP (green) in spermatocyte cilia of WT, cep162, cep2901, and cep162; cep2901 flies, and the quantification of their corresponding relative fluorescence intensities is shown on the right. Cep290-C::GFP completely lose TZ localization in the cep162 and cep2901 double mutant spermatocytes. Centriole/basal body is marked with γ-Tubulin (red). The error bars represent the mean ± SD, n = 50. (B) Immunostaining of Cep162-FL::GFP (green) and Cep162-C::GFP (green) in WT or cep2901 spermatocyte cilia. The cartoon shows Cep162 signals in WT or Cep2901. Centriole/basal body is marked with γ-Tubulin (red). (C) Cep162 signals are grossly normal in cep2901 mutant antennae. 21A6 (blue) marks the cilia base; Actin (red) marks the ciliated region. (D) Immunostaining of Cep290-C::GFP (green) in WT, cep162, cep2901, and cep162; cep2901 antennae, and the quantification of their corresponding relative fluorescence intensities is shown on the right. Notably, Cep290-C::GFP signal is significantly reduced in cep162; cep2901 double mutants. 21A6 (blue) marks the ciliary base; Actin (red) marks the cilia region. The error bars represent the mean ± SD, n = 50. Scale bars, 4 μm (A, B), 5 μm (C, D), Zoom, 1 μm (C, D). The data underlying this figure can be found in S1 Data. BB, basal body; WT, wild type. https://doi.org/10.1371/journal.pbio.3002330.g004 Cep162 genetically interacts with Cby-Fam92 module to initiate ciliogenesis Since ciliogenesis initiation was completely abolished in cep131 and cby double mutants [43], we speculated that cep162 and cby double mutants should have a similar phenotype. Indeed, cep162; cby flies showed much more severe cilia-related defects than either single mutant. The cep162; cby flies were severely uncoordinated, unable to walk and fly. Hearing assay indicated that hearing was completely lost in double mutants (Fig 5A). Morphological examination of auditory cilia revealed a failure to form cilia in auditory organ (Fig 5B). In spermatocytes, the percentage of abnormal extensions of the centriole microtubules, labeled by CG6652, increased from 12.6% in cep162 single mutants, or 47.6% in cby single mutants to 81.1% in cep162; cby double mutants, indicating a strong synthetic defect in the initiation of ciliogenesis (Fig 5C). In addition, similar to cep131; cby double mutants, sperm flagella in cep162; cby spermatids were severely affected, with almost no axoneme observed in TEM analysis (Fig 5D). Consistent with this observation, the signals of Cep290, Dzip1, Mks1, and Mks6 were completely lost from the tips of BBs in the double mutants (Fig 5E and 5F). All these results indicate that the cep162; cby mutants mimic the phenotype of cep131; cby double mutants previously reported. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 5. Cep162 genetically interacts with Cby-Fam92 module to initiate ciliogenesis. (A) cep162; cby flies completely lose their hearing and negative geotaxis. (B) Living images of cilia morphology in pupal antennal auditory neurons in WT flies and cep162; cby mutants. Cilia are completely lost in cep162; cby mutants. Sensory neurons are labeled by nompC-Gal4/UAS::GFP (green), and cilia are localized at the tip of dendrites. (C) Compared to WT or cby single mutants, in cep162; cby mutants, the percentage of spermatocyte cilia with aberrant extensions of CG6652 signal is significantly increased. Basal bodies are marked with γ-Tubulin (red). (D) Compared to WT which has 64 flagella in each spermatid cyst, few normal flagella are observed in spermatid cysts of cep162; cby mutants. (E) Cep290, Dzip1, Cby, and Mks1 are absent from the TZ in cep162; cby mutant spermatocytes. Right panel show the quantification of corresponding relative fluorescence intensities. Centriole/basal body is marked with γ-Tubulin (red). The error bars represent the mean ± SD, n = 30. (F) Cep290, Dzip1, Cby, and Mks1 are absent from the TZ in cep162; cby mutant sensory neuron. Right panel show the quantification of corresponding relative fluorescence intensities. 21A6 (blue) marks the ciliary base; Actin (red) marks the ciliated region. The error bars represent the mean ± SD, n = 30. Scale bars, 5 μm (B, C, F), 2 μm (D, E), Zoom, 1 μm (F). The data underlying this figure can be found in S1 Data. TZ, transition zone; WT, wild type. https://doi.org/10.1371/journal.pbio.3002330.g005 As Cby and Fam92 function together in a module to regulate ciliogenesis [30,47,48], we speculated that the combined mutation of Cep162 with Fam92 might also lead to synthetic ciliary defects. Indeed, as expected, the cep162; fam92 flies showed much more severe defects in walk and fly than either single mutant alone, and Cep290 was also completely lost in spermatocyte cilia of the double mutants (S5 Fig). Cby-Fam92 module is required for the association of the N-terminus of Cep290 with the membrane The synthetic defects observed in Cep162/Cep131 and Cby/Fam92 double mutants could likely be attributed to the complete loss of Cep290 signal at the BB. Considering that Cep131-Cep162 module is required for the localization of Cep290 C-terminus, and the localization pattern of Cep290 N-terminus is similar to that of Cby near the membrane, we speculated that Cby-Fam92 module might play a role in promoting the BB localization of Cep290 N-terminus. To exclude the effect of Cep290 C-terminus on our analysis, we combined deletions of Cep290 C-terminus (cep290ΔC) with Cby, and checked the localization of overexpressed Cep290-N::GFP. Indeed, the TZ signal of Cep290-N::GFP was almost completely lost in cby; cep290ΔC double mutants compared with that in cby or cep290ΔC single mutants (Fig 6A), indicating that Cby does play a role in targeting the Cep290 N-terminus to the TZ. Notably, unlike Cep290-N::GFP, Cep290-C::GFP was still able to localize to the tips of basal bodies in cby; cep290ΔC double mutants (Fig 6B), suggesting that Cby has a specific role on the localization of Cep290-N::GFP. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 6. The Cby-Fam92 module is required for the BB localization of the N-terminus of Cep290. (A) Immunostaining of the exogenous Cep290-N::GFP (green) in spermatocyte cilia of WT, cby, cep290ΔC, cby; cep290ΔC testis and the quantification of corresponding relative fluorescence intensities is shown on the right. The signal of Cep290-C::GFP is almost lost at the tips of centrioles in cby; cep290ΔC double mutants. Centriole/basal body is marked with γ-Tubulin (red). The error bars represent the mean ± SD, n = 60. (B) Immunostaining of the exogenous Cep290-C::GFP (green) in spermatocyte cilia of WT and cby; cep290ΔC testis. Cep290-C::GFP was localized to the axoneme in cby; cep290ΔC mutants. Centriole/basal body is marked with γ-Tubulin (red). (C) Immunostaining of the endogenous Cep290 in WT, cby, cep290ΔC, and cby; cep290ΔC testis and the quantification of corresponding relative fluorescence intensities were shown on the right. The C-terminal truncated Cep290 form fails to locate to the tips of centrioles in cby; cep290ΔC spermatocytes. Centriole/basal body is marked with γ-Tubulin (red). The error bars represent the mean ± SD, n = 30. (D) The endogenous Cep290 C-terminal truncated form fails to locate to the cilia base in antennal auditory neurons of cby; cep290ΔC antenna. Right panel shows the quantification of corresponding relative fluorescence intensities. 21A6 (blue) marks the cilia base; Actin (red) marks the ciliated region. The error bars represent the mean ± SD, n = 60. (E) In cby; cep290ΔC mutants, the percentage of spermatocyte cilia with aberrant extension of CG6652 signal reaches about 79.8%. Axoneme is marked with CG6652. Centriole/basal body is marked with γ-Tubulin (red). The error bars represent the mean ± SD, n = 30. (F) Dzip1 and Mks1 are absent from the TZ in spermatocyte cilia of cby; cep290ΔC mutants. Centriole/basal body is marked with γ-Tubulin (red). The error bars represent the mean ± SD, n = 30. (G) Dzip1 and Mks1 fail to locate to the cilia base in antennal auditory neurons of cby; cep290ΔC antenna. 21A6 (blue) marks the cilia base; Actin (red) marks the ciliated region. Scale bars, 2 μm (A, C, F), 5 μm (B, D, E, G), Zoom, 1 μm (D, G). The data underlying this figure can be found in S1 Data. BB, basal body; TZ, transition zone; WT, wild type. https://doi.org/10.1371/journal.pbio.3002330.g006 Given that Cby affects the localization of Cep290 N-terminus, we reasoned that endogenous truncated Cep290 may not be able to target to TZ in cby; cep290ΔC double mutants. Therefore, we examined the Cep290 signal in spermatocytes using our Cep290 antibody. In cby single mutant, Cep290 signal was slightly decreased. In cep290ΔC single mutant, Cep290 N-terminus is expressed at lower levels as previously described [29] but is still retained at the tip of BB. However, consistent with our hypothesis, this TZ signal of remaining Cep290 truncated form was completely lost in spermatocytes of cby; cep290ΔC double mutants (Fig 6C). Similar results were also observed in sensory cilia, where Cep290 signal was also completely lost in cby; cep290ΔC double mutants (Fig 6D). As well, we demonstrated that Cep290 signal was also completely lost in fam92; cep290ΔC double mutants (S5 Fig). All these results indicated that Cby-Fam92 module specifically promotes the anchoring Cep290 N-terminus to the membrane. Since Cep290 was completely lost from BBs in cby; cep290ΔC, the initiation of ciliogenesis should be completely blocked like in cby; cep131 or cby; cep162 double mutants. In fact, as expected, the proportion of abnormal extensions of CG6652 at the tips of the centrioles reached about 79.8%, indicating that most of the BBs do not anchor to the plasma membrane (Fig 6E). In addition, Dzip1 and Mks1 were completely lost from the tips of BBs in both ciliated cell types (Fig 6F and 6G). All these results indicate that the cby; cep290ΔC mutant mimics the phenotype of cep290 null mutant. Cep131 is required for the basal body localization of Cep290 C-terminus To understand the function of the Cep131 in ciliogenesis, we generated the cep1311 (C terminal deletion) mutant flies using the CRISPR-Cas9 system (S1A–S1C Fig). Similar to reported cep131 mutant [42], our cep1311 flies exhibited typical symptoms related to ciliary defects, and were severely uncoordinated during walking and flying. Consistent with previous reports that cep131 single mutant fly has mild defects in the initiation of ciliogenesis [42], we observed that approximately 33.6% of cep1311 spermatocyte centrioles showed abnormal elongation of the microtubules labeled by CG6652::GFP (S1D Fig), a Drosophila spermatocyte-specific phenotype associated with defects in BB docking and TZ membrane cap formation. Consistent with this observation, the signal of the ciliogenesis initiation regulators Cep290, Dzip1, and Cby, as well as the TZ marker Mks1 were notably reduced at the basal bodies of spermatocytes of cep1311 mutants, although they were still detectable (Fig 1A). Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 1. Cep131 is required for the BB localization of Cep290-C terminus. (A) Localization of various TZ proteins in spermatocyte cilia of WT flies and cep131 mutants and quantifications of corresponding relative fluorescence intensities. In cep131 mutants, the signals of Cep290, Dzip1, Cby::GFP, and Mks1::GFP are significantly reduced compared to WT. Importantly, Cep290-C::GFP signal is almost completely lost in cep131 mutants. The BB is labeled with γ-Tubulin (red). The error bars represent the mean ± SD, n = 30. (B) The localization of various TZ proteins in auditory cilia of WT flies and cep131 mutants. Similar to what we observed in spermatocyte cilia, the signals of Cep290, Dzip1, and Cby::GFP are significantly decreased, and Cep290-C::GFP is completely lost at the base of the sensory cilia. 21A6 (blue) marks the cilia base, Actin (red) marks the ciliated region. The error bars represent the mean ± SD, n = 30. Scale bars: 2 μm (A), 5 μm (B, full-scale images on the left), 1 μm (B, insets or zoomed in areas on the right). The data underlying this figure can be found in S1 Data. BB, basal body; TZ, transition zone; WT, wild type. https://doi.org/10.1371/journal.pbio.3002330.g001 Cep290 is the most upstream protein known in the initiation of ciliogenesis in Drosophila [29,30]. Works in fly and mammalian cells have suggested that Cep290 bridges the ciliary axoneme and the membrane, with its C-terminus associated with microtubule doublets and N-terminus associated with the membrane [29,33,34]. Previously, we have demonstrated that both N-terminal truncation (Cep290-N, aa 1–650) and C-terminal truncation (Cep290-C, aa 1385 to the end at 1978) of Drosophila Cep290 are capable of localizing to the TZ, independently of each other [29]. 3D-SIM microscopy revealed that Cep290-C::GFP localizes close to the axoneme, whereas Cep290-N::GFP localizes close to the membrane, displaying a noticeably larger diameter [29]. Given that Cep131 localizes to the lumen of TZ [43], we hypothesized that Cep131 might have a role in regulating the localization of Cep290 C-terminus. As expected, we found that the signal of Cep290-C::GFP signal was nearly completely lost in spermatocyte cilia of cep131 mutants, whereas the signal intensity of Cep290-N::GFP was similar to that in wild-type (WT) (Fig 1A). Interestingly, we noticed that both the diameter of endogenous Cep290 N-terminus (labeled by anti-Cep290 antibody against aa 292–541 in its N-terminus) and Cep290-N::GFP in cep131 mutants were significantly smaller than that in WT, suggesting a potential alternation in the conformation of Cep290 in cep131 mutants (Figs 1A and S1E). Collectively, our results indicate that Cep131 specifically regulates the localization of Cep290 C-terminus in spermatocyte cilia. To determine whether our observation is specific to spermatocyte cilia, we focused on sensory cilia, another type of cilia in Drosophila. Similar to our observation in spermatocytes, the signal intensities of Cep290, Cep290-N, Dzip1, Cby, and Mks1 were mildly affected, whereas Cep290-C::GFP was almost completely lost from the basal bodies in auditory cilia of cep131 mutants (Fig 1B). Hence, Cep131 also promotes the binding of Cep290 C-terminus to the axoneme in sensory cilia. Cep162 bridges Cep131 and Cep290 Next, we wondered whether Cep131 directly interacts with Cep290. However, no interaction between Cep131 and Cep290 was observed in our yeast two-hybrid (Y2H) assay (S2A Fig), suggesting that additional proteins might mediate the functional interaction between Cep131 and Cep290. In mammalian cells, it has been reported that Cep162 localizes to the centriole distal end and interacts with Cep290 to promote its association with microtubules [44]. Protein homology search using NCBI protein–protein BLAST identified CG42699 as the sole homolog of Cep162 in Drosophila (S3 Fig). Consistent with reports in mammalian cells, Y2H assay showed that CG42669 interacts with Drosophila Cep290 (S2A Fig). Interestingly, our Y2H assay also revealed that CG42699 interacts with Drosophila Cep131 (Fig 2A and 2B). Specifically, the C-terminal half of CG42699 (aa 448 to the end at 897) interacts with Cep131, while the N-terminal half does not. GST pull-down assay further confirmed the direct interaction between Cep131 and Cep162 C-terminus, and showed that Cep162 C-terminus interacts with both the N-terminal half (aa 1–549) and C-terminal half (aa 550 to the end at 1114) of Cep131 (S2B Fig). Subsequent analysis using Y2H showed that Cep162 C-terminus does not interact with the middle region of Cep131 (aa 481–781), but it does interacts with both the N-terminus (aa 1–480) and C-terminus (aa 782 to the end at 1114) (S2C Fig), indicating the presence of 2 binding sites for Cep162 in Cep131. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 2. Cep131 interacts with and recruits Cep162 to centriole tips and the TZ. (A, B) Cep131 directly interacts with Cep162. (A) Schematic representation of full-length (Cep162-FL) or truncated (Cep162-N, Cep162-C) Cep162 proteins used for interaction assays in B. (B) Cep131 interacts with Cep162-FL and Cep162-C, but not Cep162-N in the Y2H assay. The upper panel shows the presence of Y2H plasmids as evidenced by colony growth on SD-Leu-Trp plates. The lower panel shows the positive interaction between Cep131 and Cep162 as evidenced by colony growth on SD-Ade-Leu-Trp-His plates. (C) Immunostaining of Cep162::GFP (green) in spermatocyte cilia in cep131 mutants. Quantification of Cep162 signal intensity at the ciliary base is shown on the right. γ-Tubulin (red) labels the centriole/basal body. The error bars represent the mean ± SD, n = 30. (D) Immunostaining of Cep162::GFP (green), Actin (red), and 21A6 (blue) in auditory cilia or olfactory cilia in WT or cep131 mutants. Quantification of Cep162 signal intensity at the ciliary base is shown on the right. The error bars represent the mean ± SD, n = 30. (E) The subcellular localization of exogenous Cep162::GFP during spermatogenesis. From spermatogonia to late spermatocytes, Cep162 is localized at the tip of centriole/basal body. In round spermatids, as flagella elongates and the ciliary cap moves away from the BBs, Cep162 migrates with the ring centriole labeled by γ-Tubulin (arrowhead) and no signal is maintained at the BBs (arrow). During subsequent spermatid flagellar elongation, Cep162 signal disappears from the ciliary cap base labeled with Fbf1 (a transition fiber protein, red). γ-Tubulin (red) labels the centriole/basal body, axoneme is marked with Ac-Tub (magenta) and nuclei are marked with DAPI (blue). (F) The localization of GFP-tagged Cep162 N-terminus (1–447 aa) and C-terminus (448–897 aa) in spermatocytes. γ-Tubulin was used to label the BBs (red). (G) 3D-SIM images of Cep131::GFP, Cep162::GFP, Cep290-C::GFP, or Cep290-N::GFP co-immunostained with antibody against Dzip1 (red). The plots of the signal intensity are shown on the right, respectively. (H) Graph showing the radial diameter of Cep131, Cep162, Cep290-N, and Cep290-C. (I) Schematic diagrams of localization pattern of Cep131 and Cep162 in the cross section of TZ. Scale bars: 5 μm (C, D), 2 μm (F), 500 nm (G), Zoom, 1 μm (D, E). The data underlying this figure can be found in S1 Data. BB, basal body; TZ, transition zone; WT, wild type. https://doi.org/10.1371/journal.pbio.3002330.g002 The function of CG42699/Cep162 in fly is unknown yet. We constructed a transgenic fly strain expressing Cep162::GFP under the control of its endogenous promoter to examine its subcellular localization in testis and ciliated sensory neurons. We observed that Cep162 was localized to the BB in all types of cilia (Fig 2C–2E), indicating that the subcellular localization of Cep162 in Drosophila is conserved. Interestingly, we found that the BB signal of Cep162 was completely lost in cep131 mutants (Fig 2C and 2D), indicating that Cep131 plays a critical role in recruiting Cep162. Notably, using transgenic flies expressing a GFP-tagged C-terminal fragment of Cep162 (Cep162-C::GFP) encompassing amino acids 448–897, we observed that Cep162 C-terminus alone was able to target to the BB (Fig 2F). Conversely, the GFP-tagged N-terminal fragment Cep162-N::GFP comprising amino acids 1–447 failed to localize to the BB. These results indicate that the C-terminus of Cep162 is critical for BB targeting. This characteristic is conserved in mammals, as it has been reported previously that the centrosome localization of mammalian Cep162 also depends on its C-terminus [44]. In spermatocyte cilia, Cep162::GFP was localized at the tips of the basal bodies (Fig 2E). To more accurately determine the localization of Cep162, we performed 3D-SIM and examined its spatial distribution with respect to other BB proteins. As shown in Fig 2G, both Cep131::GFP and Cep162::GFP were surrounded by the TZ protein Dzip1. Notably, the plot profile of fluorescence showed that the distribution of Cep162 was bimodal, while the distribution of Cep131 has a single peak. Consistent with this observation, the average radial diameter of Cep162 signal was larger than that formed by Cep131 (Fig 2G and 2H), suggesting that Cep162 surrounds Cep131 (Fig 2I). Furthermore, we calculated the average radial diameters of Cep290-C::GFP and Cep290-N::GFP signals, and found that Cep290-C::GFP was close to Cep162, and formed a smaller diameter domain than that formed by Cep290-N (Fig 2G–2I). Taken together, our 3D-SIM spatial distribution data support the role of Cep162 in mediating the connection between Cep131 and Cep290 C-terminus at the TZ. In round spermatids, the TZ starts migrating along the growing axoneme. Cep162::GFP also migrated away from the BB with the ring centriole labeled by γ-tubulin, but its signal was gradually decreased and eventually completely disappeared from the ciliary cap base (labeled by Fbf1) in elongating spermatids (Fig 2E). This behavior is similar to that of Cep131 but different from other TZ proteins [43]. Such unique temporal localization pattern of Cep131 and Cep162 suggests a specific role in the initiation stage of ciliogenesis, but not as a constitutive component of the mature TZ. In addition, we noticed that unlike other TZ proteins, Cep162 was localized to the distal end of centrioles in spermatogonium (Fig 2E), suggesting that it was recruited to the centriole before cilia formation. Cep162 acts downstream of Cep131 to regulate ciliogenesis To elucidate the role of Cep162 in fly ciliogenesis, we designed 2 gRNA to knockout Cep162 using the CRISPR-Cas9 system. We obtained a deletion mutant line, cep1621 (c.981-1306Del), in which the C-terminus of Cep162 was lost due to reading frame shift caused by the deletion (Figs 3A, S4A and S4B). cep1621 mutants were viable, but showed defects in cilia-related behaviors such as movement and hearing, which could be effectively rescued by expression of Cep162 (Fig 3B). Examination of the cilia morphology in auditory organ showed that about 20.2% of cilia were missing or very short in cep162 mutants (Fig 3C). In addition, TEM analysis showed that missing spermatids were frequently observed in the cysts of cep162 testes (Fig 3D). Therefore, Cep162 is indeed a key component for ciliogenesis in Drosophila. Notably, our TEM images revealed evident mitochondrial abnormalities in cep162 spermatids, with some axonemes displaying very small or entirely lost mitochondrial derivatives (highlighted in Fig 3D). Additionally, the sizes of mitochondrial derivatives in the cep162 mutant vary, contrasting with the overall even size observed in WT. How Cep162 affects mitochondria dynamics remain unclear. Interestingly, a recent study by Bauerly and colleagues reported the impact of cilia-related gene on mitochondria dynamics during Drosophila spermatogenesis [45]. Mutants in dynein-related ciliary genes exhibited the absence or reduced size of minor mitochondrial derivatives. Therefore, the influence of cilia-related genes on mitochondria is not a singular occurrence. While the underlying mechanism is presently unclear, further investigating is needed. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 3. cep162 mutant mimics the phenotype of cep131 mutant. (A) Generation of cep162 deletion mutants. Schematics show the genomic (upper panel) and protein (lower panel) structures of Cep162, along with the predicted protein products of cep1621 mutant (Cep1621_p.(Glu327_Met436delfsTer24)). Arrows point to 2 gRNA targeting sites. cep1621 mutant has a deletion in cDNA from nt 981 to 1306, resulting in a reading frame shift and C-terminus loss. (B) Analysis of hearing and negative geotaxis of cep162 mutants. cep1621 flies show mild hearing defects. The retraction index indicates the larval response to a 1k Hz tone. The box shows the median and interquartile range; n = 25. The percentage of cep1621 flies passing the 8 cm high scale was significantly lower than that of WT flies. The error bars represent the mean ± SD, n = 50. (C) Living images of cilia morphology in antennal auditory organ of WT fly and cep162 mutant pupae. Sensory neurons were labeled by nompC-Gal4/UAS::GFP (green), cilia are localized at the tip of dendrites. Sensory cilia are lost in partial sensory neurons (white asterisks) of cep1621 mutant. The graph on the right shows the percentage of sensory neurons with ciliary defects. (D) Representative TEM images of elongating spermatid cysts in WT and cep162 mutants. There are 64 spermatids per cyst in WT, whereas the number of spermatids per cyst is reduced in cep162 mutant. (E) Immunostaining of Cep131, Cep290, Dzip1, Cby, Mks1, and Mks6 in WT or cep1621 testis. The quantification of the TZ protein intensities is shown on the lower panel. Unlike other TZ proteins, the localization of Cep131 in the TZ is normal. The error bars represent the mean ± SD, n = 30. Scale bars, 5 μm (C), 1 μm (D), 2 μm (E). The data underlying this figure can be found in S1 Data. TZ, transition zone; WT, wild type. https://doi.org/10.1371/journal.pbio.3002330.g003 In spermatocyte cilia of cep162 mutants, similar to cep131 mutants, there was a significantly reduction in the signal intensities of Cep290, Dzip1, Cby, and TZ proteins Mks1 and Mks6 (Fig 3E). Additionally, we observed abnormally extended CG6652::GFP signals in 12.6% of spermatocyte centrioles (S4C Fig), indicating impaired BB docking in cep162 mutants. Live imaging of the connection between the BB and the plasma membrane further confirmed the defective BB docking in some round spermatids (S4D Fig). Importantly, we found that the BB localization of Cep131 was normal in cep162 mutants (Fig 3E). Collectively, our data indicate that cep162 and cep131 mutants exhibit similar phenotypes, with Cep131 being necessary for recruiting Cep162, whereas the reverse is not true. Cep162 is required for the correct localization of C-terminus of Cep290 We then asked whether Cep162 is the downstream protein of Cep131 responsible for regulating the localization of Cep290 C-terminus to the TZ. In fact, the signal of overexpressed Cep290-C::GFP in cep162 mutant spermatocytes was significantly reduced compared to WT (Fig 4A), but a certain amount of signal could still be observed. Notably, such residual Cep290-C::GFP signal was much stronger than that observed in cep131 mutants (Figs 1A and 3E). Mammalian Cep290 C-terminal fragment was previously shown to interact with itself or to Cep290 N-terminal fragment, forming homodimers or heterodimers [46]. As part of endogenous Cep290 was still able to localize to the TZ in cep162 mutants, we therefore speculated that overexpressed Cep290-C::GFP might bind to remaining endogenous Cep290. To exclude this possibility, we generated the cep162 and cep2901 double mutant. Previously, we have shown that the TZ assembly is completely blocked and that centriole/basal body microtubules extend abnormally in cep2901 mutant [29]. Interestingly, we observed that both Cep162-FL::GFP and Cep162-C::GFP were localized along the abnormally extended microtubules in cep2901 single mutant (Fig 4B), suggesting that Cep162 can recognize microtubules independently of Cep290. Intriguingly, in cep2901 single mutants, Cep290-C::GFP showed a similar localization pattern as Cep162::GFP along the abnormally extended microtubules (Fig 4A), demonstrating that Cep290-C::GFP does not need full length Cep290 to be targeted to the axoneme. Furthermore, the signal of Cep290-C::GFP was significantly reduced in spermatocytes of cep162; cep2901 double mutants (Fig 4A), despite the persistence of abnormal microtubule extensions indicated by Ac-tub (S4E Fig) in these double mutants. This observation suggests that Cep162 regulates the association between Cep290 C-terminus with microtubules in spermatocyte cilia. Collectively, our results indicate that the localization of Cep290-C to the axoneme is mediated by Cep162, while it can still be retained by endogenous Cep290 in cep162 mutants. Similar results were also observed in sensory cilia, where the deletion of Cep290 did not affect the targeting of Cep162-FL or Cep162-C to the ciliary base in sensory neurons (Fig 4C), but Cep290-C::GFP was missing from the TZ in cep162; cep2901 sensory neurons (Fig 4D). Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 4. Cep162 is required for the BB localization of Cep290 C-terminus. (A) Immunostaining of Cep290-C::GFP (green) in spermatocyte cilia of WT, cep162, cep2901, and cep162; cep2901 flies, and the quantification of their corresponding relative fluorescence intensities is shown on the right. Cep290-C::GFP completely lose TZ localization in the cep162 and cep2901 double mutant spermatocytes. Centriole/basal body is marked with γ-Tubulin (red). The error bars represent the mean ± SD, n = 50. (B) Immunostaining of Cep162-FL::GFP (green) and Cep162-C::GFP (green) in WT or cep2901 spermatocyte cilia. The cartoon shows Cep162 signals in WT or Cep2901. Centriole/basal body is marked with γ-Tubulin (red). (C) Cep162 signals are grossly normal in cep2901 mutant antennae. 21A6 (blue) marks the cilia base; Actin (red) marks the ciliated region. (D) Immunostaining of Cep290-C::GFP (green) in WT, cep162, cep2901, and cep162; cep2901 antennae, and the quantification of their corresponding relative fluorescence intensities is shown on the right. Notably, Cep290-C::GFP signal is significantly reduced in cep162; cep2901 double mutants. 21A6 (blue) marks the ciliary base; Actin (red) marks the cilia region. The error bars represent the mean ± SD, n = 50. Scale bars, 4 μm (A, B), 5 μm (C, D), Zoom, 1 μm (C, D). The data underlying this figure can be found in S1 Data. BB, basal body; WT, wild type. https://doi.org/10.1371/journal.pbio.3002330.g004 Cep162 genetically interacts with Cby-Fam92 module to initiate ciliogenesis Since ciliogenesis initiation was completely abolished in cep131 and cby double mutants [43], we speculated that cep162 and cby double mutants should have a similar phenotype. Indeed, cep162; cby flies showed much more severe cilia-related defects than either single mutant. The cep162; cby flies were severely uncoordinated, unable to walk and fly. Hearing assay indicated that hearing was completely lost in double mutants (Fig 5A). Morphological examination of auditory cilia revealed a failure to form cilia in auditory organ (Fig 5B). In spermatocytes, the percentage of abnormal extensions of the centriole microtubules, labeled by CG6652, increased from 12.6% in cep162 single mutants, or 47.6% in cby single mutants to 81.1% in cep162; cby double mutants, indicating a strong synthetic defect in the initiation of ciliogenesis (Fig 5C). In addition, similar to cep131; cby double mutants, sperm flagella in cep162; cby spermatids were severely affected, with almost no axoneme observed in TEM analysis (Fig 5D). Consistent with this observation, the signals of Cep290, Dzip1, Mks1, and Mks6 were completely lost from the tips of BBs in the double mutants (Fig 5E and 5F). All these results indicate that the cep162; cby mutants mimic the phenotype of cep131; cby double mutants previously reported. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 5. Cep162 genetically interacts with Cby-Fam92 module to initiate ciliogenesis. (A) cep162; cby flies completely lose their hearing and negative geotaxis. (B) Living images of cilia morphology in pupal antennal auditory neurons in WT flies and cep162; cby mutants. Cilia are completely lost in cep162; cby mutants. Sensory neurons are labeled by nompC-Gal4/UAS::GFP (green), and cilia are localized at the tip of dendrites. (C) Compared to WT or cby single mutants, in cep162; cby mutants, the percentage of spermatocyte cilia with aberrant extensions of CG6652 signal is significantly increased. Basal bodies are marked with γ-Tubulin (red). (D) Compared to WT which has 64 flagella in each spermatid cyst, few normal flagella are observed in spermatid cysts of cep162; cby mutants. (E) Cep290, Dzip1, Cby, and Mks1 are absent from the TZ in cep162; cby mutant spermatocytes. Right panel show the quantification of corresponding relative fluorescence intensities. Centriole/basal body is marked with γ-Tubulin (red). The error bars represent the mean ± SD, n = 30. (F) Cep290, Dzip1, Cby, and Mks1 are absent from the TZ in cep162; cby mutant sensory neuron. Right panel show the quantification of corresponding relative fluorescence intensities. 21A6 (blue) marks the ciliary base; Actin (red) marks the ciliated region. The error bars represent the mean ± SD, n = 30. Scale bars, 5 μm (B, C, F), 2 μm (D, E), Zoom, 1 μm (F). The data underlying this figure can be found in S1 Data. TZ, transition zone; WT, wild type. https://doi.org/10.1371/journal.pbio.3002330.g005 As Cby and Fam92 function together in a module to regulate ciliogenesis [30,47,48], we speculated that the combined mutation of Cep162 with Fam92 might also lead to synthetic ciliary defects. Indeed, as expected, the cep162; fam92 flies showed much more severe defects in walk and fly than either single mutant alone, and Cep290 was also completely lost in spermatocyte cilia of the double mutants (S5 Fig). Cby-Fam92 module is required for the association of the N-terminus of Cep290 with the membrane The synthetic defects observed in Cep162/Cep131 and Cby/Fam92 double mutants could likely be attributed to the complete loss of Cep290 signal at the BB. Considering that Cep131-Cep162 module is required for the localization of Cep290 C-terminus, and the localization pattern of Cep290 N-terminus is similar to that of Cby near the membrane, we speculated that Cby-Fam92 module might play a role in promoting the BB localization of Cep290 N-terminus. To exclude the effect of Cep290 C-terminus on our analysis, we combined deletions of Cep290 C-terminus (cep290ΔC) with Cby, and checked the localization of overexpressed Cep290-N::GFP. Indeed, the TZ signal of Cep290-N::GFP was almost completely lost in cby; cep290ΔC double mutants compared with that in cby or cep290ΔC single mutants (Fig 6A), indicating that Cby does play a role in targeting the Cep290 N-terminus to the TZ. Notably, unlike Cep290-N::GFP, Cep290-C::GFP was still able to localize to the tips of basal bodies in cby; cep290ΔC double mutants (Fig 6B), suggesting that Cby has a specific role on the localization of Cep290-N::GFP. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 6. The Cby-Fam92 module is required for the BB localization of the N-terminus of Cep290. (A) Immunostaining of the exogenous Cep290-N::GFP (green) in spermatocyte cilia of WT, cby, cep290ΔC, cby; cep290ΔC testis and the quantification of corresponding relative fluorescence intensities is shown on the right. The signal of Cep290-C::GFP is almost lost at the tips of centrioles in cby; cep290ΔC double mutants. Centriole/basal body is marked with γ-Tubulin (red). The error bars represent the mean ± SD, n = 60. (B) Immunostaining of the exogenous Cep290-C::GFP (green) in spermatocyte cilia of WT and cby; cep290ΔC testis. Cep290-C::GFP was localized to the axoneme in cby; cep290ΔC mutants. Centriole/basal body is marked with γ-Tubulin (red). (C) Immunostaining of the endogenous Cep290 in WT, cby, cep290ΔC, and cby; cep290ΔC testis and the quantification of corresponding relative fluorescence intensities were shown on the right. The C-terminal truncated Cep290 form fails to locate to the tips of centrioles in cby; cep290ΔC spermatocytes. Centriole/basal body is marked with γ-Tubulin (red). The error bars represent the mean ± SD, n = 30. (D) The endogenous Cep290 C-terminal truncated form fails to locate to the cilia base in antennal auditory neurons of cby; cep290ΔC antenna. Right panel shows the quantification of corresponding relative fluorescence intensities. 21A6 (blue) marks the cilia base; Actin (red) marks the ciliated region. The error bars represent the mean ± SD, n = 60. (E) In cby; cep290ΔC mutants, the percentage of spermatocyte cilia with aberrant extension of CG6652 signal reaches about 79.8%. Axoneme is marked with CG6652. Centriole/basal body is marked with γ-Tubulin (red). The error bars represent the mean ± SD, n = 30. (F) Dzip1 and Mks1 are absent from the TZ in spermatocyte cilia of cby; cep290ΔC mutants. Centriole/basal body is marked with γ-Tubulin (red). The error bars represent the mean ± SD, n = 30. (G) Dzip1 and Mks1 fail to locate to the cilia base in antennal auditory neurons of cby; cep290ΔC antenna. 21A6 (blue) marks the cilia base; Actin (red) marks the ciliated region. Scale bars, 2 μm (A, C, F), 5 μm (B, D, E, G), Zoom, 1 μm (D, G). The data underlying this figure can be found in S1 Data. BB, basal body; TZ, transition zone; WT, wild type. https://doi.org/10.1371/journal.pbio.3002330.g006 Given that Cby affects the localization of Cep290 N-terminus, we reasoned that endogenous truncated Cep290 may not be able to target to TZ in cby; cep290ΔC double mutants. Therefore, we examined the Cep290 signal in spermatocytes using our Cep290 antibody. In cby single mutant, Cep290 signal was slightly decreased. In cep290ΔC single mutant, Cep290 N-terminus is expressed at lower levels as previously described [29] but is still retained at the tip of BB. However, consistent with our hypothesis, this TZ signal of remaining Cep290 truncated form was completely lost in spermatocytes of cby; cep290ΔC double mutants (Fig 6C). Similar results were also observed in sensory cilia, where Cep290 signal was also completely lost in cby; cep290ΔC double mutants (Fig 6D). As well, we demonstrated that Cep290 signal was also completely lost in fam92; cep290ΔC double mutants (S5 Fig). All these results indicated that Cby-Fam92 module specifically promotes the anchoring Cep290 N-terminus to the membrane. Since Cep290 was completely lost from BBs in cby; cep290ΔC, the initiation of ciliogenesis should be completely blocked like in cby; cep131 or cby; cep162 double mutants. In fact, as expected, the proportion of abnormal extensions of CG6652 at the tips of the centrioles reached about 79.8%, indicating that most of the BBs do not anchor to the plasma membrane (Fig 6E). In addition, Dzip1 and Mks1 were completely lost from the tips of BBs in both ciliated cell types (Fig 6F and 6G). All these results indicate that the cby; cep290ΔC mutant mimics the phenotype of cep290 null mutant. Discussion The TZ protein Cep290 bridges the ciliary membrane and the axonemal microtubules with its N-terminus close to the membrane and its C-terminus close to the axonemal microtubules [33,34]. Previous studies by us and others have shown that the N-terminus of Cep290 acts upstream of Dzip1-Cby-Fam92 module and is essential for ciliogenesis initiation and ciliary bud formation in Drosophila [29]. Here, we discover that Cep131-Cep162 module functions upstream of Cep290 and regulates the association of Cep290-C-terminus with axonemal microtubules to initiate ciliogenesis. Taken together, we propose that the initial process of ciliogenesis in Drosophila spermatocyte is as follows (Fig 7A and 7B): in spermatogonia, Cep131 localizes to the distal end of centriole and recruits Cep162 to the centriole; when the centriole starts to grow a cilium in spermatocytes, Cep162 recruits Cep290 and promotes the binding of Cep290-C to the axoneme; subsequently, the conformation of Cep290 changes from a closed to an open state, and the N-terminus of Cep290 recruits Dzip1-Cby-Fam92 module to start early ciliary membrane formation and ciliary bud formation. On the other hand, Cby-Fam92-mediated ciliary membrane formation has a positive feedback effect on promoting the association of Cep290 N-terminus with the ciliary membrane. Given that similar results were observed in sensory cilia, this ordered model is not exclusive to spermatocyte cilia. Moreover, all proteins (Cep131, Cep162, Cep290, Cby, Fam92) involved in this ordered pathway are conserved and play a critical role in TZ assembly in mammals [24,37,44,48], suggesting a potential conservation of this model in mammals. However, it should been noted that Cep290 has been shown to have different properties across different species [9], and certain animal models carrying mutations associated with patients have displayed milder symptoms [49,50], tissue and species-specific functions of Cep290 can not be ruled out. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 7. Models for ciliogenesis initiation and the mechanism of CEP290 localization in Drosophila. (A) Model for ciliogenesis initiation in Drosophila. ① In spermatogonia, Cep131 localizes to the distal end of centriole and recruits Cep162. ② When the centriole starts to grow cilia in spermatocytes, Cep162 recruits Cep290 and promotes the binding of Cep290-C to the axoneme. ③ Subsequently, the conformation of Cep290 changes from a closed to an open state, and the N-terminus of Cep290 recruits Dzip1-Cby-Fam92 module; ④ and then start early ciliary membrane formation and ciliary bud formation. (B) The cooperative model for Cep290 localization. Cep131-Cep162 module together with Cby-Fam92 module regulates the localization of Cep290 at the TZ. Mechanistically, Cep131 recruits Cep162 which mediates the association of C-terminus of Cep290 to microtubule. The N-terminus of Cep290 recruits Dzip1-Cby-Fam92 module to start early ciliary membrane formation and ciliary bud formation, whereas Cby-Fam92-mediated ciliary membrane formation has a positive feedback effect on promoting the association of Cep290 N-terminus with the ciliary membrane. TZ, transition zone. https://doi.org/10.1371/journal.pbio.3002330.g007 Previously, Drivas and colleagues have proposed a conformation change model of Cep290 during ciliogenesis [33]. According to their model, Cep290 is initially maintained in a closed and inhibited state by its N and C termini [33]. During ciliogenesis, Cep290 undergoes a conformational change, transitioning from a closed to an open state. This conformational change allows Cep290’s membrane-binding and microtubule-binding domains to become accessible, enabling the recruitment of additional interacting partners to initiate ciliogenesis. The validity of this conformational change model is supported by 2 pieces of evidence. Firstly, the N-terminal fragment of Cep290 was found to interact with the C-terminal fragment of Cep290 in vertebrate [46], providing evidence for their association. Secondly, both the N and C termini of Cep290 have been shown to play regulatory roles in the localization and function of Cep290 in both mammals and Drosophila [29,33]. Interestingly, we observed that the diameter of Cep290-N::GFP in cep131 mutants were significantly smaller than that in WT (S1E Fig), suggesting that the conformation of Cep290 may still be closed in cep131 mutants, implying the involvement of Cep131 in conformation change of Cep290. However, direct structural evidence for the conformational change of Cep290 is currently lacking. Investigating this aspect and exploring the role of Cep131 in it will be a promising research direction in the future. Notably, Cep290 has its own microtubule-binding domain and membrane-binding amphipathic α-helix motif [33]; therefore, the role of Cep131-Cep162 and Cby-Fam92 modules might just be to facilitate and enhance the efficiency of Cep290 connection with the axoneme and the membrane. Given the partial defects in ciliogenesis in the absence of Cep131 or Cep162, we speculate that the microtubule binding ability of Cep290 may be sufficient to build the TZ in absence of either Cep131 or Cep162. We propose that as long as Cep290, even in limited amounts, is initially able to localize to the BB, initiation of ciliogenesis can then proceed through the mutual recruitment of Cep290 and Dzip1-Cby-Fam92 module. This is supported by our previous observation that even the N-terminus of Cep290 alone, when properly localized, can promote cilliogenesis initiation [29]. This model provides a possible explanation for the wide range of pathological phenotypes associated with mutations in Cep290, ranging from isolated blindness to lethality, making it challenging to establish a clear genotype–phenotype correlation. We anticipate that particular developmental and cellular context, or the presence of genetic modifiers may introduce variability in cilium assembly or function by affecting various complexes required to stabilize Cep290 at the TZ. Our work not only uncovers the molecular mechanism of the synthetic interaction between Cby-Fam92 module and Cep131-Cep162 module in Cep290 recruitment and ciliogenesis initiation, but also reveals a novel molecular function of Cep131 in ciliogenesis, which recruits Cep162 to promote the binding of the C-terminus of Cep290 to the axoneme. We demonstrated that the Cep131-Cep162 module promotes Cep290 C-terminus binding to the axoneme, whereas Cby-Fam92 module promotes and stabilizes the localization of Cep290 N-terminus on the membrane, therefore, the concurrent mutation of these 2 modules collectively leads to the failure of Cep290 to localize to the TZ and blocks the initiation of ciliogenesis. Cep290 is an intriguing ciliopathy gene. More than 130 mutations have been identified, but their associated phenotypes can be dramatically different, suggesting that there may be genetic modifiers involved in the development of their phenotypes [32,51]. Therefore, identification of the modifier genes has become an important element to understand the pathogenesis of Cep290-related ciliopathies. While Cep131 and Cby have not yet been associated with ciliopathy, mutations in Cep162 and Fam92A have been reported to be linked to human cilia-related diseases [52,53]. Our observation of the synthetic defects in ciliogenesis initiation in cby; cep290ΔC double mutants suggests that Cby or Fam92 may be genetic modifiers of disease mutations in Cep290 C-terminus. Similarly, it is possible that Cep131 or Cep162 may serve as genetic modifier for disease mutations in Cep290 N-terminus. Materials and methods Fly stocks Transgenic flies of Cby::GFP, Cep131::GFP, Cep290-N::GFP, Cep290-C::GFP, CG6652::GFP, Mks1::GFP, Mks6::GFP, and nompC-Gal4;UAS::GFP were previously reported [29]. Cep162::GFP, Cep162-N::GFP and Cep162-C::GFP transgenic flies were generated in this paper. The cDNA of Cep162 or Cep162 C-terminus and its endogenous promoter were cloned and inserted into the PJFRC2 vector, and the plasmids were then used to construct transgenic flies by the Core Facility of Drosophila Resources and Technology, Shanghai Institute of Biochemistry and Cell Biology (SIBCB), Chinese Academy of Sciences (CAS). w1118 flies were used as wild type. cep290ΔC, cep2901, cby and fam92 mutant flies have been described previously [29]. All experiments were conducted at 25°C. Generation of deletion mutants in Drosophila Generation of cep131 and cep162 mutants were performed as previously reported [29]. Briefly, the mutant for Cep131 and Cep162 were generated by the CRISPR/Cas9-mediated gene targeting system. The gRNA expression plasmids were generated by inserting the targeting sequences into the PU6-BbsI-chiRNA vector using PCR. To increase the efficiency of generating fragment deletion mutants that facilitate mutant identification, 2 gRNAs were injected together into Cas9-expressing embryos. Primers used to construct gRNA vector: Cep162 g1 F: 5′-GTCGCTCTGAAGGCGAACGGGTTTTAGAGCTAGAAATAGC-3′; Cep162 g1 R: 5′-CCGTTCGCCTTCAGAGCGACCGACGTTAAATTGAAAATAGG-3′; Cep162 g2 F: 5′-TCAGTATGCGCTCCATCTCGGTTTTAGAGCTAGAAATAGC-3′; Cep162 g2 R: 5′-CGAGATGGAGCGCATACTGACGACGTTAAATTGAAAATAGG-3′; Cep131 g1 F: 5′-CAAGCACAAGCCAGGACTGG GTTTTAGAGCTAGAAATAGC-3′: Cep131 g1 R: 5′-CCAGTCCTGGCTTGTGCTTGCGACGTTAAATTGAAAATAGG-3′: Cep131 g2 F: 5′-CTCCCTCTGCGAGAAGGTGGGTTTTAGAGCTAGAAATAGC-3′: Cep131 g2 R: 5′-CCACCTTCTCGCAGAGGGAGCGACGTTAAATTGAAAATAGG-3′. Primers used to identify mutants: Cep162 F: 5′-ATATTTTCGCGAGCTGAGGACAC-3′; Cep162 R: 5′-GCGATGTGAGTCTCATATTTGGC-3′; Cep131 F: 5′-CATCAGTGTGGGCAGCCTACG-3′; Cep131 R: 5′-GCGAATGCTAGTCTCGATCTGC-3′. Immunofluorescence For IF staining of antennae or testes, 36 to 48 h after puparium formation (APF), pupae were collected and their antennae or testes were dissected with forceps in PBS. Antennae or testes were transferred to the center of a coverslip, and then gently covered by a slide over the coverslip. The slide was dipped into liquid nitrogen for 30 s, and the coverslip was immediately removed with a blade. The specimens were then fixed using methanol (−20°C) for 15 min, followed by acetone (−20°C) for 10 min. To block nonspecific binding, the specimens were incubated for 1 h in blocking buffer (0.1% Triton X-100, 3% bovine serum albumen in PBS). The primary antibodies were applied overnight in a moisture chamber at 4°C, and then the secondary antibodies were applied for 3 h at room temperature. Antibodies The primary antibodies used were as follows: Rabbit anti-Dzip1 (aa 451–737), Rabbit anti-Cep290 (aa 292–541), and Rabbit anti-Fbf1 (aa 1–336) antibodies were generated at YOUKE Biotech, rabbit anti-GFP (1:500, ab290, Abcam), mouse anti-GFP (1:200, 11814460001, Roche), mouse anti-Ac-tub (1:500, T6973, Sigma-Aldrich), mouse anti-γ-Tubulin (1:500, T5326, Sigma-Aldrich), and mouse anti-21A6 (1:200, AB528449, DSHB). The following secondary antibodies were used: goat anti-mouse Alexa Fluor 488 (1:1,000, A-11001, Invitrogen), goat anti-rabbit Alexa Fluor 594 (1:1,000, A1000701, Invitrogen), goat anti-rabbit Alexa Fluor 488 (1:1,000, A-11006, Invitrogen), goat anti-mouse IgG1 Alexa Fluor 488 (1:1,000, A-21121, Invitrogen), goat anti-mouse Alexa Fluor 594 (1:1,000, A-11007, Invitrogen), goat anti-mouse Alexa Fluor 647 (1:1,000, A-21242, Invitrogen). Microscopy and image analysis For IF staining, images were taken on a fluorescence microscope (Nikon Ti) with a 100× (1.4 NA) oil-immersion objective, or a confocal microscope (Leica Stellaris 5) with a 63× (1.4 NA) oil-immersion objective, or the Delta Vision OMX SR (GE Healthcare) with a 60× (1.42 NA) oil-immersion objective. Confocal images were acquired as Z-stacks (0.5 to 0.8 μm for Z-step size and 3 to 5 for number of steps) using xzy scan pattern. The sections of 3D-SIM images were acquired at 0.125 μm Z-steps (20 steps) and the raw data were reconstructed by using softWoRx software (GE Healthcare). Images were quantified for the pixel density using ImageJ (National Institutes of Health). For quantification of the pixel density, images were taken using equal microscopy settings. The pixel density values were calculated by the sum pixel density values in a defined region subtracting the sum pixel density values in an area close to the defined region. All images assembled into figures using Photoshop (CS5, Adobe). Transmission electron microscopy Samples were prepared for electron microscopy as previously described [54]. Briefly, samples were incubated in 2.5% glutaraldehyde/0.2 M phosphate buffer on ice for 24 h, postfixed in 2% OsO4/0.1 M phosphate buffer on ice, dehydrated with ethanol, and embedded in epoxy resin. Selected areas were sectioned using an ultramicrotome. Ultrathin sections were stained with uranyl acetate and lead citrate, and examined with Hitachi H-7650 transmission electron microscopy at 80 kV. Negative geotaxis assay Virgin flies were collected and cultured in fresh medium for 3 to 5 days. A total of 50 flies were sorted into 5 measuring vials of 10 each, and then tapping flies to the bottom of the vials and counting the number of flies that climbed over the 8 cm high bar within 10 s. Each group was repeated 3 times. Larval hearing assay Third instar larvae were divided into 5 groups of 5, placed on an agar plate above the speaker, and stimulated with 1k Hz sound every 30 s, the number of larvae with contractile responds on the head or body within 1 s after stimulation were counted. Each group was repeated 5 times. Yeast two-hybrid assay Cep131-FL (1–1114 aa), Cep131-N1 (1–480 aa), Cep131-M1 (481–781 aa), Cep131-C1 (782–1114 aa), Cep162-N (1–447 aa), Cep162-C (448–897 aa), Cep290-N (1–887 aa), and Cep290-ΔN (888–1978 aa) were introduced into either pGBKT7 or pGADT7 vector. Clones in pGADT7 and pGBKT7 were transformed into yeast strain AH109 (Takara Bio). The yeasts were grown on SD-Leu-Trp plates at 30°C. After 3 to 4 days incubation, 2 independent positive colonies were picked and diluted with TE buffer (10 mM Tris-HCl, 1 mM EDTA (pH 7.5)), and then transferred to SD-Ade-Leu-Trp-His or SD-Leu-Trp-His plates with 3-amino-1,2,4-triazole and incubated for 5 days at 30°C. GST pull-down assay To generate bacterial expression plasmids for His-Cep162-N (1–447 aa), His-Cep162-C (448–897 aa), His-Cep131-N (1–549 aa), His-Cep131-C (550–1114 aa), the cDNA fragments encoding the indicated amino acids were amplified by PCR and subcloned into the pET28a or pGEX-4 T-1 vector. The proteins were expressed using BL21 (DE3) E. coli. strain with IPTG induction and purified with Glutathione-agarose beads or Ni-resin (Yeasen). Purified His-Cep131 truncations were incubated with immobilized GST or GST-Cep162 truncations in the binding buffer (25 mM Tris-HCl at pH 7.4, 150 mM NaCl, 0.5% Triton X-100, 1 mM dithiothreitol, 10% glycerol, and protease inhibitors) at 4°C for 4 h. After incubation, the beads were washed 3 times with washing buffer (binding buffer with 20 mM imidazole) and then boiled for 10 min in 1 × SDS loading buffer. The protein samples were then separated by SDS-PAGE gels and transferred to the PVDF membrane for either immunoblotting with His antibody (Invitrogen) or staining with Ponceau S. Primary antibodies were used at a dilution of 1/1,000, and secondary antibodies were used at a dilution of 1/2,000. All uncropped images can be found in S1 Raw Image. Statistics Data were analyzed and graphed using Microsoft Excel or Graphpad Prism. Unless otherwise indicated, all error bars represent the standard deviation (SD) of the mean, and the statistical significance between data was assessed with an unpaired two-tailed Student’s T tests. Differences between data were considered statistically significant when P ≤ 0.05. Fly stocks Transgenic flies of Cby::GFP, Cep131::GFP, Cep290-N::GFP, Cep290-C::GFP, CG6652::GFP, Mks1::GFP, Mks6::GFP, and nompC-Gal4;UAS::GFP were previously reported [29]. Cep162::GFP, Cep162-N::GFP and Cep162-C::GFP transgenic flies were generated in this paper. The cDNA of Cep162 or Cep162 C-terminus and its endogenous promoter were cloned and inserted into the PJFRC2 vector, and the plasmids were then used to construct transgenic flies by the Core Facility of Drosophila Resources and Technology, Shanghai Institute of Biochemistry and Cell Biology (SIBCB), Chinese Academy of Sciences (CAS). w1118 flies were used as wild type. cep290ΔC, cep2901, cby and fam92 mutant flies have been described previously [29]. All experiments were conducted at 25°C. Generation of deletion mutants in Drosophila Generation of cep131 and cep162 mutants were performed as previously reported [29]. Briefly, the mutant for Cep131 and Cep162 were generated by the CRISPR/Cas9-mediated gene targeting system. The gRNA expression plasmids were generated by inserting the targeting sequences into the PU6-BbsI-chiRNA vector using PCR. To increase the efficiency of generating fragment deletion mutants that facilitate mutant identification, 2 gRNAs were injected together into Cas9-expressing embryos. Primers used to construct gRNA vector: Cep162 g1 F: 5′-GTCGCTCTGAAGGCGAACGGGTTTTAGAGCTAGAAATAGC-3′; Cep162 g1 R: 5′-CCGTTCGCCTTCAGAGCGACCGACGTTAAATTGAAAATAGG-3′; Cep162 g2 F: 5′-TCAGTATGCGCTCCATCTCGGTTTTAGAGCTAGAAATAGC-3′; Cep162 g2 R: 5′-CGAGATGGAGCGCATACTGACGACGTTAAATTGAAAATAGG-3′; Cep131 g1 F: 5′-CAAGCACAAGCCAGGACTGG GTTTTAGAGCTAGAAATAGC-3′: Cep131 g1 R: 5′-CCAGTCCTGGCTTGTGCTTGCGACGTTAAATTGAAAATAGG-3′: Cep131 g2 F: 5′-CTCCCTCTGCGAGAAGGTGGGTTTTAGAGCTAGAAATAGC-3′: Cep131 g2 R: 5′-CCACCTTCTCGCAGAGGGAGCGACGTTAAATTGAAAATAGG-3′. Primers used to identify mutants: Cep162 F: 5′-ATATTTTCGCGAGCTGAGGACAC-3′; Cep162 R: 5′-GCGATGTGAGTCTCATATTTGGC-3′; Cep131 F: 5′-CATCAGTGTGGGCAGCCTACG-3′; Cep131 R: 5′-GCGAATGCTAGTCTCGATCTGC-3′. Immunofluorescence For IF staining of antennae or testes, 36 to 48 h after puparium formation (APF), pupae were collected and their antennae or testes were dissected with forceps in PBS. Antennae or testes were transferred to the center of a coverslip, and then gently covered by a slide over the coverslip. The slide was dipped into liquid nitrogen for 30 s, and the coverslip was immediately removed with a blade. The specimens were then fixed using methanol (−20°C) for 15 min, followed by acetone (−20°C) for 10 min. To block nonspecific binding, the specimens were incubated for 1 h in blocking buffer (0.1% Triton X-100, 3% bovine serum albumen in PBS). The primary antibodies were applied overnight in a moisture chamber at 4°C, and then the secondary antibodies were applied for 3 h at room temperature. Antibodies The primary antibodies used were as follows: Rabbit anti-Dzip1 (aa 451–737), Rabbit anti-Cep290 (aa 292–541), and Rabbit anti-Fbf1 (aa 1–336) antibodies were generated at YOUKE Biotech, rabbit anti-GFP (1:500, ab290, Abcam), mouse anti-GFP (1:200, 11814460001, Roche), mouse anti-Ac-tub (1:500, T6973, Sigma-Aldrich), mouse anti-γ-Tubulin (1:500, T5326, Sigma-Aldrich), and mouse anti-21A6 (1:200, AB528449, DSHB). The following secondary antibodies were used: goat anti-mouse Alexa Fluor 488 (1:1,000, A-11001, Invitrogen), goat anti-rabbit Alexa Fluor 594 (1:1,000, A1000701, Invitrogen), goat anti-rabbit Alexa Fluor 488 (1:1,000, A-11006, Invitrogen), goat anti-mouse IgG1 Alexa Fluor 488 (1:1,000, A-21121, Invitrogen), goat anti-mouse Alexa Fluor 594 (1:1,000, A-11007, Invitrogen), goat anti-mouse Alexa Fluor 647 (1:1,000, A-21242, Invitrogen). Microscopy and image analysis For IF staining, images were taken on a fluorescence microscope (Nikon Ti) with a 100× (1.4 NA) oil-immersion objective, or a confocal microscope (Leica Stellaris 5) with a 63× (1.4 NA) oil-immersion objective, or the Delta Vision OMX SR (GE Healthcare) with a 60× (1.42 NA) oil-immersion objective. Confocal images were acquired as Z-stacks (0.5 to 0.8 μm for Z-step size and 3 to 5 for number of steps) using xzy scan pattern. The sections of 3D-SIM images were acquired at 0.125 μm Z-steps (20 steps) and the raw data were reconstructed by using softWoRx software (GE Healthcare). Images were quantified for the pixel density using ImageJ (National Institutes of Health). For quantification of the pixel density, images were taken using equal microscopy settings. The pixel density values were calculated by the sum pixel density values in a defined region subtracting the sum pixel density values in an area close to the defined region. All images assembled into figures using Photoshop (CS5, Adobe). Transmission electron microscopy Samples were prepared for electron microscopy as previously described [54]. Briefly, samples were incubated in 2.5% glutaraldehyde/0.2 M phosphate buffer on ice for 24 h, postfixed in 2% OsO4/0.1 M phosphate buffer on ice, dehydrated with ethanol, and embedded in epoxy resin. Selected areas were sectioned using an ultramicrotome. Ultrathin sections were stained with uranyl acetate and lead citrate, and examined with Hitachi H-7650 transmission electron microscopy at 80 kV. Negative geotaxis assay Virgin flies were collected and cultured in fresh medium for 3 to 5 days. A total of 50 flies were sorted into 5 measuring vials of 10 each, and then tapping flies to the bottom of the vials and counting the number of flies that climbed over the 8 cm high bar within 10 s. Each group was repeated 3 times. Larval hearing assay Third instar larvae were divided into 5 groups of 5, placed on an agar plate above the speaker, and stimulated with 1k Hz sound every 30 s, the number of larvae with contractile responds on the head or body within 1 s after stimulation were counted. Each group was repeated 5 times. Yeast two-hybrid assay Cep131-FL (1–1114 aa), Cep131-N1 (1–480 aa), Cep131-M1 (481–781 aa), Cep131-C1 (782–1114 aa), Cep162-N (1–447 aa), Cep162-C (448–897 aa), Cep290-N (1–887 aa), and Cep290-ΔN (888–1978 aa) were introduced into either pGBKT7 or pGADT7 vector. Clones in pGADT7 and pGBKT7 were transformed into yeast strain AH109 (Takara Bio). The yeasts were grown on SD-Leu-Trp plates at 30°C. After 3 to 4 days incubation, 2 independent positive colonies were picked and diluted with TE buffer (10 mM Tris-HCl, 1 mM EDTA (pH 7.5)), and then transferred to SD-Ade-Leu-Trp-His or SD-Leu-Trp-His plates with 3-amino-1,2,4-triazole and incubated for 5 days at 30°C. GST pull-down assay To generate bacterial expression plasmids for His-Cep162-N (1–447 aa), His-Cep162-C (448–897 aa), His-Cep131-N (1–549 aa), His-Cep131-C (550–1114 aa), the cDNA fragments encoding the indicated amino acids were amplified by PCR and subcloned into the pET28a or pGEX-4 T-1 vector. The proteins were expressed using BL21 (DE3) E. coli. strain with IPTG induction and purified with Glutathione-agarose beads or Ni-resin (Yeasen). Purified His-Cep131 truncations were incubated with immobilized GST or GST-Cep162 truncations in the binding buffer (25 mM Tris-HCl at pH 7.4, 150 mM NaCl, 0.5% Triton X-100, 1 mM dithiothreitol, 10% glycerol, and protease inhibitors) at 4°C for 4 h. After incubation, the beads were washed 3 times with washing buffer (binding buffer with 20 mM imidazole) and then boiled for 10 min in 1 × SDS loading buffer. The protein samples were then separated by SDS-PAGE gels and transferred to the PVDF membrane for either immunoblotting with His antibody (Invitrogen) or staining with Ponceau S. Primary antibodies were used at a dilution of 1/1,000, and secondary antibodies were used at a dilution of 1/2,000. All uncropped images can be found in S1 Raw Image. Statistics Data were analyzed and graphed using Microsoft Excel or Graphpad Prism. Unless otherwise indicated, all error bars represent the standard deviation (SD) of the mean, and the statistical significance between data was assessed with an unpaired two-tailed Student’s T tests. Differences between data were considered statistically significant when P ≤ 0.05. Supporting information S1 Fig. Identification and phenotype analysis of cep131 mutants. (A) Diagram showing of the generation of cep131 mutants. Schematics show the genomic (upper panel) and protein (lower panel) structures of Cep131, along with the predict protein product of Cep1311 mutant (Cep1311_p.(His480_Leu708delfsTer61)). Two arrows represent gRNA target sites. cep1311, a frameshift line, has a deletion in cDNA from nt 1439 to 2123, resulting in a reading frame shift and C-terminus loss. (B) Genotyping of cep131 mutants using PCR. The PCR amplification products were 1471 bp long for w1118 and 603 bp long for cep131 flies. (C) Sequence confirmation of the deletion in cep131 mutant. Primers used for sequence are marked with orange frames. The locations of 2 gRNAs used for mutant generation are underlined with black. Red and blue frames label the boundary of deletion cep131 in mutant. (D) Immunostaining of CG6652 in spermatocyte cilia of WT or cep1311 testis. CG6652 (green) marks the ciliary axoneme. In cep1311, a few centrioles have over elongated CG6652 signals (arrows). Centriole/basal body is marked with γ-Tubulin (red). Scale bars, 5 μm. (E) Quantification of the radial distance of Cep290 or Cep290-N. In cep131 mutants, the radial distance of Cep290 or Cep290-N were significantly reduced compared to WT. The error bars represent the mean ± SD, n = 60. The data underlying this figure can be found in S1 Data. https://doi.org/10.1371/journal.pbio.3002330.s001 (TIF) S2 Fig. Yeast two-hybrid assay of the interaction between Cep290 and Cep162 or Cep131. (A) In Y2H assay, Cep290 interacts with Cep162 (CG42699), but not Cep131 in Drosophila. LW: Selective media SD-Leu-Trp plates. LWH: Selective media SD-Leu-Trp-His plates. LWHA: SD-Ade-Leu-Trp-His plates. (B) The GST pull-down assay confirmed the direct interaction between Cep131 and Cep162 in vitro. GST, GST-Cep162-N, and GST-Cep162-C recombinant proteins were pulled down with His-Cep131-N or His-Cep131-C proteins. (C) In Y2H assay, Cep162 interacts with Cep131-N1 and Cep131-C1, but not Cep131-M1. LW: Selective media SD-Leu-Trp plates. LWH: Selective media SD-Leu-Trp-His plates. LWHA: SD-Ade-Leu-Trp-His plates. https://doi.org/10.1371/journal.pbio.3002330.s002 (TIF) S3 Fig. Protein sequence alignment of the C-terminus of human Cep162, mice Cep162, and Drosophila CG42699. CG42699 is the only one significant alignment sequence when searching for homology of human or mice Cep162 protein in Drosophila melanogaster. https://doi.org/10.1371/journal.pbio.3002330.s003 (TIF) S4 Fig. Cep162 is required for BBs docking to the plasma membrane in spermatocytes. (A) Sequence confirmation of the deletion in cep162 mutant. Primers used for sequence are marked with orange frames. The locations of 2 gRNAs used for mutant generation are underlined with black. Red and blue frames label the boundary of deletion cep162 in mutant. (B) Genotyping of cep162 flies using PCR. The amplification products were 881 bp long for WT and 555 bp long for mutants. (C) Immunostaining of CG6652 in spermatocyte cilia of WT or cep1621 testis. CG6652 (green) marks the ciliary axoneme. In cep1621, a few centrioles have over elongated CG6652 signals (arrows). (D) Live imaging of the connection between the ciliary cap and the plasma membrane in WT flies and cep162 mutants. The plasma membrane (PM) was labeled with CellMask, and the BBs were marked by UNC-GFP. Defective connection between the BBs and the membrane was observed in some spermatids of cep1311 mutant (arrows). (E) Spermatocytes showing abnormal acetylated-tubulin extensions in centriole/BBs in cep162; cep2901 mutants. Scale bars, 5 μm (C, E), 10 μm (D). https://doi.org/10.1371/journal.pbio.3002330.s004 (TIF) S5 Fig. Synthetic defects in Cep290 localization in cep162; fam92 or fam92; cep290ΔC mutants. Cep290 are absent from the TZ in the spermatocyte of cep162; fam92 or fam92; cep290ΔC mutants. The BB was labeled with γ-Tubulin (red). The error bars represent the mean ± SD, n = 30. Scale bars, 2 μm. The data underlying this figure can be found in S1 Data. https://doi.org/10.1371/journal.pbio.3002330.s005 (TIF) S1 Raw Image. Full scans of western blots. https://doi.org/10.1371/journal.pbio.3002330.s006 (TIF) S1 Data. Raw data underlies the figures. https://doi.org/10.1371/journal.pbio.3002330.s007 (XLSX) Acknowledgments We thank Dr. Wei Zhang from Tsinghua University for strains. We thank core facilities of Drosophila Resource and Technology (SIBCB, CAS), confocal imaging core facilities (SIAT, CAS), and EM core facilities (SIPPE, CAS) for their technical support.

Wikman, Patrik;Salmela, Viljami;Sjöblom, Eetu;Leminen, Miika;Laine, Matti;Alho, Kimmo

doi: 10.1371/journal.pbio.3002534pmid: 38466713

Introduction Humans effortlessly recognise and separate auditory objects in complex sound environments. This ability relies on hierarchical neural processing in the auditory ventral “what” stream, where sequential processing stages extract and integrate increasingly complex object attributes [1,2]—starting with processing of simple features (e.g., frequency) in the primary auditory cortex, progressing to complex acoustic structures (e.g., frequency-modulated sweeps) in secondary areas and selectivity for complete auditory objects in the anterior superior temporal cortex [3–6]. The ventral stream terminates in the anterior temporal and inferior frontal cortex where sound category and semantic information is apparently stored [7–9]. In the absence of spatial cues, there are usually only subtle differences in the vocal attributes that separate concurrent speakers from each other [10]. Therefore, top-down modulation facilitated by selective attention plays a significant role in separating relevant speech objects from irrelevant background speech [10,11]. This top-down modulation is classically assumed to enhance the gain [12–15] or the accuracy [16–18] of responses in neuronal populations processing the relevant sounds. More intricate theories suggest that attention also affects predictive mechanisms in sensory cortices [19] or that attentional modulation arises as neural networks adapt to specific tasks in various contexts [20–23]. Recent methodological advances in electrocorticography (ECoG) [15,24–27], magnetoencephalography (MEG) [11,28], and electroencephalography (EEG) [25,29,30] have revealed that attention enhances neuronal tracking of speech sounds. This amplification is concordant with modulation of both early (i.e., within 100 ms; e.g., [31]) and late (after 100 ms; e.g., [32,33]) neural response curves to sound envelope changes, consistent with the view that selective attention shifts neuronal processing in low-level auditory and higher-level speech-sensitive regions towards the features of the attended speaker [24,31,34,35]. These methods, however, lack spatial precision. That is, ECoG studies are limited by the extent of the implanted electrodes, while MEG/EEG source localisation is relatively inaccurate especially in the case of simultaneously firing neuronal populations [36]. In contrast, functional magnetic resonance imaging (fMRI) provides better spatial resolution, revealing that selective attention to cocktail-party speech modulates information processing in not only low-level auditory regions but also in extensive superior temporal, inferior parietal, and inferior frontal brain regions (e.g., [37–42]). Furthermore, multivariate pattern analyses on cocktail party fMRI data have indicated that neuronal populations that show differential responses during selective attention to speech are distributed globally in disparate cortical regions [38,43]. Yet, fMRI has limitations in estimating the timing of these modulations. Therefore, some fMRI studies have employed a combination of language modelling and multivariate analysis of fMRI responses to address the temporal limitations of fMRI when tracking continuous speech [43,44]. However, here we opted for a different approach by utilising EEG-fMRI fusion [45,46]. This technique allows us to overcome the spatial limitations of EEG and the temporal constraints of fMRI, enabling us to estimate the spatiotemporal characteristics of selective attention to audiovisual (AV) speech. In the present paradigm, participants watched video clips of dialogues between 2 speakers (dialogue stream) with a distracting speech stream played in the background (background stream; Fig 1A). To increase attentional demands, we modulated the auditory quality of the dialogue stream with noise-vocoding [47] and visual quality in the videos by masking [48]. We also modulated the semantic coherence of the dialogue stream (Fig 1B and 1C). We employed a fully factorial design where participants performed 2 different tasks: (1) attend speech task, where the participants attended to the AV dialogue while ignoring the background speech; and (2) ignore speech task, where the participants ignored both the dialogue and the background speech, and instead counted rotations of a white cross presented visually near the mouth of either speaker. This enabled us to study the effect of selective attention (difference between the attend speech task and the ignore speech task) on both the relevant speech stream (dialogue stream) and the irrelevant background (background stream). We expected that attending to the dialogues would increase speech envelope reconstruction (SER) accuracy of the dialogue stream and enhance related early and late neural temporal response components, with opposite effects for the background speech stream. Based on results from our previous fMRI study [38], attentional modulation of the dialogue stream SER accuracy was expected to be temporally variable, changing from line-to-line in a nonlinear fashion. Additionally, we expected SER accuracy to be greatest for dialogues with good audiovisual quality [49,50] and coherent semantics [51]. However, we also wanted to determine whether this applies only to attended speech. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 1. The AV cocktail party paradigm. (A) Participants underwent either EEG (n = 19) or fMRI (n = 19) recordings while watching AV video clips of dialogues consisting of 7 lines (dialogue stream) with a continuous audiobook (background stream) played in the background. Participants performed 2 tasks: (1) an attend speech task where they attended to the dialogue while ignoring background speech; and (2) ignore speech task where they ignored all speech and counted rotations of a cross presented below the neck of the talker. Dialogues were either semantically coherent or incoherent (B), and the audio quality varied with different levels of noise-vocoding (C). Additionally, visual quality was manipulated with dynamic white noise masking (D). AV, audiovisual; EEG, electroencephalography; fMRI, functional magnetic resonance imaging. https://doi.org/10.1371/journal.pbio.3002534.g001 Using SER on the EEG data (Fig 2), we replicated earlier findings that neuronal tracking is amplified for attended speech (see e.g., [29]). Importantly, however, we found that this amplification was not temporally uniform: the tracking amplitude abated linearly with the procession of the spoken line. Further, neuronal tracking of attended speech displayed nonlinear fluctuations over the course of the dialogue, similar to those previously reported with fMRI [38]. We discuss how such temporal dynamics may arise due to interactions between prediction and attention and other nonlinear plastic effects in speech processing circuits [19,20]. To evaluate the minute temporal modulation of selective attention, we estimated neural temporal response functions (TRFs) for the EEG data separately for both speech streams (Fig 3). Finally, we performed EEG-fMRI fusion: Based on representational similarity analysis (RSA), we identified brain regions in the fMRI data that contained representational structures similar to those calculated from TRFs, resulting in a TRF-fMRI correlation time series for each brain region (Fig 4 and S1 and S2 Videos, www.mv.helsinki.fi/home/jkaurama/vdialog/, www.mv.helsinki.fi/home/jkaurama/vbook/). This analysis indicated that attention facilitates recurrent feedforward-feedback loops in the ventral processing stream (see [2]). Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 2. Schematic illustration of SER and SER results. (A) Participants heard and saw AV dialogues with overlapping background speech, i.e., a mixed auditory signal. SER was employed to assess neural tracking of the dialogue and background stream. First, we extracted the amplitude envelope for both speech streams. Then, using data from all 128 EEG channels, we separately reconstructed the amplitude envelopes for the dialogue and background stream. To assess the accuracy of neural tracking, we correlated the reconstructed speech with its corresponding envelope and compared this to correlations with the opposite envelope. Accuracy values in B and D represent Δr (r-difference scores) between direct correlations and across-reconstruction correlations. (B) SER accuracy exhibited a significant linear temporal decrease within each line of the attended dialogue stream. (C) Our prior fMRI study [38] demonstrated that attention-related modulation changed from line-to-line in a nonlinear fashion (the red coloured regions, which we named primary speech network in our previous study), the other colours indicate networks where this temporal modulation effect showed another pattern (see [38] for details). (D) SER accuracy displayed a similar nonlinear temporal pattern as fMRI (C), but specifically for the attended speech. This trend was observed in both univariate SER accuracy analysis (left) and multivariate SVM decoding (middle; details in “Decoding analysis of SER accuracies”). Participants’ SER accuracy was predicted based on their behavioural performance for the attended dialogue stream (right), and this prediction (beta-weight) inversely followed SER accuracy. Error bars indicate ± SEMs. Code and processed EEG data used to generate this figure are archived on the Open Science Framework; HTTPS://DOI.ORG/10.17605/OSF.IO/AGXTH. Data frames are available in S1 Data. AV, audiovisual; EEG, electroencephalography; fMRI, functional magnetic resonance imaging; SER, speech envelope reconstruction; SVM, support vector machine. https://doi.org/10.1371/journal.pbio.3002534.g002 Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 3. Schematic of TRF estimation and TRF results. (A) TRFs were estimated using the same speech amplitude envelopes as in our SER analysis, separately for the dialogue and background streams. (B) Average TRFs over frontocentral electrodes, with points indicating significant differences between the 2 TRFs (paired permutation t test df = 18, note the 2 streams have separate y-scales). (C) Left: RDMs were constructed using TRFs for all 16 conditions (first the 8 attend speech conditions and thereafter the 8 ignore speech conditions). This involved pairwise correlations for each condition combination at each time point across EEG channels. The upper left corner shows the average TRF RDMs for both dialogue and background streams. The plot in the left corner displays the correlation between an attentional task model (attend speech vs. ignore speech, att. vs. ign.) and the 2 TRF RDM time series, with significant points displayed below the plot (FDR corrected, one-sample t test, df = 19). Right: Similar to TRFs, fMRI RDMs were constructed using searchlight SVM decoding across the 16 conditions, resulting in voxel-specific RDMs. Regions with above-average correlations between the attentional task model and fMRI RDMs are displayed (HPC parcellation). Shading indicates ± SEM. Code and processed EEG and fMRI data used to generate this figure are archived on the Open Science Framework; HTTPS://DOI.ORG/10.17605/OSF.IO/AGXTH. EEG, electroencephalography; RDM, representational dissimilarity matrix; SER, speech envelope reconstruction; SVM, support vector machine; TRF, temporal response function. https://doi.org/10.1371/journal.pbio.3002534.g003 Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 4. Schematic illustration of TRF-fMRI fusion and results. TRFs were separately estimated for the dialogue (upper part) and background streams (lower part) for each combination of semantic coherence and audiovisual quality and EEG channel. Average TRFs are displayed for frontocentral electrodes in the middle column (attend speech: red, ignore speech: blue). We constructed TRF RDMs for each time point by correlating each EEG channel TRF pairwise across conditions and participants. Similar fMRI RDMs were constructed based on SVM decoding between the 16 condition pairs from fMRI data. Thus, we constructed similar RDMs for the EEG and the fMRI, allowing us to fuse information from both datasets by correlating vectorised TRF RDMs with fMRI RDMs, controlling for task and opposite speech stream TRF RDMs (see Fig 3C). To identify fMRI activations which corresponded to TRF RDMs at different time points, we conducted one-sample t tests (df = 18, FDR corrected) averaged across the HCP parcellation ROIs. Six time points of this TRF-fMRI RSA analysis are displayed for both the dialogue (upper part) and background streams (lower part) on the right side of the figure. For the full-time series, refer to S1 and S2 Videos. Code and processed EEG and fMRI data used to generate this figure are archived on the Open Science Framework; HTTPS://DOI.ORG/10.17605/OSF.IO/AGXTH. EEG, electroencephalography; fMRI, functional magnetic resonance imaging; RDM, representational dissimilarity matrix; ROI, region-of-interest; RSA, representational similarity analysis; SVM, support vector machine; TRF, temporal response function. https://doi.org/10.1371/journal.pbio.3002534.g004 Results Attentional modulation of speech envelope reconstruction accuracy fluctuates during the course of the dialogue We used the accuracy of SER to study how selective attention and our other experimental manipulations affected the neuronal tracking of AV cocktail party speech. We employed a 2 × 2 × 2 × 2 within-subjects factorial design, where participants performed 3 runs including all possible combinations of the Attentional Task (attend, ignore), Auditory Quality (good, poor), Visual Quality (good, poor), and Semantic Coherence (coherent, incoherent). To control for stimulus effects, each dialogue/background speech segment (across all conditions and runs) was unique (i.e., each was heard only once). The condition order and the dialogue allocated to each condition varied between participants (see “Procedure”). Because the dialogue stream comprised audiovisual speech, while the background speech comprised purely auditory speech, spoken by a speaker different from the ones having the dialogue, the main comparisons were conducted separately within speech streams. In brief, multidimensional transfer functions were estimated based on all EEG channels for the dialogue streams and the background streams separately for each combination of Attentional Task, Semantic Coherence, Auditory Quality, Visual Quality, Line of the speech stream (1–7), and Segment of the line (1–4). Thereafter, the accuracy of the speech reconstruction was assessed by correlating the reconstruction with its corresponding speech envelope and correcting for spurious correlations (see “First-level analysis of EEG-data” and “Univariate analysis of EEG data” for details, and Fig 2A). The strength of the correlation between the SER and its corresponding speech envelope is generally assumed to reflect the accuracy of neuronal entrainment to the input speech [24]. We used linear mixed models to assess the effects of the repeated factors Attentional Task, Semantic Coherence, Auditory Quality, and Visual Quality on SER accuracy, separately for the dialogue stream and the background stream. SER accuracy for the dialogue stream was significantly modulated by Attentional Task (F1,18.7 = 67.2, p < 0.001, η2 = 0.78). That is, SER accuracy was higher in the task where participants attentively listened to the dialogue streams (mean Δr = 0.14, SEM = 0.004) than in the task where they ignored the dialogue stream (mean Δr = 0.06, SEM = 0.005). This is in line with previous studies that have shown that selective attention to a specific speech stream strongly increases neuronal tracking of that speech stream compared to the ignored speech streams [29,51,52]. Please refer to S1 Text and S1 and S2 Figs for all other significant effects in these linear mixed models and their correspondences to the behavioural performance results (e.g., effects related to Semantic Coherence, Auditory, and Visual Quality). In our previous study utilising fMRI, we reported that in the auditory cortex attention-related modulations changed in a linear-quadratic fashion during the dialogue (i.e., increased in the beginning of the dialogue and abated thereafter; see Fig 2C, [38]). Neuronal tracking has been suggested to be most strongly affected by neuronal processes in the superior temporal cortex [29], and thus we expected similar temporal effects here. However, unlike with fMRI, here we could assess whether the previously reported temporal modulations were due to the processing of the attended or the ignored speech stream because they are separable in the EEG data. Further, utilising the EEG data in the present study also allowed us to evaluate whether attentional modulation changes within each line in a similar fashion as between the lines, which was not possible in our previous study due to the temporal resolution of fMRI. Thus, first we examined within-line effects, i.e., whether SER accuracy changed within the line (lines were divided into 4 equal length segments). As seen in Fig 2B, there was a significant linear decrease in SER accuracy throughout the line for the dialogue stream when it was attended, and to some extent for the dialogue stream when it was ignored but not for any other combinations of Attentional Task and Speech Stream (Fig 2B; significant Condition × Segment interaction, F9,116 = 11.2, p < 0.001, η2 = 0.46; linear mixed model with the repeated factors Condition (attend speech task dialogue stream; attend speech task background stream; ignore speech task dialogue stream; ignore speech task background stream) and Segment (1–4)). Next, we analysed whether attention changed SER accuracy from line-to-line in a similar nonlinear fashion as previously seen in our fMRI study [38]. As seen in Fig 2D (left), SER accuracy showed a similar temporal profile changing from line-to-line as previously observed with fMRI. In other words, SER accuracy increased during the first lines of the dialogue and abated towards the end. Further, this temporal effect was only evident when the participants selectively attended to the dialogue stream (Fig 2D left; significant Condition × Line Number interaction, F18,137.3 = 2.4, p < 0.002, η2 = 0.24; linear mixed model with the repeated factors Condition (attend speech task dialogue stream; attend speech task background stream; ignore speech task dialogue stream; ignore speech task background stream) and Line Number (1–7)). Next, we considered the possibility that the slow temporal effects (i.e., line-to-line effects) we found in the SER accuracy data were only evident when analysing SER accuracies separately for each speech stream. That is, it might be that weaker neuronal tracking of the dialogue stream causes a similar concordant change in the neural tracking of the background stream, and thus the contrast between the 2 streams remained constant throughout the dialogues. Therefore, we performed a multivariate analysis that integrated information from both the dialogue stream and the background stream. Specifically, we assessed whether classification of trials as belonging to the attend speech task or the ignore speech task (using the SER correlations for the dialogue streams and the background streams as input) changed over the course of the dialogue (for details, see “Decoding analysis of SER accuracies”). This analysis revealed that decoding accuracy changed in a similar fashion as SER accuracies of the attended dialogue stream alone (Fig 2D, middle; linear mixed model with Line Number as the repeated factor and decoding accuracy as the outcome, F6,25.6 = 2.7, p < 0.03, η2 = 0.39). Previous studies have shown that SER accuracy correlates positively with behavioural performance [29]. Therefore, attentional lability [53] during different parts of the dialogue could be considered a simple explanation for the slow temporal modulations. This attentional lability should also be observed in the behavioural performance. However, there was no significant change in the behavioural performance in the attend speech task across the lines of the dialogue (generalised linear model with Line Number as a repeated factor; χ26 = 9.8, p > 0.13, see also S2 Fig right, and [38]). Furthermore, unlike previous reports [29], we found no significant general association between SER accuracy in the attend speech task (dialogue stream) and behavioural performance. However, we found that the association between performance and SER accuracy changed over the lines of the dialogue (Fig 2D right, linear mixed model with Line Number as a repeated factor; F6, 28.6 = 2.9, p < 0.02, η2 = 0.37). This temporal profile was inverse to the temporal profile of SER accuracy. That is, behavioural performance was negatively associated with the lines that showed the highest SER accuracy and positively associated with the lines that showed the lowest accuracy. Attention modulates temporal response functions of the attended and ignored speech streams Speech reconstruction analysis has the advantage of maximising the power of finding effects in the EEG data, because it integrates information across channels and time points to estimate the optimal reconstruction of the sound stimulus. This, however, has the drawback of losing timing and location information in the neural signatures. Therefore, we also performed encoding modelling, separately for each combination of speech stream and listening condition (Fig 3A). In this model, the speech envelope was used as a regressor in a ridge regressor model, performed separately for data from each EEG channel (see “Univariate analysis of EEG data”). The output of this analysis is a TRF, which describes the convolution in time needed to translate the speech envelope into the EEG data. With some caveats, TRFs can be conceptualised as event-related potentials (ERPs) to a continuous variable, which here is the continuous speech amplitude envelope, and where the timescale refers to time lags to the speech signal (see “First-level analysis of EEG data”) [11]. The caveats are that TRFs are filtered more heavily than standard ERPs (we used passband of 0.5 to 10 Hz) and the choice of regularisation smears exact temporal information, and due to this, the estimated timing of neural events cannot be assumed as exact as for standard ERPs. We analysed whether selective attention modulates TRFs in frontocentral electrodes (optimal for picking up auditory cortex attention effects; e.g., [54,55]) separately for the dialogue stream (Fig 3B, left) and the background stream (Fig 3B, right). Selective attention significantly enhanced the TRFs for the dialogue stream (i.e., there was a significant main effect of Attentional Task, i.e., attend speech > ignore speech) and this effect was present at 2 intervals between 0 and 800 ms, first at ca. 50 to 100 ms and then at ca. 200 to 400 ms after sound envelope changes (paired permutation t tests, df = 18). This is consistent with previous ERP and TRF studies showing that attention modulates auditory processing of speech relatively early (i.e., within 100 ms), but that the strongest modulation is found at later time points [11,31,35,56–58]. Selective attention also changed both the timing and the amplitude of the TRF to the background stream (at ca. 50 to 200 ms; paired permutation t tests, df = 18). It is important to note that the background stream was ignored in all conditions. However, during the attend speech task, the participants had to actively suppress the background stream, while in the ignore speech task they focused on visual stimuli, designed to automatically keep attention away from all speech streams. Thus, since especially early components of the TRF response likely originate from the auditory cortex [35], it could be expected that the early components of the TRF to the background stream would be smaller for the attend speech task than the ignore speech task. However, our results indicated the reverse. This pattern might arise if participants had involuntary momentary lapses of attention to the wrong speech stream [11] during the attend speech task, causing enhancements also in the background speech stream. We find this unlikely, however, because such lapses would likely cause more variance in the background stream TRFs, rather than the change in amplitude seen in the present results. Furthermore, previous studies using the same paradigm [42] have found that in general, participants do not remember topics of the background stream. EEG TRF–fMRI fusion reveals that attention facilitates several feedforward-feedback loops related to the processing of cocktail-party speech Next, we performed multivariate RSA on the TRFs. TRFs were estimated for each condition (16 conditions, 8 attend speech task, 8 ignore speech task, for the exact order of the conditions, see “Multivariate analysis of TRFs”) and channel (128 channels) separately for the dialogue and the background streams. Thereafter, for each sample of the TRFs (128 Hz, ca. 8 ms samples), we performed pairwise correlations across the EEG channels for each condition pair to construct dissimilarity matrices (1-r). This resulted in 1 TRF representational dissimilarity matrix (RDM) for each time point of each speech stream. An RDM is a geometrical description of the data, showing an assembly of all pairwise dissimilarities across neural responses or model predictions to different stimuli or experimental conditions [59]. As can be seen in Fig 3C (upper left corner), especially in the dialogue stream, the attend speech task conditions are generally similar to each other and dissimilar to the ignore speech task conditions, i.e., there is an effect of Attentional Task. To test when this effect was significant, we constructed a model matrix for the main effect of Attentional Task (Fig 3C, upper left corner) and correlated this model with the TRF RDMs for each time point (one-sample t test, FDR corrected across time points; for other model correlations, see S3 Fig). This analysis revealed that selective attention modulated TRFs throughout almost the whole time range from 100 ms (after the sound envelope changes) onwards (Fig 3C, lower left corner). For the background stream, there were significant correlations with the attentional task model between 150 and 300 ms. We also performed the same RSA analysis on our fMRI data that used the same paradigm but different participants (the same fMRI data were used for the TRF-fMRI fusion, see below). Here, dissimilarity matrices were generated based on pairwise searchlight SVM decoding between the 16 conditions (see “Decoding analysis on the fMRI data” for details). Fig 3C (lower right) shows the regions (averaged for each region of the human connectome project atlas, HCP parcellation [60]), where the correlation with the attentional task model was above average (i.e., r > 0.34, the threshold for significance was r > 0.08). This analysis shows that information that distinguishes the attend speech task from the ignore speech is contained globally in the brain (see also, [38]) which also probably partly explains why the attentional task model correlated with TRF RDM matrices throughout the time interval. To gain an understanding of how the TRF RDMs corresponded to the fMRI RDMs, we performed EEG-fMRI fusion [45,46], using TRF RDMs, estimated using the EEG data, and fMRI RDMs (see section “TRF-fMRI fusion” for details). This was achieved by correlating each TRF RDM with the fMRI RDMs averaged in each region-of-interest (ROI) from the HPC parcellation. Because the fMRI RDMs integrate the differences for both the dialogue and background stream, while the TRF RDMs separate these effects, we corrected the TRF-fMRI fusion for the TRF RDMs of the opposite speech stream. That is, the dialogue stream TRF-fMRI fusion was corrected for the background stream TRF RDMs and vice versa. We also corrected for the main effect of Attentional Task in the TRF-fMRI fusion analysis because this effect was global in both the TRF and the fMRI responses (see above; Fig 3B and 3C), and thus masks subtle differences between different regions. Thus, the main effect of Attentional Task (Fig 3B and 3C) does not contribute to the TRF-fMRI fusion, and the correspondences arise instead due to other more complicated correspondences between the EEG and fMRI RDMs (see below). As seen in Fig 4 (upper right corner) and S1 Video (www.mv.helsinki.fi/home/jkaurama/vdialog/) for the dialogue stream, the first significant correlations (one-sample t test, df = 18, FDR-corrected) between the TRFs and fMRI RDMs arose at ca. 16 ms after sound envelope changes in dorsolateral and dorsomedial frontal regions. Hereafter, correlations arose at ca. 30 ms in posterior auditory regions and slowly thereafter in anterior auditory cortical regions. After 150 ms correlations arose in the anterior temporal lobe and slowly spread (ca. 250 ms) back to auditory, frontal, and speech processing regions in an anterior–posterior fashion. A second anterior–posterior sweep in the auditory cortex occurred starting at ca. 450 ms. Note that the timing information in this analysis is derived entirely from the TRF data and the spatial information from the fMRI data. For the background stream (Fig 4, bottom right corner, S2 Video, www.mv.helsinki.fi/home/jkaurama/vbook/), there were initially correlations in the visual cortex. At ca. 140 ms, there were correlations in the dorsolateral frontal cortex and then at ca. 300 ms correlations arose in the auditory cortex moving in a posterior–anterior fashion. This effect may relate to suppression of the background stream (cf. [61]). The TRF-fMRI correlation patterns for the background stream were lateralised to the left, while for the dialogue stream they were bilateral, which is consistent with earlier neuroimaging studies [62]. It is important to note that these TRF-fMRI fusion patterns were not due to a main effect of Attentional Task (Att. versus ign.) since this effect was controlled for in the analysis. Furthermore, as can be seen in S3 Fig, no other main effect model or interaction model yielded FDR-corrected significant results. However, based on the uncorrected results (S3 Fig), it seems that the correlations were mostly influenced by interactions between Attentional Task (Att. versus ign.) and stimulus features. However, many of the correlation patterns in the TRF-fMRI fusion analysis likely arose due to idiosyncratic differences between the different task conditions at different time points of speech processing. For time series in different a priori regions of interest, please see S4 Fig. Attentional modulation of speech envelope reconstruction accuracy fluctuates during the course of the dialogue We used the accuracy of SER to study how selective attention and our other experimental manipulations affected the neuronal tracking of AV cocktail party speech. We employed a 2 × 2 × 2 × 2 within-subjects factorial design, where participants performed 3 runs including all possible combinations of the Attentional Task (attend, ignore), Auditory Quality (good, poor), Visual Quality (good, poor), and Semantic Coherence (coherent, incoherent). To control for stimulus effects, each dialogue/background speech segment (across all conditions and runs) was unique (i.e., each was heard only once). The condition order and the dialogue allocated to each condition varied between participants (see “Procedure”). Because the dialogue stream comprised audiovisual speech, while the background speech comprised purely auditory speech, spoken by a speaker different from the ones having the dialogue, the main comparisons were conducted separately within speech streams. In brief, multidimensional transfer functions were estimated based on all EEG channels for the dialogue streams and the background streams separately for each combination of Attentional Task, Semantic Coherence, Auditory Quality, Visual Quality, Line of the speech stream (1–7), and Segment of the line (1–4). Thereafter, the accuracy of the speech reconstruction was assessed by correlating the reconstruction with its corresponding speech envelope and correcting for spurious correlations (see “First-level analysis of EEG-data” and “Univariate analysis of EEG data” for details, and Fig 2A). The strength of the correlation between the SER and its corresponding speech envelope is generally assumed to reflect the accuracy of neuronal entrainment to the input speech [24]. We used linear mixed models to assess the effects of the repeated factors Attentional Task, Semantic Coherence, Auditory Quality, and Visual Quality on SER accuracy, separately for the dialogue stream and the background stream. SER accuracy for the dialogue stream was significantly modulated by Attentional Task (F1,18.7 = 67.2, p < 0.001, η2 = 0.78). That is, SER accuracy was higher in the task where participants attentively listened to the dialogue streams (mean Δr = 0.14, SEM = 0.004) than in the task where they ignored the dialogue stream (mean Δr = 0.06, SEM = 0.005). This is in line with previous studies that have shown that selective attention to a specific speech stream strongly increases neuronal tracking of that speech stream compared to the ignored speech streams [29,51,52]. Please refer to S1 Text and S1 and S2 Figs for all other significant effects in these linear mixed models and their correspondences to the behavioural performance results (e.g., effects related to Semantic Coherence, Auditory, and Visual Quality). In our previous study utilising fMRI, we reported that in the auditory cortex attention-related modulations changed in a linear-quadratic fashion during the dialogue (i.e., increased in the beginning of the dialogue and abated thereafter; see Fig 2C, [38]). Neuronal tracking has been suggested to be most strongly affected by neuronal processes in the superior temporal cortex [29], and thus we expected similar temporal effects here. However, unlike with fMRI, here we could assess whether the previously reported temporal modulations were due to the processing of the attended or the ignored speech stream because they are separable in the EEG data. Further, utilising the EEG data in the present study also allowed us to evaluate whether attentional modulation changes within each line in a similar fashion as between the lines, which was not possible in our previous study due to the temporal resolution of fMRI. Thus, first we examined within-line effects, i.e., whether SER accuracy changed within the line (lines were divided into 4 equal length segments). As seen in Fig 2B, there was a significant linear decrease in SER accuracy throughout the line for the dialogue stream when it was attended, and to some extent for the dialogue stream when it was ignored but not for any other combinations of Attentional Task and Speech Stream (Fig 2B; significant Condition × Segment interaction, F9,116 = 11.2, p < 0.001, η2 = 0.46; linear mixed model with the repeated factors Condition (attend speech task dialogue stream; attend speech task background stream; ignore speech task dialogue stream; ignore speech task background stream) and Segment (1–4)). Next, we analysed whether attention changed SER accuracy from line-to-line in a similar nonlinear fashion as previously seen in our fMRI study [38]. As seen in Fig 2D (left), SER accuracy showed a similar temporal profile changing from line-to-line as previously observed with fMRI. In other words, SER accuracy increased during the first lines of the dialogue and abated towards the end. Further, this temporal effect was only evident when the participants selectively attended to the dialogue stream (Fig 2D left; significant Condition × Line Number interaction, F18,137.3 = 2.4, p < 0.002, η2 = 0.24; linear mixed model with the repeated factors Condition (attend speech task dialogue stream; attend speech task background stream; ignore speech task dialogue stream; ignore speech task background stream) and Line Number (1–7)). Next, we considered the possibility that the slow temporal effects (i.e., line-to-line effects) we found in the SER accuracy data were only evident when analysing SER accuracies separately for each speech stream. That is, it might be that weaker neuronal tracking of the dialogue stream causes a similar concordant change in the neural tracking of the background stream, and thus the contrast between the 2 streams remained constant throughout the dialogues. Therefore, we performed a multivariate analysis that integrated information from both the dialogue stream and the background stream. Specifically, we assessed whether classification of trials as belonging to the attend speech task or the ignore speech task (using the SER correlations for the dialogue streams and the background streams as input) changed over the course of the dialogue (for details, see “Decoding analysis of SER accuracies”). This analysis revealed that decoding accuracy changed in a similar fashion as SER accuracies of the attended dialogue stream alone (Fig 2D, middle; linear mixed model with Line Number as the repeated factor and decoding accuracy as the outcome, F6,25.6 = 2.7, p < 0.03, η2 = 0.39). Previous studies have shown that SER accuracy correlates positively with behavioural performance [29]. Therefore, attentional lability [53] during different parts of the dialogue could be considered a simple explanation for the slow temporal modulations. This attentional lability should also be observed in the behavioural performance. However, there was no significant change in the behavioural performance in the attend speech task across the lines of the dialogue (generalised linear model with Line Number as a repeated factor; χ26 = 9.8, p > 0.13, see also S2 Fig right, and [38]). Furthermore, unlike previous reports [29], we found no significant general association between SER accuracy in the attend speech task (dialogue stream) and behavioural performance. However, we found that the association between performance and SER accuracy changed over the lines of the dialogue (Fig 2D right, linear mixed model with Line Number as a repeated factor; F6, 28.6 = 2.9, p < 0.02, η2 = 0.37). This temporal profile was inverse to the temporal profile of SER accuracy. That is, behavioural performance was negatively associated with the lines that showed the highest SER accuracy and positively associated with the lines that showed the lowest accuracy. Attention modulates temporal response functions of the attended and ignored speech streams Speech reconstruction analysis has the advantage of maximising the power of finding effects in the EEG data, because it integrates information across channels and time points to estimate the optimal reconstruction of the sound stimulus. This, however, has the drawback of losing timing and location information in the neural signatures. Therefore, we also performed encoding modelling, separately for each combination of speech stream and listening condition (Fig 3A). In this model, the speech envelope was used as a regressor in a ridge regressor model, performed separately for data from each EEG channel (see “Univariate analysis of EEG data”). The output of this analysis is a TRF, which describes the convolution in time needed to translate the speech envelope into the EEG data. With some caveats, TRFs can be conceptualised as event-related potentials (ERPs) to a continuous variable, which here is the continuous speech amplitude envelope, and where the timescale refers to time lags to the speech signal (see “First-level analysis of EEG data”) [11]. The caveats are that TRFs are filtered more heavily than standard ERPs (we used passband of 0.5 to 10 Hz) and the choice of regularisation smears exact temporal information, and due to this, the estimated timing of neural events cannot be assumed as exact as for standard ERPs. We analysed whether selective attention modulates TRFs in frontocentral electrodes (optimal for picking up auditory cortex attention effects; e.g., [54,55]) separately for the dialogue stream (Fig 3B, left) and the background stream (Fig 3B, right). Selective attention significantly enhanced the TRFs for the dialogue stream (i.e., there was a significant main effect of Attentional Task, i.e., attend speech > ignore speech) and this effect was present at 2 intervals between 0 and 800 ms, first at ca. 50 to 100 ms and then at ca. 200 to 400 ms after sound envelope changes (paired permutation t tests, df = 18). This is consistent with previous ERP and TRF studies showing that attention modulates auditory processing of speech relatively early (i.e., within 100 ms), but that the strongest modulation is found at later time points [11,31,35,56–58]. Selective attention also changed both the timing and the amplitude of the TRF to the background stream (at ca. 50 to 200 ms; paired permutation t tests, df = 18). It is important to note that the background stream was ignored in all conditions. However, during the attend speech task, the participants had to actively suppress the background stream, while in the ignore speech task they focused on visual stimuli, designed to automatically keep attention away from all speech streams. Thus, since especially early components of the TRF response likely originate from the auditory cortex [35], it could be expected that the early components of the TRF to the background stream would be smaller for the attend speech task than the ignore speech task. However, our results indicated the reverse. This pattern might arise if participants had involuntary momentary lapses of attention to the wrong speech stream [11] during the attend speech task, causing enhancements also in the background speech stream. We find this unlikely, however, because such lapses would likely cause more variance in the background stream TRFs, rather than the change in amplitude seen in the present results. Furthermore, previous studies using the same paradigm [42] have found that in general, participants do not remember topics of the background stream. EEG TRF–fMRI fusion reveals that attention facilitates several feedforward-feedback loops related to the processing of cocktail-party speech Next, we performed multivariate RSA on the TRFs. TRFs were estimated for each condition (16 conditions, 8 attend speech task, 8 ignore speech task, for the exact order of the conditions, see “Multivariate analysis of TRFs”) and channel (128 channels) separately for the dialogue and the background streams. Thereafter, for each sample of the TRFs (128 Hz, ca. 8 ms samples), we performed pairwise correlations across the EEG channels for each condition pair to construct dissimilarity matrices (1-r). This resulted in 1 TRF representational dissimilarity matrix (RDM) for each time point of each speech stream. An RDM is a geometrical description of the data, showing an assembly of all pairwise dissimilarities across neural responses or model predictions to different stimuli or experimental conditions [59]. As can be seen in Fig 3C (upper left corner), especially in the dialogue stream, the attend speech task conditions are generally similar to each other and dissimilar to the ignore speech task conditions, i.e., there is an effect of Attentional Task. To test when this effect was significant, we constructed a model matrix for the main effect of Attentional Task (Fig 3C, upper left corner) and correlated this model with the TRF RDMs for each time point (one-sample t test, FDR corrected across time points; for other model correlations, see S3 Fig). This analysis revealed that selective attention modulated TRFs throughout almost the whole time range from 100 ms (after the sound envelope changes) onwards (Fig 3C, lower left corner). For the background stream, there were significant correlations with the attentional task model between 150 and 300 ms. We also performed the same RSA analysis on our fMRI data that used the same paradigm but different participants (the same fMRI data were used for the TRF-fMRI fusion, see below). Here, dissimilarity matrices were generated based on pairwise searchlight SVM decoding between the 16 conditions (see “Decoding analysis on the fMRI data” for details). Fig 3C (lower right) shows the regions (averaged for each region of the human connectome project atlas, HCP parcellation [60]), where the correlation with the attentional task model was above average (i.e., r > 0.34, the threshold for significance was r > 0.08). This analysis shows that information that distinguishes the attend speech task from the ignore speech is contained globally in the brain (see also, [38]) which also probably partly explains why the attentional task model correlated with TRF RDM matrices throughout the time interval. To gain an understanding of how the TRF RDMs corresponded to the fMRI RDMs, we performed EEG-fMRI fusion [45,46], using TRF RDMs, estimated using the EEG data, and fMRI RDMs (see section “TRF-fMRI fusion” for details). This was achieved by correlating each TRF RDM with the fMRI RDMs averaged in each region-of-interest (ROI) from the HPC parcellation. Because the fMRI RDMs integrate the differences for both the dialogue and background stream, while the TRF RDMs separate these effects, we corrected the TRF-fMRI fusion for the TRF RDMs of the opposite speech stream. That is, the dialogue stream TRF-fMRI fusion was corrected for the background stream TRF RDMs and vice versa. We also corrected for the main effect of Attentional Task in the TRF-fMRI fusion analysis because this effect was global in both the TRF and the fMRI responses (see above; Fig 3B and 3C), and thus masks subtle differences between different regions. Thus, the main effect of Attentional Task (Fig 3B and 3C) does not contribute to the TRF-fMRI fusion, and the correspondences arise instead due to other more complicated correspondences between the EEG and fMRI RDMs (see below). As seen in Fig 4 (upper right corner) and S1 Video (www.mv.helsinki.fi/home/jkaurama/vdialog/) for the dialogue stream, the first significant correlations (one-sample t test, df = 18, FDR-corrected) between the TRFs and fMRI RDMs arose at ca. 16 ms after sound envelope changes in dorsolateral and dorsomedial frontal regions. Hereafter, correlations arose at ca. 30 ms in posterior auditory regions and slowly thereafter in anterior auditory cortical regions. After 150 ms correlations arose in the anterior temporal lobe and slowly spread (ca. 250 ms) back to auditory, frontal, and speech processing regions in an anterior–posterior fashion. A second anterior–posterior sweep in the auditory cortex occurred starting at ca. 450 ms. Note that the timing information in this analysis is derived entirely from the TRF data and the spatial information from the fMRI data. For the background stream (Fig 4, bottom right corner, S2 Video, www.mv.helsinki.fi/home/jkaurama/vbook/), there were initially correlations in the visual cortex. At ca. 140 ms, there were correlations in the dorsolateral frontal cortex and then at ca. 300 ms correlations arose in the auditory cortex moving in a posterior–anterior fashion. This effect may relate to suppression of the background stream (cf. [61]). The TRF-fMRI correlation patterns for the background stream were lateralised to the left, while for the dialogue stream they were bilateral, which is consistent with earlier neuroimaging studies [62]. It is important to note that these TRF-fMRI fusion patterns were not due to a main effect of Attentional Task (Att. versus ign.) since this effect was controlled for in the analysis. Furthermore, as can be seen in S3 Fig, no other main effect model or interaction model yielded FDR-corrected significant results. However, based on the uncorrected results (S3 Fig), it seems that the correlations were mostly influenced by interactions between Attentional Task (Att. versus ign.) and stimulus features. However, many of the correlation patterns in the TRF-fMRI fusion analysis likely arose due to idiosyncratic differences between the different task conditions at different time points of speech processing. For time series in different a priori regions of interest, please see S4 Fig. Discussion Our SER analyses on the EEG data replicated that selective attention enhances neural tracking of attended speech [25,29,30]. Similarly, we replicated that attending to a specific speech stream enhances its EEG TRFs, both at early latencies (ca. 30–150 ms, e.g., [31]) and later latencies (ca. 200–400 ms, e.g., [35]). These findings are consistent with the view that selective attention increases the contrast between attended speech and distracting speech through top-down neural signals, which propagate from higher-level cortical regions to sensory regions and serve to enhance the gain of neurons that process the relevant speech [12–14]. While this is a likely explanation for some of our observations, we find it highly unlikely that this model exhaustively explains how attention modulates sensory processing in the auditory cortex, which we will discuss below. Although the background stream was always ignored, TRFs for the background stream were both temporally expedited and amplified when participants listened to the dialogue stream compared to when both streams were ignored. Thus, it seems that selective attention not only enhances the processing of relevant speech but also modulates the processing of the actively ignored distracting speech (for similar findings, see [11,63]). Such modulations might reflect active suppression of auditory cortex neurons processing attributes of distracting speech, which has been suggested as a complementary mechanism to increase the contrast between attended sounds and ignored sounds [56,63–65]. Alternatively, the effects may reflect that early processing of the background stream cannot be suppressed when attending to speech [32], attention enhancements partially spread to the background stream [66] or attention fluctuates between the 2 streams [11]. As the background speech is affected by manipulations of the attended speech, future studies could do manipulations for both sound streams (dialogue and background streams) and alternate the focus of attention between the 2 streams. Later studies utilising source localisation and/or intracranial measurements could also reveal both the spatial and laminar attributes of these effects and the neural populations contributing to them. RSA of the EEG data for both the dialogue stream and the background stream revealed that selective attention strongly modulated TRFs at several latencies that have not been reported in previous studies. Note that unlike the univariate TRF analyses the RSA analyses utilised all EEG-channels and were thus expected to find significant patterns in the channels outside those usually studied in auditory attention studies (e.g., frontocentral electrodes). Corroborating this, RSA analyses on the fMRI data showed that attention modulated information processing in extensive cortical fields not limited to regions associated with speech processing or executive functions (see also [63]). Thus, these results cast doubt on models that highlight simple interactions between frontal and sensory neural networks as origins for selective attentional effects. Rather, our results suggest that selective attention modulates a multitude of different subprocesses widely distributed in the brain (see also [38]). Using RSA, we performed TRF-fMRI fusion, which showed that attentional modulation of information flow between sensory regions and higher-level regions displayed reliable spatial and temporal characteristics. The earliest modulations were found in the lateral, medial, and inferior frontal cortices at around 8 to 16 ms. This is consistent with earlier MEG source localisation of attentional effects on speech related TRFs [11] and might reflect preparatory signals biasing the attentional speech processing (e.g., when the quality of the sensory input is poor). Thereafter, information flow generally followed the ventral stream model [2], with information processing first being modulated in the secondary auditory cortex (around 30 ms), continuing anteriorly to the superior temporal cortex and finally to the anterior temporal lobe (at around 150 ms). At later latencies (after 200 ms), several back-propagating loops of information flow between the anterior temporal cortex, frontal cortex, and the auditory cortex can be discerned. This suggest that information flow during active processing of cocktail-party speech is associated with reverberant bidirectional (feedforward-feedback) informational flow from sensory regions to regions associated with semantic [8], syntactic [9], and executive functions [67], within the ventral processing stream. As previously mentioned, we found that attention enhanced the neural tracking of the attended speech. However, this modulation was not uniform in time, i.e., the SER accuracy linearly decreased within the line of the dialogue (ca. 5 s long). This type of decrease could be explained within a predictive coding framework [68], assuming that information accumulates as the line proceeds, which constrains prediction error (PE) in neural networks [51,69]. Importantly, however, we found that decreases in SER were most consistently observed for attended speech. Thus, if the SER temporal profile is explained by predictive mechanisms, such mechanisms seem to depend on selective attention. Indeed, some current models postulate that predictive coding mechanisms and selective auditory attention interact during attentive processing of sensory information (cf. [19,70,71]). In the model proposed by Schröger and colleagues [19], the attentional processing of relevant sounds is biased in the auditory cortex through recurrent loops, with higher-order processing networks establishing an “attentional trace” which maximally distinguishes the features of the attended sounds from the features of the irrelevant sounds. In this model, selective attention improves the precision and gain of PEs generated by neurons encoding the attended stimuli. These enhanced error signals are concurrently sent to regions at the higher level of the processing hierarchy, which in turn send stronger modulatory signals to lower levels of the hierarchy. Thus, attention may influence feedback/feedforward loops, which interact with, for example, the predictability of the input. This model seems to explain quite well the present linear decrease effects. The model also gives a framework for understanding our TRF-fMRI fusion results, suggesting that the recurrent feedforward/feedback loops reflect the propagation of PE from the lower level of the hierarchy to the next level, on the one hand, and correcting predictive signals from the higher level to the lower level, on the other. We also found that the strength by which selective attention enhances neural tracking of speech changes on a slow temporal scale (from line-to-line of the dialogue, Fig 2C). In contrast to the linear decrease seen within a line, the neural tracking first increased up to the middle of the dialogue, and thereafter decreased towards the end of the dialogue. Similar slow fluctuations of attentional effects have been previously described using fMRI [38,39] and behavioural experiments (e.g., [72]). From the predictive coding framework, it could be postulated that such a temporal profile would arise if the ability of attention to maximally increase the gain of PEs takes time to build up, causing an initial increase in SER. The subsequent decrease could be explained, as for the within-line effects, by predictions becoming more stable towards the end of the dialogue. This account, however, fails to explain why there is no indication of such a delay in facilitating attentional processes within the line. Furthermore, based on this account, behavioural performance would be expected to improve as the dialogue proceeds and the model of the heard speech becomes increasingly accurate. We did not, however, find any evidence for such changes in the behavioural performance data. Importantly, in our previous publication on the fMRI data, we reported similar slow temporal changes of attention-related modulations in the superior temporal cortex [38]. In that paper, we suggested that the temporal modulations arose due to recruitment of additional neuronal resources in speech networks that may aid in automatising speech processing. This account is based on the model proposed by Kilgard [20], originally used to explain why attention and plasticity initially recruit neurons in the sensory cortex, which after task automatisation no longer participate in the task. Several animal studies have shown that attentional tasks cause transient–persistent plastic changes in auditory neuronal response profiles (e.g., [73,74]). The conundrum, however, has been that some studies have indicated that behavioural performance accuracy persist after the original plastic changes have subsided [75]. Therefore, Kilgard proposed that when the task is initially learned, all possible neuronal networks that may be useful to solve the task at hand are recruited. Gradually, the unnecessary, less informative neuronal networks are pruned out, and the most efficient network ends up performing the task (sparse coding). Thus, the slow temporal profiles seen in the current study may reflect that in the sensory cortex, neurons that may help in building the attentional trace are initially recruited and subsequently pruned out to encode information in a maximally sparse manner. This account would also explain the present perplexing performance-SER association (Fig 2C). That is, we found that behavioural performance predicted SER accuracy negatively in the middle of the dialogue when SER accuracy was strongest and positively when accuracy was weakest. Thus, it may be that behavioural associations were negative in the middle of the dialogue because at this point, neuronal resources processing the speech may not necessarily help in performing the task, while towards the end of the dialogue, unnecessary units are pruned out and the association between SER and performance returns to positive. Models of the auditory system have generally overlooked how factors like attention and active tasks influence the processing of sounds in neural networks. This oversight relies on the premise that attention simply changes neuronal response gain. Our results, however, highlight that the enhanced neuronal tracking of attended speech is not necessarily uniformly associated with more accurate representation of the attended speech (see e.g., [29]) but changes as a function of time due to predictive and/or other nonlinear plastic mechanisms in sensory cortex. We argue that the approach to selective attention needs to be updated to reflect recent views on how cognition is organised in neural systems (see e.g., [76]). Instead of mechanistic models where higher-level networks enhance gain mechanisms in sensory neurons, attention could be modelled as a collection of temporally changing processes that route activity in distributed neural networks according to behavioural demands. These findings may offer key insights in improving dynamic computational models of selective attention in noisy conversational settings (see e.g., [77]). Current AI platforms struggle to match human listeners and deliver unsatisfactory performance. Later multi- and single unit recordings in the auditory cortex could test the hypothesis that attention both changes the gain of neuronal populations and initially recruit neuronal resources that may aid in the performance of the task that are later discarded due to optimisation of task performance. Methods Experimental model and study participant details Participants. EEG data were collected from 20 adult university students at the University of Helsinki and Aalto University (11 females, age range 19 to 28 years, mean 23.4 years). One participant was excluded due to a technical problem with the EEG data acquisition. fMRI data were collected from a separate sample of adult university students at the University of Helsinki and Aalto University comprising 23 adult participants (14 females, age range 19 to 30 years, mean 24.3 years). fMRI data were excluded based on preestablished criteria. Two participants were excluded due to excessive head motion (>5 mm) and 2 participants due to anatomical anomalies that affected coregistration. Thus, data from 19 participants were used in the analyses. The fMRI data has been previously analysed and published in [38] but in the present manuscript, the data were analysed differently, yielding previously unreported results, e.g., fusion with the EEG-data. All participants were monolingual native Finnish speakers, and they did not have any self-reported neurological or psychiatric diseases. In addition, they had self-reported normal hearing and normal or corrected-to-normal vision. All participants were right-handed, and this was confirmed by the Edinburgh Handedness Inventory [78]. Ethics statement. The studies involving human participants were reviewed and approved by Ethics Review Board in the Humanities and Social and Behavioral Sciences, University of Helsinki (number: 14/2017). The research follows the ethical guidelines of the Declaration of Helsinki. The participants provided their written informed consent to participate in this study. Written informed consent was obtained for the sharing of processed anomynised data from each participant. The 2 people visible in Fig 1 and the photographer gave written consent for the publication of the identifiable images under the Creative Commons By 4.0. license. Method details Preparation of stimulus materials. The stimuli comprised dialogues between 2 (female and male) native Finnish speakers. Written informed consent has been obtained from the individual(s) for the publication of any potentially identifiable images or data included in this manuscript (see also [38–40,42]). The dialogue topics were about neutral everyday subjects such as the weather. The dialogues comprised 7 lines (ca. 5.4 s of duration) followed by a ca. 3 s break (2.9 to 4.3 s), resulting in a total length of 55 to 65 s (mean 59.2 s) for each dialogue. The speakers spoke their lines in an alternating fashion; the female talker started the conversation in half of the video clips. The original dialogues [42] were recorded so that the talkers sat next to one another with their faces slightly tilted towards each other (see Fig 1A). For more details on the recordings, see [42]. In both the EEG and the fMRI experiment, we used 24 of the original dialogues for the coherent context conditions. The rest of the dialogues were used to construct 24 new dialogues for the incoherent context conditions. These semantically incoherent dialogues were constructed by shuffling lines from different dialogues of the 36 original dialogues. Dialogues were chosen based on the location and posture of the speakers so that there would be minimal visual transition between each line of the shuffled dialogues. Because slight differences in lighting and posture of the speakers, we divided the videos into pools of 6 videos that were maximally similar. In the semantically incoherent dialogues, each of the 5 lines were from a separate dialogue, and the remaining 2 from one dialogue. To secure that all lines were equally unpredictable, we ensured that the 2 lines from the same original dialogue were separated by at least 4 other lines. The semantically incoherent dialogues were constructed by first removing the audio stream from the video, whereafter the video image was edited with Adobe Premiere Pro CC software with the morph-cut function (Adobe, San Jose, California, United States of America). To prevent participants from noticing these changes, the transition from one dialogue to another always occurred on the side where the talker was silent (see [38], Supplementary video material 1–8; https://osf.io/agxth/). The lighting was edited to fade small differences between the different clips. Two small grey squares (size 1.5° × 1.5°) were added to the videos below the faces of the speakers. A white cross (height 0.5°) was placed in the middle of the square below the face of the talker who was speaking at that given moment. This cross faded out immediately as the talker ended their line and reappeared 1.5 s later. Thus, most of the time, there were 2 crosses present in the video (see [38], Supplementary Video material 1–8; unlike in our experiments, these videos have English subtitles). In the visual control task, the disappearance of the cross indicated that the participant should turn their attention to the other side of the video frame. The cross changed from a plus sign (+) to multiplication sign (×) or vice versa, randomly 9 to 15 times during each dialogue. The cross rotated only on the side where the talker of the dialogue was speaking. During each of the 7 lines, the cross rotated 1 to 4 times, i.e., every 1.25 to 2.5 s. The audio streams were noise-vocoded before adding the audio streams back to the videos [42]. This was achieved by dividing the audio streams into 4 (poor auditory conditions) and 16 (good auditory conditions) logarithmically spaced frequency bands between 0.3 and 5 kHz using Praat software [version 6.0.27, 47]. The talkers’ F0 (frequencies 0 to 0.3 kHz) was unchanged (see [42] for details). To manipulate the amount of visual speech seen by the participants, we added a dynamic white noise masker onto the speakers’ faces (see [42]). Finally, the poor and good quality audio files were recombined with the poor and good visual quality videos with a custom Matlab script. As the final step, we added a continuous background stream to the dialogues. We used a freely available audiobook about cultural history (a Finnish translation of The Autumn of the Middle Ages by Johan Huizinga, distributed online by YLE, the Finnish Broadcasting company), read by a female native Finnish professional actor. The F0 of the reader was lowered to 0.16 kHz and the audiobook was low-pass filtered at 5.0 kHz [42]. Procedure. The videos, including the dialogue stream and background stream described above, were used in our 16 experimental conditions defined by Attentional Task (attend speech, ignore speech; Att. versus ign.), Semantic Coherence (coherent, incoherent), Auditory Quality (good, poor), and Visual Quality (good, poor). We presented 3 runs, each containing 8 of the 24 coherent video clips (in all coherent context conditions) and 8 of the 24 incoherent video clips (in all incoherent context conditions). Thus, all the participants were presented with all the 48 dialogues. Every other run started with the attend speech task, and every other with the ignore speech task. Within the functional runs, the attend speech task and the ignore speech task were presented in an alternating order. The order of conditions and dialogues presented was pseudorandomised. Because we could not entirely randomise the videos into the 16 conditions per run, we used the Latin square to construct 4 different versions of the experiment (see Suppl. Table 3 in [38]). Stimulus presentation was controlled by using Presentation 20.0–22.0 software (Neurobehavioral Systems, Berkeley, California, USA). The auditory stimuli were presented binaurally through insert earphones (Sensimetrics model S14; Sensimetrics, Malden, Massachusetts, USA). Before the experiment, the audio volume was set to a comfortable level individually for each participant. It was approximately 75 to 86 dB SPL at the ear drum. During EEG, the video clips (size 26° × 15°) were presented in the middle of a 24-inch LCD monitor (HP Compaq LA2405x; HP, Palo Alto, California, USA) that was at ca. 40 cm from the eyes of the participant. During fMRI, the video clips (size 26° × 15°) were projected onto a mirror attached to the head coil and presented in the middle of the screen. Videos were presented on a uniform grey background. In the middle of each run, there was a break of 40 s. During the break, the participants were asked to rest and focus on a fixation cross (located in the middle of the screen, height 0.5°). The distracting audiobook (presented with a sound intensity 3 dB lower than the voices of the viewed male and female speakers) started randomly 0.5 to 2 s before video onset and stopped at the offset of the video. The differences in dialogue durations were compensated by inserting periods with a fixation cross between the instruction and the onset of the dialogue, keeping the overall trial durations constant. Tasks. During the attend speech task, the participants were asked to attend to the 2 speakers having a discussion in the videos while ignoring the background speech. After every dialogue, the participants were presented with 7 statements relating to the occurrence of a topic in each line from the dialogue by pressing the “Yes” or “No” button on a response pad with their right index or middle finger. Questions were for example, “Did the boy drop his phone?”, “Was there a cat on the table?”. A new statement was presented every 2 s. After the 7 statements, the participants were provided with feedback on their performance (number of correct responses). During the ignore speech task, the participants were asked to attend to the fixation cross presented in the videos and calculate how many times the cross rotated from a multiplication sign (×) to a plus sign (+) and vice versa. Every time the cross disappeared, the participants were supposed to shift their attention to the other fixation cross on the other side of the frame. The participants were instructed to actively ignore all speech stimuli, i.e., the dialogues and the audiobook. At the end of the video, the participants were presented with 7 statements about the rotating cross (“Did the cross turn X times?”, the X being between 9 and 15 in an ascending order). As in the attend speech task, the response was given by pressing either the “Yes” or “No” button on a response pad. If the participants were unsure, they were instructed to answer “Yes” to all the alternatives they deemed possible. After the 7 statements, the participants received feedback on their performance (number of correct responses). Additional task. After completing the 3 runs, the participants were presented with an additional run consisting of a single dialogue and one set of 7 questions (note only in the EEG experiment). The dialogue employed in this extra run was the one of the 12 original coherent dialogues that were used to create the 24 incoherent dialogues (i.e., these 12 dialogues had not been seen/heard by the participants in their coherent form in the present experiment). The purpose of this additional run was to evaluate how much the participants processed the semantics of the dialogues they were instructed to ignore during the visual control task. The participants were presented with a dialogue video, and they were told to complete the visual control task and hence ignore the dialogue while counting fixation cross rotations. At the end of the video, they were, however, instructed to answer 7 yes-no questions about the lines of the dialogue. The dialogues in this additional run were presented with good auditory and visual qualities and with a coherent semantic context as this was considered the type of conversation that would be the hardest to ignore. This task concluded the experiment. Thus, the additional task was completed by 19 participants participating in the EEG experiment. For results on this task, please see [38]. Pre-trial. Before the experiment, all participants practised the tasks. In the practice phase, the participants performed the attend speech task and the ignore speech task, using a coherent dialogue not included in the actual experiment. The dialogue was presented with different auditory and visual qualities. Data acquisition. The EEG data were collected at the Department of Psychology and Logopedics, University of Helsinki, in a soundproof and electrically shielded EEG laboratory. The data were registered separately for each of the 3 runs of each participant, and the overall duration of the EEG measurements was approximately 1.5 h per participant. The EEG data were recorded with a BrainVision actiCHamp amplifier (128 channels) and a BrainVision actiCAP snap electrode cap with an actiCAP slim electrode set of 128 active electrodes (Brain Products GmbH, Gilching, Germany). The electrode layout was an extended version of the International 10–20 system, and recording reference was at FCz. The amplifier bandwidth was 0 to 140 Hz and the sampling rate was 500 Hz. The EEG data were recorded with BrainVision Recorder (version 1.21.0402–1.22.0002; Brain Products GmbH, Gilching, Germany). Electrode impedances were checked prior to recording, and they were below 10 kΩ for most electrodes for most participants. When needed, worsened impedances were enhanced in-between the experimental runs. For a detailed description of the fMRI acquisition, see [38]. We report the parameters used in brief in Table 1. Download: PPT PowerPoint slide PNG larger image TIFF original image Table 1. MRI-acquisition parameters used in the fMRI data collection (see [38] for details). https://doi.org/10.1371/journal.pbio.3002534.t001 Quantification and statistical analysis Analysis of behavioural data. The total number of questions in the experiment was 336 (48 dialogues × 7 lines). We registered the number of correct answers in each task block. Misses were treated as incorrect button presses. The mean task performance and standard error of mean were used to establish that the participants were performing the task as expected. To analyse participants’ performance (EEG/fMRI experiment) during the attend speech and ignore speech task, 2 separate repeated-measures analyses of variance (ANOVA) were computed with 3 factors: Semantic Coherence (coherent, incoherent), Auditory Quality (good, poor), and Visual Quality (good, poor). ANOVAs were chosen instead of linear mixed models for these analyses to yield comparable results to those reported for the fMRI experiment, which are not reported in the present manuscript, but can be found in [38]. We also analysed the performance line-by-line to evaluate whether participants’ performance changed during each dialogue (only performed for the attend speech condition performance data gathered in the EEG experiment). Here, we used a generalised linear model (identity link function) with the participant added as a random effect (including intercept) and the effect of line was modelled as a categorical repeated measure. The model was run using maximum likelihood estimation with a maximum of 100 iterations to converge. Statistical analyses were carried out with IBM 18 SPSS Statistics 25 (IBM SPSS, Armonk, New York, USA) software and the results were visualised with Python (Mathworks, Natick, Massachusetts, USA). Preprocessing of EEG data. EEG data preprocessing was carried out using the MNE Python 0.22 [79]. All channels were referenced to an average reference. Next, the data were manually inspected for channels that would subsequently be interpolated. At least one of the following criteria had to be met for a channel to be chosen for interpolation, and the criterion had to be present persistently throughout at least one of the 3 experimental runs. The criteria were a flat line response, high-frequency deviation, electrode pop artifacts, and body movement artifacts. The deviant channels were temporarily removed from the data. An independent component analysis (ICA) was fitted on the concatenated runs for each participant separately, using MNE-ICA (picard-type). For each participant, we defined 2 to 4 components to be denoised that were classified as either blinks, lateral eye movements, or heartbeats. Thereafter, the raw data from each run was denoised using MNE.ica.apply. After this, the formerly chosen deviant channels were interpolated, the data was bandpass filtered (0.5 to 10 Hz) using the mne.raw.filter function, with the “firwin” option (default settings). This function computes the coefficients of a finite impulse response filter (hamming window). Thereafter, the data were down sampled to 128 Hz for the TRF analyses and 64 Hz for the speech reconstruction analyses. Thereafter, the EEG time series were cut into 6.5 s epochs based on the dialogue speech trials (see below). First-level analysis of EEG data. To estimate the neural response to the 2 speech streams (dialogue stream and background stream), we performed speech tracking, using both an encoding and a decoding approach. In the encoding approach, we estimated TRFs for each EEG channel. In the decoding approach, we reconstructed the speech using data pooled across all 128 EEG channels (SER analysis). The rationale for TRF estimation has been described in detail elsewhere (see e.g., [80]). In brief, TRFs constitute linear transfer functions describing the relationship between features of the stimulus function (S) and the response function (R; i.e., the EEG channel data). Stimulus features were constructed by extracting sound amplitude envelopes separately for the dialogue stream and the background stream using a Hilbert transform. The envelopes were band-pass filtered (0 to 10 Hz) and down sampled to 128 Hz for TRFs and 64 Hz for SERs (a lower sampling rate was chosen to speed up analysis for SERs). Thereafter, the envelopes were cut into separate lines (6.5 s) for both sound streams. In the encoding approach, 2 separate TRFs were estimated per EEG channel (dialogue and background; Fig 2A). These TRFs can be conceptualised as a linear composition of partially overlapping neural responses at different time lags (τ) to a continuous stimulus, and they are therefore conceptually similar to ERPs ([11]). We estimated TRFs with the receptive field function (MNE-python: based on the mTRF toolbox that utilises ridge regression), with time lags −200 to 800 ms, and a common regularisation parameter (λ) of 105 ([80]; see Fig 2A). Note that the regularisation parameter used affects the shape and amplitude of the TRF curves (for simulations, see e.g., [11]) and therefore we chose a common regularisation parameter (based on [80]) and used it in all conditions and participants. In the decoding analysis, a multidimensional transfer function was estimated using all EEG channels as input (R) in an attempt to reconstruct separately the dialogue stream and the background stream amplitude modulations (see Fig 3A) using the receptive field function (MNE-python), with time lags (τ) of −200 to 0 ms and a common regularisation parameter (λ) of 104 [80]. Unlike the encoding analysis, this analysis yields stimulus construction for each time point of the stimulus function (see Fig 3A). Both models used a leave-one-out approach, where in each iteration all trials (except one) are selected to train the model (train set), which was then used to predict either the neural response at each EEG channel (TRF) or the speech envelope of speech streams (SER) in the left-out trial (test set). This procedure was repeated with a different train ‐ test partition in each iteration averaged over all iterations. Univariate analysis of EEG data. For the TRFs, we tested whether Attentional Task (i.e., attend speech task versus ignore speech task) modulated the TRFs averaged across 7 frontocentral electrodes (Cz, FCC1h, FCC2h, FC1, FC2, FFC1h, and FFC2h; Fig 3A), separately for the dialogue stream and the background stream for each time bin employing permutation paired t tests (20,000 permutations) using custom scripts written in Python. For SER, in accordance with [29] we calculated correlations (Pearson) between the original speech stream envelopes and their reconstructions. Thereafter, we cross-reconstruction correlated the stimulus reconstructions and the stimulus envelopes (e.g., correlation between the dialogue speech amplitude and the background speech reconstruction which should be close to zero (Fig 3A)). Finally, to estimate SER accuracy we used correlation difference scores (Δr; between the correct reconstruction correlations and the cross-reconstruction correlations (Fig 3B)). For segment-level analysis, we divided the stimulus envelopes and the stimulus reconstructions into 4 segments of equal length and calculated correlations based on these instead of the full line. The SER accuracies were analysed with different linear mixed models using IBM 18 SPSS Statistics 25. All models included the participant as a random effect and intercept. For repeated factors, the diagonal covariance structure was chosen. If there was more than one repeated factor, a random slope was added for all repeated main effects and interactions using the variance components method. The models were estimated using restricted maximum likelihood estimation with a maximum iteration number of 100 to converge and df-estimation was performed using Satterthwaite. Because SPSS does not produce effect size estimates for the fixed effects in linear mixed models, we used the formula partial η2 = F × df1 / (F × df1 + df2) [81] to approximate effect sizes where applicable. To analyse how performance in individual trials affected the dialogue stream reconstruction during the attend speech conditions, we performed a similar two-level analysis commonly used when analysing fMRI data [38]. First, we created separately for each participant a linear regression model with response (correct or incorrect) in each trial as the predictor and SER accuracy for that trial as output. Because there were not enough incorrect responses in any one sub-condition (e.g., coherent, good auditory, and good visual quality), all trials were pooled across the 8 stimulus conditions. However, because the quality manipulations might affect performance—SER accuracy associations, we added Semantic Coherence, Auditory Quality, and Visual Quality as confounds in this model. The β-weight for response was thereafter taken to the second-level analysis. The second-level analysis was a similar linear mixed model as described above with line entered as the repeated predictor and 5% trimming used for the output variable to remove noise due to the paucity of incorrect trials. Decoding analysis on the fMRI data. The preprocessing and first-level analysis pipeline for the fMRI data were the same as that described in detail in [38] (for a brief description, see Table 2). Download: PPT PowerPoint slide PNG larger image TIFF original image Table 2. fMRI preprocessing parameters and first-level GLM specifications (see [38] for details). https://doi.org/10.1371/journal.pbio.3002534.t002 Support vector machine (SVM) decoding with leave-one-run-out cross-validation [82] was used to classify each pair of the 16 conditions (Task (attend speech task, ignore speech task) × Semantic Coherence (coherent, incoherent) × Auditory Quality (good, poor) × Visual Quality (good, poor)) in the fMRI data. Each line constituted an exemplar and each voxel a feature in the analysis. The SVM was conducted with the decoding toolbox (TDT, [83]) using the beta images from the first-level GLM in the participants’ anatomical space. We used searchlight-based decoding [84] with a radius of 6 mm (isotropic), and with default settings of TDT; L2-norm SVM with regularising parameter C = 1 running in LIBSVM [85]. The resulting accuracy maps for each condition pair were thereafter projected to the Freesurfer average (fsaverage) using the participants’ own Freesurfer surface (surface smoothing: 5 mm2 full-width half maximum smoothing). The pairwise decoding accuracies were averaged within each 360 ROIs (HCP parcellation [60]) and RDMs ([86]) were constructed for each subject and each ROI. All RDMs in this study are displayed rank-scaled and the conditions are ordered so that the 8 attend speech task conditions are first and the ignore speech task are second. The coherence and quality conditions are in the following order (coherent: co, incoherent: inco, good: g, poor: p, visual quality: v, auditory quality: a; co-gv-ga, inco-gv-ga, co-pv-ga, inco-pv-ga, co-gv-pa, inco-gv-pa, co-pv-pa, inco-pv-pa). The fMRI RDMs were compared with the attentional task model (Fig 3C). First, model and data RDMs were vectorised (lower triangular) and then correlated (Spearman r) with each other for each ROI and each participant. The statistical significance of the mean correlation above zero was tested with right-tailed t test, FDR-corrected for 360 ROIs. Decoding analysis of SER accuracies. SVM decoding with leave-one-run-out cross-validation [82] was used to classify SER correlations as either belonging to the attend speech task or the ignore speech task. Each line constituted an exemplar and the 4 correlations used to define SER accuracies in the univariate analyses (see above) were used as features (r: reconstruction of the dialogue stream envelope × the dialogue stream envelope, reconstruction of the dialogue stream envelope × the background stream envelope, reconstruction of the background stream envelope × the background stream envelope, reconstruction of the background stream envelope × the dialogue stream envelope). The SVM was conducted with the decoding toolbox (TDT, [84]) using the SER correlations from each participant, with default settings of TDT; L2-norm SVM with regularising parameter C = 1 running in LIBSVM [85], 100 iterations. The resulting accuracies were thereafter analysed using linear mixed models. Multivariate analysis of TRFs. RDMs were constructed from the dialogue TRFs as well as background speech TRFs separately for each time point by calculating 1-r (Spearman) of all conditions across all channels. Like fMRI, the TRF RDMs were compared to model RDMs (see S3 Fig). First, model and data RDMs were vectorised (lower triangular) and then correlated (Spearman r) with each other for each time point and each subject. The statistical significance of the mean correlation above zero was tested with right-tailed t test, FDR-corrected for 100 time points. TRF-fMRI fusion. Representational similarity analysis was used to combine EEG and fMRI data [45,46]. The TRF RDMs for 100 time points (0 to 800 ms) were correlated (Spearman r) with the 360 fMRI RDMs. Prior to correlations, the TRF RDMs were averaged across subjects to reduce noise in the data. Furthermore, partial correlation (Spearman r) was used, and the effect of task and background speech was controlled for when fusing dialogue TRFs and fMRI, and the effect of task and dialogue was controlled for when fusing background speech TRFs and fMRI. The statistical significance of the mean correlation above zero was tested with right-tailed t tests, FDR-correction was applied for time points, ROIs, and models (task and dialogue/background speech). Experimental model and study participant details Participants. EEG data were collected from 20 adult university students at the University of Helsinki and Aalto University (11 females, age range 19 to 28 years, mean 23.4 years). One participant was excluded due to a technical problem with the EEG data acquisition. fMRI data were collected from a separate sample of adult university students at the University of Helsinki and Aalto University comprising 23 adult participants (14 females, age range 19 to 30 years, mean 24.3 years). fMRI data were excluded based on preestablished criteria. Two participants were excluded due to excessive head motion (>5 mm) and 2 participants due to anatomical anomalies that affected coregistration. Thus, data from 19 participants were used in the analyses. The fMRI data has been previously analysed and published in [38] but in the present manuscript, the data were analysed differently, yielding previously unreported results, e.g., fusion with the EEG-data. All participants were monolingual native Finnish speakers, and they did not have any self-reported neurological or psychiatric diseases. In addition, they had self-reported normal hearing and normal or corrected-to-normal vision. All participants were right-handed, and this was confirmed by the Edinburgh Handedness Inventory [78]. Ethics statement. The studies involving human participants were reviewed and approved by Ethics Review Board in the Humanities and Social and Behavioral Sciences, University of Helsinki (number: 14/2017). The research follows the ethical guidelines of the Declaration of Helsinki. The participants provided their written informed consent to participate in this study. Written informed consent was obtained for the sharing of processed anomynised data from each participant. The 2 people visible in Fig 1 and the photographer gave written consent for the publication of the identifiable images under the Creative Commons By 4.0. license. Participants. EEG data were collected from 20 adult university students at the University of Helsinki and Aalto University (11 females, age range 19 to 28 years, mean 23.4 years). One participant was excluded due to a technical problem with the EEG data acquisition. fMRI data were collected from a separate sample of adult university students at the University of Helsinki and Aalto University comprising 23 adult participants (14 females, age range 19 to 30 years, mean 24.3 years). fMRI data were excluded based on preestablished criteria. Two participants were excluded due to excessive head motion (>5 mm) and 2 participants due to anatomical anomalies that affected coregistration. Thus, data from 19 participants were used in the analyses. The fMRI data has been previously analysed and published in [38] but in the present manuscript, the data were analysed differently, yielding previously unreported results, e.g., fusion with the EEG-data. All participants were monolingual native Finnish speakers, and they did not have any self-reported neurological or psychiatric diseases. In addition, they had self-reported normal hearing and normal or corrected-to-normal vision. All participants were right-handed, and this was confirmed by the Edinburgh Handedness Inventory [78]. Ethics statement. The studies involving human participants were reviewed and approved by Ethics Review Board in the Humanities and Social and Behavioral Sciences, University of Helsinki (number: 14/2017). The research follows the ethical guidelines of the Declaration of Helsinki. The participants provided their written informed consent to participate in this study. Written informed consent was obtained for the sharing of processed anomynised data from each participant. The 2 people visible in Fig 1 and the photographer gave written consent for the publication of the identifiable images under the Creative Commons By 4.0. license. Method details Preparation of stimulus materials. The stimuli comprised dialogues between 2 (female and male) native Finnish speakers. Written informed consent has been obtained from the individual(s) for the publication of any potentially identifiable images or data included in this manuscript (see also [38–40,42]). The dialogue topics were about neutral everyday subjects such as the weather. The dialogues comprised 7 lines (ca. 5.4 s of duration) followed by a ca. 3 s break (2.9 to 4.3 s), resulting in a total length of 55 to 65 s (mean 59.2 s) for each dialogue. The speakers spoke their lines in an alternating fashion; the female talker started the conversation in half of the video clips. The original dialogues [42] were recorded so that the talkers sat next to one another with their faces slightly tilted towards each other (see Fig 1A). For more details on the recordings, see [42]. In both the EEG and the fMRI experiment, we used 24 of the original dialogues for the coherent context conditions. The rest of the dialogues were used to construct 24 new dialogues for the incoherent context conditions. These semantically incoherent dialogues were constructed by shuffling lines from different dialogues of the 36 original dialogues. Dialogues were chosen based on the location and posture of the speakers so that there would be minimal visual transition between each line of the shuffled dialogues. Because slight differences in lighting and posture of the speakers, we divided the videos into pools of 6 videos that were maximally similar. In the semantically incoherent dialogues, each of the 5 lines were from a separate dialogue, and the remaining 2 from one dialogue. To secure that all lines were equally unpredictable, we ensured that the 2 lines from the same original dialogue were separated by at least 4 other lines. The semantically incoherent dialogues were constructed by first removing the audio stream from the video, whereafter the video image was edited with Adobe Premiere Pro CC software with the morph-cut function (Adobe, San Jose, California, United States of America). To prevent participants from noticing these changes, the transition from one dialogue to another always occurred on the side where the talker was silent (see [38], Supplementary video material 1–8; https://osf.io/agxth/). The lighting was edited to fade small differences between the different clips. Two small grey squares (size 1.5° × 1.5°) were added to the videos below the faces of the speakers. A white cross (height 0.5°) was placed in the middle of the square below the face of the talker who was speaking at that given moment. This cross faded out immediately as the talker ended their line and reappeared 1.5 s later. Thus, most of the time, there were 2 crosses present in the video (see [38], Supplementary Video material 1–8; unlike in our experiments, these videos have English subtitles). In the visual control task, the disappearance of the cross indicated that the participant should turn their attention to the other side of the video frame. The cross changed from a plus sign (+) to multiplication sign (×) or vice versa, randomly 9 to 15 times during each dialogue. The cross rotated only on the side where the talker of the dialogue was speaking. During each of the 7 lines, the cross rotated 1 to 4 times, i.e., every 1.25 to 2.5 s. The audio streams were noise-vocoded before adding the audio streams back to the videos [42]. This was achieved by dividing the audio streams into 4 (poor auditory conditions) and 16 (good auditory conditions) logarithmically spaced frequency bands between 0.3 and 5 kHz using Praat software [version 6.0.27, 47]. The talkers’ F0 (frequencies 0 to 0.3 kHz) was unchanged (see [42] for details). To manipulate the amount of visual speech seen by the participants, we added a dynamic white noise masker onto the speakers’ faces (see [42]). Finally, the poor and good quality audio files were recombined with the poor and good visual quality videos with a custom Matlab script. As the final step, we added a continuous background stream to the dialogues. We used a freely available audiobook about cultural history (a Finnish translation of The Autumn of the Middle Ages by Johan Huizinga, distributed online by YLE, the Finnish Broadcasting company), read by a female native Finnish professional actor. The F0 of the reader was lowered to 0.16 kHz and the audiobook was low-pass filtered at 5.0 kHz [42]. Procedure. The videos, including the dialogue stream and background stream described above, were used in our 16 experimental conditions defined by Attentional Task (attend speech, ignore speech; Att. versus ign.), Semantic Coherence (coherent, incoherent), Auditory Quality (good, poor), and Visual Quality (good, poor). We presented 3 runs, each containing 8 of the 24 coherent video clips (in all coherent context conditions) and 8 of the 24 incoherent video clips (in all incoherent context conditions). Thus, all the participants were presented with all the 48 dialogues. Every other run started with the attend speech task, and every other with the ignore speech task. Within the functional runs, the attend speech task and the ignore speech task were presented in an alternating order. The order of conditions and dialogues presented was pseudorandomised. Because we could not entirely randomise the videos into the 16 conditions per run, we used the Latin square to construct 4 different versions of the experiment (see Suppl. Table 3 in [38]). Stimulus presentation was controlled by using Presentation 20.0–22.0 software (Neurobehavioral Systems, Berkeley, California, USA). The auditory stimuli were presented binaurally through insert earphones (Sensimetrics model S14; Sensimetrics, Malden, Massachusetts, USA). Before the experiment, the audio volume was set to a comfortable level individually for each participant. It was approximately 75 to 86 dB SPL at the ear drum. During EEG, the video clips (size 26° × 15°) were presented in the middle of a 24-inch LCD monitor (HP Compaq LA2405x; HP, Palo Alto, California, USA) that was at ca. 40 cm from the eyes of the participant. During fMRI, the video clips (size 26° × 15°) were projected onto a mirror attached to the head coil and presented in the middle of the screen. Videos were presented on a uniform grey background. In the middle of each run, there was a break of 40 s. During the break, the participants were asked to rest and focus on a fixation cross (located in the middle of the screen, height 0.5°). The distracting audiobook (presented with a sound intensity 3 dB lower than the voices of the viewed male and female speakers) started randomly 0.5 to 2 s before video onset and stopped at the offset of the video. The differences in dialogue durations were compensated by inserting periods with a fixation cross between the instruction and the onset of the dialogue, keeping the overall trial durations constant. Tasks. During the attend speech task, the participants were asked to attend to the 2 speakers having a discussion in the videos while ignoring the background speech. After every dialogue, the participants were presented with 7 statements relating to the occurrence of a topic in each line from the dialogue by pressing the “Yes” or “No” button on a response pad with their right index or middle finger. Questions were for example, “Did the boy drop his phone?”, “Was there a cat on the table?”. A new statement was presented every 2 s. After the 7 statements, the participants were provided with feedback on their performance (number of correct responses). During the ignore speech task, the participants were asked to attend to the fixation cross presented in the videos and calculate how many times the cross rotated from a multiplication sign (×) to a plus sign (+) and vice versa. Every time the cross disappeared, the participants were supposed to shift their attention to the other fixation cross on the other side of the frame. The participants were instructed to actively ignore all speech stimuli, i.e., the dialogues and the audiobook. At the end of the video, the participants were presented with 7 statements about the rotating cross (“Did the cross turn X times?”, the X being between 9 and 15 in an ascending order). As in the attend speech task, the response was given by pressing either the “Yes” or “No” button on a response pad. If the participants were unsure, they were instructed to answer “Yes” to all the alternatives they deemed possible. After the 7 statements, the participants received feedback on their performance (number of correct responses). Additional task. After completing the 3 runs, the participants were presented with an additional run consisting of a single dialogue and one set of 7 questions (note only in the EEG experiment). The dialogue employed in this extra run was the one of the 12 original coherent dialogues that were used to create the 24 incoherent dialogues (i.e., these 12 dialogues had not been seen/heard by the participants in their coherent form in the present experiment). The purpose of this additional run was to evaluate how much the participants processed the semantics of the dialogues they were instructed to ignore during the visual control task. The participants were presented with a dialogue video, and they were told to complete the visual control task and hence ignore the dialogue while counting fixation cross rotations. At the end of the video, they were, however, instructed to answer 7 yes-no questions about the lines of the dialogue. The dialogues in this additional run were presented with good auditory and visual qualities and with a coherent semantic context as this was considered the type of conversation that would be the hardest to ignore. This task concluded the experiment. Thus, the additional task was completed by 19 participants participating in the EEG experiment. For results on this task, please see [38]. Pre-trial. Before the experiment, all participants practised the tasks. In the practice phase, the participants performed the attend speech task and the ignore speech task, using a coherent dialogue not included in the actual experiment. The dialogue was presented with different auditory and visual qualities. Data acquisition. The EEG data were collected at the Department of Psychology and Logopedics, University of Helsinki, in a soundproof and electrically shielded EEG laboratory. The data were registered separately for each of the 3 runs of each participant, and the overall duration of the EEG measurements was approximately 1.5 h per participant. The EEG data were recorded with a BrainVision actiCHamp amplifier (128 channels) and a BrainVision actiCAP snap electrode cap with an actiCAP slim electrode set of 128 active electrodes (Brain Products GmbH, Gilching, Germany). The electrode layout was an extended version of the International 10–20 system, and recording reference was at FCz. The amplifier bandwidth was 0 to 140 Hz and the sampling rate was 500 Hz. The EEG data were recorded with BrainVision Recorder (version 1.21.0402–1.22.0002; Brain Products GmbH, Gilching, Germany). Electrode impedances were checked prior to recording, and they were below 10 kΩ for most electrodes for most participants. When needed, worsened impedances were enhanced in-between the experimental runs. For a detailed description of the fMRI acquisition, see [38]. We report the parameters used in brief in Table 1. Download: PPT PowerPoint slide PNG larger image TIFF original image Table 1. MRI-acquisition parameters used in the fMRI data collection (see [38] for details). https://doi.org/10.1371/journal.pbio.3002534.t001 Preparation of stimulus materials. The stimuli comprised dialogues between 2 (female and male) native Finnish speakers. Written informed consent has been obtained from the individual(s) for the publication of any potentially identifiable images or data included in this manuscript (see also [38–40,42]). The dialogue topics were about neutral everyday subjects such as the weather. The dialogues comprised 7 lines (ca. 5.4 s of duration) followed by a ca. 3 s break (2.9 to 4.3 s), resulting in a total length of 55 to 65 s (mean 59.2 s) for each dialogue. The speakers spoke their lines in an alternating fashion; the female talker started the conversation in half of the video clips. The original dialogues [42] were recorded so that the talkers sat next to one another with their faces slightly tilted towards each other (see Fig 1A). For more details on the recordings, see [42]. In both the EEG and the fMRI experiment, we used 24 of the original dialogues for the coherent context conditions. The rest of the dialogues were used to construct 24 new dialogues for the incoherent context conditions. These semantically incoherent dialogues were constructed by shuffling lines from different dialogues of the 36 original dialogues. Dialogues were chosen based on the location and posture of the speakers so that there would be minimal visual transition between each line of the shuffled dialogues. Because slight differences in lighting and posture of the speakers, we divided the videos into pools of 6 videos that were maximally similar. In the semantically incoherent dialogues, each of the 5 lines were from a separate dialogue, and the remaining 2 from one dialogue. To secure that all lines were equally unpredictable, we ensured that the 2 lines from the same original dialogue were separated by at least 4 other lines. The semantically incoherent dialogues were constructed by first removing the audio stream from the video, whereafter the video image was edited with Adobe Premiere Pro CC software with the morph-cut function (Adobe, San Jose, California, United States of America). To prevent participants from noticing these changes, the transition from one dialogue to another always occurred on the side where the talker was silent (see [38], Supplementary video material 1–8; https://osf.io/agxth/). The lighting was edited to fade small differences between the different clips. Two small grey squares (size 1.5° × 1.5°) were added to the videos below the faces of the speakers. A white cross (height 0.5°) was placed in the middle of the square below the face of the talker who was speaking at that given moment. This cross faded out immediately as the talker ended their line and reappeared 1.5 s later. Thus, most of the time, there were 2 crosses present in the video (see [38], Supplementary Video material 1–8; unlike in our experiments, these videos have English subtitles). In the visual control task, the disappearance of the cross indicated that the participant should turn their attention to the other side of the video frame. The cross changed from a plus sign (+) to multiplication sign (×) or vice versa, randomly 9 to 15 times during each dialogue. The cross rotated only on the side where the talker of the dialogue was speaking. During each of the 7 lines, the cross rotated 1 to 4 times, i.e., every 1.25 to 2.5 s. The audio streams were noise-vocoded before adding the audio streams back to the videos [42]. This was achieved by dividing the audio streams into 4 (poor auditory conditions) and 16 (good auditory conditions) logarithmically spaced frequency bands between 0.3 and 5 kHz using Praat software [version 6.0.27, 47]. The talkers’ F0 (frequencies 0 to 0.3 kHz) was unchanged (see [42] for details). To manipulate the amount of visual speech seen by the participants, we added a dynamic white noise masker onto the speakers’ faces (see [42]). Finally, the poor and good quality audio files were recombined with the poor and good visual quality videos with a custom Matlab script. As the final step, we added a continuous background stream to the dialogues. We used a freely available audiobook about cultural history (a Finnish translation of The Autumn of the Middle Ages by Johan Huizinga, distributed online by YLE, the Finnish Broadcasting company), read by a female native Finnish professional actor. The F0 of the reader was lowered to 0.16 kHz and the audiobook was low-pass filtered at 5.0 kHz [42]. Procedure. The videos, including the dialogue stream and background stream described above, were used in our 16 experimental conditions defined by Attentional Task (attend speech, ignore speech; Att. versus ign.), Semantic Coherence (coherent, incoherent), Auditory Quality (good, poor), and Visual Quality (good, poor). We presented 3 runs, each containing 8 of the 24 coherent video clips (in all coherent context conditions) and 8 of the 24 incoherent video clips (in all incoherent context conditions). Thus, all the participants were presented with all the 48 dialogues. Every other run started with the attend speech task, and every other with the ignore speech task. Within the functional runs, the attend speech task and the ignore speech task were presented in an alternating order. The order of conditions and dialogues presented was pseudorandomised. Because we could not entirely randomise the videos into the 16 conditions per run, we used the Latin square to construct 4 different versions of the experiment (see Suppl. Table 3 in [38]). Stimulus presentation was controlled by using Presentation 20.0–22.0 software (Neurobehavioral Systems, Berkeley, California, USA). The auditory stimuli were presented binaurally through insert earphones (Sensimetrics model S14; Sensimetrics, Malden, Massachusetts, USA). Before the experiment, the audio volume was set to a comfortable level individually for each participant. It was approximately 75 to 86 dB SPL at the ear drum. During EEG, the video clips (size 26° × 15°) were presented in the middle of a 24-inch LCD monitor (HP Compaq LA2405x; HP, Palo Alto, California, USA) that was at ca. 40 cm from the eyes of the participant. During fMRI, the video clips (size 26° × 15°) were projected onto a mirror attached to the head coil and presented in the middle of the screen. Videos were presented on a uniform grey background. In the middle of each run, there was a break of 40 s. During the break, the participants were asked to rest and focus on a fixation cross (located in the middle of the screen, height 0.5°). The distracting audiobook (presented with a sound intensity 3 dB lower than the voices of the viewed male and female speakers) started randomly 0.5 to 2 s before video onset and stopped at the offset of the video. The differences in dialogue durations were compensated by inserting periods with a fixation cross between the instruction and the onset of the dialogue, keeping the overall trial durations constant. Tasks. During the attend speech task, the participants were asked to attend to the 2 speakers having a discussion in the videos while ignoring the background speech. After every dialogue, the participants were presented with 7 statements relating to the occurrence of a topic in each line from the dialogue by pressing the “Yes” or “No” button on a response pad with their right index or middle finger. Questions were for example, “Did the boy drop his phone?”, “Was there a cat on the table?”. A new statement was presented every 2 s. After the 7 statements, the participants were provided with feedback on their performance (number of correct responses). During the ignore speech task, the participants were asked to attend to the fixation cross presented in the videos and calculate how many times the cross rotated from a multiplication sign (×) to a plus sign (+) and vice versa. Every time the cross disappeared, the participants were supposed to shift their attention to the other fixation cross on the other side of the frame. The participants were instructed to actively ignore all speech stimuli, i.e., the dialogues and the audiobook. At the end of the video, the participants were presented with 7 statements about the rotating cross (“Did the cross turn X times?”, the X being between 9 and 15 in an ascending order). As in the attend speech task, the response was given by pressing either the “Yes” or “No” button on a response pad. If the participants were unsure, they were instructed to answer “Yes” to all the alternatives they deemed possible. After the 7 statements, the participants received feedback on their performance (number of correct responses). Additional task. After completing the 3 runs, the participants were presented with an additional run consisting of a single dialogue and one set of 7 questions (note only in the EEG experiment). The dialogue employed in this extra run was the one of the 12 original coherent dialogues that were used to create the 24 incoherent dialogues (i.e., these 12 dialogues had not been seen/heard by the participants in their coherent form in the present experiment). The purpose of this additional run was to evaluate how much the participants processed the semantics of the dialogues they were instructed to ignore during the visual control task. The participants were presented with a dialogue video, and they were told to complete the visual control task and hence ignore the dialogue while counting fixation cross rotations. At the end of the video, they were, however, instructed to answer 7 yes-no questions about the lines of the dialogue. The dialogues in this additional run were presented with good auditory and visual qualities and with a coherent semantic context as this was considered the type of conversation that would be the hardest to ignore. This task concluded the experiment. Thus, the additional task was completed by 19 participants participating in the EEG experiment. For results on this task, please see [38]. Pre-trial. Before the experiment, all participants practised the tasks. In the practice phase, the participants performed the attend speech task and the ignore speech task, using a coherent dialogue not included in the actual experiment. The dialogue was presented with different auditory and visual qualities. Data acquisition. The EEG data were collected at the Department of Psychology and Logopedics, University of Helsinki, in a soundproof and electrically shielded EEG laboratory. The data were registered separately for each of the 3 runs of each participant, and the overall duration of the EEG measurements was approximately 1.5 h per participant. The EEG data were recorded with a BrainVision actiCHamp amplifier (128 channels) and a BrainVision actiCAP snap electrode cap with an actiCAP slim electrode set of 128 active electrodes (Brain Products GmbH, Gilching, Germany). The electrode layout was an extended version of the International 10–20 system, and recording reference was at FCz. The amplifier bandwidth was 0 to 140 Hz and the sampling rate was 500 Hz. The EEG data were recorded with BrainVision Recorder (version 1.21.0402–1.22.0002; Brain Products GmbH, Gilching, Germany). Electrode impedances were checked prior to recording, and they were below 10 kΩ for most electrodes for most participants. When needed, worsened impedances were enhanced in-between the experimental runs. For a detailed description of the fMRI acquisition, see [38]. We report the parameters used in brief in Table 1. Download: PPT PowerPoint slide PNG larger image TIFF original image Table 1. MRI-acquisition parameters used in the fMRI data collection (see [38] for details). https://doi.org/10.1371/journal.pbio.3002534.t001 Quantification and statistical analysis Analysis of behavioural data. The total number of questions in the experiment was 336 (48 dialogues × 7 lines). We registered the number of correct answers in each task block. Misses were treated as incorrect button presses. The mean task performance and standard error of mean were used to establish that the participants were performing the task as expected. To analyse participants’ performance (EEG/fMRI experiment) during the attend speech and ignore speech task, 2 separate repeated-measures analyses of variance (ANOVA) were computed with 3 factors: Semantic Coherence (coherent, incoherent), Auditory Quality (good, poor), and Visual Quality (good, poor). ANOVAs were chosen instead of linear mixed models for these analyses to yield comparable results to those reported for the fMRI experiment, which are not reported in the present manuscript, but can be found in [38]. We also analysed the performance line-by-line to evaluate whether participants’ performance changed during each dialogue (only performed for the attend speech condition performance data gathered in the EEG experiment). Here, we used a generalised linear model (identity link function) with the participant added as a random effect (including intercept) and the effect of line was modelled as a categorical repeated measure. The model was run using maximum likelihood estimation with a maximum of 100 iterations to converge. Statistical analyses were carried out with IBM 18 SPSS Statistics 25 (IBM SPSS, Armonk, New York, USA) software and the results were visualised with Python (Mathworks, Natick, Massachusetts, USA). Preprocessing of EEG data. EEG data preprocessing was carried out using the MNE Python 0.22 [79]. All channels were referenced to an average reference. Next, the data were manually inspected for channels that would subsequently be interpolated. At least one of the following criteria had to be met for a channel to be chosen for interpolation, and the criterion had to be present persistently throughout at least one of the 3 experimental runs. The criteria were a flat line response, high-frequency deviation, electrode pop artifacts, and body movement artifacts. The deviant channels were temporarily removed from the data. An independent component analysis (ICA) was fitted on the concatenated runs for each participant separately, using MNE-ICA (picard-type). For each participant, we defined 2 to 4 components to be denoised that were classified as either blinks, lateral eye movements, or heartbeats. Thereafter, the raw data from each run was denoised using MNE.ica.apply. After this, the formerly chosen deviant channels were interpolated, the data was bandpass filtered (0.5 to 10 Hz) using the mne.raw.filter function, with the “firwin” option (default settings). This function computes the coefficients of a finite impulse response filter (hamming window). Thereafter, the data were down sampled to 128 Hz for the TRF analyses and 64 Hz for the speech reconstruction analyses. Thereafter, the EEG time series were cut into 6.5 s epochs based on the dialogue speech trials (see below). First-level analysis of EEG data. To estimate the neural response to the 2 speech streams (dialogue stream and background stream), we performed speech tracking, using both an encoding and a decoding approach. In the encoding approach, we estimated TRFs for each EEG channel. In the decoding approach, we reconstructed the speech using data pooled across all 128 EEG channels (SER analysis). The rationale for TRF estimation has been described in detail elsewhere (see e.g., [80]). In brief, TRFs constitute linear transfer functions describing the relationship between features of the stimulus function (S) and the response function (R; i.e., the EEG channel data). Stimulus features were constructed by extracting sound amplitude envelopes separately for the dialogue stream and the background stream using a Hilbert transform. The envelopes were band-pass filtered (0 to 10 Hz) and down sampled to 128 Hz for TRFs and 64 Hz for SERs (a lower sampling rate was chosen to speed up analysis for SERs). Thereafter, the envelopes were cut into separate lines (6.5 s) for both sound streams. In the encoding approach, 2 separate TRFs were estimated per EEG channel (dialogue and background; Fig 2A). These TRFs can be conceptualised as a linear composition of partially overlapping neural responses at different time lags (τ) to a continuous stimulus, and they are therefore conceptually similar to ERPs ([11]). We estimated TRFs with the receptive field function (MNE-python: based on the mTRF toolbox that utilises ridge regression), with time lags −200 to 800 ms, and a common regularisation parameter (λ) of 105 ([80]; see Fig 2A). Note that the regularisation parameter used affects the shape and amplitude of the TRF curves (for simulations, see e.g., [11]) and therefore we chose a common regularisation parameter (based on [80]) and used it in all conditions and participants. In the decoding analysis, a multidimensional transfer function was estimated using all EEG channels as input (R) in an attempt to reconstruct separately the dialogue stream and the background stream amplitude modulations (see Fig 3A) using the receptive field function (MNE-python), with time lags (τ) of −200 to 0 ms and a common regularisation parameter (λ) of 104 [80]. Unlike the encoding analysis, this analysis yields stimulus construction for each time point of the stimulus function (see Fig 3A). Both models used a leave-one-out approach, where in each iteration all trials (except one) are selected to train the model (train set), which was then used to predict either the neural response at each EEG channel (TRF) or the speech envelope of speech streams (SER) in the left-out trial (test set). This procedure was repeated with a different train ‐ test partition in each iteration averaged over all iterations. Univariate analysis of EEG data. For the TRFs, we tested whether Attentional Task (i.e., attend speech task versus ignore speech task) modulated the TRFs averaged across 7 frontocentral electrodes (Cz, FCC1h, FCC2h, FC1, FC2, FFC1h, and FFC2h; Fig 3A), separately for the dialogue stream and the background stream for each time bin employing permutation paired t tests (20,000 permutations) using custom scripts written in Python. For SER, in accordance with [29] we calculated correlations (Pearson) between the original speech stream envelopes and their reconstructions. Thereafter, we cross-reconstruction correlated the stimulus reconstructions and the stimulus envelopes (e.g., correlation between the dialogue speech amplitude and the background speech reconstruction which should be close to zero (Fig 3A)). Finally, to estimate SER accuracy we used correlation difference scores (Δr; between the correct reconstruction correlations and the cross-reconstruction correlations (Fig 3B)). For segment-level analysis, we divided the stimulus envelopes and the stimulus reconstructions into 4 segments of equal length and calculated correlations based on these instead of the full line. The SER accuracies were analysed with different linear mixed models using IBM 18 SPSS Statistics 25. All models included the participant as a random effect and intercept. For repeated factors, the diagonal covariance structure was chosen. If there was more than one repeated factor, a random slope was added for all repeated main effects and interactions using the variance components method. The models were estimated using restricted maximum likelihood estimation with a maximum iteration number of 100 to converge and df-estimation was performed using Satterthwaite. Because SPSS does not produce effect size estimates for the fixed effects in linear mixed models, we used the formula partial η2 = F × df1 / (F × df1 + df2) [81] to approximate effect sizes where applicable. To analyse how performance in individual trials affected the dialogue stream reconstruction during the attend speech conditions, we performed a similar two-level analysis commonly used when analysing fMRI data [38]. First, we created separately for each participant a linear regression model with response (correct or incorrect) in each trial as the predictor and SER accuracy for that trial as output. Because there were not enough incorrect responses in any one sub-condition (e.g., coherent, good auditory, and good visual quality), all trials were pooled across the 8 stimulus conditions. However, because the quality manipulations might affect performance—SER accuracy associations, we added Semantic Coherence, Auditory Quality, and Visual Quality as confounds in this model. The β-weight for response was thereafter taken to the second-level analysis. The second-level analysis was a similar linear mixed model as described above with line entered as the repeated predictor and 5% trimming used for the output variable to remove noise due to the paucity of incorrect trials. Decoding analysis on the fMRI data. The preprocessing and first-level analysis pipeline for the fMRI data were the same as that described in detail in [38] (for a brief description, see Table 2). Download: PPT PowerPoint slide PNG larger image TIFF original image Table 2. fMRI preprocessing parameters and first-level GLM specifications (see [38] for details). https://doi.org/10.1371/journal.pbio.3002534.t002 Support vector machine (SVM) decoding with leave-one-run-out cross-validation [82] was used to classify each pair of the 16 conditions (Task (attend speech task, ignore speech task) × Semantic Coherence (coherent, incoherent) × Auditory Quality (good, poor) × Visual Quality (good, poor)) in the fMRI data. Each line constituted an exemplar and each voxel a feature in the analysis. The SVM was conducted with the decoding toolbox (TDT, [83]) using the beta images from the first-level GLM in the participants’ anatomical space. We used searchlight-based decoding [84] with a radius of 6 mm (isotropic), and with default settings of TDT; L2-norm SVM with regularising parameter C = 1 running in LIBSVM [85]. The resulting accuracy maps for each condition pair were thereafter projected to the Freesurfer average (fsaverage) using the participants’ own Freesurfer surface (surface smoothing: 5 mm2 full-width half maximum smoothing). The pairwise decoding accuracies were averaged within each 360 ROIs (HCP parcellation [60]) and RDMs ([86]) were constructed for each subject and each ROI. All RDMs in this study are displayed rank-scaled and the conditions are ordered so that the 8 attend speech task conditions are first and the ignore speech task are second. The coherence and quality conditions are in the following order (coherent: co, incoherent: inco, good: g, poor: p, visual quality: v, auditory quality: a; co-gv-ga, inco-gv-ga, co-pv-ga, inco-pv-ga, co-gv-pa, inco-gv-pa, co-pv-pa, inco-pv-pa). The fMRI RDMs were compared with the attentional task model (Fig 3C). First, model and data RDMs were vectorised (lower triangular) and then correlated (Spearman r) with each other for each ROI and each participant. The statistical significance of the mean correlation above zero was tested with right-tailed t test, FDR-corrected for 360 ROIs. Decoding analysis of SER accuracies. SVM decoding with leave-one-run-out cross-validation [82] was used to classify SER correlations as either belonging to the attend speech task or the ignore speech task. Each line constituted an exemplar and the 4 correlations used to define SER accuracies in the univariate analyses (see above) were used as features (r: reconstruction of the dialogue stream envelope × the dialogue stream envelope, reconstruction of the dialogue stream envelope × the background stream envelope, reconstruction of the background stream envelope × the background stream envelope, reconstruction of the background stream envelope × the dialogue stream envelope). The SVM was conducted with the decoding toolbox (TDT, [84]) using the SER correlations from each participant, with default settings of TDT; L2-norm SVM with regularising parameter C = 1 running in LIBSVM [85], 100 iterations. The resulting accuracies were thereafter analysed using linear mixed models. Multivariate analysis of TRFs. RDMs were constructed from the dialogue TRFs as well as background speech TRFs separately for each time point by calculating 1-r (Spearman) of all conditions across all channels. Like fMRI, the TRF RDMs were compared to model RDMs (see S3 Fig). First, model and data RDMs were vectorised (lower triangular) and then correlated (Spearman r) with each other for each time point and each subject. The statistical significance of the mean correlation above zero was tested with right-tailed t test, FDR-corrected for 100 time points. TRF-fMRI fusion. Representational similarity analysis was used to combine EEG and fMRI data [45,46]. The TRF RDMs for 100 time points (0 to 800 ms) were correlated (Spearman r) with the 360 fMRI RDMs. Prior to correlations, the TRF RDMs were averaged across subjects to reduce noise in the data. Furthermore, partial correlation (Spearman r) was used, and the effect of task and background speech was controlled for when fusing dialogue TRFs and fMRI, and the effect of task and dialogue was controlled for when fusing background speech TRFs and fMRI. The statistical significance of the mean correlation above zero was tested with right-tailed t tests, FDR-correction was applied for time points, ROIs, and models (task and dialogue/background speech). Analysis of behavioural data. The total number of questions in the experiment was 336 (48 dialogues × 7 lines). We registered the number of correct answers in each task block. Misses were treated as incorrect button presses. The mean task performance and standard error of mean were used to establish that the participants were performing the task as expected. To analyse participants’ performance (EEG/fMRI experiment) during the attend speech and ignore speech task, 2 separate repeated-measures analyses of variance (ANOVA) were computed with 3 factors: Semantic Coherence (coherent, incoherent), Auditory Quality (good, poor), and Visual Quality (good, poor). ANOVAs were chosen instead of linear mixed models for these analyses to yield comparable results to those reported for the fMRI experiment, which are not reported in the present manuscript, but can be found in [38]. We also analysed the performance line-by-line to evaluate whether participants’ performance changed during each dialogue (only performed for the attend speech condition performance data gathered in the EEG experiment). Here, we used a generalised linear model (identity link function) with the participant added as a random effect (including intercept) and the effect of line was modelled as a categorical repeated measure. The model was run using maximum likelihood estimation with a maximum of 100 iterations to converge. Statistical analyses were carried out with IBM 18 SPSS Statistics 25 (IBM SPSS, Armonk, New York, USA) software and the results were visualised with Python (Mathworks, Natick, Massachusetts, USA). Preprocessing of EEG data. EEG data preprocessing was carried out using the MNE Python 0.22 [79]. All channels were referenced to an average reference. Next, the data were manually inspected for channels that would subsequently be interpolated. At least one of the following criteria had to be met for a channel to be chosen for interpolation, and the criterion had to be present persistently throughout at least one of the 3 experimental runs. The criteria were a flat line response, high-frequency deviation, electrode pop artifacts, and body movement artifacts. The deviant channels were temporarily removed from the data. An independent component analysis (ICA) was fitted on the concatenated runs for each participant separately, using MNE-ICA (picard-type). For each participant, we defined 2 to 4 components to be denoised that were classified as either blinks, lateral eye movements, or heartbeats. Thereafter, the raw data from each run was denoised using MNE.ica.apply. After this, the formerly chosen deviant channels were interpolated, the data was bandpass filtered (0.5 to 10 Hz) using the mne.raw.filter function, with the “firwin” option (default settings). This function computes the coefficients of a finite impulse response filter (hamming window). Thereafter, the data were down sampled to 128 Hz for the TRF analyses and 64 Hz for the speech reconstruction analyses. Thereafter, the EEG time series were cut into 6.5 s epochs based on the dialogue speech trials (see below). First-level analysis of EEG data. To estimate the neural response to the 2 speech streams (dialogue stream and background stream), we performed speech tracking, using both an encoding and a decoding approach. In the encoding approach, we estimated TRFs for each EEG channel. In the decoding approach, we reconstructed the speech using data pooled across all 128 EEG channels (SER analysis). The rationale for TRF estimation has been described in detail elsewhere (see e.g., [80]). In brief, TRFs constitute linear transfer functions describing the relationship between features of the stimulus function (S) and the response function (R; i.e., the EEG channel data). Stimulus features were constructed by extracting sound amplitude envelopes separately for the dialogue stream and the background stream using a Hilbert transform. The envelopes were band-pass filtered (0 to 10 Hz) and down sampled to 128 Hz for TRFs and 64 Hz for SERs (a lower sampling rate was chosen to speed up analysis for SERs). Thereafter, the envelopes were cut into separate lines (6.5 s) for both sound streams. In the encoding approach, 2 separate TRFs were estimated per EEG channel (dialogue and background; Fig 2A). These TRFs can be conceptualised as a linear composition of partially overlapping neural responses at different time lags (τ) to a continuous stimulus, and they are therefore conceptually similar to ERPs ([11]). We estimated TRFs with the receptive field function (MNE-python: based on the mTRF toolbox that utilises ridge regression), with time lags −200 to 800 ms, and a common regularisation parameter (λ) of 105 ([80]; see Fig 2A). Note that the regularisation parameter used affects the shape and amplitude of the TRF curves (for simulations, see e.g., [11]) and therefore we chose a common regularisation parameter (based on [80]) and used it in all conditions and participants. In the decoding analysis, a multidimensional transfer function was estimated using all EEG channels as input (R) in an attempt to reconstruct separately the dialogue stream and the background stream amplitude modulations (see Fig 3A) using the receptive field function (MNE-python), with time lags (τ) of −200 to 0 ms and a common regularisation parameter (λ) of 104 [80]. Unlike the encoding analysis, this analysis yields stimulus construction for each time point of the stimulus function (see Fig 3A). Both models used a leave-one-out approach, where in each iteration all trials (except one) are selected to train the model (train set), which was then used to predict either the neural response at each EEG channel (TRF) or the speech envelope of speech streams (SER) in the left-out trial (test set). This procedure was repeated with a different train ‐ test partition in each iteration averaged over all iterations. Univariate analysis of EEG data. For the TRFs, we tested whether Attentional Task (i.e., attend speech task versus ignore speech task) modulated the TRFs averaged across 7 frontocentral electrodes (Cz, FCC1h, FCC2h, FC1, FC2, FFC1h, and FFC2h; Fig 3A), separately for the dialogue stream and the background stream for each time bin employing permutation paired t tests (20,000 permutations) using custom scripts written in Python. For SER, in accordance with [29] we calculated correlations (Pearson) between the original speech stream envelopes and their reconstructions. Thereafter, we cross-reconstruction correlated the stimulus reconstructions and the stimulus envelopes (e.g., correlation between the dialogue speech amplitude and the background speech reconstruction which should be close to zero (Fig 3A)). Finally, to estimate SER accuracy we used correlation difference scores (Δr; between the correct reconstruction correlations and the cross-reconstruction correlations (Fig 3B)). For segment-level analysis, we divided the stimulus envelopes and the stimulus reconstructions into 4 segments of equal length and calculated correlations based on these instead of the full line. The SER accuracies were analysed with different linear mixed models using IBM 18 SPSS Statistics 25. All models included the participant as a random effect and intercept. For repeated factors, the diagonal covariance structure was chosen. If there was more than one repeated factor, a random slope was added for all repeated main effects and interactions using the variance components method. The models were estimated using restricted maximum likelihood estimation with a maximum iteration number of 100 to converge and df-estimation was performed using Satterthwaite. Because SPSS does not produce effect size estimates for the fixed effects in linear mixed models, we used the formula partial η2 = F × df1 / (F × df1 + df2) [81] to approximate effect sizes where applicable. To analyse how performance in individual trials affected the dialogue stream reconstruction during the attend speech conditions, we performed a similar two-level analysis commonly used when analysing fMRI data [38]. First, we created separately for each participant a linear regression model with response (correct or incorrect) in each trial as the predictor and SER accuracy for that trial as output. Because there were not enough incorrect responses in any one sub-condition (e.g., coherent, good auditory, and good visual quality), all trials were pooled across the 8 stimulus conditions. However, because the quality manipulations might affect performance—SER accuracy associations, we added Semantic Coherence, Auditory Quality, and Visual Quality as confounds in this model. The β-weight for response was thereafter taken to the second-level analysis. The second-level analysis was a similar linear mixed model as described above with line entered as the repeated predictor and 5% trimming used for the output variable to remove noise due to the paucity of incorrect trials. Decoding analysis on the fMRI data. The preprocessing and first-level analysis pipeline for the fMRI data were the same as that described in detail in [38] (for a brief description, see Table 2). Download: PPT PowerPoint slide PNG larger image TIFF original image Table 2. fMRI preprocessing parameters and first-level GLM specifications (see [38] for details). https://doi.org/10.1371/journal.pbio.3002534.t002 Support vector machine (SVM) decoding with leave-one-run-out cross-validation [82] was used to classify each pair of the 16 conditions (Task (attend speech task, ignore speech task) × Semantic Coherence (coherent, incoherent) × Auditory Quality (good, poor) × Visual Quality (good, poor)) in the fMRI data. Each line constituted an exemplar and each voxel a feature in the analysis. The SVM was conducted with the decoding toolbox (TDT, [83]) using the beta images from the first-level GLM in the participants’ anatomical space. We used searchlight-based decoding [84] with a radius of 6 mm (isotropic), and with default settings of TDT; L2-norm SVM with regularising parameter C = 1 running in LIBSVM [85]. The resulting accuracy maps for each condition pair were thereafter projected to the Freesurfer average (fsaverage) using the participants’ own Freesurfer surface (surface smoothing: 5 mm2 full-width half maximum smoothing). The pairwise decoding accuracies were averaged within each 360 ROIs (HCP parcellation [60]) and RDMs ([86]) were constructed for each subject and each ROI. All RDMs in this study are displayed rank-scaled and the conditions are ordered so that the 8 attend speech task conditions are first and the ignore speech task are second. The coherence and quality conditions are in the following order (coherent: co, incoherent: inco, good: g, poor: p, visual quality: v, auditory quality: a; co-gv-ga, inco-gv-ga, co-pv-ga, inco-pv-ga, co-gv-pa, inco-gv-pa, co-pv-pa, inco-pv-pa). The fMRI RDMs were compared with the attentional task model (Fig 3C). First, model and data RDMs were vectorised (lower triangular) and then correlated (Spearman r) with each other for each ROI and each participant. The statistical significance of the mean correlation above zero was tested with right-tailed t test, FDR-corrected for 360 ROIs. Decoding analysis of SER accuracies. SVM decoding with leave-one-run-out cross-validation [82] was used to classify SER correlations as either belonging to the attend speech task or the ignore speech task. Each line constituted an exemplar and the 4 correlations used to define SER accuracies in the univariate analyses (see above) were used as features (r: reconstruction of the dialogue stream envelope × the dialogue stream envelope, reconstruction of the dialogue stream envelope × the background stream envelope, reconstruction of the background stream envelope × the background stream envelope, reconstruction of the background stream envelope × the dialogue stream envelope). The SVM was conducted with the decoding toolbox (TDT, [84]) using the SER correlations from each participant, with default settings of TDT; L2-norm SVM with regularising parameter C = 1 running in LIBSVM [85], 100 iterations. The resulting accuracies were thereafter analysed using linear mixed models. Multivariate analysis of TRFs. RDMs were constructed from the dialogue TRFs as well as background speech TRFs separately for each time point by calculating 1-r (Spearman) of all conditions across all channels. Like fMRI, the TRF RDMs were compared to model RDMs (see S3 Fig). First, model and data RDMs were vectorised (lower triangular) and then correlated (Spearman r) with each other for each time point and each subject. The statistical significance of the mean correlation above zero was tested with right-tailed t test, FDR-corrected for 100 time points. TRF-fMRI fusion. Representational similarity analysis was used to combine EEG and fMRI data [45,46]. The TRF RDMs for 100 time points (0 to 800 ms) were correlated (Spearman r) with the 360 fMRI RDMs. Prior to correlations, the TRF RDMs were averaged across subjects to reduce noise in the data. Furthermore, partial correlation (Spearman r) was used, and the effect of task and background speech was controlled for when fusing dialogue TRFs and fMRI, and the effect of task and dialogue was controlled for when fusing background speech TRFs and fMRI. The statistical significance of the mean correlation above zero was tested with right-tailed t tests, FDR-correction was applied for time points, ROIs, and models (task and dialogue/background speech). Supporting information S1 Video. Video illustration of the full-time series TRF-fMRI fusion and results for the dialogue stream. TRFs were separately estimated for the dialogue streams for each combination of semantic coherence and audiovisual quality and EEG channel. Average TRFs are displayed for frontocentral electrodes in the middle column (attend speech: red, ignore speech: blue). We constructed TRF RDMs for each time point by correlating each EEG channel TRF pairwise across conditions and participants. Similar fMRI RDMs were constructed based on SVM decoding between the 16 condition pairs from fMRI data. Thus, we constructed similar RDMs for the EEG and the fMRI, allowing us to fuse information from both datasets by correlating vectorised TRF RDMs with fMRI RDMs, controlling for task and opposite speech stream TRF RDMs (see Fig 3C). To identify fMRI activations which corresponded to TRF RDMs at different time points, we conducted one-sample t tests (df = 18, FDR corrected) averaged across HCP parcellation ROIs. Code and processed EEG and fMRI data used to generate the video are archived on the Open Science Framework; HTTPS://DOI.ORG/10.17605/OSF.IO/AGXTH. https://doi.org/10.1371/journal.pbio.3002534.s001 (AVI) S2 Video. Video illustration of the full-time series TRF-fMRI fusion and results for the background stream. TRFs were separately estimated for the background streams for each combination of semantic coherence and audiovisual quality and EEG channel. Average TRFs are displayed for frontocentral electrodes in the middle column (attend speech: red, ignore speech: blue). We constructed TRF RDMs for each time point by correlating each EEG channel TRF pairwise across conditions and participants. Similar fMRI RDMs were constructed based on SVM decoding between the 16 condition pairs from fMRI data. Thus, we constructed similar RDMs for the EEG and the fMRI, allowing us to fuse information from both datasets by correlating vectorised TRF RDMs with fMRI RDMs, controlling for task and opposite speech stream TRF RDMs (see Fig 3C). To identify fMRI activations which corresponded to TRF RDMs at different time points, we conducted one-sample t tests (df = 18, FDR corrected) averaged across HCP parcellation ROIs. Code and processed EEG and fMRI data used to generate the video are archived on the Open Science Framework; HTTPS://DOI.ORG/10.17605/OSF.IO/AGXTH. https://doi.org/10.1371/journal.pbio.3002534.s002 (AVI) S1 Data. Data frames to reproduce plots displayed in Fig 2B–2D. We used the IBM 18 SPSS Statistics 25 (IBM SPSS, Armonk, New York, USA), UNIANOVA method, and EMMEANS to derive cell means and error terms. In Fig 2B, dependent variable was SER and factors were Segment and Condition. In Fig 2C, dependent variable was Percent_signal_change and factors were Line. In Fig 2D (left), dependent variable was SER and factors were Line and Condition. In Fig 2D (middle), dependent variable was Decoding_accuracy and factors were Line. In Fig 2D (right), dependent variable was Beta_weight_trimmed and factors were Line, note that the untrimmed Beta_weight is also provided. https://doi.org/10.1371/journal.pbio.3002534.s003 (XLSX) S2 Data. Data frames to reproduce plot displayed in S1 Fig. We used the BM 18 SPSS Statistics 25 (IBM SPSS, Armonk, New York, USA), UNIANOVA method, and EMMEANS to derive cell means and error terms. The dependent variable was SER and factors were Coherence, Auditory_quality, and Visual_quality. https://doi.org/10.1371/journal.pbio.3002534.s004 (XLSX) S3 Data. Data frames to reproduce plot displayed in S2 Fig. We used the IBM 18 SPSS Statistics 25 (IBM SPSS, Armonk, New York, USA), UNIANOVA method, and EMMEANS to derive cell means and error terms. In S2 Fig (left), the dependent variable was Mean_performance_percent and factors were Coherence, Auditory_quality, and Visual_quality. In S2 Fig (middle), the dependent variable was Mean_performance_percent and factors were Coherence, Auditory_quality, and Visual_quality. In S2 Fig (left), the dependent variable was Mean_performance_percent and factors were line. https://doi.org/10.1371/journal.pbio.3002534.s005 (XLSX) S1 Text. Supplementary results text. https://doi.org/10.1371/journal.pbio.3002534.s006 (DOCX) S1 Fig. SER-accuracy changed depending on Attentional task, Semantic Coherence, Auditory, and Visual Quality. SER accuracy was estimated using SER-correlation difference scores (see 4.9.4, for details; here the difference between the attend task and the ignore task is displayed). Error bars denote ± SEM. Abbreviations: p, poor; g, good; inco, incoherent; co, coherent; V, visual quality. Code and processed EEG data used to generate this figure are archived on the Open Science Framework; HTTPS://DOI.ORG/10.17605/OSF.IO/AGXTH. Data frames are available in S2 Data. https://doi.org/10.1371/journal.pbio.3002534.s007 (TIFF) S2 Fig. Behavioural results. Performance (percentage of correct answers) for the attend speech task (left), ignore speech task (middle). The rightmost plot shows the percentage of correct answers for the attend speech task by number of lines in the dialogue. Error bars denote ± SEM. Abbreviations: p, poor; g, good; inco, incoherent; co, coherent; V, visual Quality. Data frames are available in S3 Data. https://doi.org/10.1371/journal.pbio.3002534.s008 (TIFF) S3 Fig. Representational dissimilarity (RDM) model correlations for the temporal response function (TRF) RDM time series separately for dialogue and background speech streams. We constructed all possible main effect and interaction RDM models and correlated them with the speech stream RDM time series (see section 4.11). The attentional task (Att. vs. ign.) model yielded significant correlations (FDR corrected) for both the dialogue and the background stream TRF-RDM time series. No other model (including those not displayed here) yielded significant FDR-corrected correlations. Here, we display some of the models that yielded reliable uncorrected, p < 0.05 correlation. Comparing the plots displayed with the TRF-fMRI fusion displayed in Fig 4 and S1–S2 Videos reveals that early effects for the dialogue stream (i.e., 0–150 ms) were affected by an interaction between Att. vs. ign. × Auditory Quality × Visual Quality. The effects between 250–450 were affected by a main effect of Visual Quality, and interactions between Att. vs. ign. × Auditory Quality, Att. vs. ign. × Coherence × Auditory Quality and Coherence × Auditory Quality × Visual Quality. The interactions that occurred around ca. 700 ms were affected by Att. Vs. ign. × Auditory Quality × Visual Quality. For the background speech, the early effects (around 0–50 ms) were affected by a Coherence × Visual Quality and the late (around 300 ms) by Coherence × Auditory Quality × Visual Quality. Code and processed EEG and fMRI data used to generate this figure are archived on the Open Science Framework; HTTPS://DOI.ORG/10.17605/OSF.IO/AGXTH. https://doi.org/10.1371/journal.pbio.3002534.s009 (TIFF) S4 Fig. TRF-fMRI fusion results averaged for 10 different a priori selected regions of interest (ROIs). We performed TRF-fMRI fusion to reveal when there were correspondences between the information structure of the TRFs and the fMRI data (for details see main text, Fig 4 and S1–S2 Videos). Here, we display the TRF-RDM-fMRI correlation time series averaged in 10 a priori selected ROIs of the HCP parcellation (for ROI names, see [60]). We selected A1 and PBelt based on the meta-analysis on selective attention effects for speech stimuli in the auditory cortex [87]. The STS ROIs (TPoJ1, STSdp, STSda) were chosen because these regions showed the strongest effects of selective attention in [38]. The IFG ROIs [44,45] were selected because they were described as central nodes of the speech control networks in our previous fMRI studies [38–40,42]. V1 and MT was chosen because the presented speech was audiovisually presented, i.e., contained moving visual speech. The dorsolateral prefrontal region p9-46v was chosen because this region was shown to change its connectivity with the auditory cortex during selective attention to speech in our previous analyses on the fMRI data [38]. Correlations are displayed for the dialogue and the background sound stream as well as the attentional task model (temporally demeaned). Code and processed EEG and fMRI data used to generate this figure are archived on the Open Science Framework; HTTPS://DOI.ORG/10.17605/OSF.IO/AGXTH. https://doi.org/10.1371/journal.pbio.3002534.s010 (TIFF) Acknowledgments We would like to thank Viivi Kanerva, Elisa Sahari, and Artturi Ylinen for help with the gathering of the EEG and fMRI data and Ilkka Muukkonen for consultation with the decoding analyses.

Bussi, Claudio;Gutierrez, Maximiliano G.

doi: 10.1371/journal.pbio.3002576pmid: 38517908

Lysosome heterogeneity at the intracellular level is well documented and related to many factors. The positioning of a lysosome within the cell is not random; it is strategic for lysosomal function. For example, perinuclear lysosomes are often involved in degradation, whereas peripheral lysosomes are involved in plasma membrane repair [1]. The size of a lysosome can also differ, influenced by factors such as cellular metabolic needs or external stimuli [2]. Shape is important, and tubular lysosomes have been implicated in a wide range of cellular functions. Adding to this complexity, proteolytic activity and ion concentration have crucial roles in shaping lysosome heterogeneity [1,2]. Biochemical differences in individual lysosomes will also affect the outcomes after damage and leakage of contents. In fact, there is compelling evidence that not all lysosomes undergo damage and/or repair, which suggests the presence of an intrinsic factor that impacts membrane stability [3]. The diversity of lysosome biochemical properties indicates a range of functions beyond degradation, from antigen presentation to cell death regulation [1], and poses a critical challenge: can we correlate these distinct states with specific functions at the individual lysosome level? To effectively tackle the challenge of lysosome heterogeneity and its correlation with specific cellular functions, we propose the adoption of a “lysosome states” framework. This approach advocates for a detailed classification of lysosomes on the basis of their molecular signature, functional capabilities, localization, and morphological characteristics, each considered at the individual organelle level. By capturing the unique features of each individual lysosome within a larger interconnected network, this framework transcends the limitations of a “one size fits all” model of lysosome function and dynamics. Most biochemical and cellular characterizations of lysosomes have been conducted using standard tumor-derived cell lines. However, significant differences already exist among these cell lines [4], and even more pronounced distinctions emerge when comparing them to primary differentiated cells [5,6]. Data from our group and others show that the endolysosomal system in these standard cell line models differs from that in differentiated cells [5,6] and in proliferating but non-tumor-derived cells, such as stem cells [7]. These cell type-dependent variations jeopardize broad generalizations of lysosome function and dynamics, emphasizing the importance of considering the diversity inherent in different cell types for a more comprehensive understanding of lysosome biology. Although expanding the range of cell models used in lysosome research would indeed be beneficial, a more critical adjustment could be useful. Our primary focus should shift towards developing methodologies that can accurately account for this diversity. This approach would enable a more precise understanding of lysosome function and dynamics, reflective of the complex biological reality, thereby enhancing the reliability and applicability of lysosome studies in advancing cellular and molecular biology. The diverse compositions and activities of lysosomal enzymes observed in immune cells compared with those in conventional epithelial lines highlight the limitations of using universal models to capture lysosome function across cell types. While comparing lysosome properties across studies poses challenges, analyzing different cell types in parallel reveals significant variations in proteolytic activities, morphologies, and intracellular heterogeneity. Our analysis further reinforces this notion (Fig 1), demonstrating clear differences in lysosome morphology and activity between iPSC-derived macrophages, THP-1 cells, and commonly used epithelial cell lines. For example, macrophage lysosomes exhibit a considerably larger area compared with their epithelial counterparts, with further variations observed within epithelial cell lines themselves (Fig 1A). Lysosome number also show significant variation, with macrophages and RPE-1 cells harboring the highest numbers, and HEK-293T cells possessing the fewest (Fig 1B). Notably, functional differences in lysosome proteolytic activity are also evident, with macrophages displaying higher proteolytic capacity than epithelial cells (Fig 1C and 1D). Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 1. Single lysosome properties across different cell types. (A) Lysosome size distribution (quantified as the area based on LysoTracker-positive puncta) across different cell types. The plot displays a weighted distribution to emphasize the percentage of lysosomes larger than 0.5 μm2. (B) Density plots illustrating the number of lysosome puncta normalized to cell area across the indicated cell types. (C) Evaluation of lysosome proteolytic activity using a pan-cathepsin activity-based probe. (D) Representative images depict fluorescence intensity levels of the pan-cathepsin activity-based probe (Fire LUT scale). All quantifications were based on single-cell and single-object (puncta) segmentation using an Opera Phenix High-Content Microscope, involving n ≥ 300 cells and 3 independent experiments. All measurements were conducted simultaneously using DMEM media. Scale bar: 10 μm. (E) Evaluation of GAL-3 puncta in cells left untreated or treated with LLOMe (1 mM) for 10 min and 1 h. Representative images are included. All quantifications were based on single-cell and single-object (puncta) segmentation using an Opera Phenix High-Content Microscope, involving n ≥ 300 cells and 3 independent experiments. All measurements were conducted simultaneously using DMEM media. Scale bar: 10 μm. Plots were done using R Studio 2023.09.1. https://doi.org/10.1371/journal.pbio.3002576.g001 Adding another layer of complexity, lysosomes show heterogeneity not only across cell types but also within them, encompassing variations in membrane stability, acidification, and degradation capabilities. Recent findings challenge the paradigm that galectin-3 (GAL-3)-positive lysosomes were assumed to be destined for lysophagy [8]. These data suggest that GAL-3 is not a universal marker of membrane damage and highlight the possibility of lysosome repair circumventing GAL-3 involvement under specific conditions [9]. The dynamics of lysosome damage and repair, crucial for lysosome function, also showcase marked variations across cell types. Analyzing GAL-3 recruitment kinetics following lysosome damage in various cell types exemplifies this point (Fig 1E). While the potential role of other galectins in detecting lysosomal damage cannot be excluded [10], these findings emphasize the limitations of applying universal criteria and single markers to evaluate lysosome quality control mechanisms across diverse cell types. Each cell type might require context-specific evaluation due to distinct marker sets and mechanisms. This heterogeneity underscores the intrinsic complexity of cellular processes and necessitates context-specific evaluations when studying lysosome functions and related mechanisms. Furthermore, the discovery of several ESCRT-independent repair mechanisms alongside ESCRT-dependent pathways suggests a more intricate landscape of lysosome quality control than previously appreciated [3]. This raises intriguing questions. Does a specific repair mechanism dominate depending on cell type and stimuli? Can they coexist or target distinct lysosome populations? Identifying the specific molecular cues triggering these mechanisms remains a critical question. Considering the observed heterogeneity among lysosome functions across different cell types, these findings collectively underscore the critical need for a refined approach to study lysosome biology. The marked differences in lysosome enzymes, morphology, and activity among cell lines and primary cells further highlight the inadequacy of a “one size fits all” approach in lysosome research. Consequently, we advocate for the use of multiplex profiling approaches as a concerted effort to identify and characterize lysosome states. By leveraging advanced techniques such as 3D quantitative live-cell imaging, this initiative aims to enhance our understanding of lysosome dynamics through context-specific evaluations. To define the functions and dynamics of individual lysosomes (with the aim of understanding these states), approaches that shift away from lysosome population studies will be required. For example, although lysosome immunopurification is valuable in exploring cell-to-cell heterogeneity, its use is limited by its reliance on a single marker [11]. This dependence on marker expression, which can vary by lysosome type, cell type, and condition, hinders its accuracy in capturing individual lysosome diversity, highlighting the need to reevaluate not only our definition of a lysosome (including marker selection) but also our understanding of transient states such as repaired lysosomes, lysophagy-targeted lysosomes, and secretory lysosomes. The future of lysosome research is bright. Advances in molecular tools capable of monitoring lysosome properties at single-organelle resolution offer a promising avenue for transcending previous limitations. The advent of novel reporters and probes capable of elucidating functional lysosomal attributes, such as proteolytic activity and ion concentration, at the level of individual organelles [12] and imaging techniques for tracking single organelles, in conjunction with specific markers and reporters, offer a potent strategy for unveiling the functional heterogeneity of lysosomes and providing real-time, dynamic insights into single lysosome function. They also allow for the early detection of subtle lysosome changes, which may precede the manifestation of lysosome-related diseases. Given the complexities and inherent variability in lysosome functions across different cellular contexts, we believe the introduction of the lysosome states framework represents a significant leap forward in our understanding of lysosome dynamics and function. With each lysosome potentially serving as a sentinel of cellular health, the implications for diagnostics and treatment are substantial, charting a transformative path for future research in cell biology and pathology.

Jiang, Xiaowen;Zhao, Kun;Sun, Yi;Song, Xinyue;Yi, Chao;Xiong, Tianlong;Wang, Sen;Yu, Yi;Chen, Xiduo;Liu, Run;Yan, Xin;Antos, Christopher L.

doi: 10.1371/journal.pbio.3002565pmid: 38527087

Introduction The scaling of anatomical structures involves the coordinated control of gene transcription and intracellular communication. An increasing number of findings are linking electrophysiological changes to the regulation of developmental phenomena in different biological contexts [1,2]; as examples, transmembrane voltage potential influences eye development via regulation of Pax6 [3]; Hedgehog signaling is regulated in Drosophila wing discs by cell depolarization [4]; and inactivation of K+ inward rectifying channels leads to patterning defects in the craniofacial skeleton, vertebrate phalanges, and fly wings via disruption of BMP/Dpp signaling [5,6]. Furthermore, gain-of-function mutations in channels that facilitate outward flow of intracellular K+ are also linked to syndromes that generate craniofacial alterations, neurodevelopmental impairments, defects in the development of the limbs, etc. [7,8]. As discoveries for the involvement of electrophysiological regulation in development increase, much remains unknown about how these endogenous electrophysiological changes are integrated into the known molecular mechanisms that regulate development. In zebrafish, mutations that increase the activity of K+ channels promote allometric growth of juvenile and adult fish fins [9–13], linking intracellular K+ in the control of coordinated proportional growth of these anatomical structures. The K+ concentration is high inside cells, so the opening of K+-leak channels generally causes an outward flow of K+ [14]. In the adult fin, increasing K+-leak channel activity increases the transcription of several morphogens and promotes growth [11]. However, it remains unknown how K+ channel activity at the cell membrane is controlled and relayed during the scaling process of fish appendages, and whether a similar electrophysiological control exists in the conserved embryonic vertebrate fin/limb bud developmental program. Early embryonic limb and pectoral fin buds form at specific locations in the lateral plate mesoderm that expresses an evolutionary conserved profile of morphogens and growth factors [15]. The development of limb buds and pectoral fin buds initiate when retinoic acid (RA) indirectly promotes the distal transcription of Fgf10 [16], which in turn induces the expression of Fgf8 and the formation of the apical ectodermal ridge (AER) in the distal anterior–posterior interface of ectodermal cells of the buds [17,18]. After formation of the AER, the morphogen Sonic hedgehog (Shh) manifests in a group of bud mesenchymal cells known as a zone of polarizing activity (ZPA). The ZPA forms in the posterior bud mesenchyme and is required for growth and patterning through the activities of Shh [19,20]. Control of the AER and ZPA involve communication between these 2 regions, as well as other regional cell groups via their morphogens/growth factors (e.g., Bmp4) [21]. Similar to early limb bud development, activation of these signaling centers and other morphogens promote pectoral fin bud outgrowth [22]. For the zebrafish pectoral fin bud, growth is underway by 28 to30 hours post fertilization (hpf) [23]. During limb development, growth persist until the formation of the distal digits [24], but in zebrafish, pectoral fin bud development is limited to forming the endochondral bone, musculature, and other tissues at the proximal base of the fin [25,26]. Roughly 24 h after initiation of fin bud outgrowth, its development has transitioned into the development of a different type of appendage: the finfold, and many of the morphogen/growth factor signals have rearranged to the distal tip of the fin bud by 54 to 56 hpf to generate the finfold and then ultimately larval pectoral fins [23,27]. While the basic interactions between signaling centers and signaling molecules during early fin/limb bud development is understood, it is unknown whether changes in intracellular K+ will coordinately regulate them in a similar manner as it does during developmental/regenerative growth of fins. Using a genetic sensor for K+ [28] with Fluorescence Lifetime Microscopy (FLIM), we found that relative levels of intracellular K+ decrease throughout the early pectoral fin bud during its growth. When we transgenically overexpressed K+-leak channels that decrease in intracellular K+, we increased bud growth and coordinately increased the expression several morphogens (fgf8a, fgf10a, aldh1a2, shha, and bmp4) that control bud development. Treatment with RA, a signal that can promote growth of vertebrate appendages, decreased intracellular K+ levels in the buds. We found that RA induces rcan2 expression and that rcan2 scales fin buds and decreases intracellular K+. In addition, we found that the K+-leak channel kcnk5b is expressed in pectoral fin buds and that it is required for Rcan2-mediated scaling. We also determined that Kcnk5b promotes depolarization and that depolarization is required for enhanced fin bud growth. We further show that Kcnk5b activity requires IP3R-mediated Ca2+ release and CaMKK activity for SHH transcription in vitro and in vivo as well as for kcnk5b’s enhancement of fin bud growth. Thus, we provide a mechanism through which an important proximal morphogen (RA) can regulate the activity of a K+-leak channel via Rcan2 to promote Ca2+-mediated scaling of embryonic pectoral fins buds. Results Endogenous intracellular K+ levels decrease during fin bud growth We previously showed that Kcnk5b promotes allometric growth of adult fins via hierarchical activation of several morphogens [11]. The ability of an individual K+ channel to promote the expression of several developmental morphogens in the adult fin and larva led us to hypothesize that this phenomenon may be more broadly functional, and that it regulates the early developmental program conserved in vertebrate fin/limb buds, which is present in the developing pectoral fin buds of zebrafish [22]. Therefore, we wanted to determine whether there are endogenous changes in intracellular K+ in the developing pectoral fin bud. To measure intracellular K+, we used an established FRET-based genetic sensor (KIRIN1) that can detect changes specifically in intracellular K+ levels [28]. To overcome the limitations associated with FRET measurements in vivo, we used FLIM, because FLIM does not rely on intensity-based ratios, rather, FLIM detects changes in the intrinsic decay rate of a donor fluorophore (CPF) when it undergoes FRET to the acceptor fluorophore (YFP) within the sensor (Fig 1Aa and 1Ab). One can directly quantify the accumulated decay profile of a fluorophore (CFP) after excitation by laser pulses [29] (Fig 1Ba and 1Bb); consequently, we could detect relative differences in intracellular K+ caused by increases in K+-leak channel expression (S1Aa and S1Ab Fig), or by posttranslational regulation of a K+-leak channel (S1B Fig) or by different K+-leak channel mutants that have different activities (S1C Fig) [11]. These experiments confirmed that we can detect changes in relative intracellular K+ levels with the KIRIN1 sensor using FLIM. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 1. Intracellular K+ decreases during pectoral fin bud growth. (A) FRET-based mechanism of detection of intracellular K+ by KIRIN1 sensor and application of FLIM assessment. Changes in FRET affect the fluorescence lifetime of the donor fluorophore CFP that is excited by a 2-Photon laser pulse (a). A reduction of intracellular K+ is detected as higher (longer) lifetime of the CFP, while increase in K+ is detected as its lower (shorter) lifetime (b). (B) We depict these lifetime (nanoseconds, ns) relationships of higher lifetime equating to lower intracellular K+ (a) and the inverse (original) as 1/lifetime (ns-1) (b) in order to more easily relate the trend changes with what the FLIM-based measurements are indicating for intracellular K+. (C) Design of K+ sensor (KIRIN1) transgene for transgenic zebrafish line (a). Heat-shock method for inducing transgene during different developmental time points of the fin bud and subsequently FLIM-FRET (FILM) imaging it (b). (D) FLIM measurements of the CFP in confocal sections of ectoderm (Ecto) and mesenchyme (Mesen) in pectoral fin buds at 48 hpf using a FRET-based sensor for K+. Measurements from a transgene containing only CFP that is unable to FRET with YFP. (E) Time course of FLIM measurements of intracellular K+ in the ectodermal and mesenchymal tissues of the developing pectoral fin buds. (F) Time course of growth measurements of the developing fin buds. (G–I) Z-stacks of confocal FLIM planes in fin buds at 32 hpf (G), 48 hpf (H), and 56 hpf (I). Z-stacks cover distances of 15.2 μm (32 hpf), 18.36 μm (48 hpf), and 10.3 μm (56 hpf). The lifetime value of each pixel in the ectoderm (white arrowhead) and the mesenchyme (asterisk) is represented by colors in a rainbow scale from 2.8 ns (blue) to 3.4 ns (red). (J) Generation of fin buds that mosaically express the KIRIN1 sensor and either mCherry or kcnk5b-mCherry (mosaic for 2 transgenes) and proposed outcomes. (Ka,b) Images of cells in fin buds of 56 hpf embryos harboring hsp70:mCherry (red asterisks) and cells lacking the mCherry transgene (white asterisks). Confocal plane for mCherry fluorescence (a). FLIM image of KIRIN1+ cells of the same confocal plane (b). (La,b) Images of cells in fin buds of 56 hpf embryos harboring hsp70:kcnk5b-mCherry (red asterisks) and cells lacking the transgene (white asterisks). Confocal plane for mCherry fluorescence (a). FLIM image of KIRIN1+ cells of the same confocal plane (b). (M) FLIM measurements in fin buds of 56 hpf embryos of indicated cell categories; “near” indicates cells next or distant to mCherry-positive (mCherry+) or kcnk5b-mCherry-positive (kcnk5b+) cells. The graphs are depicted as 1/lifetime (ns-1) to more easily relate the portrayal of the values to the related change in intracellular K+ (see S1A Fig). Experiments were repeated 3 or more time (N ≥ 3). For FLIM, we measured 2 or 3 locations in each tissue of 1 fin bud per embryo. Each measured value is represented as a data point (D, M), except for E, in which the data points are presented as averages and standard deviations of all the measurements (≥11) at each time point. For fin bud size measurements (F), each data point represents 1 fin bud per embryo. P values represent statistical analysis by Student’s two-tailed t test. P values >0.05 are designated as “not significant” (NS). Numerical data used in this figure are included in S1 Data. FLIM, Fluorescence Lifetime Microscopy; hpf, hours post fertilization. https://doi.org/10.1371/journal.pbio.3002565.g001 To assess the intracellular K+ in vivo, we generated a heat-shock-inducible transgenic reporter Tg[hsp70:CFP-KIRIN1-YFP] (hsp70:KIRIN1) line (Fig 1Ca) and used it to detect relative differences in K+ levels via FLIM 6 h after heat shock (Fig 1Cb). FLIM measurements in the pectoral fin bud showed relatively lower intracellular K+ levels in the mesenchyme compared to ectoderm at 48 hpf (Fig 1D), a time point in which the growth of the pectoral fin buds is underway. We subsequently measured several time points during pectoral fin bud outgrowth from the same animals, and we observed an increase in intracellular K+ in both the ectoderm and the mesenchyme between 32 and 34 hpf, but afterwards, intracellular K+ levels decreased to their lowest levels at around 48 hpf (Fig 1E). Intracellular K+ levels increased again between 54 and 56 hpf (Fig 1E). From the plotted time points, we observed that although mesenchyme K+ levels remained lower than the ectoderm, the changes in K+ in both tissues were coupled (Fig 1E). These lifetime measurements were not influenced by differences in intensity levels (S2D–S2H Fig) or by the method for immobilizing the embryos (S2I–S2O Fig). To ascertain whether there is a relationship between relative changes in intracellular K+ and growth, we measured fin bud sizes between 30 hpf to 56 hpf. From bud area measurements, we observed incremental growth from 30 hpf to 56 hpf (Fig 1F). We also observed 2 pauses in average growth between 32 hpf and 34 hpf and 54 hpf and 56 hpf (Fig 1F) that correlated with time points in which intracellular K+ increased (Fig 1E). Together, these results suggested a coordinated regulation of intracellular K+ that relates to fin bud growth. To visualize the relative distribution of intracellular K+ during pectoral fin bud development, we represented the lifetime values along a rainbow scale: the lower the value (the higher intracellular K+ levels), the more towards blue; conversely, the higher values (the lower intracellular K+ levels), the more towards red. From the colorized individual confocal FLIM planes (S3A–S3C Fig) and 3D Z-stacks of FLIM planes through fin buds at 32 hpf (Figs 1Ga–1Ge and S3A), at 48 hpf (Figs 1Ha–1He and S3B), and at 56 hpf (Figs 1Ia–1Ie and S3C), we observed clear differences in K+ levels between the mesenchyme and ectoderm, but few regional differences within each tissue. We also observed global decreases in intracellular K+ between 32 hpf (Figs 1Ga–1Ge and S3A) and 48 hpf (Figs 1Ha–1He and S3B) when the growth of the bud is high. Conversely, we observed global increases in K+ between 48 hpf and 56 hpf (Figs 1Ia–1Ie and S3C) when growth reduces. From these observations, we propose that there is a decrease in intracellular K+ during fin bud growth due to increases in K+-leak channel activity, and these changes are linked to the growth of the bud. The lack of regional differences in intracellular K+ in either the mesenchyme or ectoderm suggested equilibration of K+ between the cells of each tissue. To test this hypothesis, we generated double mosaic embryos by injecting recipient embryos with the heat-shock-inducible hsp70:KIRIN1 and either kcnk5b-mCherry transgene (hsp70:kcnk5b-mCherry) or with mCherry (hsp70:mCherry) transgene as a control (Fig 1J). We identified mCherry-expressing cells (Fig 1K) or kcnk5b-mCherry cells (Fig 1L) in fin buds and then assessed intracellular K+ in mCherry-positive cells (Fig 1Ka, 1La, and 1M and red asterisks in Fig 1Kb and 1Lb) and the surrounding mCherry-negative cells (Fig 1K and 1L, white asterisks). Compared to cells that harbored the mCherry transgene (Fig 1K, red asterisks), we observed that cells expressing kcnk5b-mCherry (Fig 1L, red asterisks) and cells surrounding the transgene-expressing cells displayed decreases in K+ levels (Figs 1M and S3D). These data indicated that changes in intracellular K+ are shared such that cells with more K+-leak channel activity can decrease the intracellular K+ of neighboring cells. Overexpression of a K+-leak channel in early fin bud coordinately enhances the expression of the important morphogens to scale proximodistal growth The coordinated decrease in the endogenous levels of K+ in the developing pectoral fin bud suggested that intracellular K+ has a role in bud development. Consequently, we wished to know whether decreasing intracellular K+ is sufficient to enhance the scale of the buds. To decrease intracellular K+, we overexpressed Kcnk5b or Kcnk10 K+-leak channel by heat-shock induction of the Tg[hsp70:kcnk5b-GFP] [11] or Tg[hsp70:kcnk10a-GFP] transgenic lines. Six hours after transgene induction at 48 hpf (S4A–S4C Fig), we measured the growth area of the fin buds and standardized the measurements of each bud to the area of the eye (Fig 2A and 2B) or the otic vesicle (S4D and S4E Fig). From these analyses, we observed enhanced growth of pectoral fin buds caused by kcnk5b or kcnk10 compared to heat-shocked non-transgenic siblings and transgenic Tg[hsp70:GFP] control groups at 32 hpf, 48 hpf, and 56 hpf when standardized to the area of the eye (Fig 2C) or of the area of the otic vesicle (S4F Fig). Together, these data indicated that reducing intracellular K+ increases growth of the buds. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 2. K+-leak channels enhance the growth of pectoral fin buds. (A, B) Brightfield images of a post-heat-shocked 48 hpf non-transgenic embryo (A) or post-heat-shocked 48 hpf hsp70:kcnk5b-GFP transgenic embryo (B). The area of the fin bud (black-dotted line). (C) Measurements of pectoral fin bud areas of heat-shocked groups of non-transgenic (AB), Tg[hsp70:GFP], Tg[hsp70:kcnk5b-GFP], and Tg[hsp70:kcnk10-GFP]. The non-transgenic and Tg[hsp70:kcnk5b-GFP] embryos are siblings. Each measured bud was standardized to the eye area (red-dotted circles in A,B) in the same embryo. (D–S) Expression of the indicated morphogens. (D, G, J, M) Lateral views of whole-mount in situs of heat-shock control non-transgenic embryos at 48 hpf for fgf8a (D), fgf10a (G), shha (at 52 hpf) (J), aldh1a2 (M) and dorsal view of aldh1a2 (O) or lateral views of heat-shocked Tg[hsp70:kcnk5b-GFP] embryos at 48 hpf for fgf8a (E), fgf10a (H), shha (at 52 hpf) (K), aldh1a2 (N), and dorsal view of aldh1a2 (P). White arrows indicate proximal expression of aldh1a2. qRT-PCR measurements from fin buds for fgf8a (F), fgf10a (I), shha (L), and aldh1a2 (Q). Lateral views of in situs of heat-shocked non-transgenic fin bud for bmp4 (R) and msx2b (U) and of heat-shocked Tg[hsp70:kcnk5b-GFP] embryos at 48 hpf for bmp4 (S) and msx2b (V). qRT-PCR measurements from fin buds for bmp4 (T), msx2b (W) the in situ experiments were repeated 3 or more time (N ≥ 3). Each in situ repeat contains 6–12 embryos per replicate. For the fin bud size measurements, we measured 1 fin bud and eye per embryo. We measured at least 4 embryos per experimental repeat. For the qRT-PCRs, we collected 80 fin bud samples per isolation. We assessed gene expression with 3 or more isolations. Each isolation was assessed in duplicate or triplicate. Each measured value is represented as a data point. P values represent statistical analysis by Student’s two-tailed t test. P values ≥0.05 are designated as “not significant” (NS). The scale bars are 0.5 mm (A, B), 100 μm (D–P, R–S,U–T). Numerical data used in this figure are included in S2 Data. hpf, hours post fertilization. https://doi.org/10.1371/journal.pbio.3002565.g002 The increase in pectoral fin bud size from K+-leak channel overexpression indicated that the developmental gene program was affected. We therefore examined the expression of selected morphogens known to control early bud growth at 48 hpf. We observed that compared to heat-shocked control siblings (Fig 2D, 2G, 2J, 2M and 2O), transgenic expression of kcnk5b increased the expression of fgf8a (Fig 2D and 2E), fgf10a (Fig 2G and 2H), shha (Fig 2J and 2K), and aldh1a2 (Fig 2M–2P). We observed similar up-regulation of these genes from qRT-PCR analyses of fin buds (Fig 2F, 2I, 2L and 2Q). Because previous findings show that BMP signaling in vertebrate and invertebrate appendages is affected by K+ channels [5,6], we also assessed the expression of an important BMP ligand (bmp4) and its down-stream target msx2b. We observed that compared to in situ controls (Fig 2R and 2U), kcnk5b increased the expression of both genes in the in buds (Fig 2S and 2V). We also observed similar up-regulation of these genes by qRT-PCR (Fig 2T and 2W). Thus, the endogenous decreases in intracellular K+ during fin bud development (Fig 1), the coordinated increase in the expression of several important morphogens by increasing the expression of the K+-leak channel kcnk5b (Fig 2D–2W), and the enhanced growth the pectoral fin bud by kcnk5b or kcnk10a overexpression (Fig 2C) all implicated intracellular K+ as an integral part of the fin bud growth control, specifically, decreased intracellular K+ augmented bud proportional growth. Retinoic acid decreases intracellular K+ via an Rcan2-mediated mechanism that scales fin buds The decrease in intracellular K+ during pectoral fin bud growth suggested that K+ levels might be responsive to developmental signals that regulate growth. One such signal is RA [30–32]. Therefore, we examined whether RA stimulation influenced intracellular K+ levels in developing fin buds at 32 hpf by FLIM measurements of the KIRIN1 transgene 6 h after its heat-shock induction. Compared to treatment with the solvent DMSO (Fig 3A and 3C), we observed decreases in K+ after 6 h of treatment with 200 nM RA (Fig 3B and 3C), both in ectoderm and mesenchyme cells (Fig 3D). To test whether the changes in intracellular K+ were specific to RA, we assessed the effect of thyroid hormone (TH), another nuclear hormone receptor mechanism. Treatment with TH showed no significant differences in K+ levels compared to DMSO-treated controls (S5C Fig). These data indicated that RA-mediated signaling is sufficient to decrease intracellular K+ in fin bud tissues and that this effect is not induced by all nuclear hormones. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 3. Retinoic acid decreases intracellular K+ in pectoral fin buds via transcriptional activation of the calcineurin inhibitor rcan2. (A, B) FLIM images of developing fin buds from embryos treated for 6 h with DSMO-treated (A) or with 200 nM RA-treated (B) 32 hpf embryos. (C) FLIM measurements from cells in developing buds at 32 hpf of the indicated treatment groups. (D) FLIM measurements of cells in the ectoderm (Ecto) or the mesenchyme (Mesen) of buds at 32 hpf treated either with DMSO or RA. (E) In situ images of dorsal view (a) and lateral view (b) for kcnk5b expression in 48 hpf embryos. Pectoral fin buds (pfb), ectoderm (e), and mesenchyme (m). (F) Representative lateral view images of pectoral fin buds of in situ experiments for kcnk5b expression at 32 hpf (a), 56 hpf (b), and 72 hpf (c). (G) Representative lateral views of pectoral fin buds after in situ experiments for rcan2 expression at 32 hpf (a), 34 hpf (b), 48 hpf (c), and 56 hpf (d) embryos. (H) qRT-PCR of rcan2 from isolated pectoral fin buds of 48–52 hpf embryos treated with DMSO or 200 nM RA for 6 h. (I) In situ for rcan2 expression in fin buds of 48 hpf embryos after 6 h DMSO treatment. Ectoderm is indicated by (e), and mesenchyme is indicated by (m). (J) In situ images for rcan2 expression in fin buds of 48 hpf embryo after 6 h 200 nM RA treatment. Ectoderm (e), mesenchyme (m). (K, L) Pectoral fin buds of heat-shocked non-transgenic embryo (K) or heat-shocked transgenic hsp70:rcan2-mCherry sibling (L). (M) Measured fin buds were standardized to the area of an eye in the same embryo. (N) FLIM measurements of intracellular K+ in pectoral fin bud cells from heat-shocked non-transgenic and mCherry controls and transgenic rcan2-mCherry-expressing siblings. (O–R) Representative fin buds of embryos expressing empty sgRNA vector “EV” (O), rcan2 sgRNA “KO” (P), mCherry mRNA with rcan2 sgRNA “KO+mCherry” (Q), mutated rescue rcan2 mRNA with rcan2 sgRNA “KO+rcan2” (R). (S) Ratios of fin bud areas to eye areas of embryos of the indicated experimental groups. (T) FLIM measurements of intracellular K+ in the pectoral fin bud cells in control and CRISPR-Cas9 knockout of rcan2 in embryos. Experiments were repeated 3 or more time (N ≥ 3). For FLIM, we measured 2 or 3 locations in each tissue of 1 fin bud per embryo. Each measured value is represented as a data point (C, D, N, T). For the fin bud size measurements, we measured 1 fin bud per embryo (M) and 2 fin buds per embryo (S). For the qRT-PCR experiments, we collected 80 fin bud samples per isolation. Three or more isolations were measured. Each isolation was measured in duplicate or triplicate. Each isolation is represented as a data point. Each in situ repeat contained 6–12 embryos per replicate. P values represent statistical analysis by Student’s two-tailed t test. P values >0.05 are designated as “not significant” (NS). Scale bars represent 50 μm (E–G, I–L, O–R). Numerical data used in this figure are included in S3 Data. FLIM, Fluorescence Lifetime Microscopy; hpf, hours post fertilization; RA, retinoic acid. https://doi.org/10.1371/journal.pbio.3002565.g003 RA regulates gene transcription via specific intracellular receptors [33]. Consequently, the relative reduction in K+ by RA suggests that RA regulates the transcription of one or more K+ channels. Because Kcnk5b is involved in adult fin scaling [9], we performed in situ hybridization experiments to determine whether kcnk5b is present in growing fin buds of 48 hpf embryos. We observed expression primarily in distal pharyngeal pouches (Fig 3Ea), and the mesenchyme of the growing fin buds (Fig 3Ea and 3Eb). We subsequently assessed kcnk5b expression during fin bud growth at 34 hpf, at 56 hpf, and at 72hpf, a time point in which bud growth has already ceased. We observed kcnk5b expression primarily in the mesenchyme at 32 hpf (Fig 3Fa) and at 56 hpf (Fig 3Fb), but by 72 hpf, the channel expression was difficult to detect in the bud (Fig 3Fc). Because treatment with RA decreases intracellular K+ levels, we reckoned that RA may be enhancing kcnk5b transcription. Therefore, we tested whether RA-mediated decrease in intracellular K+ was related to an up-regulation of kcnk5b expression. We did not observe an increase in RA-treated embryos by qRT-PCR (S5A Fig) despite a significant increase in a known RA-activated gene (S5B Fig) [34]. We previously showed that inhibition of calcineurin increases the activity of Kcnk5b to decrease intracellular K+ and enhance the scale of the adult fin [11]. Since Kcnk5b is present in the embryonic pectoral fin bud (Fig 3E and 3F), we hypothesized that RA may decrease intracellular K+ by posttranslational regulation of Kcnk5b through the inhibition of calcineurin. RCAN proteins are well-documented in vivo inhibitors of calcineurin [35–37]. From in situ hybridization experiments, we detected expression at 32 hpf (Fig 3Ga), at 34 hpf (Fig 3Gb), and 48 hpf (Fig 3Gc). However, rcan2 expression reduced by 56 hpf (Fig 3Gd). Because RA treatment decreased intracellular K+ and Rcan2 inhibits calcineurin, which subsequently could suppress calcineurin-mediated inhibition of Kcnk5b, we tested whether rcan2 expression is altered by RA. We observed that compared to DMSO controls, RA treatment increased rcan2 expression by qRT-PCR (Fig 3H), and from in situ hybridization experiments, we observed staining in DMSO-treated buds (Fig 3I) that became more intense after RA treatment (Figs 3J and S5D). We also observed similar phenomena in the adult zebrafish fins: RA treatment decreased intracellular K+ levels (S5G Fig) and increased rcan2 transcription (S5H Fig). Furthermore, we detected rcan2 transcripts and protein in the blastemas and distal regenerating interray tissue of regenerating fins (S5I–S5K Fig), where growth occurs. We did not observe protein expression in fin rays immediately after amputation when regenerative growth has not yet commenced (S5J and S5K Fig). The ability of RA to up-regulate rcan2, which inhibits a phosphatase that suppresses the K+-leak channel Kcnk5b (present and scales fin buds and adult fins [9,11]) led us to hypothesize that Rcan2 may enhance proportional growth of fin buds. To test whether increasing Rcan2 enhances pectoral fin bud growth, we generated a fish line harboring the hsp70:rcan2-mCherry transgene to overexpress rcan2 by heat-shock induction (S5L–S5O Fig). Compared to heat-shocked non-transgenic control siblings (Fig 3K and 3M) and overexpressed mCherry (Fig 3M), induction of rcan2 enhanced the growth area of the fin buds (Fig 3L and 3M). We also observed enhanced growth by transgenic overexpression of rcan2 during adult fin regeneration (S5P–S5V Fig). To determine whether rcan2-enhanced growth correlated with reduced intracellular K+, we generated double-transgenic fish that harbored hsp70:rcan2-mCherry and hsp70:KIRIN1 transgenes. After heat-shock induction of the transgenes, we observed that rcan2 expression decreased intracellular K+ levels compared to controls (Fig 3N). Together, these data showed that induction of rcan2 expression was sufficient to enhance growth in the fin bud and adult fin as well as reduce their intracellular K+ levels in vivo. We then assessed whether targeting rcan2 by CRISPR-Cas9 (S5W–S5Z Fig) affected fin bud size and intracellular K+. Compared to control embryos (Fig 3O and 3S), we observed that rcan2-targeted embryos displayed reduced fin bud growth (Fig 3P, 3Q and 3S), which could be rescued by overexpression of rcan2 mRNA (Fig 3R and 3S) that harbored mutations in its wobble bases to impair interactions between the overexpressed mRNA and the genome-targeting sgRNA (S5Za and S5Zb Fig). We also assessed caudal fins of CRISPR-Cas9-targeted juvenile fish, which also displayed shorter fins compared to wild type (S5A and S5B Fig). When we assessed the effect of targeting rcan2 on intracellular K+ levels by targeting rcan2 in Tg[hsp70:KIRIN1] embryos, we observed that targeting rcan2 increased intracellular K+ (Fig 3T) and rescue with rcan2 mRNA returned intracellular K+ back to control levels (Fig 3T). Together, these results from embryonic fin buds and post-metamorphosis caudal fins indicated the involvement of rcan2 in K+-channel-mediated scaling of embryonic and adult appendages. Rcan2-mediated scaling requires Kcnk5b activity via its Serine 345 Calcineurin limits fin proportional growth by inhibiting Kcnk5b through serine 345 [11]. Since Rcan2 inhibits calcineurin [35–37], and kcnk5b is expressed in the developing pectoral fin buds (Fig 3E and 3F), we first assessed the importance of kcnk5b in fin bud growth by CRISPR-targeting kcnk5b (S6A–S6G Fig). Compared to controls (Fig 4A and 4E), fin buds of kcnk5b-targeted embryos were reduced in size by 48 hpf (Fig 4B and 4E). This phenotype was rescued by overexpression of kcnk5b mRNA (Fig 4D and 4E) that was mutated to impair interaction with the targeting sgRNA (S6H–S6J Fig). When we overexpressed rcan2 in kcnk5b-targeted embryos, we observed that while rcan2 overexpression in non-targeted embryos enhanced fin bud growth (Fig 4F, 4G and 4I), rcan2 required kcnk5b to enhance growth (Fig 4H and 4I). We previously showed that calcineurin inhibits Kcnk5b via serine 345 in the cytoplasmic tail of the channel [11], so we tested whether the kcnk5bS345A (the mutant with the lowest K+-leak activity) blunts rcan2 enhancement of fin bud growth in vivo. Compared to control embryos overexpressing rcan2 with the wild-type channel (kcnk5bS345) (Fig 4J and 4L), the calcineurin-dephosphorylated mimic kcnk5bS345A channel reduced rcan2-mediated fin bud growth (Fig 4K and 4L). Together, these data indicated that rcan2-enhanced growth requires Kcnk5b and that the Kcnk5b channel that mimics calcineurin-mediated inhibition of the channel prevents the enhanced growth caused by overexpression of the endogenous calcineurin inhibitor rcan2. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 4. Rcan2-mediated decrease in intracellular K+ involves Kcnk5b. (A–D) Representative 48 hpf embryos and enlarged panels of fin buds of control Cas9 and empty sgRNA vector (EV) embryo (A), CRISPR-targeted kcnk5b embryo (B), CRISPR-targeted kcnk5b embryo overexpressing GFP (C), CRISPR-targeted kcnk5b embryo overexpressing kcnk5b*-GFP that harbors altered wobble bases to impair interaction with targeting sgRNA (D). (E) Fin bud-to-eye area ratios of the indicated genotypes. (F–I) Representative 48 hpf embryos and enlarged panels of fin buds of control Cas9 and empty sgRNA vector (EV) non-Tg embryo (F), rcan2-mCherry-expressing embryo and enlarged panel of the fin bud (G) CRISPR-targeted kcnk5b embryo overexpressing rcan2-mCherry (H). (I) Fin bud-to-eye area ratios of the indicated genotypes. (J, K) Representative 48 hpf embryos and enlarged panel of the fin buds expressing wild-type kcnk5bS345-GFP and rcan2-mCherry (J), or expressing calcineurin-dephosphorylated mimic kcnk5bS345A-GFP and rcan2-mCherry (K). (L) Fin bud-to-eye ratios show that rcan2-mediated enlargement of fin buds is impaired by kcnk5bS345A mutant. Experiments were repeated 3 or more time (N ≥ 3). For the fin bud size measurements, we measured 1 fin bud (E, I) or 2 fin buds (L) per embryo at 48 hpf and not at a later time points to avoid incorporating measurements of the finfold growth that start around 56 hpf. Each measured value is represented as a data point. P values represent statistical analysis by Student’s two-tailed t test. P values >0.05 are designated as “not significant” (NS). Scale bars equal 100 μm (A–D, F–H, J, K). Numerical data used in this figure are included in S4 Data. hpf, hours post fertilization. https://doi.org/10.1371/journal.pbio.3002565.g004 Kcnk5b-enhanced growth involves cell depolarization Kcnk5b is a two-pore K+ leak channel whose activity alters the electrical membrane potential at the plasma membrane of cells. The decrease in intracellular K+ during fin bud growth (Fig 1) and the importance of Kcnk5b for growth (Fig 4) suggest there are important K+-associated changes in membrane potential during fin bud development. To determine whether membrane potential alters during fin bud development, we used DiSBAC2(3), a dye that increases its fluorescence as it enters cells when channels open to depolarize the cells [38]. We assessed DiSBAC2(3) fluorescence using time-correlated single photon counting photodetectors (same used for FLIM) that count the emitted photons per pixel in a confocal plane. From measurements at different locations in confocal planes of several fin buds, we observed the least amount of depolarization at 32 hpf in the ectoderm (Fig 5A) and the mesenchyme (Fig 5B). Afterward, depolarization increased by 42 hpf with average highest levels at 48 hpf and 56 hpf (Fig 5A and 5B). While the averages increased, these averages represented a broad distribution of depolarization levels particularly at 48 hpf and 56 hpf. To visualize the distribution of relative differences in membrane potential at the selected time points, we depicted the photon-count information of each pixel in mid-level confocal planes of the fin buds using a rainbow color scale in which red represented the highest levels of DiSBAC2(3) fluorescence (depolarization) and green to blue depicted lower levels. We observed that 32 hpf consistently showed the lowest levels of depolarization (Fig 5C’ and 5C”). DiSBAC2(3) fluorescence incrementally increased throughout the fin buds at 42 hpf (Fig 5D’), 48 hpf (Fig 5E’), and 56 hpf (Fig 5F’). As indicated by the variance in our measurements, we also observed fin buds at these time points that displayed lower levels of fluorescence or variegated patterns high and low fluorescence (Fig 5D”, 5E” and 5F”). We posit that the observed variation in relative membrane potential values after 32 hpf may stem from oscillations around the average membrane potential, although we can not rule out differences in dye penetration. In any case, the combined measurement data show a collective increase in depolarization. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 5. Kcnk5b enhanced depolarization is required for enhanced fin bud growth. (A, B) DiSBAC2(3) fluorescence measurements of the ectoderm (A) and mesenchyme cells (B) of developing pectoral fin buds at 32, 42, 48, and 56 hpf. (C–F) Confocal images of developing fin buds displaying the intensities fluorescence as photon counts per pixel at 32 hpf (C), 42 hpf (D), 48 hpf (E), and 56 hpf (F). Colors represent the range of counted photons per pixel. Blue representing the lowest level of counted photons, and red representing their highest counts (up to 750 photons or more). Images representing high photon count (C’, D’, E’, F’) and low photon count (C”, D”, E”, F”). The total exposure range was set at 1,500 counts (1.5). (G) Distribution of counted photons from DiSBAC2(3) in the confocal plane of the representative fin bud at 32 hpf. (H) Distribution of counted photons from DiSBAC2(3) in the confocal plane of the representative fin bud at 32 hpf. (I) Assessment of DiSBAC2(3) fluorescence intensity as counted photons for fin buds expressing mCherry or kcnk5b-mCherry. (J) Assessment of DiSBAC2(3) fluorescence intensity of mCherry-expressing or kcnk5b-mCherry fin buds at 32 hpf after treating for 4 h as indicated: ethanol (Et), DMSO, 10 μm vinpocetine (vin), 40 μm dibucaine (dib). (K) Representative image of pectoral fin bud of non-transgenic 48 hpf AB fish after heat shock at 32 hpf and start of treatment at 36 hpf with the drug solvents Ethanol (Et) and DMSO. (L) Representative image of pectoral fin buds of 48 hpf AB fish heat shocked at 32 hpf and start of treatment with 10 μm Vinpocetine (Vin) and 40 μm Dibucane (Dib) at 36 hpf. (M) Representative image of pectoral fin buds of 48 hpf transgenic Tg[hsp70:kcnk5b-GFP] after heat shock at 32 hpf and start of treatment with EtOH and DMSO at 36 hpf. (N) Representative image of pectoral fin buds of 48 hpf transgenic Tg[hsp70:kcnk5b-GFP] after heat shock at 32 hpf and start of treatment with 10 μm Vin and 40 μm Dib at 36 hpf. (O) Assessment of pectoral fin bud growth at 48 hpf expressing either AB and kcnk5b-GFP in the indicated treatment groups. Experiments were repeated 3 or more time (N ≥ 3). Each repeat contained 6 or more embryos; one fin bud was measured per embryo. For the DiSBAC2(3) fluorescence measurements, 6 independent points were measured from different 4 locations in each fin bud, distal, anterior, posterior, and proximal, and then averaged to represent a data point (A, B, I, J). For the fin bud size measurements, we measured 1 fin bud and eye per embryo. We measured at least 15 embryos per repeat, and each measurement is 1 data point (O). P values represent statistical analysis by Student’s two-tailed t test. P values >0.05 are designated as “not significant” (NS). Scale bars equal 100 μm. Numerical data used in this figure are included in S5 Data. hpf, hours post fertilization. https://doi.org/10.1371/journal.pbio.3002565.g005 To assess the effect of Kcnk5b-mediated decrease in intercellular K+ on the membrane potential, we overexpressed kcnk5b-mCherry or mCherry at 32 hpf, the time point with the lowest depolarization levels (Fig 5A–5C), and then measured membrane potential. We observed that compared to mCherry-expressing fin buds (Fig 5G and 5I), overexpression of kcnk5b-mCherry significantly increased depolarization (Fig 5H and 5I). These results indicated that Kcnk5b activity promotes depolarization. The observed increase in depolarization caused by kcnk5b, a leak channel that should promote hyperpolarization, suggested that other channels whose activity causes depolarization, such as Na+ channels, were involved. We therefore examined whether Kcnk5b-mediated depolarization required Na+-channel activity. We expressed either control mCherry or kcnk5b-mCherry and then treated these fish with the Na+-channel inhibitors Vinpocetine (a broad voltage-gated sodium channel inhibitor, including the TTX-insensitive channels) and Dibucane (broad sodium channel inhibition) (S7E–S7J Fig). We observed that inhibition of Na+ channels decreased depolarization in the fin bud as well as prevented Kcnk5b-induced increase in depolarization (Fig 5J). To test whether Na+-channel-mediated depolarization is required for kcnk5b-enchanced growth, we overexpressed kcnk5b-GFP and assess the effect of impairing depolarization using the Na+-channel inhibitors. Compared to the enhanced fin bud sizes of control-treated kcnk5b-GFP-transgenic fish (Fig 5M and 5O), treatment of kcnk5b-GFP-expressing fish with the Na+-channel inhibitors impaired Kcnk5b-induced fin bud growth (Fig 5N and 5O). Together, these data indicated that Kcnk5b activity promotes depolarization via Na+ channels and that Kcnk5b-induce depolarization is required for Kcnk5b-enhanced growth. IP3R-mediated Ca2+ release is required for Kcnk5b-induced shha expression and fin bud scaling Our observations that K+-leak channels increase the expression of important morphogens suggested that the channels are doing so through one or more signaling mechanisms. Since the intracellular accumulation of second messengers is involved in many signaling mechanisms, we assessed whether Kcnk5b activity alters the levels of particular second messengers. We used HEK293 cells for an initial assessment, because we previously observed that Kcnk5b induces SHH in these cells [11], a morphogen important for the development of early fin/limb buds [39]. To determine whether cAMP or cGMP levels change in response to Kcnk5b, we used FLIM with an established FRET-based sensor for each [40,41]. Compared to controls groups of unstimulated cells or cells stimulated with forskolin that produces cAMP (S8A Fig) or cells stimulated with SNAP to produce cGMP (S8B Fig), Kcnk5b did not significantly alter the intracellular levels of cAMP or cGMP (S8A and S8B Fig). We then assessed intracellular Ca2+ using the GCaMP6s sensor [42], and we observed that compared to cells transfected with control plasmid (Fig 6A and 6D), transfection with Kcnk5b led to significant increases in Ca2+ (Fig 6B and 6D). An inactive mutant version of the Kcnk5b channel, Kcnk5bMut (S8C Fig), also did not increase GCaMP6s activity (Fig 6C and 6D). To further assess the relationship between Kcnk5b and intracellular Ca2+, we transfected cells with GCaMP6s and either with Kcnk5b-mCherry or with mCherry. We subsequently cultured the cells for 24 h, FACS sorted both transfection groups for mCherry fluorescence, and then plated each group of sorted cells at 100% confluency (Fig 6Ea and 6Eb). mCherry-transfected cells that did not consistently display GCaMP6s fluorescence (Fig 6Ea, 6Ec, 6Ee, 6Eg arrow versus asterisk and 6F), while Kcnk5b-mCherry+ cells always showed high GCaMP6s activity (Fig 6Eb, 6Ed, 6Ef, 6Eh asterisks and 6F). We observed this phenomenon 100% of the time in all confocal images (Fig 6G). Together, these results indicate that Kcnk5b activity promotes a rise in intracellular Ca2+. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 6. Kcnk5b activity induces IP3R-mediated Ca2+ release from the ER. (A) GCaMP6s fluorescence in HEK293 cells. (B) GCaMP6s fluorescence in HEK293 cells expressing Kcnk5-mCherry. (C) GCaMP6s fluorescence in HEK293 cells expressing Kcnk5BSM-mCherry. (D) Fluorescence measurements of indicated groups. (E) Representative confocal images of mCherry-transfected, GCaMP6s-transfected cells showing brightfield with merged fluorescence from mCherry and GCaMP6s (a) or mCherry (c) or GCaMP6s (e) or merged mCherry-GCaMP6s (g), and Kcnk5b-mCherry-transfected, GCaMP6s-transfected cells: brightfield with merged fluorescence from mCherry and GCaMP6s (b) or Kcnk5b-mCherry (d) or GCaMP6s (f) or merged mCherry-GCaMP6s (h). (F) Measurements of GCaMP6s fluorescence intensity from the indicated experimental transfection groups. (G) Percents of mCherry-GCaMP6s-double positive over the total number of mCherry-positive cells in each group relate the frequency of GCaMP6s-positive cells in the mCherry or Kcnk5b-mCherry groups. (H) GCaMP6s fluorescence in the pectoral fin buds of transgenic fish harboring both Tg[Cca.actb:GCaMP6s] and Tg[hsp70:kcnk5b-mCherry]. (I) GCaMP6s fluorescence in embryos harboring the stable transgenic fish line Tg[Cca.actb:GCaMP6s] that mosaically express patches of mCherry or kcnk5b-mCherry (designated mCherry+) for the ectoderm (a) and mesenchyme (b). (J) Diagram of IP3R inhibition by 2-APB. (K) qRT-PCR of SHH expression in HEK293 cells transfected either with GFP or Kcnk5b-GFP after 20-h treatment with 2-APB at the indicated concentrations. (L) Assessment of pectoral fin bud size at 48 hpf after 4 h of treatment with 13 μm 2-APB. (M) Expression of shha in pectoral fin buds of heat-shocked non-transgenic sibling embryos after 4-h treatment with DMSO (a) or 13 μm 2-APB (b), of heat-shocked transgenic Tg[hsp70:kcnk5b-mCherry] siblings after 4-h treatment with DMSO (c) or 13 μm 2-APB (d). (N) Pixel area of in situ staining of shha in the indicated treatment groups. (O) qRT-PCR of shha expression in isolated fin buds of the indicated groups. Experiments were repeated 3 or more time (N ≥ 3). For cell culture experiments, each repeat contained duplicate or triplicate wells, and 10 or more cells were measured per well. Each data point represents 1 cell (D, F, G). For fin bud fluorescence measurements, we measured 2 or 3 locations in each tissue of 1 fin bud per embryo (H, I). For fin bud area measurements, we measured the area of 1 fin bud and eye per embryo. We measured at least 4 embryos per repeat (L). For the qRT-PCR experiments, we collected 80 fin bud samples per isolation. Three or more isolations were measured. Each isolation was measured in duplicate or triplicate. Each measured value is represented as a data point. P values represent statistical analysis by Student’s two-tailed t test. P values >0.05 are designated as “not significant” (NS). Scale bars equal 100 μm (A–C), 10 μm (E), 0.5 μm (M). Numerical data used in this figure are included in S6 Data. ER, endoplasmic reticulum; hpf, hours post fertilization. https://doi.org/10.1371/journal.pbio.3002565.g006 To determine whether Kcnk5b has the same effect on intercellular Ca2+ levels in vivo, we generated double-transgenic fish that harbored Tg[hsp70:kcnk5b-mCherry] and a CGaMP6s Ca2+ reporter under the control of the β-actin promoter Tg[Cca.actb:GCaMP6s] [43]. We assessed intracellular Ca2+ in the fin buds of heat-shocked transgenic kcnk5b-mCherry embryos at 48 hpf and their non-transgenic siblings as controls. We observed that Ca2+ levels were higher in kcnk5b-mCherry compared to non-transgenic embryos (Fig 6H). We observed similar results when we assessed mosaic embryos harboring the stable transgenic fish line Tg[Cca.actb:GCaMP6s] that mosaically express patches of mCherry or kcnk5b-mCherry (Fig 6Ia and 6Ib). Together, these data indicated that Kcnk5b normally promotes the increase in intracellular Ca2+ levels in vivo, and they suggested Ca2+ mediates the growth-inducing effect of Kcnk5b. Increases in Ca2+ caused by the decrease in intracellular K+ could occur from extracellular sources via Ca2+ channels in the plasma membrane and/or from intracellular sources such as the endoplasmic reticulum (ER) [44]. To determine the importance of these 2 sources, we tested whether inhibiting release of Ca2+ from each source impaired Kcnk5b-induced transcription of SHH in HEK293 cells. After inhibiting T- and L-type Ca2+ channels on the plasma membrane (S8D Fig), we did not detect a significant effect on Kcnk5b-induced SHH transcription (S8E and S8F Fig). However, after inhibiting IP3R-mediated release of Ca2+ from the ER (Fig 6J), we observed decreased Kcnk5b-induced SHH transcription dose-dependently (Fig 6K). To determine whether IP3R-mediated Ca2+ release was required for the Kcnk5b-enhanced growth of pectoral fin buds, we inhibited IP3R activity for 4 h and assessed bud growth in heat-shocked non-transgenic and transgenic Tg[hsp70:kcnk5b-GFP] siblings. We observed that the IP3R inhibitor blocked the growth phenotype caused by expression of kcnk5b at 48 hpf (Fig 6L). This decrease in growth was associated with the impairment of Kcnk5b-induced increase of shha expression domains (Fig 6M and 6N) and qRT-PCR (Fig 6O). The importance of Ca2+ release for shha expression and fin bud growth indicated that one or more Ca2+-dependent kinases were involved. Therefore, we tested which of the Ca2+-activated CaM kinases were required for Kcnk5b-enhanced expression of SHH in HEK293 cells. While we observed no significant effects by inhibiting CaMKII and CaMKIV (S9A Fig), we did observe that inhibition of CaMKK impaired Kcnk5b-enhanced SHH expression (Fig 7A). We then tested whether inhibiting CaMKK in the fin buds has similar effects on shha expression in vivo. We observed that inhibiting CaMKK decreased the expression of shha by in situ (Fig 7B and 7C) and by qRT-PCR (Fig 7D). Inhibiting CaMKK also impaired Kcnk5b-enhanced growth (Fig 7E). Furthermore, overexpressing the human CaMKK2 (Fig 7F) or the zebrafish camkk1b (Fig 7G) was sufficient to increase SHH expression in HEK cells. Together, these results indicated that Ca2+ is required for Kcnk5b-induced transcription of shha and that CaMKK is an important part of Ca2+ regulation of shha expression and pectoral fin bud growth. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 7. Kcnk5b requires CaMKK for growth and SHH/shha expression. (A) SHH expression in HEK293 cells transfected with GFP or Kcnk5b-GFP after 20-h treatment at the indicated concentrations of the CaMKK inhibitor STO-609 at the indicated concentrations. (B) Expression of shha in pectoral fin buds from heat-shocked non-transgenic siblings after 6-h treatment with DMSO (a) or 24 μm STO-609 (b) and from transgenic Tg[hsp70:kcnk5b-mCherry] siblings after treatment with DMSO (c) or STO-609 (d). (C) The ratios of each in situ staining area of shha to the area of its corresponding fin bud for the indicated groups. (D) qRT-PCR of shha expression in isolated fin buds in the indicated control and experimental groups. (E) Graphed assessment of fin-bud-area-to-eye-area measurement ratios. (F, G) Difference in SHH expression in HEK293 cells transfected with mCherry or human CaMKK2-mCherry (F) or camkk1b-mCherry (G). Experiments were repeated 3 or more time (N ≥ 3). For cell qRT-PCR experiments, each RNA isolation per well was measured in duplicate or triplicate. Each measured value is represented as a data point (A, F, G). For fin bud fluorescence measurements, we measured 2 or 3 locations in each tissue of 1 fin bud per embryo. For fin bud area measurements, we measured the area of 1 fin bud and eye per embryo. We measured at least 4 embryos per repeat (E). For fin bud qRT-PCR experiments, we collected 80 fin bud samples per isolation. Three or more isolations were measured. Each isolation was measured in duplicate or triplicate samples (D). Each measured value is represented as a data point. P values represent statistical analysis by Student’s two-tailed t test. P values >0.05 are designated as “not significant” (NS). Scale bars equal 0.5 μm (B). Numerical data used in this figure are included in S7 Data. https://doi.org/10.1371/journal.pbio.3002565.g007 Endogenous intracellular K+ levels decrease during fin bud growth We previously showed that Kcnk5b promotes allometric growth of adult fins via hierarchical activation of several morphogens [11]. The ability of an individual K+ channel to promote the expression of several developmental morphogens in the adult fin and larva led us to hypothesize that this phenomenon may be more broadly functional, and that it regulates the early developmental program conserved in vertebrate fin/limb buds, which is present in the developing pectoral fin buds of zebrafish [22]. Therefore, we wanted to determine whether there are endogenous changes in intracellular K+ in the developing pectoral fin bud. To measure intracellular K+, we used an established FRET-based genetic sensor (KIRIN1) that can detect changes specifically in intracellular K+ levels [28]. To overcome the limitations associated with FRET measurements in vivo, we used FLIM, because FLIM does not rely on intensity-based ratios, rather, FLIM detects changes in the intrinsic decay rate of a donor fluorophore (CPF) when it undergoes FRET to the acceptor fluorophore (YFP) within the sensor (Fig 1Aa and 1Ab). One can directly quantify the accumulated decay profile of a fluorophore (CFP) after excitation by laser pulses [29] (Fig 1Ba and 1Bb); consequently, we could detect relative differences in intracellular K+ caused by increases in K+-leak channel expression (S1Aa and S1Ab Fig), or by posttranslational regulation of a K+-leak channel (S1B Fig) or by different K+-leak channel mutants that have different activities (S1C Fig) [11]. These experiments confirmed that we can detect changes in relative intracellular K+ levels with the KIRIN1 sensor using FLIM. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 1. Intracellular K+ decreases during pectoral fin bud growth. (A) FRET-based mechanism of detection of intracellular K+ by KIRIN1 sensor and application of FLIM assessment. Changes in FRET affect the fluorescence lifetime of the donor fluorophore CFP that is excited by a 2-Photon laser pulse (a). A reduction of intracellular K+ is detected as higher (longer) lifetime of the CFP, while increase in K+ is detected as its lower (shorter) lifetime (b). (B) We depict these lifetime (nanoseconds, ns) relationships of higher lifetime equating to lower intracellular K+ (a) and the inverse (original) as 1/lifetime (ns-1) (b) in order to more easily relate the trend changes with what the FLIM-based measurements are indicating for intracellular K+. (C) Design of K+ sensor (KIRIN1) transgene for transgenic zebrafish line (a). Heat-shock method for inducing transgene during different developmental time points of the fin bud and subsequently FLIM-FRET (FILM) imaging it (b). (D) FLIM measurements of the CFP in confocal sections of ectoderm (Ecto) and mesenchyme (Mesen) in pectoral fin buds at 48 hpf using a FRET-based sensor for K+. Measurements from a transgene containing only CFP that is unable to FRET with YFP. (E) Time course of FLIM measurements of intracellular K+ in the ectodermal and mesenchymal tissues of the developing pectoral fin buds. (F) Time course of growth measurements of the developing fin buds. (G–I) Z-stacks of confocal FLIM planes in fin buds at 32 hpf (G), 48 hpf (H), and 56 hpf (I). Z-stacks cover distances of 15.2 μm (32 hpf), 18.36 μm (48 hpf), and 10.3 μm (56 hpf). The lifetime value of each pixel in the ectoderm (white arrowhead) and the mesenchyme (asterisk) is represented by colors in a rainbow scale from 2.8 ns (blue) to 3.4 ns (red). (J) Generation of fin buds that mosaically express the KIRIN1 sensor and either mCherry or kcnk5b-mCherry (mosaic for 2 transgenes) and proposed outcomes. (Ka,b) Images of cells in fin buds of 56 hpf embryos harboring hsp70:mCherry (red asterisks) and cells lacking the mCherry transgene (white asterisks). Confocal plane for mCherry fluorescence (a). FLIM image of KIRIN1+ cells of the same confocal plane (b). (La,b) Images of cells in fin buds of 56 hpf embryos harboring hsp70:kcnk5b-mCherry (red asterisks) and cells lacking the transgene (white asterisks). Confocal plane for mCherry fluorescence (a). FLIM image of KIRIN1+ cells of the same confocal plane (b). (M) FLIM measurements in fin buds of 56 hpf embryos of indicated cell categories; “near” indicates cells next or distant to mCherry-positive (mCherry+) or kcnk5b-mCherry-positive (kcnk5b+) cells. The graphs are depicted as 1/lifetime (ns-1) to more easily relate the portrayal of the values to the related change in intracellular K+ (see S1A Fig). Experiments were repeated 3 or more time (N ≥ 3). For FLIM, we measured 2 or 3 locations in each tissue of 1 fin bud per embryo. Each measured value is represented as a data point (D, M), except for E, in which the data points are presented as averages and standard deviations of all the measurements (≥11) at each time point. For fin bud size measurements (F), each data point represents 1 fin bud per embryo. P values represent statistical analysis by Student’s two-tailed t test. P values >0.05 are designated as “not significant” (NS). Numerical data used in this figure are included in S1 Data. FLIM, Fluorescence Lifetime Microscopy; hpf, hours post fertilization. https://doi.org/10.1371/journal.pbio.3002565.g001 To assess the intracellular K+ in vivo, we generated a heat-shock-inducible transgenic reporter Tg[hsp70:CFP-KIRIN1-YFP] (hsp70:KIRIN1) line (Fig 1Ca) and used it to detect relative differences in K+ levels via FLIM 6 h after heat shock (Fig 1Cb). FLIM measurements in the pectoral fin bud showed relatively lower intracellular K+ levels in the mesenchyme compared to ectoderm at 48 hpf (Fig 1D), a time point in which the growth of the pectoral fin buds is underway. We subsequently measured several time points during pectoral fin bud outgrowth from the same animals, and we observed an increase in intracellular K+ in both the ectoderm and the mesenchyme between 32 and 34 hpf, but afterwards, intracellular K+ levels decreased to their lowest levels at around 48 hpf (Fig 1E). Intracellular K+ levels increased again between 54 and 56 hpf (Fig 1E). From the plotted time points, we observed that although mesenchyme K+ levels remained lower than the ectoderm, the changes in K+ in both tissues were coupled (Fig 1E). These lifetime measurements were not influenced by differences in intensity levels (S2D–S2H Fig) or by the method for immobilizing the embryos (S2I–S2O Fig). To ascertain whether there is a relationship between relative changes in intracellular K+ and growth, we measured fin bud sizes between 30 hpf to 56 hpf. From bud area measurements, we observed incremental growth from 30 hpf to 56 hpf (Fig 1F). We also observed 2 pauses in average growth between 32 hpf and 34 hpf and 54 hpf and 56 hpf (Fig 1F) that correlated with time points in which intracellular K+ increased (Fig 1E). Together, these results suggested a coordinated regulation of intracellular K+ that relates to fin bud growth. To visualize the relative distribution of intracellular K+ during pectoral fin bud development, we represented the lifetime values along a rainbow scale: the lower the value (the higher intracellular K+ levels), the more towards blue; conversely, the higher values (the lower intracellular K+ levels), the more towards red. From the colorized individual confocal FLIM planes (S3A–S3C Fig) and 3D Z-stacks of FLIM planes through fin buds at 32 hpf (Figs 1Ga–1Ge and S3A), at 48 hpf (Figs 1Ha–1He and S3B), and at 56 hpf (Figs 1Ia–1Ie and S3C), we observed clear differences in K+ levels between the mesenchyme and ectoderm, but few regional differences within each tissue. We also observed global decreases in intracellular K+ between 32 hpf (Figs 1Ga–1Ge and S3A) and 48 hpf (Figs 1Ha–1He and S3B) when the growth of the bud is high. Conversely, we observed global increases in K+ between 48 hpf and 56 hpf (Figs 1Ia–1Ie and S3C) when growth reduces. From these observations, we propose that there is a decrease in intracellular K+ during fin bud growth due to increases in K+-leak channel activity, and these changes are linked to the growth of the bud. The lack of regional differences in intracellular K+ in either the mesenchyme or ectoderm suggested equilibration of K+ between the cells of each tissue. To test this hypothesis, we generated double mosaic embryos by injecting recipient embryos with the heat-shock-inducible hsp70:KIRIN1 and either kcnk5b-mCherry transgene (hsp70:kcnk5b-mCherry) or with mCherry (hsp70:mCherry) transgene as a control (Fig 1J). We identified mCherry-expressing cells (Fig 1K) or kcnk5b-mCherry cells (Fig 1L) in fin buds and then assessed intracellular K+ in mCherry-positive cells (Fig 1Ka, 1La, and 1M and red asterisks in Fig 1Kb and 1Lb) and the surrounding mCherry-negative cells (Fig 1K and 1L, white asterisks). Compared to cells that harbored the mCherry transgene (Fig 1K, red asterisks), we observed that cells expressing kcnk5b-mCherry (Fig 1L, red asterisks) and cells surrounding the transgene-expressing cells displayed decreases in K+ levels (Figs 1M and S3D). These data indicated that changes in intracellular K+ are shared such that cells with more K+-leak channel activity can decrease the intracellular K+ of neighboring cells. Overexpression of a K+-leak channel in early fin bud coordinately enhances the expression of the important morphogens to scale proximodistal growth The coordinated decrease in the endogenous levels of K+ in the developing pectoral fin bud suggested that intracellular K+ has a role in bud development. Consequently, we wished to know whether decreasing intracellular K+ is sufficient to enhance the scale of the buds. To decrease intracellular K+, we overexpressed Kcnk5b or Kcnk10 K+-leak channel by heat-shock induction of the Tg[hsp70:kcnk5b-GFP] [11] or Tg[hsp70:kcnk10a-GFP] transgenic lines. Six hours after transgene induction at 48 hpf (S4A–S4C Fig), we measured the growth area of the fin buds and standardized the measurements of each bud to the area of the eye (Fig 2A and 2B) or the otic vesicle (S4D and S4E Fig). From these analyses, we observed enhanced growth of pectoral fin buds caused by kcnk5b or kcnk10 compared to heat-shocked non-transgenic siblings and transgenic Tg[hsp70:GFP] control groups at 32 hpf, 48 hpf, and 56 hpf when standardized to the area of the eye (Fig 2C) or of the area of the otic vesicle (S4F Fig). Together, these data indicated that reducing intracellular K+ increases growth of the buds. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 2. K+-leak channels enhance the growth of pectoral fin buds. (A, B) Brightfield images of a post-heat-shocked 48 hpf non-transgenic embryo (A) or post-heat-shocked 48 hpf hsp70:kcnk5b-GFP transgenic embryo (B). The area of the fin bud (black-dotted line). (C) Measurements of pectoral fin bud areas of heat-shocked groups of non-transgenic (AB), Tg[hsp70:GFP], Tg[hsp70:kcnk5b-GFP], and Tg[hsp70:kcnk10-GFP]. The non-transgenic and Tg[hsp70:kcnk5b-GFP] embryos are siblings. Each measured bud was standardized to the eye area (red-dotted circles in A,B) in the same embryo. (D–S) Expression of the indicated morphogens. (D, G, J, M) Lateral views of whole-mount in situs of heat-shock control non-transgenic embryos at 48 hpf for fgf8a (D), fgf10a (G), shha (at 52 hpf) (J), aldh1a2 (M) and dorsal view of aldh1a2 (O) or lateral views of heat-shocked Tg[hsp70:kcnk5b-GFP] embryos at 48 hpf for fgf8a (E), fgf10a (H), shha (at 52 hpf) (K), aldh1a2 (N), and dorsal view of aldh1a2 (P). White arrows indicate proximal expression of aldh1a2. qRT-PCR measurements from fin buds for fgf8a (F), fgf10a (I), shha (L), and aldh1a2 (Q). Lateral views of in situs of heat-shocked non-transgenic fin bud for bmp4 (R) and msx2b (U) and of heat-shocked Tg[hsp70:kcnk5b-GFP] embryos at 48 hpf for bmp4 (S) and msx2b (V). qRT-PCR measurements from fin buds for bmp4 (T), msx2b (W) the in situ experiments were repeated 3 or more time (N ≥ 3). Each in situ repeat contains 6–12 embryos per replicate. For the fin bud size measurements, we measured 1 fin bud and eye per embryo. We measured at least 4 embryos per experimental repeat. For the qRT-PCRs, we collected 80 fin bud samples per isolation. We assessed gene expression with 3 or more isolations. Each isolation was assessed in duplicate or triplicate. Each measured value is represented as a data point. P values represent statistical analysis by Student’s two-tailed t test. P values ≥0.05 are designated as “not significant” (NS). The scale bars are 0.5 mm (A, B), 100 μm (D–P, R–S,U–T). Numerical data used in this figure are included in S2 Data. hpf, hours post fertilization. https://doi.org/10.1371/journal.pbio.3002565.g002 The increase in pectoral fin bud size from K+-leak channel overexpression indicated that the developmental gene program was affected. We therefore examined the expression of selected morphogens known to control early bud growth at 48 hpf. We observed that compared to heat-shocked control siblings (Fig 2D, 2G, 2J, 2M and 2O), transgenic expression of kcnk5b increased the expression of fgf8a (Fig 2D and 2E), fgf10a (Fig 2G and 2H), shha (Fig 2J and 2K), and aldh1a2 (Fig 2M–2P). We observed similar up-regulation of these genes from qRT-PCR analyses of fin buds (Fig 2F, 2I, 2L and 2Q). Because previous findings show that BMP signaling in vertebrate and invertebrate appendages is affected by K+ channels [5,6], we also assessed the expression of an important BMP ligand (bmp4) and its down-stream target msx2b. We observed that compared to in situ controls (Fig 2R and 2U), kcnk5b increased the expression of both genes in the in buds (Fig 2S and 2V). We also observed similar up-regulation of these genes by qRT-PCR (Fig 2T and 2W). Thus, the endogenous decreases in intracellular K+ during fin bud development (Fig 1), the coordinated increase in the expression of several important morphogens by increasing the expression of the K+-leak channel kcnk5b (Fig 2D–2W), and the enhanced growth the pectoral fin bud by kcnk5b or kcnk10a overexpression (Fig 2C) all implicated intracellular K+ as an integral part of the fin bud growth control, specifically, decreased intracellular K+ augmented bud proportional growth. Retinoic acid decreases intracellular K+ via an Rcan2-mediated mechanism that scales fin buds The decrease in intracellular K+ during pectoral fin bud growth suggested that K+ levels might be responsive to developmental signals that regulate growth. One such signal is RA [30–32]. Therefore, we examined whether RA stimulation influenced intracellular K+ levels in developing fin buds at 32 hpf by FLIM measurements of the KIRIN1 transgene 6 h after its heat-shock induction. Compared to treatment with the solvent DMSO (Fig 3A and 3C), we observed decreases in K+ after 6 h of treatment with 200 nM RA (Fig 3B and 3C), both in ectoderm and mesenchyme cells (Fig 3D). To test whether the changes in intracellular K+ were specific to RA, we assessed the effect of thyroid hormone (TH), another nuclear hormone receptor mechanism. Treatment with TH showed no significant differences in K+ levels compared to DMSO-treated controls (S5C Fig). These data indicated that RA-mediated signaling is sufficient to decrease intracellular K+ in fin bud tissues and that this effect is not induced by all nuclear hormones. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 3. Retinoic acid decreases intracellular K+ in pectoral fin buds via transcriptional activation of the calcineurin inhibitor rcan2. (A, B) FLIM images of developing fin buds from embryos treated for 6 h with DSMO-treated (A) or with 200 nM RA-treated (B) 32 hpf embryos. (C) FLIM measurements from cells in developing buds at 32 hpf of the indicated treatment groups. (D) FLIM measurements of cells in the ectoderm (Ecto) or the mesenchyme (Mesen) of buds at 32 hpf treated either with DMSO or RA. (E) In situ images of dorsal view (a) and lateral view (b) for kcnk5b expression in 48 hpf embryos. Pectoral fin buds (pfb), ectoderm (e), and mesenchyme (m). (F) Representative lateral view images of pectoral fin buds of in situ experiments for kcnk5b expression at 32 hpf (a), 56 hpf (b), and 72 hpf (c). (G) Representative lateral views of pectoral fin buds after in situ experiments for rcan2 expression at 32 hpf (a), 34 hpf (b), 48 hpf (c), and 56 hpf (d) embryos. (H) qRT-PCR of rcan2 from isolated pectoral fin buds of 48–52 hpf embryos treated with DMSO or 200 nM RA for 6 h. (I) In situ for rcan2 expression in fin buds of 48 hpf embryos after 6 h DMSO treatment. Ectoderm is indicated by (e), and mesenchyme is indicated by (m). (J) In situ images for rcan2 expression in fin buds of 48 hpf embryo after 6 h 200 nM RA treatment. Ectoderm (e), mesenchyme (m). (K, L) Pectoral fin buds of heat-shocked non-transgenic embryo (K) or heat-shocked transgenic hsp70:rcan2-mCherry sibling (L). (M) Measured fin buds were standardized to the area of an eye in the same embryo. (N) FLIM measurements of intracellular K+ in pectoral fin bud cells from heat-shocked non-transgenic and mCherry controls and transgenic rcan2-mCherry-expressing siblings. (O–R) Representative fin buds of embryos expressing empty sgRNA vector “EV” (O), rcan2 sgRNA “KO” (P), mCherry mRNA with rcan2 sgRNA “KO+mCherry” (Q), mutated rescue rcan2 mRNA with rcan2 sgRNA “KO+rcan2” (R). (S) Ratios of fin bud areas to eye areas of embryos of the indicated experimental groups. (T) FLIM measurements of intracellular K+ in the pectoral fin bud cells in control and CRISPR-Cas9 knockout of rcan2 in embryos. Experiments were repeated 3 or more time (N ≥ 3). For FLIM, we measured 2 or 3 locations in each tissue of 1 fin bud per embryo. Each measured value is represented as a data point (C, D, N, T). For the fin bud size measurements, we measured 1 fin bud per embryo (M) and 2 fin buds per embryo (S). For the qRT-PCR experiments, we collected 80 fin bud samples per isolation. Three or more isolations were measured. Each isolation was measured in duplicate or triplicate. Each isolation is represented as a data point. Each in situ repeat contained 6–12 embryos per replicate. P values represent statistical analysis by Student’s two-tailed t test. P values >0.05 are designated as “not significant” (NS). Scale bars represent 50 μm (E–G, I–L, O–R). Numerical data used in this figure are included in S3 Data. FLIM, Fluorescence Lifetime Microscopy; hpf, hours post fertilization; RA, retinoic acid. https://doi.org/10.1371/journal.pbio.3002565.g003 RA regulates gene transcription via specific intracellular receptors [33]. Consequently, the relative reduction in K+ by RA suggests that RA regulates the transcription of one or more K+ channels. Because Kcnk5b is involved in adult fin scaling [9], we performed in situ hybridization experiments to determine whether kcnk5b is present in growing fin buds of 48 hpf embryos. We observed expression primarily in distal pharyngeal pouches (Fig 3Ea), and the mesenchyme of the growing fin buds (Fig 3Ea and 3Eb). We subsequently assessed kcnk5b expression during fin bud growth at 34 hpf, at 56 hpf, and at 72hpf, a time point in which bud growth has already ceased. We observed kcnk5b expression primarily in the mesenchyme at 32 hpf (Fig 3Fa) and at 56 hpf (Fig 3Fb), but by 72 hpf, the channel expression was difficult to detect in the bud (Fig 3Fc). Because treatment with RA decreases intracellular K+ levels, we reckoned that RA may be enhancing kcnk5b transcription. Therefore, we tested whether RA-mediated decrease in intracellular K+ was related to an up-regulation of kcnk5b expression. We did not observe an increase in RA-treated embryos by qRT-PCR (S5A Fig) despite a significant increase in a known RA-activated gene (S5B Fig) [34]. We previously showed that inhibition of calcineurin increases the activity of Kcnk5b to decrease intracellular K+ and enhance the scale of the adult fin [11]. Since Kcnk5b is present in the embryonic pectoral fin bud (Fig 3E and 3F), we hypothesized that RA may decrease intracellular K+ by posttranslational regulation of Kcnk5b through the inhibition of calcineurin. RCAN proteins are well-documented in vivo inhibitors of calcineurin [35–37]. From in situ hybridization experiments, we detected expression at 32 hpf (Fig 3Ga), at 34 hpf (Fig 3Gb), and 48 hpf (Fig 3Gc). However, rcan2 expression reduced by 56 hpf (Fig 3Gd). Because RA treatment decreased intracellular K+ and Rcan2 inhibits calcineurin, which subsequently could suppress calcineurin-mediated inhibition of Kcnk5b, we tested whether rcan2 expression is altered by RA. We observed that compared to DMSO controls, RA treatment increased rcan2 expression by qRT-PCR (Fig 3H), and from in situ hybridization experiments, we observed staining in DMSO-treated buds (Fig 3I) that became more intense after RA treatment (Figs 3J and S5D). We also observed similar phenomena in the adult zebrafish fins: RA treatment decreased intracellular K+ levels (S5G Fig) and increased rcan2 transcription (S5H Fig). Furthermore, we detected rcan2 transcripts and protein in the blastemas and distal regenerating interray tissue of regenerating fins (S5I–S5K Fig), where growth occurs. We did not observe protein expression in fin rays immediately after amputation when regenerative growth has not yet commenced (S5J and S5K Fig). The ability of RA to up-regulate rcan2, which inhibits a phosphatase that suppresses the K+-leak channel Kcnk5b (present and scales fin buds and adult fins [9,11]) led us to hypothesize that Rcan2 may enhance proportional growth of fin buds. To test whether increasing Rcan2 enhances pectoral fin bud growth, we generated a fish line harboring the hsp70:rcan2-mCherry transgene to overexpress rcan2 by heat-shock induction (S5L–S5O Fig). Compared to heat-shocked non-transgenic control siblings (Fig 3K and 3M) and overexpressed mCherry (Fig 3M), induction of rcan2 enhanced the growth area of the fin buds (Fig 3L and 3M). We also observed enhanced growth by transgenic overexpression of rcan2 during adult fin regeneration (S5P–S5V Fig). To determine whether rcan2-enhanced growth correlated with reduced intracellular K+, we generated double-transgenic fish that harbored hsp70:rcan2-mCherry and hsp70:KIRIN1 transgenes. After heat-shock induction of the transgenes, we observed that rcan2 expression decreased intracellular K+ levels compared to controls (Fig 3N). Together, these data showed that induction of rcan2 expression was sufficient to enhance growth in the fin bud and adult fin as well as reduce their intracellular K+ levels in vivo. We then assessed whether targeting rcan2 by CRISPR-Cas9 (S5W–S5Z Fig) affected fin bud size and intracellular K+. Compared to control embryos (Fig 3O and 3S), we observed that rcan2-targeted embryos displayed reduced fin bud growth (Fig 3P, 3Q and 3S), which could be rescued by overexpression of rcan2 mRNA (Fig 3R and 3S) that harbored mutations in its wobble bases to impair interactions between the overexpressed mRNA and the genome-targeting sgRNA (S5Za and S5Zb Fig). We also assessed caudal fins of CRISPR-Cas9-targeted juvenile fish, which also displayed shorter fins compared to wild type (S5A and S5B Fig). When we assessed the effect of targeting rcan2 on intracellular K+ levels by targeting rcan2 in Tg[hsp70:KIRIN1] embryos, we observed that targeting rcan2 increased intracellular K+ (Fig 3T) and rescue with rcan2 mRNA returned intracellular K+ back to control levels (Fig 3T). Together, these results from embryonic fin buds and post-metamorphosis caudal fins indicated the involvement of rcan2 in K+-channel-mediated scaling of embryonic and adult appendages. Rcan2-mediated scaling requires Kcnk5b activity via its Serine 345 Calcineurin limits fin proportional growth by inhibiting Kcnk5b through serine 345 [11]. Since Rcan2 inhibits calcineurin [35–37], and kcnk5b is expressed in the developing pectoral fin buds (Fig 3E and 3F), we first assessed the importance of kcnk5b in fin bud growth by CRISPR-targeting kcnk5b (S6A–S6G Fig). Compared to controls (Fig 4A and 4E), fin buds of kcnk5b-targeted embryos were reduced in size by 48 hpf (Fig 4B and 4E). This phenotype was rescued by overexpression of kcnk5b mRNA (Fig 4D and 4E) that was mutated to impair interaction with the targeting sgRNA (S6H–S6J Fig). When we overexpressed rcan2 in kcnk5b-targeted embryos, we observed that while rcan2 overexpression in non-targeted embryos enhanced fin bud growth (Fig 4F, 4G and 4I), rcan2 required kcnk5b to enhance growth (Fig 4H and 4I). We previously showed that calcineurin inhibits Kcnk5b via serine 345 in the cytoplasmic tail of the channel [11], so we tested whether the kcnk5bS345A (the mutant with the lowest K+-leak activity) blunts rcan2 enhancement of fin bud growth in vivo. Compared to control embryos overexpressing rcan2 with the wild-type channel (kcnk5bS345) (Fig 4J and 4L), the calcineurin-dephosphorylated mimic kcnk5bS345A channel reduced rcan2-mediated fin bud growth (Fig 4K and 4L). Together, these data indicated that rcan2-enhanced growth requires Kcnk5b and that the Kcnk5b channel that mimics calcineurin-mediated inhibition of the channel prevents the enhanced growth caused by overexpression of the endogenous calcineurin inhibitor rcan2. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 4. Rcan2-mediated decrease in intracellular K+ involves Kcnk5b. (A–D) Representative 48 hpf embryos and enlarged panels of fin buds of control Cas9 and empty sgRNA vector (EV) embryo (A), CRISPR-targeted kcnk5b embryo (B), CRISPR-targeted kcnk5b embryo overexpressing GFP (C), CRISPR-targeted kcnk5b embryo overexpressing kcnk5b*-GFP that harbors altered wobble bases to impair interaction with targeting sgRNA (D). (E) Fin bud-to-eye area ratios of the indicated genotypes. (F–I) Representative 48 hpf embryos and enlarged panels of fin buds of control Cas9 and empty sgRNA vector (EV) non-Tg embryo (F), rcan2-mCherry-expressing embryo and enlarged panel of the fin bud (G) CRISPR-targeted kcnk5b embryo overexpressing rcan2-mCherry (H). (I) Fin bud-to-eye area ratios of the indicated genotypes. (J, K) Representative 48 hpf embryos and enlarged panel of the fin buds expressing wild-type kcnk5bS345-GFP and rcan2-mCherry (J), or expressing calcineurin-dephosphorylated mimic kcnk5bS345A-GFP and rcan2-mCherry (K). (L) Fin bud-to-eye ratios show that rcan2-mediated enlargement of fin buds is impaired by kcnk5bS345A mutant. Experiments were repeated 3 or more time (N ≥ 3). For the fin bud size measurements, we measured 1 fin bud (E, I) or 2 fin buds (L) per embryo at 48 hpf and not at a later time points to avoid incorporating measurements of the finfold growth that start around 56 hpf. Each measured value is represented as a data point. P values represent statistical analysis by Student’s two-tailed t test. P values >0.05 are designated as “not significant” (NS). Scale bars equal 100 μm (A–D, F–H, J, K). Numerical data used in this figure are included in S4 Data. hpf, hours post fertilization. https://doi.org/10.1371/journal.pbio.3002565.g004 Kcnk5b-enhanced growth involves cell depolarization Kcnk5b is a two-pore K+ leak channel whose activity alters the electrical membrane potential at the plasma membrane of cells. The decrease in intracellular K+ during fin bud growth (Fig 1) and the importance of Kcnk5b for growth (Fig 4) suggest there are important K+-associated changes in membrane potential during fin bud development. To determine whether membrane potential alters during fin bud development, we used DiSBAC2(3), a dye that increases its fluorescence as it enters cells when channels open to depolarize the cells [38]. We assessed DiSBAC2(3) fluorescence using time-correlated single photon counting photodetectors (same used for FLIM) that count the emitted photons per pixel in a confocal plane. From measurements at different locations in confocal planes of several fin buds, we observed the least amount of depolarization at 32 hpf in the ectoderm (Fig 5A) and the mesenchyme (Fig 5B). Afterward, depolarization increased by 42 hpf with average highest levels at 48 hpf and 56 hpf (Fig 5A and 5B). While the averages increased, these averages represented a broad distribution of depolarization levels particularly at 48 hpf and 56 hpf. To visualize the distribution of relative differences in membrane potential at the selected time points, we depicted the photon-count information of each pixel in mid-level confocal planes of the fin buds using a rainbow color scale in which red represented the highest levels of DiSBAC2(3) fluorescence (depolarization) and green to blue depicted lower levels. We observed that 32 hpf consistently showed the lowest levels of depolarization (Fig 5C’ and 5C”). DiSBAC2(3) fluorescence incrementally increased throughout the fin buds at 42 hpf (Fig 5D’), 48 hpf (Fig 5E’), and 56 hpf (Fig 5F’). As indicated by the variance in our measurements, we also observed fin buds at these time points that displayed lower levels of fluorescence or variegated patterns high and low fluorescence (Fig 5D”, 5E” and 5F”). We posit that the observed variation in relative membrane potential values after 32 hpf may stem from oscillations around the average membrane potential, although we can not rule out differences in dye penetration. In any case, the combined measurement data show a collective increase in depolarization. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 5. Kcnk5b enhanced depolarization is required for enhanced fin bud growth. (A, B) DiSBAC2(3) fluorescence measurements of the ectoderm (A) and mesenchyme cells (B) of developing pectoral fin buds at 32, 42, 48, and 56 hpf. (C–F) Confocal images of developing fin buds displaying the intensities fluorescence as photon counts per pixel at 32 hpf (C), 42 hpf (D), 48 hpf (E), and 56 hpf (F). Colors represent the range of counted photons per pixel. Blue representing the lowest level of counted photons, and red representing their highest counts (up to 750 photons or more). Images representing high photon count (C’, D’, E’, F’) and low photon count (C”, D”, E”, F”). The total exposure range was set at 1,500 counts (1.5). (G) Distribution of counted photons from DiSBAC2(3) in the confocal plane of the representative fin bud at 32 hpf. (H) Distribution of counted photons from DiSBAC2(3) in the confocal plane of the representative fin bud at 32 hpf. (I) Assessment of DiSBAC2(3) fluorescence intensity as counted photons for fin buds expressing mCherry or kcnk5b-mCherry. (J) Assessment of DiSBAC2(3) fluorescence intensity of mCherry-expressing or kcnk5b-mCherry fin buds at 32 hpf after treating for 4 h as indicated: ethanol (Et), DMSO, 10 μm vinpocetine (vin), 40 μm dibucaine (dib). (K) Representative image of pectoral fin bud of non-transgenic 48 hpf AB fish after heat shock at 32 hpf and start of treatment at 36 hpf with the drug solvents Ethanol (Et) and DMSO. (L) Representative image of pectoral fin buds of 48 hpf AB fish heat shocked at 32 hpf and start of treatment with 10 μm Vinpocetine (Vin) and 40 μm Dibucane (Dib) at 36 hpf. (M) Representative image of pectoral fin buds of 48 hpf transgenic Tg[hsp70:kcnk5b-GFP] after heat shock at 32 hpf and start of treatment with EtOH and DMSO at 36 hpf. (N) Representative image of pectoral fin buds of 48 hpf transgenic Tg[hsp70:kcnk5b-GFP] after heat shock at 32 hpf and start of treatment with 10 μm Vin and 40 μm Dib at 36 hpf. (O) Assessment of pectoral fin bud growth at 48 hpf expressing either AB and kcnk5b-GFP in the indicated treatment groups. Experiments were repeated 3 or more time (N ≥ 3). Each repeat contained 6 or more embryos; one fin bud was measured per embryo. For the DiSBAC2(3) fluorescence measurements, 6 independent points were measured from different 4 locations in each fin bud, distal, anterior, posterior, and proximal, and then averaged to represent a data point (A, B, I, J). For the fin bud size measurements, we measured 1 fin bud and eye per embryo. We measured at least 15 embryos per repeat, and each measurement is 1 data point (O). P values represent statistical analysis by Student’s two-tailed t test. P values >0.05 are designated as “not significant” (NS). Scale bars equal 100 μm. Numerical data used in this figure are included in S5 Data. hpf, hours post fertilization. https://doi.org/10.1371/journal.pbio.3002565.g005 To assess the effect of Kcnk5b-mediated decrease in intercellular K+ on the membrane potential, we overexpressed kcnk5b-mCherry or mCherry at 32 hpf, the time point with the lowest depolarization levels (Fig 5A–5C), and then measured membrane potential. We observed that compared to mCherry-expressing fin buds (Fig 5G and 5I), overexpression of kcnk5b-mCherry significantly increased depolarization (Fig 5H and 5I). These results indicated that Kcnk5b activity promotes depolarization. The observed increase in depolarization caused by kcnk5b, a leak channel that should promote hyperpolarization, suggested that other channels whose activity causes depolarization, such as Na+ channels, were involved. We therefore examined whether Kcnk5b-mediated depolarization required Na+-channel activity. We expressed either control mCherry or kcnk5b-mCherry and then treated these fish with the Na+-channel inhibitors Vinpocetine (a broad voltage-gated sodium channel inhibitor, including the TTX-insensitive channels) and Dibucane (broad sodium channel inhibition) (S7E–S7J Fig). We observed that inhibition of Na+ channels decreased depolarization in the fin bud as well as prevented Kcnk5b-induced increase in depolarization (Fig 5J). To test whether Na+-channel-mediated depolarization is required for kcnk5b-enchanced growth, we overexpressed kcnk5b-GFP and assess the effect of impairing depolarization using the Na+-channel inhibitors. Compared to the enhanced fin bud sizes of control-treated kcnk5b-GFP-transgenic fish (Fig 5M and 5O), treatment of kcnk5b-GFP-expressing fish with the Na+-channel inhibitors impaired Kcnk5b-induced fin bud growth (Fig 5N and 5O). Together, these data indicated that Kcnk5b activity promotes depolarization via Na+ channels and that Kcnk5b-induce depolarization is required for Kcnk5b-enhanced growth. IP3R-mediated Ca2+ release is required for Kcnk5b-induced shha expression and fin bud scaling Our observations that K+-leak channels increase the expression of important morphogens suggested that the channels are doing so through one or more signaling mechanisms. Since the intracellular accumulation of second messengers is involved in many signaling mechanisms, we assessed whether Kcnk5b activity alters the levels of particular second messengers. We used HEK293 cells for an initial assessment, because we previously observed that Kcnk5b induces SHH in these cells [11], a morphogen important for the development of early fin/limb buds [39]. To determine whether cAMP or cGMP levels change in response to Kcnk5b, we used FLIM with an established FRET-based sensor for each [40,41]. Compared to controls groups of unstimulated cells or cells stimulated with forskolin that produces cAMP (S8A Fig) or cells stimulated with SNAP to produce cGMP (S8B Fig), Kcnk5b did not significantly alter the intracellular levels of cAMP or cGMP (S8A and S8B Fig). We then assessed intracellular Ca2+ using the GCaMP6s sensor [42], and we observed that compared to cells transfected with control plasmid (Fig 6A and 6D), transfection with Kcnk5b led to significant increases in Ca2+ (Fig 6B and 6D). An inactive mutant version of the Kcnk5b channel, Kcnk5bMut (S8C Fig), also did not increase GCaMP6s activity (Fig 6C and 6D). To further assess the relationship between Kcnk5b and intracellular Ca2+, we transfected cells with GCaMP6s and either with Kcnk5b-mCherry or with mCherry. We subsequently cultured the cells for 24 h, FACS sorted both transfection groups for mCherry fluorescence, and then plated each group of sorted cells at 100% confluency (Fig 6Ea and 6Eb). mCherry-transfected cells that did not consistently display GCaMP6s fluorescence (Fig 6Ea, 6Ec, 6Ee, 6Eg arrow versus asterisk and 6F), while Kcnk5b-mCherry+ cells always showed high GCaMP6s activity (Fig 6Eb, 6Ed, 6Ef, 6Eh asterisks and 6F). We observed this phenomenon 100% of the time in all confocal images (Fig 6G). Together, these results indicate that Kcnk5b activity promotes a rise in intracellular Ca2+. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 6. Kcnk5b activity induces IP3R-mediated Ca2+ release from the ER. (A) GCaMP6s fluorescence in HEK293 cells. (B) GCaMP6s fluorescence in HEK293 cells expressing Kcnk5-mCherry. (C) GCaMP6s fluorescence in HEK293 cells expressing Kcnk5BSM-mCherry. (D) Fluorescence measurements of indicated groups. (E) Representative confocal images of mCherry-transfected, GCaMP6s-transfected cells showing brightfield with merged fluorescence from mCherry and GCaMP6s (a) or mCherry (c) or GCaMP6s (e) or merged mCherry-GCaMP6s (g), and Kcnk5b-mCherry-transfected, GCaMP6s-transfected cells: brightfield with merged fluorescence from mCherry and GCaMP6s (b) or Kcnk5b-mCherry (d) or GCaMP6s (f) or merged mCherry-GCaMP6s (h). (F) Measurements of GCaMP6s fluorescence intensity from the indicated experimental transfection groups. (G) Percents of mCherry-GCaMP6s-double positive over the total number of mCherry-positive cells in each group relate the frequency of GCaMP6s-positive cells in the mCherry or Kcnk5b-mCherry groups. (H) GCaMP6s fluorescence in the pectoral fin buds of transgenic fish harboring both Tg[Cca.actb:GCaMP6s] and Tg[hsp70:kcnk5b-mCherry]. (I) GCaMP6s fluorescence in embryos harboring the stable transgenic fish line Tg[Cca.actb:GCaMP6s] that mosaically express patches of mCherry or kcnk5b-mCherry (designated mCherry+) for the ectoderm (a) and mesenchyme (b). (J) Diagram of IP3R inhibition by 2-APB. (K) qRT-PCR of SHH expression in HEK293 cells transfected either with GFP or Kcnk5b-GFP after 20-h treatment with 2-APB at the indicated concentrations. (L) Assessment of pectoral fin bud size at 48 hpf after 4 h of treatment with 13 μm 2-APB. (M) Expression of shha in pectoral fin buds of heat-shocked non-transgenic sibling embryos after 4-h treatment with DMSO (a) or 13 μm 2-APB (b), of heat-shocked transgenic Tg[hsp70:kcnk5b-mCherry] siblings after 4-h treatment with DMSO (c) or 13 μm 2-APB (d). (N) Pixel area of in situ staining of shha in the indicated treatment groups. (O) qRT-PCR of shha expression in isolated fin buds of the indicated groups. Experiments were repeated 3 or more time (N ≥ 3). For cell culture experiments, each repeat contained duplicate or triplicate wells, and 10 or more cells were measured per well. Each data point represents 1 cell (D, F, G). For fin bud fluorescence measurements, we measured 2 or 3 locations in each tissue of 1 fin bud per embryo (H, I). For fin bud area measurements, we measured the area of 1 fin bud and eye per embryo. We measured at least 4 embryos per repeat (L). For the qRT-PCR experiments, we collected 80 fin bud samples per isolation. Three or more isolations were measured. Each isolation was measured in duplicate or triplicate. Each measured value is represented as a data point. P values represent statistical analysis by Student’s two-tailed t test. P values >0.05 are designated as “not significant” (NS). Scale bars equal 100 μm (A–C), 10 μm (E), 0.5 μm (M). Numerical data used in this figure are included in S6 Data. ER, endoplasmic reticulum; hpf, hours post fertilization. https://doi.org/10.1371/journal.pbio.3002565.g006 To determine whether Kcnk5b has the same effect on intercellular Ca2+ levels in vivo, we generated double-transgenic fish that harbored Tg[hsp70:kcnk5b-mCherry] and a CGaMP6s Ca2+ reporter under the control of the β-actin promoter Tg[Cca.actb:GCaMP6s] [43]. We assessed intracellular Ca2+ in the fin buds of heat-shocked transgenic kcnk5b-mCherry embryos at 48 hpf and their non-transgenic siblings as controls. We observed that Ca2+ levels were higher in kcnk5b-mCherry compared to non-transgenic embryos (Fig 6H). We observed similar results when we assessed mosaic embryos harboring the stable transgenic fish line Tg[Cca.actb:GCaMP6s] that mosaically express patches of mCherry or kcnk5b-mCherry (Fig 6Ia and 6Ib). Together, these data indicated that Kcnk5b normally promotes the increase in intracellular Ca2+ levels in vivo, and they suggested Ca2+ mediates the growth-inducing effect of Kcnk5b. Increases in Ca2+ caused by the decrease in intracellular K+ could occur from extracellular sources via Ca2+ channels in the plasma membrane and/or from intracellular sources such as the endoplasmic reticulum (ER) [44]. To determine the importance of these 2 sources, we tested whether inhibiting release of Ca2+ from each source impaired Kcnk5b-induced transcription of SHH in HEK293 cells. After inhibiting T- and L-type Ca2+ channels on the plasma membrane (S8D Fig), we did not detect a significant effect on Kcnk5b-induced SHH transcription (S8E and S8F Fig). However, after inhibiting IP3R-mediated release of Ca2+ from the ER (Fig 6J), we observed decreased Kcnk5b-induced SHH transcription dose-dependently (Fig 6K). To determine whether IP3R-mediated Ca2+ release was required for the Kcnk5b-enhanced growth of pectoral fin buds, we inhibited IP3R activity for 4 h and assessed bud growth in heat-shocked non-transgenic and transgenic Tg[hsp70:kcnk5b-GFP] siblings. We observed that the IP3R inhibitor blocked the growth phenotype caused by expression of kcnk5b at 48 hpf (Fig 6L). This decrease in growth was associated with the impairment of Kcnk5b-induced increase of shha expression domains (Fig 6M and 6N) and qRT-PCR (Fig 6O). The importance of Ca2+ release for shha expression and fin bud growth indicated that one or more Ca2+-dependent kinases were involved. Therefore, we tested which of the Ca2+-activated CaM kinases were required for Kcnk5b-enhanced expression of SHH in HEK293 cells. While we observed no significant effects by inhibiting CaMKII and CaMKIV (S9A Fig), we did observe that inhibition of CaMKK impaired Kcnk5b-enhanced SHH expression (Fig 7A). We then tested whether inhibiting CaMKK in the fin buds has similar effects on shha expression in vivo. We observed that inhibiting CaMKK decreased the expression of shha by in situ (Fig 7B and 7C) and by qRT-PCR (Fig 7D). Inhibiting CaMKK also impaired Kcnk5b-enhanced growth (Fig 7E). Furthermore, overexpressing the human CaMKK2 (Fig 7F) or the zebrafish camkk1b (Fig 7G) was sufficient to increase SHH expression in HEK cells. Together, these results indicated that Ca2+ is required for Kcnk5b-induced transcription of shha and that CaMKK is an important part of Ca2+ regulation of shha expression and pectoral fin bud growth. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 7. Kcnk5b requires CaMKK for growth and SHH/shha expression. (A) SHH expression in HEK293 cells transfected with GFP or Kcnk5b-GFP after 20-h treatment at the indicated concentrations of the CaMKK inhibitor STO-609 at the indicated concentrations. (B) Expression of shha in pectoral fin buds from heat-shocked non-transgenic siblings after 6-h treatment with DMSO (a) or 24 μm STO-609 (b) and from transgenic Tg[hsp70:kcnk5b-mCherry] siblings after treatment with DMSO (c) or STO-609 (d). (C) The ratios of each in situ staining area of shha to the area of its corresponding fin bud for the indicated groups. (D) qRT-PCR of shha expression in isolated fin buds in the indicated control and experimental groups. (E) Graphed assessment of fin-bud-area-to-eye-area measurement ratios. (F, G) Difference in SHH expression in HEK293 cells transfected with mCherry or human CaMKK2-mCherry (F) or camkk1b-mCherry (G). Experiments were repeated 3 or more time (N ≥ 3). For cell qRT-PCR experiments, each RNA isolation per well was measured in duplicate or triplicate. Each measured value is represented as a data point (A, F, G). For fin bud fluorescence measurements, we measured 2 or 3 locations in each tissue of 1 fin bud per embryo. For fin bud area measurements, we measured the area of 1 fin bud and eye per embryo. We measured at least 4 embryos per repeat (E). For fin bud qRT-PCR experiments, we collected 80 fin bud samples per isolation. Three or more isolations were measured. Each isolation was measured in duplicate or triplicate samples (D). Each measured value is represented as a data point. P values represent statistical analysis by Student’s two-tailed t test. P values >0.05 are designated as “not significant” (NS). Scale bars equal 0.5 μm (B). Numerical data used in this figure are included in S7 Data. https://doi.org/10.1371/journal.pbio.3002565.g007 Discussion K+-channel activity as a mechanism for regulating growth of vertebrate fin/limb buds The fish mutants longfin, another long fin, kcn13j, and rapunzel demonstrate that increasing the expression or the posttranslational activity of a K+ channel (Kcnh2a, Kcn13j, Kcnk5b, or Kcc4a) induces allometric growth of adult fin [9,10,12,13]. The observations that different K+ channels enhance growth support the conclusion that changes in intracellular K+ is what alters scaling. While Kcnk5b and other K+-leak channels reduce intracellular K+ levels [45], Kcn13j K+-rectifying channels facilitate K+ entry into the cells [46]. An explanation for why channels that release K+ and a channel that restores intracellular K+ both promote growth is current evidence that suggests that K+ leak channels act directly in tissues of the fin, while the K+-rectifying channel acts in the dermomyotome of somites to affect the later growth of adult fins [13]. The value of our in vivo FLIM measurements of intracellular K+ is that they provide direct evidence of a decrease in relative intracellular K+ levels in pectoral fin bud tissues during outgrowth (Fig 1D–1I), which supports the conclusion that growth of the pectoral fin buds involves an overall reduction of intracellular K+ in bud tissues. This conclusion is also supported by multiple observations: (1) the enhanced proportions from transgenic overexpression of different K+-leak channels (Fig 2); (2) the expression of Kcnk5b during the fin bud outgrowth and its down regulation as fin bud development ceases (Fig 3E and 3F); (3) the smaller fin bud size in CRISPR-targeted kcnk5b embryos (Fig 4A–4E); (4) the decrease in intracellular K+ by RA treatment (Fig 3A–3D); (5) the enhanced bud size (Fig 3K–3M); (6) the decreased K+ (Fig 3N) from rcan2 overexpression (which enhances growth); (7) rcan2-induced enhanced growth requires kcnk5b (Fig 4); and (8) CRISPR targeting of rcan2 both decreased bud size and increased K+ (Fig 3S and 3T). We also observed rcan2 expression and its similar effects on scaling and regulating intracellular K+ in adult caudal fins. The endogenous expression profiles of kcnk5b (Fig 3E and 3F) and rcan2 (Fig 3G) show that both are present during the growth of the fin bud. Rcan2 is also present in the adult fin blastema where it can regulate Kcnk5b and enhance growth (S5I–S5K Fig). While the accumulated data indicate that the expression of rcan2 coincides with kcnk5b in fin bud and fin growth, between 32 hpf and 34 hpf, we observed an increase in intracellular K+ (Fig 1E) despite the presence of rcan2 and kcnk5b. We posit that there is another mechanism regulating Kcnk5b or another K+ channel or K+ pumps that facilitates the increase in K+. We propose that the increase in K+ between 32 hpf and 34 hpf (Fig 1E) explains the pause in bud growth during this period (Fig 1F). We suspect the kcnk5b expression at 56 hpf (Fig 3Fb) means that this channel is involved in other phenomena such as early tissue differentiation, while the decreased rcan2 at 56 hpf (Fig 3Gd) indicates that the RA-Rcan2 mechanism no longer promotes Kcnk5b activity and, consequently, intracellular K+ levels increase. Zebrafish fin bud development occurs within a 24-h period, by which time it has transitioned into finfold growth, which initiates out of the distal-most region of the bud to form the pectoral fins of the larval fish. Because this transition between the conserved vertebrate fin/limb bud developmental stage, we primarily focused our characterizations to 48 hpf in order to define the contribution of intracellular K+ regulation in the scaling of the conserved embryonic bud structure. Thus, the changes in fin bud size caused by knockout or overexpression of kcnk5b or rcan2 are not large, but they are biologically significant. We attribute the variance that we observed at the measured time points to natural variations in growth within the narrow window of pectoral fin bud development. There are several signaling centers/zones within the developing fin buds [22], yet we observed few distinctions in the distribution of relative intracellular K+ levels in the early buds (Figs 1G–1I and S3). We propose this is due to gap junctions. Gap junctions are important for limb bud development and fin growth [47–49], and the shared distribution of relatively similar K+ levels via gap junctions can explain the coordinated regulation of each morphogen throughout the bud. We observed that the primary difference in relative levels of intracellular K+ exists between the mesenchyme and the ectoderm (Fig 1D and 1E). These differences may be due to an ECM barrier between the tissues, fewer gap-junctional connections and/or differences in molecular signals that influence K+ channel expression/activity. Because cells use intracellular K+ to control resting membrane potential [45,50], any differences in K+ levels between these tissues suggest that they have distinct electrophysiological properties. In regards to the in vivo electrophysiology, since K+-leak channels generally result in the outward flow of K+ ions, this should hyperpolarize the membrane potential as the cytoplasmic side becomes more negatively charged from the outflow of positive-charged ions. However, we observed that depolarization increases during outgrowth (Fig 5A–5F) while intracellular K+ decreases (Fig 1E). Furthermore, when we overexpress kcnk5b in the fin buds, we increased depolarization (Fig 5G–5I). We propose that the increased K+-channel activity first hyperpolarizes the membrane potential, and this increase in membrane potential immediately activates Na+ channels to depolarize the cells. This hypothesis fits with our findings that we impair the increase in depolarization (Fig 5J) as well as Kcnk5b-enhanced growth with Na+ channel inhibitors (Fig 5K–5O). We posit that the rise in intracellular K+ by 56 hpf directly facilitates the observed depolarization from the accumulation of K+ at the membrane. Integration of Kcnk5b into a mechanism of fin bud development and growth The coordinated changes in intracellular K+ (Fig 1) argue that there are common mechanisms regulating K+ dynamics. We identified retinoic acid-regulated signaling as one such mechanism (Fig 3), since RA was sufficient to decrease intracellular K+ in the ectoderm and mesenchyme (Fig 3A–3D). We previously showed that calcineurin inhibition induces allometric growth of adult fins [51] by increasing Kcnk5b activity [11]. Our findings that kcnk5b and the endogenous calcineurin inhibitor regulator of calcineurin 2 (rcan2) are present in pectoral fin buds (Fig 3E–3G) and that both are important in determining the size of the fin buds (Figs 3 and 4) along with our finding that RA increases rcan2 expression (Fig 3H–3J) suggest a mechanism that involves calcineurin-mediated antagonism of Kcnk5b activity to scale the buds (Figs 3 and 4). Our results support this model: (1) overexpression of the calcineurin inhibitor rcan2 decreased intracellular K+ and increases growth of fin buds (Fig 3M and 3N) and of adult fins (S5AA and S5BB); (2) overexpression of kcnk5b (which decreases intracellular K+) enhanced the developmental transcription and growth of the fin buds (Fig 2B–2W) and of adult fins [11]; and (3) a reduction in Kcnk5b activity by knockout or by point mutation impairs rcan2-induced bud growth (Fig 4F–4L). The expression patterns of rcan2 and kcnk5b were primarily in the mesenchyme of the buds (Fig 3E–3G). We interpreted these findings to mean that intracellular K+ is higher in the ectoderm because of very low rcan2 and kcnk5b levels or because of the absence of these 2 genes and the presence of another RA-regulated K+ channel in the ectoderm. Differences in the expression of K+ channels between the ectoderm and the mesenchyme can explain their differences in K+ levels (Fig 1D and 1E). RA does regulate the expression other morphogens and growth factors in fin/limb buds, so the linked changes in intracellular K+ between the ectoderm and mesenchyme can be coupled through alternative RA-mediated activities that affect the activity or expression of other K+-channels in the ectoderm. The coordinated regulation of intracellular K+ had some specificity, since RA did regulate intracellular K+ while a different nuclear hormone (thyroid hormone) did not have any effect (S5C Fig). We interpret these results to mean that TH3 signaling does not directly regulate the Kcnk5b-mediated growth mechanism even though thyroid hormones in other biological contexts can induce Rcan2 and/or promote growth [52,53]. We propose that any thyroid-mediated growth occurs via another molecular mechanism and not this electrophysiological one. Alternatively, there are endogenous factors present that limit the effects of this hormone, since it can promote metamorphosis [54], which needs occur later as the fish ends its larval stage [55]. Another possibility is that in early embryonic fin bud and homeostatic growth of the adult fin, the binding sites that thyroid nuclear hormone receptors require are either not accessible or are absent from the regulatory regions of the zebrafish rcan2. Considering the importance of thyroid hormone-mediated expression of rcan2 in the differentiation of osteoblasts [56], and considering that growth of the fin bud and regenerating fin needs cell proliferation before tissue differentiation, our findings may highlight one of these possibilities. K+-leak channels scale using Ca2+ An important question is how do K+-leak channels scale fin buds. Part of the answer involves IP3R-induced Ca2+ release from the ER. However, our previous findings showed that inhibition of the Ca2+-dependent phosphatase calcineurin increases the activity of Kcnk5b [11]. While this finding appears incongruent with our current finding that IP3R-mediated increase in intracellular Ca2+ is required for growth, the observed changes in SHH expression from our IP3R inhibition experiments (Fig 6K) offer an explanation: we observed that milder inhibition of IP3R (Fig 6K, 30 μm 2-APB) enhances Kcnk5b-induced expression of SHH, which we posit as reducing the pool of Ca2+ needed for calcineurin’s inhibition of Kcnk5b, while greater IP3R inhibition decreases SHH (Fig 6K, ≥105 μm 2-APB, and 6M–6O) by impairing other Ca2+-dependent enzymes needed for SHH expression, such as CaMKK (Fig 7). Our findings are in line with other evidence that point to the importance of Ca2+ in the growth of appendages. The L-type Ca2+ channel Cav1.2 can cause syndactyly, in which the bones of the digits improperly fuse, when mutations cause this channel to stay open longer and increase Ca2+ in the sarcoplasmic reticulum (muscle ER) [57]. Conversely, knock-out of Cav1.2 in the limb mesenchyme leads to shorter limbs due to impaired skeletal development [58]. In Drosophila wing discs, disruption of proteins that maintain ER Ca2+ stores—such as the Serca2 Ca2+ pump, the Orai Ca2+ channel in the plasma membrane, the Best2 Cl- channel in the ER membrane, or Stim, a scaffold protein that colocalizes Orai and Best2—leads to mispatterned, stunted adult wings [59]. The ER has the largest intracellular Ca2+ store, and the release of Ca2+ from the ER into the cytoplasm and active pumping of Ca2+ back into the ER occurs at regulated frequencies to generate oscillating cytoplasmic waves [60]. Ca2+ oscillations coordinate mesenchyme cell movement in the developing buds of feathers [61], and a similar phenomenon may regulate the growth of the fin buds [12,62]. Our observation that K+ is shared between cells (Fig 1K–1M) prompts questions about which ions are involved in the coordinated control of genes and allometric growth (Fig 2). We propose that the sustained transcription that is needed for prolonged allometric growth involves a sustained stimulus. Based on our observations, the decrease in intracellular K+ remains relatively constant as the fin bud grows (Fig 1E and 1F). Decreases in K+ could increase intracellular Ca2+ by increasing the amplitude or the duration of Ca2+ release from the ER. We posit that it enhances duration, since CaMKK activity is required, and this enzyme needs sustained durations of Ca2+ that achieve its two-step activation process: Ca2+ must be present long enough to interact with calmodulin and then allow the Ca2+-calmodulin complex to activate CaMKK. We conclude that K+ is an overarching long-term signal that adjusts IP3R-mediated Ca2+ release to regulate growth. Intracellular K+ in coordinated regulation of morphogens during development An important question is why use K+ channels to scale structures. One possibility is that specific K+ channels have interactions with specific growth-regulating receptors. The Thromboxane receptor interacts with the K+ miniK channel to regulate the channel’s activity [63]. Trimeric GPCRs and other membrane-associated signaling molecules interact with channels to impact channel function [64–66]. However, we currently believe that specific channel–receptor interactions do not explain our observations, because the similar allometric growth phenotypes can be induced by different K+ channels (Figs 2C and S4D–S4F) [9–12] that likely do not interact with the same growth-promoting receptors. A second possibility is that changing intracellular K+ levels alters the electrophysiology of cells to promote growth. Changes in intracellular K+ are known to alter the electrophysiological properties of cells [45], and such changes could alter the activities of pro-growth transmembrane receptors or membrane-associated signaling cascade components without direct interactions. A related possibility is that intracellular K+ is distributed throughout the cytoplasm, so changes in intracellular K+ could influence factors beyond the plasma membrane, such as the IP3R. Changes in K+ could alter activities of other transduction cascade components by influencing ionic interaction with charged amino acids in proteins or between them. Such a K+-mediated regulatory mechanism would not necessarily be an on-off switch, but could be an amplifier that augments the activities of signaling components that enhance the expression of existing developmental signals (Fig 2D–2W). The ionic “amplifier” mechanism fits with the observation that CRISPR-targeting of kcnk5b did not prevent growth of the fin buds; instead, it just reduced their proportions (Fig 4A–4E). K+ channels in the broader context of development and disease In a broader context of development, there are several findings that link different K+ outward-flow channels to human syndromes that harbor limb defects. The voltage-gated KCNH1 (Kv10.1), the two-pore channel KCNK4 (TRAAK/TREK) and the small-conductance Ca2+-activated KCNN3 (SK3/KCa2.3) are all K+ channels whose increase in activities can lead to hypoplasia/aplasia of the distal phalanges, as well as lead to alterations in cranial-facial features and neuropathies [8,67–70]. Conversely, mutations that impair KCNK9 activity produce bilateral hand contractures and talipes equinovarus feet [71,72]. Comparison between these findings and our findings shows the diversity in the physiological activities of K+ channels that decrease intracellular K+. While the different phenotypes can manifest from growth defects, we suspect that the differences are due to tissue-specific activities, since even the same channel in different cells can facilitate the transcription of different genes [11]. In addition to channels involved in outward K+ flow, inwardly rectifying K+ channels are important for physiology. While many discoveries link their importance to behavioral phenomena [73], mutations in KCNJ6 (GIRK2) or KCNJ13 (Kir7.1) result in severe cranial-facial malformations along with intellectual disabilities [74,75]. However, it is unclear how many of the detrimental phenotypes are due to KCNKJ6’s dysfunctional K+ flow, since the only characterized mutation that causes the loss of K+ selectivity, also gains Ca2+ permeability [76]. In regards to growth, increased expression of Kcnj2 is linked to hypoplasia of distal digit structures [77], and increased activity of another inward rectifying channel Kcnj13 has also been linked to enhanced growth of the adult fins, although it appears to initiate this defect via its embryonic activities [13]. K+ channels are also linked to tumor formation and cancer. Several different cancers harbor up-regulated expression and/or activity of K+-leak channels. Expression and activity of KCNK5 is up-regulated in some breast cancer cell lines. Signaling from Estrogen Receptor-α (ERα) has been found to promote breast cancer [78]. ERα signaling can up-regulate the expression of KCNK5, and blocking KCNK5 activity impairs the cell proliferation caused by activated ERα [79]. KCNK9 expression is also elevated in a number of breast cancer tumors, and experimentally overexpressing this channel promotes tumor formation in vivo [80]. A link between KCNN4 up-regulation and cell proliferation has been shown in smooth muscle cells of the vasculature [81]. Several cancer cell lines harbor elevated expression of KCNH1, and increasing this channel’s expression in cells can transform them into cancer-like cells [82]. KCNH5 is highly up-regulated in medulloblastomas, and targeted down-regulation of this channel reduced blastoma growth in vivo [83]. Tumor tissues and cell lines can also have reduced expression of the potassium channel regulator KCNRG [84], which reduces K+ currents across the plasma membrane [85,86] and reduces cell proliferation [84]. It is hypothesized that KCNRG reduces K+ channel expression, thereby limiting the release of intracellular K+ through in the plasma membrane [86]. Along these lines, decreased expression of Kv1.3, a rectifying K+ channel involved in restoring intracellular K+ levels, was observed to be down-regulated in breast adenocarcinoma cell line MCF-7 [87]. Our data are consistent with other finding that show that K+ channels that reduce intracellular K+ to promote growth. However, ours and others’ findings also indicate that K+ channel activity regulates more than cell proliferation, since K+ channel activity regulates the transcription of several morphogens in different regions of the developing buds and adult fins [11,88]. If Kcnk channels acted solely as oncogenes in the fin buds and adult fins, then they would likely produce tumors rather than foster coordinated allometric growth of the entire anatomical structures [9,11] or reversed polarity of regenerative outgrowth in adult fins [88]. Furthermore, the coordinated patterning alterations in craniofacial and limb structures by defective K+ channels also suggest that these channels regulate more than cell proliferation. The findings that individual dysfunctional K+ channels produce compound defects suggest that K+ channels have diverse activities via different molecular mechanisms. However, in the majority of cases, it remains unclear how these channels are involved. Ultimately, to get a better understanding for how specific K+ channels regulate the formation and growth of specific tissues and what regulates their channel contributions, future tissue-specific targeting of specific K+ channels is needed along with in vivo assessments of their electrophysiological activities. In summary, we propose that RA regulates intracellular K+ via an Rcan2-mediated increase in Kcnk5b activity to promote fin bud growth. The resulting decrease in intracellular K+ levels causes IP3R-mediated Ca2+ release that enhances shha transcription and other morphogens either directly or through Shha production (Fig 8). Thus, our observations integrate K+-channel-regulated scaling into known molecular controls of fin/limb bud development. While our proposed mechanism describes how one K+-leak channel is linked into developmental signals, details of the mechanisms still need to be defined. Given the diversity of ion channels and the importance of K+ in regulating the electrophysiological properties of cells, our findings may have broader implications in organ scaling and diseases that are caused by the loss of proportional growth. Continued work will determine the extent and diversity in how the electrochemical properties of cells interface with the molecular controls that govern organ development and proportional growth. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 8. Model. RA signaling induces the transcription of rcan2, an endogenous inhibitor of calcineurin, to alleviate calcineurin inhibition of Kcnk5b channel activity. Kcnk5b decreases intracellular K+ levels in the mesenchyme that increase depolarization to promote IP3R-mediated Ca2+ release from the ER. The increase in intracellular Ca2+ activates CaMKK, and both are required for the increased transcription of shha and enhanced growth of the pectoral fin buds by Kcnk5b. ER, endoplasmic reticulum; RA, retinoic acid. https://doi.org/10.1371/journal.pbio.3002565.g008 K+-channel activity as a mechanism for regulating growth of vertebrate fin/limb buds The fish mutants longfin, another long fin, kcn13j, and rapunzel demonstrate that increasing the expression or the posttranslational activity of a K+ channel (Kcnh2a, Kcn13j, Kcnk5b, or Kcc4a) induces allometric growth of adult fin [9,10,12,13]. The observations that different K+ channels enhance growth support the conclusion that changes in intracellular K+ is what alters scaling. While Kcnk5b and other K+-leak channels reduce intracellular K+ levels [45], Kcn13j K+-rectifying channels facilitate K+ entry into the cells [46]. An explanation for why channels that release K+ and a channel that restores intracellular K+ both promote growth is current evidence that suggests that K+ leak channels act directly in tissues of the fin, while the K+-rectifying channel acts in the dermomyotome of somites to affect the later growth of adult fins [13]. The value of our in vivo FLIM measurements of intracellular K+ is that they provide direct evidence of a decrease in relative intracellular K+ levels in pectoral fin bud tissues during outgrowth (Fig 1D–1I), which supports the conclusion that growth of the pectoral fin buds involves an overall reduction of intracellular K+ in bud tissues. This conclusion is also supported by multiple observations: (1) the enhanced proportions from transgenic overexpression of different K+-leak channels (Fig 2); (2) the expression of Kcnk5b during the fin bud outgrowth and its down regulation as fin bud development ceases (Fig 3E and 3F); (3) the smaller fin bud size in CRISPR-targeted kcnk5b embryos (Fig 4A–4E); (4) the decrease in intracellular K+ by RA treatment (Fig 3A–3D); (5) the enhanced bud size (Fig 3K–3M); (6) the decreased K+ (Fig 3N) from rcan2 overexpression (which enhances growth); (7) rcan2-induced enhanced growth requires kcnk5b (Fig 4); and (8) CRISPR targeting of rcan2 both decreased bud size and increased K+ (Fig 3S and 3T). We also observed rcan2 expression and its similar effects on scaling and regulating intracellular K+ in adult caudal fins. The endogenous expression profiles of kcnk5b (Fig 3E and 3F) and rcan2 (Fig 3G) show that both are present during the growth of the fin bud. Rcan2 is also present in the adult fin blastema where it can regulate Kcnk5b and enhance growth (S5I–S5K Fig). While the accumulated data indicate that the expression of rcan2 coincides with kcnk5b in fin bud and fin growth, between 32 hpf and 34 hpf, we observed an increase in intracellular K+ (Fig 1E) despite the presence of rcan2 and kcnk5b. We posit that there is another mechanism regulating Kcnk5b or another K+ channel or K+ pumps that facilitates the increase in K+. We propose that the increase in K+ between 32 hpf and 34 hpf (Fig 1E) explains the pause in bud growth during this period (Fig 1F). We suspect the kcnk5b expression at 56 hpf (Fig 3Fb) means that this channel is involved in other phenomena such as early tissue differentiation, while the decreased rcan2 at 56 hpf (Fig 3Gd) indicates that the RA-Rcan2 mechanism no longer promotes Kcnk5b activity and, consequently, intracellular K+ levels increase. Zebrafish fin bud development occurs within a 24-h period, by which time it has transitioned into finfold growth, which initiates out of the distal-most region of the bud to form the pectoral fins of the larval fish. Because this transition between the conserved vertebrate fin/limb bud developmental stage, we primarily focused our characterizations to 48 hpf in order to define the contribution of intracellular K+ regulation in the scaling of the conserved embryonic bud structure. Thus, the changes in fin bud size caused by knockout or overexpression of kcnk5b or rcan2 are not large, but they are biologically significant. We attribute the variance that we observed at the measured time points to natural variations in growth within the narrow window of pectoral fin bud development. There are several signaling centers/zones within the developing fin buds [22], yet we observed few distinctions in the distribution of relative intracellular K+ levels in the early buds (Figs 1G–1I and S3). We propose this is due to gap junctions. Gap junctions are important for limb bud development and fin growth [47–49], and the shared distribution of relatively similar K+ levels via gap junctions can explain the coordinated regulation of each morphogen throughout the bud. We observed that the primary difference in relative levels of intracellular K+ exists between the mesenchyme and the ectoderm (Fig 1D and 1E). These differences may be due to an ECM barrier between the tissues, fewer gap-junctional connections and/or differences in molecular signals that influence K+ channel expression/activity. Because cells use intracellular K+ to control resting membrane potential [45,50], any differences in K+ levels between these tissues suggest that they have distinct electrophysiological properties. In regards to the in vivo electrophysiology, since K+-leak channels generally result in the outward flow of K+ ions, this should hyperpolarize the membrane potential as the cytoplasmic side becomes more negatively charged from the outflow of positive-charged ions. However, we observed that depolarization increases during outgrowth (Fig 5A–5F) while intracellular K+ decreases (Fig 1E). Furthermore, when we overexpress kcnk5b in the fin buds, we increased depolarization (Fig 5G–5I). We propose that the increased K+-channel activity first hyperpolarizes the membrane potential, and this increase in membrane potential immediately activates Na+ channels to depolarize the cells. This hypothesis fits with our findings that we impair the increase in depolarization (Fig 5J) as well as Kcnk5b-enhanced growth with Na+ channel inhibitors (Fig 5K–5O). We posit that the rise in intracellular K+ by 56 hpf directly facilitates the observed depolarization from the accumulation of K+ at the membrane. Integration of Kcnk5b into a mechanism of fin bud development and growth The coordinated changes in intracellular K+ (Fig 1) argue that there are common mechanisms regulating K+ dynamics. We identified retinoic acid-regulated signaling as one such mechanism (Fig 3), since RA was sufficient to decrease intracellular K+ in the ectoderm and mesenchyme (Fig 3A–3D). We previously showed that calcineurin inhibition induces allometric growth of adult fins [51] by increasing Kcnk5b activity [11]. Our findings that kcnk5b and the endogenous calcineurin inhibitor regulator of calcineurin 2 (rcan2) are present in pectoral fin buds (Fig 3E–3G) and that both are important in determining the size of the fin buds (Figs 3 and 4) along with our finding that RA increases rcan2 expression (Fig 3H–3J) suggest a mechanism that involves calcineurin-mediated antagonism of Kcnk5b activity to scale the buds (Figs 3 and 4). Our results support this model: (1) overexpression of the calcineurin inhibitor rcan2 decreased intracellular K+ and increases growth of fin buds (Fig 3M and 3N) and of adult fins (S5AA and S5BB); (2) overexpression of kcnk5b (which decreases intracellular K+) enhanced the developmental transcription and growth of the fin buds (Fig 2B–2W) and of adult fins [11]; and (3) a reduction in Kcnk5b activity by knockout or by point mutation impairs rcan2-induced bud growth (Fig 4F–4L). The expression patterns of rcan2 and kcnk5b were primarily in the mesenchyme of the buds (Fig 3E–3G). We interpreted these findings to mean that intracellular K+ is higher in the ectoderm because of very low rcan2 and kcnk5b levels or because of the absence of these 2 genes and the presence of another RA-regulated K+ channel in the ectoderm. Differences in the expression of K+ channels between the ectoderm and the mesenchyme can explain their differences in K+ levels (Fig 1D and 1E). RA does regulate the expression other morphogens and growth factors in fin/limb buds, so the linked changes in intracellular K+ between the ectoderm and mesenchyme can be coupled through alternative RA-mediated activities that affect the activity or expression of other K+-channels in the ectoderm. The coordinated regulation of intracellular K+ had some specificity, since RA did regulate intracellular K+ while a different nuclear hormone (thyroid hormone) did not have any effect (S5C Fig). We interpret these results to mean that TH3 signaling does not directly regulate the Kcnk5b-mediated growth mechanism even though thyroid hormones in other biological contexts can induce Rcan2 and/or promote growth [52,53]. We propose that any thyroid-mediated growth occurs via another molecular mechanism and not this electrophysiological one. Alternatively, there are endogenous factors present that limit the effects of this hormone, since it can promote metamorphosis [54], which needs occur later as the fish ends its larval stage [55]. Another possibility is that in early embryonic fin bud and homeostatic growth of the adult fin, the binding sites that thyroid nuclear hormone receptors require are either not accessible or are absent from the regulatory regions of the zebrafish rcan2. Considering the importance of thyroid hormone-mediated expression of rcan2 in the differentiation of osteoblasts [56], and considering that growth of the fin bud and regenerating fin needs cell proliferation before tissue differentiation, our findings may highlight one of these possibilities. K+-leak channels scale using Ca2+ An important question is how do K+-leak channels scale fin buds. Part of the answer involves IP3R-induced Ca2+ release from the ER. However, our previous findings showed that inhibition of the Ca2+-dependent phosphatase calcineurin increases the activity of Kcnk5b [11]. While this finding appears incongruent with our current finding that IP3R-mediated increase in intracellular Ca2+ is required for growth, the observed changes in SHH expression from our IP3R inhibition experiments (Fig 6K) offer an explanation: we observed that milder inhibition of IP3R (Fig 6K, 30 μm 2-APB) enhances Kcnk5b-induced expression of SHH, which we posit as reducing the pool of Ca2+ needed for calcineurin’s inhibition of Kcnk5b, while greater IP3R inhibition decreases SHH (Fig 6K, ≥105 μm 2-APB, and 6M–6O) by impairing other Ca2+-dependent enzymes needed for SHH expression, such as CaMKK (Fig 7). Our findings are in line with other evidence that point to the importance of Ca2+ in the growth of appendages. The L-type Ca2+ channel Cav1.2 can cause syndactyly, in which the bones of the digits improperly fuse, when mutations cause this channel to stay open longer and increase Ca2+ in the sarcoplasmic reticulum (muscle ER) [57]. Conversely, knock-out of Cav1.2 in the limb mesenchyme leads to shorter limbs due to impaired skeletal development [58]. In Drosophila wing discs, disruption of proteins that maintain ER Ca2+ stores—such as the Serca2 Ca2+ pump, the Orai Ca2+ channel in the plasma membrane, the Best2 Cl- channel in the ER membrane, or Stim, a scaffold protein that colocalizes Orai and Best2—leads to mispatterned, stunted adult wings [59]. The ER has the largest intracellular Ca2+ store, and the release of Ca2+ from the ER into the cytoplasm and active pumping of Ca2+ back into the ER occurs at regulated frequencies to generate oscillating cytoplasmic waves [60]. Ca2+ oscillations coordinate mesenchyme cell movement in the developing buds of feathers [61], and a similar phenomenon may regulate the growth of the fin buds [12,62]. Our observation that K+ is shared between cells (Fig 1K–1M) prompts questions about which ions are involved in the coordinated control of genes and allometric growth (Fig 2). We propose that the sustained transcription that is needed for prolonged allometric growth involves a sustained stimulus. Based on our observations, the decrease in intracellular K+ remains relatively constant as the fin bud grows (Fig 1E and 1F). Decreases in K+ could increase intracellular Ca2+ by increasing the amplitude or the duration of Ca2+ release from the ER. We posit that it enhances duration, since CaMKK activity is required, and this enzyme needs sustained durations of Ca2+ that achieve its two-step activation process: Ca2+ must be present long enough to interact with calmodulin and then allow the Ca2+-calmodulin complex to activate CaMKK. We conclude that K+ is an overarching long-term signal that adjusts IP3R-mediated Ca2+ release to regulate growth. Intracellular K+ in coordinated regulation of morphogens during development An important question is why use K+ channels to scale structures. One possibility is that specific K+ channels have interactions with specific growth-regulating receptors. The Thromboxane receptor interacts with the K+ miniK channel to regulate the channel’s activity [63]. Trimeric GPCRs and other membrane-associated signaling molecules interact with channels to impact channel function [64–66]. However, we currently believe that specific channel–receptor interactions do not explain our observations, because the similar allometric growth phenotypes can be induced by different K+ channels (Figs 2C and S4D–S4F) [9–12] that likely do not interact with the same growth-promoting receptors. A second possibility is that changing intracellular K+ levels alters the electrophysiology of cells to promote growth. Changes in intracellular K+ are known to alter the electrophysiological properties of cells [45], and such changes could alter the activities of pro-growth transmembrane receptors or membrane-associated signaling cascade components without direct interactions. A related possibility is that intracellular K+ is distributed throughout the cytoplasm, so changes in intracellular K+ could influence factors beyond the plasma membrane, such as the IP3R. Changes in K+ could alter activities of other transduction cascade components by influencing ionic interaction with charged amino acids in proteins or between them. Such a K+-mediated regulatory mechanism would not necessarily be an on-off switch, but could be an amplifier that augments the activities of signaling components that enhance the expression of existing developmental signals (Fig 2D–2W). The ionic “amplifier” mechanism fits with the observation that CRISPR-targeting of kcnk5b did not prevent growth of the fin buds; instead, it just reduced their proportions (Fig 4A–4E). K+ channels in the broader context of development and disease In a broader context of development, there are several findings that link different K+ outward-flow channels to human syndromes that harbor limb defects. The voltage-gated KCNH1 (Kv10.1), the two-pore channel KCNK4 (TRAAK/TREK) and the small-conductance Ca2+-activated KCNN3 (SK3/KCa2.3) are all K+ channels whose increase in activities can lead to hypoplasia/aplasia of the distal phalanges, as well as lead to alterations in cranial-facial features and neuropathies [8,67–70]. Conversely, mutations that impair KCNK9 activity produce bilateral hand contractures and talipes equinovarus feet [71,72]. Comparison between these findings and our findings shows the diversity in the physiological activities of K+ channels that decrease intracellular K+. While the different phenotypes can manifest from growth defects, we suspect that the differences are due to tissue-specific activities, since even the same channel in different cells can facilitate the transcription of different genes [11]. In addition to channels involved in outward K+ flow, inwardly rectifying K+ channels are important for physiology. While many discoveries link their importance to behavioral phenomena [73], mutations in KCNJ6 (GIRK2) or KCNJ13 (Kir7.1) result in severe cranial-facial malformations along with intellectual disabilities [74,75]. However, it is unclear how many of the detrimental phenotypes are due to KCNKJ6’s dysfunctional K+ flow, since the only characterized mutation that causes the loss of K+ selectivity, also gains Ca2+ permeability [76]. In regards to growth, increased expression of Kcnj2 is linked to hypoplasia of distal digit structures [77], and increased activity of another inward rectifying channel Kcnj13 has also been linked to enhanced growth of the adult fins, although it appears to initiate this defect via its embryonic activities [13]. K+ channels are also linked to tumor formation and cancer. Several different cancers harbor up-regulated expression and/or activity of K+-leak channels. Expression and activity of KCNK5 is up-regulated in some breast cancer cell lines. Signaling from Estrogen Receptor-α (ERα) has been found to promote breast cancer [78]. ERα signaling can up-regulate the expression of KCNK5, and blocking KCNK5 activity impairs the cell proliferation caused by activated ERα [79]. KCNK9 expression is also elevated in a number of breast cancer tumors, and experimentally overexpressing this channel promotes tumor formation in vivo [80]. A link between KCNN4 up-regulation and cell proliferation has been shown in smooth muscle cells of the vasculature [81]. Several cancer cell lines harbor elevated expression of KCNH1, and increasing this channel’s expression in cells can transform them into cancer-like cells [82]. KCNH5 is highly up-regulated in medulloblastomas, and targeted down-regulation of this channel reduced blastoma growth in vivo [83]. Tumor tissues and cell lines can also have reduced expression of the potassium channel regulator KCNRG [84], which reduces K+ currents across the plasma membrane [85,86] and reduces cell proliferation [84]. It is hypothesized that KCNRG reduces K+ channel expression, thereby limiting the release of intracellular K+ through in the plasma membrane [86]. Along these lines, decreased expression of Kv1.3, a rectifying K+ channel involved in restoring intracellular K+ levels, was observed to be down-regulated in breast adenocarcinoma cell line MCF-7 [87]. Our data are consistent with other finding that show that K+ channels that reduce intracellular K+ to promote growth. However, ours and others’ findings also indicate that K+ channel activity regulates more than cell proliferation, since K+ channel activity regulates the transcription of several morphogens in different regions of the developing buds and adult fins [11,88]. If Kcnk channels acted solely as oncogenes in the fin buds and adult fins, then they would likely produce tumors rather than foster coordinated allometric growth of the entire anatomical structures [9,11] or reversed polarity of regenerative outgrowth in adult fins [88]. Furthermore, the coordinated patterning alterations in craniofacial and limb structures by defective K+ channels also suggest that these channels regulate more than cell proliferation. The findings that individual dysfunctional K+ channels produce compound defects suggest that K+ channels have diverse activities via different molecular mechanisms. However, in the majority of cases, it remains unclear how these channels are involved. Ultimately, to get a better understanding for how specific K+ channels regulate the formation and growth of specific tissues and what regulates their channel contributions, future tissue-specific targeting of specific K+ channels is needed along with in vivo assessments of their electrophysiological activities. In summary, we propose that RA regulates intracellular K+ via an Rcan2-mediated increase in Kcnk5b activity to promote fin bud growth. The resulting decrease in intracellular K+ levels causes IP3R-mediated Ca2+ release that enhances shha transcription and other morphogens either directly or through Shha production (Fig 8). Thus, our observations integrate K+-channel-regulated scaling into known molecular controls of fin/limb bud development. While our proposed mechanism describes how one K+-leak channel is linked into developmental signals, details of the mechanisms still need to be defined. Given the diversity of ion channels and the importance of K+ in regulating the electrophysiological properties of cells, our findings may have broader implications in organ scaling and diseases that are caused by the loss of proportional growth. Continued work will determine the extent and diversity in how the electrochemical properties of cells interface with the molecular controls that govern organ development and proportional growth. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 8. Model. RA signaling induces the transcription of rcan2, an endogenous inhibitor of calcineurin, to alleviate calcineurin inhibition of Kcnk5b channel activity. Kcnk5b decreases intracellular K+ levels in the mesenchyme that increase depolarization to promote IP3R-mediated Ca2+ release from the ER. The increase in intracellular Ca2+ activates CaMKK, and both are required for the increased transcription of shha and enhanced growth of the pectoral fin buds by Kcnk5b. ER, endoplasmic reticulum; RA, retinoic acid. https://doi.org/10.1371/journal.pbio.3002565.g008 Materials and methods Cloning Constructs were designed either by standard restriction enzyme or by homologous recombination methods. KIRIN1 was synthesized (Genewiz) and cloned into MCS region of pcDNA6-myc-6xHIS-tag plasmid (Invitrogen, V22120) or pBluescript (VWR, 95040–830) harboring the hsp70 zebrafish promoter by 2 miniTol2 sites (transgenic vector). We cloned kcnk5b-GFP, GFP, kcnk10a-GFP, kcnk5b-mCherry, CaMKK2-mCherry, camkk1b-mCherry, and mCherry, rcan2-mCherry into pcDNA-myc-6xHIS-tag or pBluescript II vector (Invitrogen AM1344) for expression in cells or fish. We cloned kcnk5b-GFP, rcan2-mCherry and mCherry into pXT7 vector (Addgene, #32995) for mRNA injection. Zebrafish husbandry AB strain fish were raised in 10 L tanks with constantly flowing water, 26°C standard light-dark cycle (Brand and colleagues) [89] HaiSheng aquarium system. Fish embryos and larva were raised in 1× E3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2, 0.33 mM MgSO4, 10%–5% Methylene Blue) until 10 to 12 dpf, then transferred to aquarium water tanks to grow. Transgenic lines established by screening for GFP, CFP, and mCherry expression after heat shock. Experiments used male and female fish equally. In general, embryos were immobilized by adding Tricane (MESAB) to E3 at a final concentration of 4 μm for 2 min (or 200 nM blebbistatin for 30 min where indicated) and then embedded in 1% low-melting agarose for 10 to 30-min imaging sessions. Where indicated, embryos were immobilized with Blebbistatin in 200 nM final concentration for 30 min instead of Tricane. Ethics statement Fish experiments were compliant to the general animal welfare guidelines and protocols (#20200903003) approved by legally authorized animal welfare committee, ShanghaiTech Animal Welfare Committee. Generation of transgenic lines Zebrafish embryos were collected at one-cell stage for plasmid injection. Transgenic lines harboring the hsp70:KIRIN1, hsp70:GFP hsp70:kcnk5b-GFP, hsp70:kcnk5b-mCherry, hsp70:mCherry, hsp70:kcnk10a-mCherry, or hsp70:rcan2-mCherry transgene plasmids were created by injecting 300 μg/μl of each construct together with mRNA of Tol2 transposase [90]. Positive embryos were screened after 37°C heat shock for 1 h. Positive embryos were brought up to adult fish and screened by crossing with wild-type fish. Then, the positive F1 and successive generations were screened by heat-shocking and identifying fluorescence expression. Adult fish were 6 months to a year old, unless indicated differently in the text. Embryonic stages that were used are indicated in the text. Growth measurements Embryos were first staged by the hour until 30 hpf and subsequently by the published head to body angles as the embryos continued develop [30 hpf (angle 85°), 32 hpf (90°), 34 hpf (97.5°), 36 hpf (105°), 38 hpf (110°), 40 hpf (115°), 44 hpf (125°), 46 hpf (130°), until 48 hpf (135°) [91]. Embryos were subsequently measured every 2 h. At each designated stage, the embryos were imaged using a Zeiss Stereoscope. The fin bud areas, eye areas, and otic vesicle areas were measured using Zen 3.4 (Blue edition) software by outlining each anatomical structure with the program’s contour tool and measuring the encompassed area. Heat-shock induction of transgenes Parent fish of heat-shock-driven transgenic lines were either outcrossed (rcan2, kcnk5b) as hemizygosity to same-strain wild-type fish or in-crossed (KIRIN1, GCaMP6s) for homozygosity. Progeny were collected in 1xE3 and raised at 28°C. All embryos were screen for their respective transgenes by a single heat shock at 37°C for 1 h at 24 hpf and selected for fluorescence. We determined that transgene expression was highest at 6 h post heat shock, so we designed our heat shock regimens accordingly. For FLIM imaging, transgenic lines Tg[hsp70:KIRIN1] embryos were heat shocked for 1 h at 37°C imaged 4 to 6 h later for each time point of the time course experiments. Tg[hsp70:kcnk5b-GFP] embryos and non-transgenic siblings were heat shocked twice: once at 36 hpf and again at 42 hpf for 1 h at 37°C to maximize target gene expression and then collected for the in situ and qRT-PCR experiments at 48 hpf. For pectoral fin bud growth experiments, Tg[hsp70:kcnk5b-GFP] and Tg[hsp70:rcan2-mCherry] embryos and non-transgenic siblings were heat shocked at 36 hpf and again at 42 hpf and then imaged at 48 hpf to maximize their expression over a significant portion of the pectoral fin bud growth period. For depolarization inhibitor treatments and subsequent DiSBaC2(3) experiments, Tg[hsp70:kcnk5b-GFP] embryos and non-transgenic siblings were heat shocked at 26 hpf and imaged at 32 hpf. For depolarization inhibitor treatments and subsequent measurements of fin bud size, Tg[hsp70:kcnk5b-GFP] embryos and non-transgenic siblings were heat shocked at 32 hpf once to avoid the heart problems induced by Na+ channel inhibition. For adult Tg[hsp70:rcan2-mCherry] and non-transgenic fish during caudal fin regeneration, the whole fish was incubated in 38.5°C aquarium water for 12 min once a day for the duration of each specific experiment. In situ hybridization mRNA probes were made from RT-PCR products isolated from 2 dpf zebrafish embryos or caudal fins from 6-month-old adults. The primer sequences for generating the published probes [92] are listed in Table 1. mRNA probes were made from RT-PCR products from fish larva (5 dpf) or regenerating (3 dpa) adult caudal fins. In vitro transcription reagents were from Promega. Embryos and tail fins were incubated in 4% PFA in 1xPBS at 4°C overnight with gentle rocking. Samples were dehydrated by incubation for 15 min in methanol at room temperature and then incubated in 100% methanol for ≥2 h at −20°C. Samples were then rehydrated using the reversed dehydration series of methanol/1xPBS solutions (75%, 50%, 25%). Samples were then incubated more than 4× in 1xPBS to remove all methanol, and subsequently incubated in 10 μm Proteinase K for 10 min at RT. Samples were then incubated 20 min in 4% PFA/1xPBS to inactivate the Proteinase K. Samples were incubated in 1xPBS 6× 10 min to remove the PFA, then incubated in pre-hybridization buffer for 3 h at 65°C. Samples were subsequently incubated in the hybridization solution containing 200 ng/ml of each mRNA probe ≥14 h at 65°C. Samples then were washed with successive wash steps to remove unbound probe and prepare for antibody incubation: once 2xSSC/75% deionized formalin at 65°C, once 2xSSC/50% deionized formalin at 65°C, once 2xSSC/25% deionized formalin at 65°C. 2xSSC at 65°C, twice 0.2xSSC at 65°C, 6 times 1xPBST (1xPBS with Tween-20), once in blocking solution [2% bovine albumin (Sigma-Aldrich, A3294-100G), 2% Sheep Serum (Meilunbio, M134510)] at RT for 4 h. Samples were incubated with Anti-digoxigenin-AP Fab Fragment (Sigma-Aldrich, 11093274910, RRID: AB_514497) in blocking solution ≥14 h at 4°C. Samples were then washed 6 times with 1xPBST, subsequently incubated in [0.1 M Tris-HCl (pH 9.5), 0.1 M NaCl, 0.05 M MgCl2] 3 times for 30 min, and then in Nitro Blue Tetrazolium (Sigma-Aldrich, N6639-1G) and 5-bromo-4-chloro-3-indolyl phosphate (Sigma-Aldrich, 136149) in [0.1 M Tris-HCl (pH 9.5), 0.1 M NaCl, 0.05 M MgCl2] at RT ≥8 h. Samples images under VS120-S6-W (OLYMPUS). For signal areas, images of the shha in situs were transferred to Fiji ImageJ (RRID SCR_003070). The stained regions were traced using “free ROI” and then quantitated using the “measure” function under the “analysis” menu to calculate the number of pixels contained in the stained area. For signal intensity analysis, the signal region and adjacent unstained regions were measured by “free ROI” selection and the mean intensity pixel values were determined from the “measure” function under the analysis menu by subtracting the mean value of the unstained region from the mean value of the signal region for each fin bud. Download: PPT PowerPoint slide PNG larger image TIFF original image Table 1. Materials table. https://doi.org/10.1371/journal.pbio.3002565.t001 qRT-PCR For transgenic expression in fish embryos, the embryos were heat-shocked twice at 6-h intervals at 32 hpf and 42 hpf to maximize chronic transgene effects on developmental gene expression. Fin buds were isolated by using tip of a micropipette to gently remove the yolk of the 48 hpf embryos. The heads were removed using the small lancet iris knife (Zhongyuan Healthcare Equipment, Cat # hyl10.31372714). The main body of the embryo was removed by using tweezers to fix the body (neural tube facing up) and cutting the fin buds and residual yolk membrane from the body on both side with a small lancet iris knife. The fin buds and residual yolk membrane were collected in the TRIZOL (CWBio, 03917). Over 40 embryos (80 fin buds) were collected for each sample group. For adult caudal fin tissues, we heat shocked adult fish once for 12 min at 38.5°C 6 h before the tissue isolation. After anesthetizing adult fish with 0.4% Tricane, we surgically isolated the distal caudal fin tissues from the fin tip including the distalmost bone bifurcation. For qRT-PCR experiments from HEK293 cells, we isolated RNA from 90% confluent 6 cm well per replicate. The mRNA from fish or cultured HEK293 cells was isolated using Trizol (CWBio, 03917). Then, 1 μg mRNA was used for the reverse transcription to cDNA using 4× gDNA wiper Mix, 5× HiScript III qRT SuperMix (Vazyme, L/N 7E350C9). qRT-PCR was performed using 2× ChamQ Universal SYBR qPCR Master Mix (Vazyme, L/N TE342F9) with QuantStudio3 machine (Thermo Fisher). The cycle procedure was at 50.0°C for 2 min, 95°C for 10 min in the stage 1; 95°C for 15 s, 55.0°C for 20 s for 40 routine in the stage 2; 95°C for 15 s, 60°C for 1 min, 95°C for 15 s in the Melt Curve. Samples were standardized using the detected ef1a expression for fish mRNA isolation and GAPDH for human cell line mRNA isolations. Please see the “Table 1” in this section for the primer sequences of each gene. Protein extraction and western blot Zebrafish were anesthetized with 0.4% Tricaine and fin tips (fin tissues from the distalmost edge of each fin to the 2 distalmost bone segments of the bone rays) were surgically isolated and homogenized in RIPA buffer with protease inhibitors (Thermo/Pierce), then incubated at 4°C for 1 h under constant shaking. Lysates were centrifuge at 13,000r for 15 min at 4°C. Protein concentration was determined using the BCA method kit (MDbio). Proteins were denatured (incubated at 95°C for 10 min in loading buffer and then transferred to ice immediately prior to loading (50 μg/lane) in a 10% SDS-PAGE Acrylamide-Bis (29:1) gel). After electrophoresis, gel contents were transferred onto 0.2 μm PVDF membrane. After transfer, the membrane incubated in 5% skim milk/0.5% TBST for 1.5 h at room temperature and then washed 3× 5 min room temperature. Blots were then incubated with Rcan2 primary antibody (1:1,000) overnight at 4°C. Blots were then wash in 1xTBST 3× 5 min and then incubated in secondary antibody (1:1,000) in blocking buffer for 2 h at room temperature. Blots were then washed 3× 5 min. Protein bands were detected using ECL chemiluminescence detection reagent (Perkin Elmer) and luminescence was detected using an Amersham Imager 600 chemiluminescence imager (General Electric). Cell culture HEK293 cell cultures (from ATCC.org) were incubated at 37°C, 5% CO2, 95% humidity in incubators (Thermo Fisher, FORMA STERI-CYCLE i160) in DMEM medium (Gibco,1199506) with 10% FBS (Gibco,1009914) and 1% penicillin streptomycin (Gibco, 15140122). The identity of the cell lines was not authenticated, and mycoplasma was not detected. Cells were split to 50% density and transfected with Lipofectamin (Invitrogen, 11668–019) 12 h later. Expression for the transfected constructs was evaluated by expression of fluorescent marker. All cultures tested negative for mycoplasma, which we tested by collecting 1 ml of DMEM cultured over 24 h with HEK293 cells, and centrifuge at 12,000 rpm for 1 min. Then, do the PCR as following primers: MGSO-TGCACCATCTGTCACTCTGTTAACCTC, GPO-3-GGGAGCAAACAGGATTAGATACCCT. FRET-FLIM detection and analysis HEK293 cells were transfected with 1 μg of pcDNA-kcnk5b-GFP; pcDNA6-mT2-CUTie-YFP [40] or pcDNA-CFP-cGi500-YFP [41]. The transgenic fish [hsp70:KIRIN1] were heat shock at 37°C for 1 h for the sensor to induce peak transgene expression 6 h later. Embryos were incubated in 250 μm PTU (1-phenyl 2-thiourea, Sigma) to prevent pigmentation starting at 24 hpf. Ten minutes before imaging, embryos were anesthetized in 4 μm MESAB (or 200 nM blebbistatin for 30 min, where indicated) and then embedded in 1% agarose before imaging. For other imaging experiments, the time points were indicated in the text or figure legends. For FLIM measurements of the adult caudal fins, adult fish (approximately 6 months old) were anesthetized with 1% MESAB and then placed on wet plastic Petri dish and imaged for 5 min and then placed back into aquarium water to regain consciousness. Measurements were from 3 different locations in the mesenchyme of caudal fins: 1 at each of the 2 lobe tips and 1 in the distal midline between the 2 lobes. Each measurement is 1 data point. For the embryonic developmental time course of fluorescence lifetime imaging (FLIM-FRET) measurements, each embryo was imaged at 32 hpf, 35 hpf, 38 hpf, 42 hpf, 48 hpf, and 56 hpf. Fluorescence lifetime imaging measurements were made by photon counting the fluorescence emission of either CFP (KIRIN1, cAMP, cGMP) using a 2-photon-confocal Hyperscope (Scientifica, United Kingdom) and PMT-hybrid 40 MOD 5 photon detectors (Picoquant, Germany). We determined exposure times for each plane based on the number of photons detected individual pixels. In each fin bud, we generally measured 3 regions within the mesenchyme and ectoderm by selecting regions of interest at distal, proximal, and central locations within the buds. Each region contained several pixels of that ranged from 1 to 3 cells in size. For each measurement, we used thresholds of several hundred photons per pixel to overcome background photon emissions and prevent any influence in the FLIM measurements. The lifetime curves were analyzed from graphically selected regions or cells in the confocal plane. The lifetime curves were determined (curve fitting) from the decay of the counted photon emissions after single laser pulses (2-photon Chameleon Ultra II Ti Sapphire laser) using the algorithm for base computations of multi-exponential reconvolution with corrections for instrument response time (instrument response factor, IRF) and removal of background signals <100/pixel in the program SymPhoTime 64, version 2.4 (Picoquant, Germany, RRID: SCR_016263). The fitted curves represented 2 component (fluorophores FRET at <10 nm and fluorophores no FRET >10 nm) equations with χ2 values of 1 ± 0.19 for accuracy. To visualize differences in spatial distribution of each pixel lifetime value, individual lifetime values were distributed along a rainbow scale in which boundary lifetime values were assigned: blue lowest boundary and red highest boundary. FRET efficiency (on which our FLIM measurements are based) using our donor-only (CFP) and sensor-based donor + acceptor (CFP-YFP) K+-detection data in Fig 1A. The resulting average efficiency was 37.9%. We entered this calculated efficiency in a standardized equation [RDA = R0 x 6√(1-Efficiency)2/Efficiency)] to calculate the effective distance between the 2 fluorophores for FRET from our measurements, (4.59 Å) and then to compare this to the reference distance of maximal FRET of CFP and YFP (4.7 Å) [93]. RDA is calculated distance between donor and acceptor fluorophores. R0 is the distance required for 50% efficiency. The closeness our calculated and reference distances indicated meaningful efficiency of the K+-sensor in our in vivo measurements. KIRIN1 sensor As previously published, KIRIN1 sensor was generated by fusing CFP and YFP to the N- and C-termini of the K+-binding protein (Kbp) from E. coli [28]. The Kbp protein normally maintains an elongated conformation that keeps the donor and acceptor fluorophores apart, which prevents FRET between the fluorophores. When K+ binds to the bacterial OsmY (BON) domain in the N-terminus and the Lysin motif (LysM) in the C terminus, it brings the 2 fluorophores together to allow FRET [28]. This results in a reduction in the lifetime of the donor (CFP) fluorophore, which can be detected by FLIM. DiSBAC2(3) staining and measurements DiSBAC2(3) was diluted to a 10 μm concentration in 1× E3 medium that contained 250 μm concentration of PTU (1-phenyl 2-thiourea, Sigma, P7629). Embryos were put into PTU at 24 hpf to prevent the pigmentation of the embryos. Three hours before imaging, embryos were incubated in diluted DiSBAC2(3) at RT under gentle rocking. They were immobilized by adding blebbistatin to a final concentration of 200 nM for 30 min before imaging, embedded in 1% low-melting agarose gel, and then imaged under a 2-photon-confocal Hyperscope (Scientifica, UK) at 900 nm to maximize the detection of the fluorescence from the DiSBAC2(3) dye and minimize fluorescence of mCherry under 2-photon stimulation (www.fpbase.org). Photon counts were detected using PMT-hybrid 40 MOD 5 photon detectors (Picoquant, Germany) that detect photons (provide a numerical count) for each pixel. We used blebbistatin, because it inhibits muscle movement by inhibiting myosin activity and does not target any ion channel. We imaged each embryo for 3 min and measured 6 pixels in specific regions in each fin bud (distal mesenchyme, anterior mesenchyme, proximal mesenchyme, and posterior mesenchyme). The 6 pixels from each region were averaged and each average represents a data point in the graph; 32 hpf were chosen for imaging the effect of kcnk5b overexpression, because of the low levels of depolarization at this stage (Fig 5A–5C). CRISPR knockouts for rcan2 and kcnk5b The ChopChop online tool (http://chopchop.cbu.uib.no/) was used to design sgRNAs to limit predicted off-target sgRNA cutting. The fourth exon of rcan2 was targeted by the sgRNA sequence is 5′-AACCGACTGGAGGTGAGCCA-3′ ordered from IDT with Alt-R gRNA modification. The first exon in kcnk5b was targeted by the sgRNA sequence 5′-TGCGATATTCCAAATCCTCG-3′ and 5′-GGTCTTGGGCGAAATTATTG-3′ ordered by IDT with Alt-R gRNA modification, and 1 μl sgRNAs (400 ng/μl) and Cas9 (500 ng/μl) protein (Genscript, Cat. #Z03469) were co-injected into single-cell zebrafish embryos. After imaging for experiments, each embryo was numbered and raised separately to 5 dpf. Genomic DNA was isolated from each 5 dpf larva individually. After genomic DNA extraction by alkaline lysis buffer, the genomic region surrounding the target sites was PCR amplified using primers forward: 5′-CAACACACTCTCTGGCTTTCAG-3′ and reverse: 5′-TGAACTGCATAGTTGAGATGGG-3′ for rcan2 or forward: 5′-ATCACCAGAAACTTGGGAGTGT-3′ and reverse: 5′-GCTTTTGCGTGAGAACTACCAT-3′ for kcnk5b and set for sequencing (Genewiz.com.cn). In vivo mRNA overexpression experiments Mutated rcan2-P2A-mcherry, kcnk5b-GFP and kcnk5bS345A-GFP were cloned into pXT7 vector (from Lin Haifan and Bao Baolong lab) and either linearized with Xba1 (Anza Thermo Scientific #ER0682) or amplified by PCR to make templates. For rcan2, the original target sequence 5′-AACCGACTGGAGGTGAGCCA-3′ was mutated to 5′-ATCCAACCGGTGGGCTTCCG-3′. For kcnk5b, the original target-2 sequence 5′-TGCGATATTCCAAATCCTCGAGG-3′ was mutated to 5′-TGCAATCTTTCAGATACTAGAAG-3′, and the original target-14 sequence 5′-GGTCTTGGGCGAAATTATTGAGG-3′ was mutated to 5′-GGTATTAGGAGAGATCATCGAAG-3′ to impair interaction between the sgRNA targeting sequences and the rescue mRNAs without changing coded amino acids. The linear template was transcribed to mRNA using T7 MESSAGE MACHINE (Invitrogen, #AM1344). Concentrations injected into embryos: 1 μl Cas9 protein, 1 μl sgRNA, 1 μl rcan2-P2A-mCherry mRNA/kcnk5b-P2A-GFP (700 ng/1 μl), mCherry mRNA (375 ng/1 μl), 1 μl empty sgRNA (T7-sgRNA scaffold, 2,000 ng/μl) or kcnk5bS345-GFP/kcnk5bS345A-GFP (200 ng/μl). Mosaic analyses in fin buds Double mosaic embryos were created by injecting 120 μg/μl of hsp70:kcnk5b-mCherry and 400 ng/μl into Tg[hsp70:KIRIN1] separately into 1–4 cell AB embryos. Embryos only expressing kcnk5b-mCherry mosaically were injected with hsp70:kcnk5b-mCherry into the transgenic Tg[Cca.actb:GCaMP6s] line. Injected embryos and larva were raised in 1xE3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2, 0.33 mM MgSO4, 10%–5% Methylene Blue). The mosaic larvae were selected by screening for mCherry and CFP expression for kcnk5 and KIRIN1 transgenes or mCherry and GFP expression for kcnk5b and GCaMP6s transgenes in the fin buds 4 to 6 h after heat-shock induction. Small molecule treatments All inhibitors (except for Vinpocetine) were dissolved in DMSO: Retinoic acid (Abcone, R26255) 10 mM stock concentration, 2-APB (IP3R inhibitor, ENZO, ALX-400-045-M100) 75 mM stock concentration, STO-609 (CaMKK inhibitor, MedChemExpress, HY-19805) 0.5 mM stock concentration, Verapamil (T-/L-type Ca2+ channel inhibitor, MedChemExpress, HU-A0064) 100 mM stock concentration, KN-62 (CaMKII/IV inhibitor, MedChemExpress, HY-13290) 1 mM stock concentration, or in sterile water Mibefradil (T-/L-type Ca2+ channel inhibitor, MedChemExpress, HY-15553A) 2.5 mM stock concentration, Forskolin (adenylyl cyclase activator, MedChemExpress, HY-15371) 25 mM, SNAP (guanylyl cyclase activator, Tocris, 1561) 100 mM stock concentrations. HEK293 cells were incubated in DMEM cell culture medium (Gibco, 1199506) containing 10% FBS (Gibco, 1009914) and 1% penicillin streptomycin (Gibco, 15140122) at 37°C, 5% CO2. Ten hours after transfection, the drugs were added to the medium to their final concentrations (as indicated in the figure legends or the specific methods section). Cells were then trypsinized and RNA was isolated as indicated for qRT-PCR. For fish embryos, we added the small molecules mentioned above to E3 to the final concentrations (200 nM RA, 13 μm 2-APB, or 24 μm STO unless indicate otherwise) and incubated the embryos 28°C. We determined the treatment times and concentrations (provided in the figure legends or text) to be the most effective on the fin bud growth and have the least effect on embryonic development. Adult fish were treated with 100 μm RA for 6 h, since we determined this concentration showed reproducible effects (S5F Fig). Vinpocetine (MedChemExpress, HY-13295) was dissolved in ethanol to make a 10 mM stock, Dibucaine (MedChemExpress, HY-B0552) was dissolved in DMSO to make a 500 mM stock. For assessing the effects of these inhibitors on depolarization, we treated embryos with 10 μm Vinpocetine and 40 μm Dibucaine for 4 h before assessment, since these concentrations showed continued reproducible inhibitory effects. For fin bud growth experiments, 10 μm Vinpocetine and 40 μm Dibucaine treatment started at 36 hpf for 12 h to maximize the effects of the inhibitor on fin bud growth by 48 hpf. GCaMP6s measurements HEK293 cells were transfected with 1 μg of CMV-GCaMP6s, 1 μg of pcDNA-kcnk5b-MCherrry or pcDNA-mCherry or pcDNA-Kcnk5bMut. The pcDNA-Kcnk5bMut harbors 2 mutations: one changing the LEEP sequence in the first transmembrane region to VKKA and substituting S345 in the C-terminal tail with alanine. This generated a dead channel. HEK293 cells for Fig 6E were transfected and sorted by SORP ARIA Fusion (BD Biosciences) for mCherry. GCaMP6s transgenic fish Tg[Cca.actb:GCaMP6s] [43] were purchased from Zebrafish Stock Center Wuhan, China (CZ1282 doi.org/10.3390/ijms22115551). One-celled embryos from the established Transgenic fish Tg[Cca.actb:GCaMP6s] [43] were injected with Tg[hsp70-kcnk5b-mCherry] and were heat shocked at 37°C for 1 h at 42 hpf and imaged at 48 hpf. All the samples were imaged by Zeiss LSM 980 upright and analyzed by ZEN-Blue (Zeiss). To standardize intensity measurements and limit influence of differences in expression, we standardized all measurements to GCaMP6s cells lacking mCherry expression in the spinal cord (body) away from the fin bud. The results are provided as GFP bud/GFP body. Patch clamping Transfected HEK293T cells were seeded on glass coverslips (Fisher Brand) and incubated in cell culture medium at 37°C, 5% CO2, 95% relative humidity for 8 to 10 h. The seeded coverslips were transferred into Tyrode’s solution (138 mM NaCl, 4 mM KCl, 2 mM CaCl2, 1 mM MgCl2, 0.33 mM NaH2PO4, 10 mM Glucose, 10 mM HEPES). Cells were assessed in the ruptured-patch whole-cell configuration of the patch-clamp technique using and EPC9 or EPC10 amplifier (HEKA) with borosilicate glass pipettes (Sutter Instruments) with 3 to 6 MΩ resistance when filled with pipette solution (130 mM glutamic acid, 10 mM KCl, 4 mM MgCl2, 10 mM HEPES, 2 mM ATP, pH to 7.2). For detecting potassium current, after gigaseal formation, cells were voltage-clamped at −80 mV. Potassium conductance was elicited by test pulses from −100 mV to 70 mV (in 10 mV increments) of 600 ms duration at a cycle length of 10 s. The resulting tracings were converted into itx files by the ABF Software (ABF Software, RRID: SCR_019222) and then analyzed using Clampfit Software (Molecular Devices, RRID: SCR_011323). Currents were measured at the end of the test pulses. Quantification and statistical analyses To select for accurate two-component decay curve fittings of the FRET sensors, we used Chi-square analyses limiting fits 0.8>χ2<1.2 values, which was calculated within the SymPhoTime 64 software, version 2.4 (Picoquant, Germany, RRID: SCR_016263). Graphs were assembled and statistical analyses were done using Microsoft Excel (Microsoft.com) or Prism9 (Graphpad.com). All statistical analyses for graphs were done using the Students t test to assess the reproducibility of any differences between 2 groups within the data sets of each experiment. P values differences of >0.05 were indicated to be not significant “NS”; otherwise, the calculated P values were provided in the figure. Cloning Constructs were designed either by standard restriction enzyme or by homologous recombination methods. KIRIN1 was synthesized (Genewiz) and cloned into MCS region of pcDNA6-myc-6xHIS-tag plasmid (Invitrogen, V22120) or pBluescript (VWR, 95040–830) harboring the hsp70 zebrafish promoter by 2 miniTol2 sites (transgenic vector). We cloned kcnk5b-GFP, GFP, kcnk10a-GFP, kcnk5b-mCherry, CaMKK2-mCherry, camkk1b-mCherry, and mCherry, rcan2-mCherry into pcDNA-myc-6xHIS-tag or pBluescript II vector (Invitrogen AM1344) for expression in cells or fish. We cloned kcnk5b-GFP, rcan2-mCherry and mCherry into pXT7 vector (Addgene, #32995) for mRNA injection. Zebrafish husbandry AB strain fish were raised in 10 L tanks with constantly flowing water, 26°C standard light-dark cycle (Brand and colleagues) [89] HaiSheng aquarium system. Fish embryos and larva were raised in 1× E3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2, 0.33 mM MgSO4, 10%–5% Methylene Blue) until 10 to 12 dpf, then transferred to aquarium water tanks to grow. Transgenic lines established by screening for GFP, CFP, and mCherry expression after heat shock. Experiments used male and female fish equally. In general, embryos were immobilized by adding Tricane (MESAB) to E3 at a final concentration of 4 μm for 2 min (or 200 nM blebbistatin for 30 min where indicated) and then embedded in 1% low-melting agarose for 10 to 30-min imaging sessions. Where indicated, embryos were immobilized with Blebbistatin in 200 nM final concentration for 30 min instead of Tricane. Ethics statement Fish experiments were compliant to the general animal welfare guidelines and protocols (#20200903003) approved by legally authorized animal welfare committee, ShanghaiTech Animal Welfare Committee. Generation of transgenic lines Zebrafish embryos were collected at one-cell stage for plasmid injection. Transgenic lines harboring the hsp70:KIRIN1, hsp70:GFP hsp70:kcnk5b-GFP, hsp70:kcnk5b-mCherry, hsp70:mCherry, hsp70:kcnk10a-mCherry, or hsp70:rcan2-mCherry transgene plasmids were created by injecting 300 μg/μl of each construct together with mRNA of Tol2 transposase [90]. Positive embryos were screened after 37°C heat shock for 1 h. Positive embryos were brought up to adult fish and screened by crossing with wild-type fish. Then, the positive F1 and successive generations were screened by heat-shocking and identifying fluorescence expression. Adult fish were 6 months to a year old, unless indicated differently in the text. Embryonic stages that were used are indicated in the text. Growth measurements Embryos were first staged by the hour until 30 hpf and subsequently by the published head to body angles as the embryos continued develop [30 hpf (angle 85°), 32 hpf (90°), 34 hpf (97.5°), 36 hpf (105°), 38 hpf (110°), 40 hpf (115°), 44 hpf (125°), 46 hpf (130°), until 48 hpf (135°) [91]. Embryos were subsequently measured every 2 h. At each designated stage, the embryos were imaged using a Zeiss Stereoscope. The fin bud areas, eye areas, and otic vesicle areas were measured using Zen 3.4 (Blue edition) software by outlining each anatomical structure with the program’s contour tool and measuring the encompassed area. Heat-shock induction of transgenes Parent fish of heat-shock-driven transgenic lines were either outcrossed (rcan2, kcnk5b) as hemizygosity to same-strain wild-type fish or in-crossed (KIRIN1, GCaMP6s) for homozygosity. Progeny were collected in 1xE3 and raised at 28°C. All embryos were screen for their respective transgenes by a single heat shock at 37°C for 1 h at 24 hpf and selected for fluorescence. We determined that transgene expression was highest at 6 h post heat shock, so we designed our heat shock regimens accordingly. For FLIM imaging, transgenic lines Tg[hsp70:KIRIN1] embryos were heat shocked for 1 h at 37°C imaged 4 to 6 h later for each time point of the time course experiments. Tg[hsp70:kcnk5b-GFP] embryos and non-transgenic siblings were heat shocked twice: once at 36 hpf and again at 42 hpf for 1 h at 37°C to maximize target gene expression and then collected for the in situ and qRT-PCR experiments at 48 hpf. For pectoral fin bud growth experiments, Tg[hsp70:kcnk5b-GFP] and Tg[hsp70:rcan2-mCherry] embryos and non-transgenic siblings were heat shocked at 36 hpf and again at 42 hpf and then imaged at 48 hpf to maximize their expression over a significant portion of the pectoral fin bud growth period. For depolarization inhibitor treatments and subsequent DiSBaC2(3) experiments, Tg[hsp70:kcnk5b-GFP] embryos and non-transgenic siblings were heat shocked at 26 hpf and imaged at 32 hpf. For depolarization inhibitor treatments and subsequent measurements of fin bud size, Tg[hsp70:kcnk5b-GFP] embryos and non-transgenic siblings were heat shocked at 32 hpf once to avoid the heart problems induced by Na+ channel inhibition. For adult Tg[hsp70:rcan2-mCherry] and non-transgenic fish during caudal fin regeneration, the whole fish was incubated in 38.5°C aquarium water for 12 min once a day for the duration of each specific experiment. In situ hybridization mRNA probes were made from RT-PCR products isolated from 2 dpf zebrafish embryos or caudal fins from 6-month-old adults. The primer sequences for generating the published probes [92] are listed in Table 1. mRNA probes were made from RT-PCR products from fish larva (5 dpf) or regenerating (3 dpa) adult caudal fins. In vitro transcription reagents were from Promega. Embryos and tail fins were incubated in 4% PFA in 1xPBS at 4°C overnight with gentle rocking. Samples were dehydrated by incubation for 15 min in methanol at room temperature and then incubated in 100% methanol for ≥2 h at −20°C. Samples were then rehydrated using the reversed dehydration series of methanol/1xPBS solutions (75%, 50%, 25%). Samples were then incubated more than 4× in 1xPBS to remove all methanol, and subsequently incubated in 10 μm Proteinase K for 10 min at RT. Samples were then incubated 20 min in 4% PFA/1xPBS to inactivate the Proteinase K. Samples were incubated in 1xPBS 6× 10 min to remove the PFA, then incubated in pre-hybridization buffer for 3 h at 65°C. Samples were subsequently incubated in the hybridization solution containing 200 ng/ml of each mRNA probe ≥14 h at 65°C. Samples then were washed with successive wash steps to remove unbound probe and prepare for antibody incubation: once 2xSSC/75% deionized formalin at 65°C, once 2xSSC/50% deionized formalin at 65°C, once 2xSSC/25% deionized formalin at 65°C. 2xSSC at 65°C, twice 0.2xSSC at 65°C, 6 times 1xPBST (1xPBS with Tween-20), once in blocking solution [2% bovine albumin (Sigma-Aldrich, A3294-100G), 2% Sheep Serum (Meilunbio, M134510)] at RT for 4 h. Samples were incubated with Anti-digoxigenin-AP Fab Fragment (Sigma-Aldrich, 11093274910, RRID: AB_514497) in blocking solution ≥14 h at 4°C. Samples were then washed 6 times with 1xPBST, subsequently incubated in [0.1 M Tris-HCl (pH 9.5), 0.1 M NaCl, 0.05 M MgCl2] 3 times for 30 min, and then in Nitro Blue Tetrazolium (Sigma-Aldrich, N6639-1G) and 5-bromo-4-chloro-3-indolyl phosphate (Sigma-Aldrich, 136149) in [0.1 M Tris-HCl (pH 9.5), 0.1 M NaCl, 0.05 M MgCl2] at RT ≥8 h. Samples images under VS120-S6-W (OLYMPUS). For signal areas, images of the shha in situs were transferred to Fiji ImageJ (RRID SCR_003070). The stained regions were traced using “free ROI” and then quantitated using the “measure” function under the “analysis” menu to calculate the number of pixels contained in the stained area. For signal intensity analysis, the signal region and adjacent unstained regions were measured by “free ROI” selection and the mean intensity pixel values were determined from the “measure” function under the analysis menu by subtracting the mean value of the unstained region from the mean value of the signal region for each fin bud. Download: PPT PowerPoint slide PNG larger image TIFF original image Table 1. Materials table. https://doi.org/10.1371/journal.pbio.3002565.t001 qRT-PCR For transgenic expression in fish embryos, the embryos were heat-shocked twice at 6-h intervals at 32 hpf and 42 hpf to maximize chronic transgene effects on developmental gene expression. Fin buds were isolated by using tip of a micropipette to gently remove the yolk of the 48 hpf embryos. The heads were removed using the small lancet iris knife (Zhongyuan Healthcare Equipment, Cat # hyl10.31372714). The main body of the embryo was removed by using tweezers to fix the body (neural tube facing up) and cutting the fin buds and residual yolk membrane from the body on both side with a small lancet iris knife. The fin buds and residual yolk membrane were collected in the TRIZOL (CWBio, 03917). Over 40 embryos (80 fin buds) were collected for each sample group. For adult caudal fin tissues, we heat shocked adult fish once for 12 min at 38.5°C 6 h before the tissue isolation. After anesthetizing adult fish with 0.4% Tricane, we surgically isolated the distal caudal fin tissues from the fin tip including the distalmost bone bifurcation. For qRT-PCR experiments from HEK293 cells, we isolated RNA from 90% confluent 6 cm well per replicate. The mRNA from fish or cultured HEK293 cells was isolated using Trizol (CWBio, 03917). Then, 1 μg mRNA was used for the reverse transcription to cDNA using 4× gDNA wiper Mix, 5× HiScript III qRT SuperMix (Vazyme, L/N 7E350C9). qRT-PCR was performed using 2× ChamQ Universal SYBR qPCR Master Mix (Vazyme, L/N TE342F9) with QuantStudio3 machine (Thermo Fisher). The cycle procedure was at 50.0°C for 2 min, 95°C for 10 min in the stage 1; 95°C for 15 s, 55.0°C for 20 s for 40 routine in the stage 2; 95°C for 15 s, 60°C for 1 min, 95°C for 15 s in the Melt Curve. Samples were standardized using the detected ef1a expression for fish mRNA isolation and GAPDH for human cell line mRNA isolations. Please see the “Table 1” in this section for the primer sequences of each gene. Protein extraction and western blot Zebrafish were anesthetized with 0.4% Tricaine and fin tips (fin tissues from the distalmost edge of each fin to the 2 distalmost bone segments of the bone rays) were surgically isolated and homogenized in RIPA buffer with protease inhibitors (Thermo/Pierce), then incubated at 4°C for 1 h under constant shaking. Lysates were centrifuge at 13,000r for 15 min at 4°C. Protein concentration was determined using the BCA method kit (MDbio). Proteins were denatured (incubated at 95°C for 10 min in loading buffer and then transferred to ice immediately prior to loading (50 μg/lane) in a 10% SDS-PAGE Acrylamide-Bis (29:1) gel). After electrophoresis, gel contents were transferred onto 0.2 μm PVDF membrane. After transfer, the membrane incubated in 5% skim milk/0.5% TBST for 1.5 h at room temperature and then washed 3× 5 min room temperature. Blots were then incubated with Rcan2 primary antibody (1:1,000) overnight at 4°C. Blots were then wash in 1xTBST 3× 5 min and then incubated in secondary antibody (1:1,000) in blocking buffer for 2 h at room temperature. Blots were then washed 3× 5 min. Protein bands were detected using ECL chemiluminescence detection reagent (Perkin Elmer) and luminescence was detected using an Amersham Imager 600 chemiluminescence imager (General Electric). Cell culture HEK293 cell cultures (from ATCC.org) were incubated at 37°C, 5% CO2, 95% humidity in incubators (Thermo Fisher, FORMA STERI-CYCLE i160) in DMEM medium (Gibco,1199506) with 10% FBS (Gibco,1009914) and 1% penicillin streptomycin (Gibco, 15140122). The identity of the cell lines was not authenticated, and mycoplasma was not detected. Cells were split to 50% density and transfected with Lipofectamin (Invitrogen, 11668–019) 12 h later. Expression for the transfected constructs was evaluated by expression of fluorescent marker. All cultures tested negative for mycoplasma, which we tested by collecting 1 ml of DMEM cultured over 24 h with HEK293 cells, and centrifuge at 12,000 rpm for 1 min. Then, do the PCR as following primers: MGSO-TGCACCATCTGTCACTCTGTTAACCTC, GPO-3-GGGAGCAAACAGGATTAGATACCCT. FRET-FLIM detection and analysis HEK293 cells were transfected with 1 μg of pcDNA-kcnk5b-GFP; pcDNA6-mT2-CUTie-YFP [40] or pcDNA-CFP-cGi500-YFP [41]. The transgenic fish [hsp70:KIRIN1] were heat shock at 37°C for 1 h for the sensor to induce peak transgene expression 6 h later. Embryos were incubated in 250 μm PTU (1-phenyl 2-thiourea, Sigma) to prevent pigmentation starting at 24 hpf. Ten minutes before imaging, embryos were anesthetized in 4 μm MESAB (or 200 nM blebbistatin for 30 min, where indicated) and then embedded in 1% agarose before imaging. For other imaging experiments, the time points were indicated in the text or figure legends. For FLIM measurements of the adult caudal fins, adult fish (approximately 6 months old) were anesthetized with 1% MESAB and then placed on wet plastic Petri dish and imaged for 5 min and then placed back into aquarium water to regain consciousness. Measurements were from 3 different locations in the mesenchyme of caudal fins: 1 at each of the 2 lobe tips and 1 in the distal midline between the 2 lobes. Each measurement is 1 data point. For the embryonic developmental time course of fluorescence lifetime imaging (FLIM-FRET) measurements, each embryo was imaged at 32 hpf, 35 hpf, 38 hpf, 42 hpf, 48 hpf, and 56 hpf. Fluorescence lifetime imaging measurements were made by photon counting the fluorescence emission of either CFP (KIRIN1, cAMP, cGMP) using a 2-photon-confocal Hyperscope (Scientifica, United Kingdom) and PMT-hybrid 40 MOD 5 photon detectors (Picoquant, Germany). We determined exposure times for each plane based on the number of photons detected individual pixels. In each fin bud, we generally measured 3 regions within the mesenchyme and ectoderm by selecting regions of interest at distal, proximal, and central locations within the buds. Each region contained several pixels of that ranged from 1 to 3 cells in size. For each measurement, we used thresholds of several hundred photons per pixel to overcome background photon emissions and prevent any influence in the FLIM measurements. The lifetime curves were analyzed from graphically selected regions or cells in the confocal plane. The lifetime curves were determined (curve fitting) from the decay of the counted photon emissions after single laser pulses (2-photon Chameleon Ultra II Ti Sapphire laser) using the algorithm for base computations of multi-exponential reconvolution with corrections for instrument response time (instrument response factor, IRF) and removal of background signals <100/pixel in the program SymPhoTime 64, version 2.4 (Picoquant, Germany, RRID: SCR_016263). The fitted curves represented 2 component (fluorophores FRET at <10 nm and fluorophores no FRET >10 nm) equations with χ2 values of 1 ± 0.19 for accuracy. To visualize differences in spatial distribution of each pixel lifetime value, individual lifetime values were distributed along a rainbow scale in which boundary lifetime values were assigned: blue lowest boundary and red highest boundary. FRET efficiency (on which our FLIM measurements are based) using our donor-only (CFP) and sensor-based donor + acceptor (CFP-YFP) K+-detection data in Fig 1A. The resulting average efficiency was 37.9%. We entered this calculated efficiency in a standardized equation [RDA = R0 x 6√(1-Efficiency)2/Efficiency)] to calculate the effective distance between the 2 fluorophores for FRET from our measurements, (4.59 Å) and then to compare this to the reference distance of maximal FRET of CFP and YFP (4.7 Å) [93]. RDA is calculated distance between donor and acceptor fluorophores. R0 is the distance required for 50% efficiency. The closeness our calculated and reference distances indicated meaningful efficiency of the K+-sensor in our in vivo measurements. KIRIN1 sensor As previously published, KIRIN1 sensor was generated by fusing CFP and YFP to the N- and C-termini of the K+-binding protein (Kbp) from E. coli [28]. The Kbp protein normally maintains an elongated conformation that keeps the donor and acceptor fluorophores apart, which prevents FRET between the fluorophores. When K+ binds to the bacterial OsmY (BON) domain in the N-terminus and the Lysin motif (LysM) in the C terminus, it brings the 2 fluorophores together to allow FRET [28]. This results in a reduction in the lifetime of the donor (CFP) fluorophore, which can be detected by FLIM. DiSBAC2(3) staining and measurements DiSBAC2(3) was diluted to a 10 μm concentration in 1× E3 medium that contained 250 μm concentration of PTU (1-phenyl 2-thiourea, Sigma, P7629). Embryos were put into PTU at 24 hpf to prevent the pigmentation of the embryos. Three hours before imaging, embryos were incubated in diluted DiSBAC2(3) at RT under gentle rocking. They were immobilized by adding blebbistatin to a final concentration of 200 nM for 30 min before imaging, embedded in 1% low-melting agarose gel, and then imaged under a 2-photon-confocal Hyperscope (Scientifica, UK) at 900 nm to maximize the detection of the fluorescence from the DiSBAC2(3) dye and minimize fluorescence of mCherry under 2-photon stimulation (www.fpbase.org). Photon counts were detected using PMT-hybrid 40 MOD 5 photon detectors (Picoquant, Germany) that detect photons (provide a numerical count) for each pixel. We used blebbistatin, because it inhibits muscle movement by inhibiting myosin activity and does not target any ion channel. We imaged each embryo for 3 min and measured 6 pixels in specific regions in each fin bud (distal mesenchyme, anterior mesenchyme, proximal mesenchyme, and posterior mesenchyme). The 6 pixels from each region were averaged and each average represents a data point in the graph; 32 hpf were chosen for imaging the effect of kcnk5b overexpression, because of the low levels of depolarization at this stage (Fig 5A–5C). CRISPR knockouts for rcan2 and kcnk5b The ChopChop online tool (http://chopchop.cbu.uib.no/) was used to design sgRNAs to limit predicted off-target sgRNA cutting. The fourth exon of rcan2 was targeted by the sgRNA sequence is 5′-AACCGACTGGAGGTGAGCCA-3′ ordered from IDT with Alt-R gRNA modification. The first exon in kcnk5b was targeted by the sgRNA sequence 5′-TGCGATATTCCAAATCCTCG-3′ and 5′-GGTCTTGGGCGAAATTATTG-3′ ordered by IDT with Alt-R gRNA modification, and 1 μl sgRNAs (400 ng/μl) and Cas9 (500 ng/μl) protein (Genscript, Cat. #Z03469) were co-injected into single-cell zebrafish embryos. After imaging for experiments, each embryo was numbered and raised separately to 5 dpf. Genomic DNA was isolated from each 5 dpf larva individually. After genomic DNA extraction by alkaline lysis buffer, the genomic region surrounding the target sites was PCR amplified using primers forward: 5′-CAACACACTCTCTGGCTTTCAG-3′ and reverse: 5′-TGAACTGCATAGTTGAGATGGG-3′ for rcan2 or forward: 5′-ATCACCAGAAACTTGGGAGTGT-3′ and reverse: 5′-GCTTTTGCGTGAGAACTACCAT-3′ for kcnk5b and set for sequencing (Genewiz.com.cn). In vivo mRNA overexpression experiments Mutated rcan2-P2A-mcherry, kcnk5b-GFP and kcnk5bS345A-GFP were cloned into pXT7 vector (from Lin Haifan and Bao Baolong lab) and either linearized with Xba1 (Anza Thermo Scientific #ER0682) or amplified by PCR to make templates. For rcan2, the original target sequence 5′-AACCGACTGGAGGTGAGCCA-3′ was mutated to 5′-ATCCAACCGGTGGGCTTCCG-3′. For kcnk5b, the original target-2 sequence 5′-TGCGATATTCCAAATCCTCGAGG-3′ was mutated to 5′-TGCAATCTTTCAGATACTAGAAG-3′, and the original target-14 sequence 5′-GGTCTTGGGCGAAATTATTGAGG-3′ was mutated to 5′-GGTATTAGGAGAGATCATCGAAG-3′ to impair interaction between the sgRNA targeting sequences and the rescue mRNAs without changing coded amino acids. The linear template was transcribed to mRNA using T7 MESSAGE MACHINE (Invitrogen, #AM1344). Concentrations injected into embryos: 1 μl Cas9 protein, 1 μl sgRNA, 1 μl rcan2-P2A-mCherry mRNA/kcnk5b-P2A-GFP (700 ng/1 μl), mCherry mRNA (375 ng/1 μl), 1 μl empty sgRNA (T7-sgRNA scaffold, 2,000 ng/μl) or kcnk5bS345-GFP/kcnk5bS345A-GFP (200 ng/μl). Mosaic analyses in fin buds Double mosaic embryos were created by injecting 120 μg/μl of hsp70:kcnk5b-mCherry and 400 ng/μl into Tg[hsp70:KIRIN1] separately into 1–4 cell AB embryos. Embryos only expressing kcnk5b-mCherry mosaically were injected with hsp70:kcnk5b-mCherry into the transgenic Tg[Cca.actb:GCaMP6s] line. Injected embryos and larva were raised in 1xE3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2, 0.33 mM MgSO4, 10%–5% Methylene Blue). The mosaic larvae were selected by screening for mCherry and CFP expression for kcnk5 and KIRIN1 transgenes or mCherry and GFP expression for kcnk5b and GCaMP6s transgenes in the fin buds 4 to 6 h after heat-shock induction. Small molecule treatments All inhibitors (except for Vinpocetine) were dissolved in DMSO: Retinoic acid (Abcone, R26255) 10 mM stock concentration, 2-APB (IP3R inhibitor, ENZO, ALX-400-045-M100) 75 mM stock concentration, STO-609 (CaMKK inhibitor, MedChemExpress, HY-19805) 0.5 mM stock concentration, Verapamil (T-/L-type Ca2+ channel inhibitor, MedChemExpress, HU-A0064) 100 mM stock concentration, KN-62 (CaMKII/IV inhibitor, MedChemExpress, HY-13290) 1 mM stock concentration, or in sterile water Mibefradil (T-/L-type Ca2+ channel inhibitor, MedChemExpress, HY-15553A) 2.5 mM stock concentration, Forskolin (adenylyl cyclase activator, MedChemExpress, HY-15371) 25 mM, SNAP (guanylyl cyclase activator, Tocris, 1561) 100 mM stock concentrations. HEK293 cells were incubated in DMEM cell culture medium (Gibco, 1199506) containing 10% FBS (Gibco, 1009914) and 1% penicillin streptomycin (Gibco, 15140122) at 37°C, 5% CO2. Ten hours after transfection, the drugs were added to the medium to their final concentrations (as indicated in the figure legends or the specific methods section). Cells were then trypsinized and RNA was isolated as indicated for qRT-PCR. For fish embryos, we added the small molecules mentioned above to E3 to the final concentrations (200 nM RA, 13 μm 2-APB, or 24 μm STO unless indicate otherwise) and incubated the embryos 28°C. We determined the treatment times and concentrations (provided in the figure legends or text) to be the most effective on the fin bud growth and have the least effect on embryonic development. Adult fish were treated with 100 μm RA for 6 h, since we determined this concentration showed reproducible effects (S5F Fig). Vinpocetine (MedChemExpress, HY-13295) was dissolved in ethanol to make a 10 mM stock, Dibucaine (MedChemExpress, HY-B0552) was dissolved in DMSO to make a 500 mM stock. For assessing the effects of these inhibitors on depolarization, we treated embryos with 10 μm Vinpocetine and 40 μm Dibucaine for 4 h before assessment, since these concentrations showed continued reproducible inhibitory effects. For fin bud growth experiments, 10 μm Vinpocetine and 40 μm Dibucaine treatment started at 36 hpf for 12 h to maximize the effects of the inhibitor on fin bud growth by 48 hpf. GCaMP6s measurements HEK293 cells were transfected with 1 μg of CMV-GCaMP6s, 1 μg of pcDNA-kcnk5b-MCherrry or pcDNA-mCherry or pcDNA-Kcnk5bMut. The pcDNA-Kcnk5bMut harbors 2 mutations: one changing the LEEP sequence in the first transmembrane region to VKKA and substituting S345 in the C-terminal tail with alanine. This generated a dead channel. HEK293 cells for Fig 6E were transfected and sorted by SORP ARIA Fusion (BD Biosciences) for mCherry. GCaMP6s transgenic fish Tg[Cca.actb:GCaMP6s] [43] were purchased from Zebrafish Stock Center Wuhan, China (CZ1282 doi.org/10.3390/ijms22115551). One-celled embryos from the established Transgenic fish Tg[Cca.actb:GCaMP6s] [43] were injected with Tg[hsp70-kcnk5b-mCherry] and were heat shocked at 37°C for 1 h at 42 hpf and imaged at 48 hpf. All the samples were imaged by Zeiss LSM 980 upright and analyzed by ZEN-Blue (Zeiss). To standardize intensity measurements and limit influence of differences in expression, we standardized all measurements to GCaMP6s cells lacking mCherry expression in the spinal cord (body) away from the fin bud. The results are provided as GFP bud/GFP body. Patch clamping Transfected HEK293T cells were seeded on glass coverslips (Fisher Brand) and incubated in cell culture medium at 37°C, 5% CO2, 95% relative humidity for 8 to 10 h. The seeded coverslips were transferred into Tyrode’s solution (138 mM NaCl, 4 mM KCl, 2 mM CaCl2, 1 mM MgCl2, 0.33 mM NaH2PO4, 10 mM Glucose, 10 mM HEPES). Cells were assessed in the ruptured-patch whole-cell configuration of the patch-clamp technique using and EPC9 or EPC10 amplifier (HEKA) with borosilicate glass pipettes (Sutter Instruments) with 3 to 6 MΩ resistance when filled with pipette solution (130 mM glutamic acid, 10 mM KCl, 4 mM MgCl2, 10 mM HEPES, 2 mM ATP, pH to 7.2). For detecting potassium current, after gigaseal formation, cells were voltage-clamped at −80 mV. Potassium conductance was elicited by test pulses from −100 mV to 70 mV (in 10 mV increments) of 600 ms duration at a cycle length of 10 s. The resulting tracings were converted into itx files by the ABF Software (ABF Software, RRID: SCR_019222) and then analyzed using Clampfit Software (Molecular Devices, RRID: SCR_011323). Currents were measured at the end of the test pulses. Quantification and statistical analyses To select for accurate two-component decay curve fittings of the FRET sensors, we used Chi-square analyses limiting fits 0.8>χ2<1.2 values, which was calculated within the SymPhoTime 64 software, version 2.4 (Picoquant, Germany, RRID: SCR_016263). Graphs were assembled and statistical analyses were done using Microsoft Excel (Microsoft.com) or Prism9 (Graphpad.com). All statistical analyses for graphs were done using the Students t test to assess the reproducibility of any differences between 2 groups within the data sets of each experiment. P values differences of >0.05 were indicated to be not significant “NS”; otherwise, the calculated P values were provided in the figure. Supporting information S1 Data. Meta data to Fig 1. https://doi.org/10.1371/journal.pbio.3002565.s001 (XLSX) S2 Data. Meta data to Fig 2. https://doi.org/10.1371/journal.pbio.3002565.s002 (XLSX) S3 Data. Meta data to Fig 3. https://doi.org/10.1371/journal.pbio.3002565.s003 (XLSX) S4 Data. Meta data to Fig 4. https://doi.org/10.1371/journal.pbio.3002565.s004 (XLSX) S5 Data. Meta data to Fig 5. https://doi.org/10.1371/journal.pbio.3002565.s005 (XLSX) S6 Data. Meta data to Fig 6. https://doi.org/10.1371/journal.pbio.3002565.s006 (XLSX) S7 Data. Meta data to Fig 7. https://doi.org/10.1371/journal.pbio.3002565.s007 (XLSX) S8 Data. Meta data to S1 Fig. https://doi.org/10.1371/journal.pbio.3002565.s008 (XLSX) S9 Data. Meta data to S2 Fig. https://doi.org/10.1371/journal.pbio.3002565.s009 (XLSX) S10 Data. Meta data to S3 Fig. https://doi.org/10.1371/journal.pbio.3002565.s010 (XLSX) S11 Data. Meta data to S4 Fig. https://doi.org/10.1371/journal.pbio.3002565.s011 (XLSX) S12 Data. Meta data to S5 Fig. https://doi.org/10.1371/journal.pbio.3002565.s012 (XLSX) S13 Data. Meta data to S8 Fig. https://doi.org/10.1371/journal.pbio.3002565.s013 (XLSX) S14 Data. Meta data to S9 Fig. https://doi.org/10.1371/journal.pbio.3002565.s014 (XLSX) S1 Fig. FLIM-FRET measurements with KIRIN1 sensor detects relative changes in intracellular K+. (Aa) Patch-clamp experiments show increased K+ currents from K+ leaking out of cells expressing CMV:kcnk5b-mCherry (blue) compared to cells expressing the control CMV:mCherry plasmid (red). (Ab) FLIM-FRET measurements of the KIRIN1 sensor expressed in the same cells that were patched in Aa detected decreases in intracellular K+ in cells expressing CMV:kcnk5b-mCherry (blue) compared to cells lacking the expression of the channel (CMV:mCherry, red). (B) FLIM-FRET measurements for intracellular K+ levels in HEK293 cells transfected with the CMV:KIRIN1 sensor and either CMV:mCherry or CMV:Kcnk5b-mCherry. Cells were either treated with DMSO or FK506. (C) FLIM-FRET measurements for intracellular K+ levels in HEK293 Cells transfected with the CMV:KIRIN1 sensor and either CMV:Kcnk5bS345A-mCherry, CMV:Kcnk5bWT-mCherry, or CMV:Kcnk5bS345E-mCherry. The measurements were converted to fold difference by dividing their lifetime measurements with the lifetime measurements of the control cells expressing the KIRIN1 sensor and mCherry. Each experiment was repeated at least 3 times (N = 3). For cell patch clamping and FLIM measurements, each data point represents 1 cell (A, B, C). For fish embryo FLIM imaging, we measured 2 or 3 locations in each tissue of 1 fin bud per embryo. P values represent statistical analysis by Student’s two-tailed t test. Numerical data used in this figure are included in S8 Data. https://doi.org/10.1371/journal.pbio.3002565.s015 (TIF) S2 Fig. FLIM-FRET measurements not affected by changes in expression levels or sedation method. (A) Heat-shock method for inducing expression of the transgenic K+ sensor at the indicated time points for subsequent FLIM-FRET measurements. (B) Fluorescent image of non-transgenic sibling 6 h after heat shock at 48 hpf. (C) Fluorescent image of transgenic Tg[hsp70:KIRIN1] 6 h after heat shock at 48 hpf. (D) Representative confocal plane through a developing pectoral fin bud of a Tg[hsp70:KIRIN1] transgenic fish shows expression of the K+-sensor transgene in cells except for the nuclei. (E) Illustration of different decay (lifetime) curves and that despite differences in initial excitation levels of the donor fluorophore (arrows), the decay rates are similar (orange decay curves) unless energy is transferred from the donor to the acceptor fluorophore by FRET in the presence of K+, which will reduce the lifetime (blue decay curve). These specific differences in lifetime can be represented by specific colors along a rainbow scale to produce an image that relates the lifetime value of each pixel in the confocal plane to provide a spatial representation of the distribution of relative K+ levels in the fin bud. (F) 3D images of density map for illumination intensity (a, b) of pectoral fin buds at 32 hpf. Comparison between 2D plane of density map for illumination at a region of high intensity (c) and the lifetime assessment of the same region in the ectoderm (d). Comparison between 2D plane of density map for illumination at a region of low intensity (e) and the lifetime assessment of the same region in the ectoderm (f). Comparison between 2D plane of density map for illumination at a region of high intensity (g) and the lifetime assessment of the same region in the mesenchyme (h). Comparison between 2D plane of density map for illumination at a region of low intensity (i) and the lifetime assessment of the same region in the ectoderm (j). (k) Graph of lifetime values of each region measured (d, f, h, j) shows that high and low differences in intensity (y-axis) do not significantly alter the lifetime values (x-axis) of the ectoderm and the mesenchyme. (G) 3D images of density map for illumination intensity (a, b) of pectoral fin buds at 48 hpf. Comparison between 2D plane of density map for illumination at a region of high intensity (c) and the lifetime assessment of the same region in the ectoderm (d). Comparison between 2D plane of density map for illumination at a region of low intensity (e) and the lifetime assessment of the same region in the ectoderm (f). Comparison between 2D plane of density map for illumination at a region of high intensity (g) and the lifetime assessment of the same region in the mesenchyme (h). Comparison between 2D plane of density map for illumination at a region of low intensity (i) and the lifetime assessment of the same region in the ectoderm (j). (k) Graph of lifetime values of each region measured (d, f, h, j) shows that high and low differences in intensity (y-axis) do not significantly alter the lifetime values (x-axis) of the ectoderm and the mesenchyme. (H) 3D images of density map for illumination intensity (a, b) of pectoral fin buds at 56 hpf. Comparison between 2D plane of density map for illumination at a region of high intensity (c) and the lifetime assessment of the same region in the ectoderm (d). Comparison between 2D plane of density map for illumination at a region of low intensity (e) and the lifetime assessment of the same region in the ectoderm (f). Comparison between 2D plane of density map for illumination at a region of high intensity (g) and the lifetime assessment of the same region in the mesenchyme (h). Comparison between 2D plane of density map for illumination at a region of low intensity (i) and the lifetime assessment of the same region in the ectoderm (j). (k) Graph of lifetime values of each region measured (d, f, h, j) shows that high and low differences in intensity (y-axis) do not significantly alter the lifetime values (x-axis) of the ectoderm and the mesenchyme. (I) Representative FLIM-FRET image of the distribution of the lifetimes of a fin bud from a 32 hpf embryo immobilized with 200 nM Blebbistatin. (J) Representative FLIM-FRET image of the distribution of the lifetimes of a fin bud from a 32 hpf embryo immobilized with 4 μm MESAB (Tricane). (K) Representative FLIM-FRET image of the distribution of the lifetimes of a fin bud from a 48 hpf embryo immobilized with 200 nM Blebbistatin. (L) Representative FLIM-FRET image of the distribution of the lifetimes of a fin bud from a 48 hpf embryo immobilized with 4 μm MESAB. (M) Representative FLIM-FRET image of the distribution of the lifetimes of a fin bud from a 56 hpf embryo immobilized with 200 nM Blebbistatin. (N) Representative FLIM-FRET image of the distribution of the lifetimes of a fin bud from a 56 hpf embryo immobilized with 4 μm MESAB. (O) Graphed FLIM-FRET measurements of several fin buds of embryos immobilized either with MESAB or Blebbistatin from the indicated time points shows no significant differences (P > 0.05) between the 2 treatments. For the FLIM measurements of Blebbistatin-treated and MESAB-treated embryos, experiments were repeated at least 3 times (N = 3), and each repeat contained 3 or more embryos. We measured 2 or 3 locations in each tissue of 1 fin bud per embryo. Each measured value is represented as a data point (O). P values represent statistical analysis by Student’s two-tailed t test. P values ≥0.05 are designated as “not significant” (NS). Scale bars equal 1 mm (B, C), 20 μm (D), 30 um (Ha-i), or as indicated in the panels. Numerical data used in this figure are included in S9 Data. https://doi.org/10.1371/journal.pbio.3002565.s016 (TIF) S3 Fig. Confocal planes of FLIM-FRET images in fin buds at different developmental time points. (A) Confocal planes in fin bud from a 32 hpf embryo. The distance between the first plane and last plane is 15.2 μm. (B) Confocal planes in a fin bud from a 48 hpf embryo. The distance between the first plane and last plane is 18.36 μm. (C) Confocal planes in a fin bud from a 56 hpf embryo. The distance between the first plane and last plane is 10.3 μm. Numbers in lower right of each panel indicate the order of the indicated confocal plane through the Z-stack. (D) FLIM measurements in fin buds of 56 hpf embryos of indicated cell categories from the transgenic KIRIN1 fish line Tg[hsp70:KIRIN1] mosaically expressing mCherry or kcnk5b-mCherry. “Adj” indicates cells adjacent to mCherry-positive (mCherry+) or kcnk5b-mCherry-positive (kcnk5b+) cells. Each experiment was repeated at least 3 times (N = 3). For fish embryo FLIM imaging, we measured 2 or 3 locations in each tissue of 1 fin bud per embryo. Each measured value is represented as a data point (D). Numerical data used in this figure are included in S10 Data. https://doi.org/10.1371/journal.pbio.3002565.s017 (TIF) S4 Fig. Overexpression of K+-leak channels and their effects on pectoral fin bud growth. (A) qRT-PCR for kcnk5b-GFP expression after single heat shock pulse of non-transgenic siblings and Tg[hsp70:kcnk5b-GFP] siblings. Each RNA sample was isolated from fin buds of 40+ embryos at 48 hpf. (B) Expression of kcnk5b-GFP in the body (a) and in the fin bud (b) of a representative Tg[hsp70:kcnk5b-GFP] 56 hpf embryo 6 h after heat shock. (C) Expression of kcnk10a-GFP in the body (a) and in the fin bud (b) of a representative Tg[hsp70:kcnk10a-GFP] 56 hpf embryo 6 h after heat shock. (D) Brightfield image of thorax region of a post-heat-shocked 48 hpf non-transgenic embryo. The area of the fin bud (highlighted by a black-dotted line). The otic vesicle was used as a size standard (highlighted by a red-dotted line). (E) Brightfield image of thorax region of a post-heat-shocked 48 hpf hsp70:kcnk5b-GFP transgenic embryo with the fin bud (black-dotted line) and otic vesicle (red-dotted line) used as a size standard. (F) Measurements of pectoral fin bud areas of heat-shocked groups of non-transgenic (Non-Tg) and Tg[hsp70:GFP] (GFP-Tg) as controls and Tg[hsp70:kcnk5b-GFP] transgenic fish line as well as the Tg[hsp70:kcnk10a-GFP] transgenic fish line. Each measured bud area was standardized to the otic vesicle area in the same embryo and each measurement is represented as a ratio of bud-area–to–otic-vesicle area. Each experiment was repeated at least 3 times (N = 3) and each repeat contained 3 for more fish. Each data point represents 1 fin bud measurement per embryo. P values represent statistical analysis by Student’s two-tailed t test. P values ≥0.05 are designated as “not significant” (NS). Numerical data used in this figure are included in S11 Data. https://doi.org/10.1371/journal.pbio.3002565.s018 (TIF) S5 Fig. Effects of nuclear hormone treatments on kcnk5b, rcan2, and control genes in pectoral fin buds and adult fins and sufficiency/necessity experiments for rcan2. (A) qRT-PCR measurements of the transcription of kcnk5b in the developing pectoral fin bud with and without retinoic acid stimulation. (B) qRT-PCR measurements of cyp26a expression in pectoral fin buds with or without 200 nM retinoic acid treatment, a gene known to be induced by retinoic acid treatment. (C) FLIM measurements from Ectoderm (Ecto) and Mesenchyme (Mesen) cells of buds at 32 hpf treated either with DMSO or 200 nM thyroid hormone for 6 h. (D) In situ staining intensity measurements from the in situs of rcan2 expression in the embryonic fin buds after DSMO or RA treatment. (E) qRT-PCR for expression of rcan2 and dio3b in adult caudal fin after treatment with DMSO or 500 nM thyroid hormone for 24 h. dio3b is a gene known to be up-regulated by thyroid hormone stimulation in the adult caudal fin. (F) RT-PCR for expression of cyp26a treated either with DMSO or 100 nM RA for 6 h in adult caudal fin. (G) FLIM measurements of intracellular K+ levels in adult caudal fin cells after 6 h treated either with DMSO or 100 nM RA. (H) qRT-PCR measurements using 2 different primer sets for rcan2 in adult caudal fins of indicated treatment groups. (I) Whole-mount in situ hybridization of a caudal fin 3 day post amputation shows rcan2 expression in the distal blastemal but absent from the proximal blastema (a). Arrowheads indicate amputation plane. Cryo cross sections through the 3 day post amputation, distal tip of a regenerating adult caudal fin in the ray (b) and interray tissues (c) after in situ hybridization for rcan2. The blue color indicates rcan2 expression. (J) Representative western blot for Rcan2 and beta-actin proteins from lysates of regenerating adult caudal fins at the indicated days post amputation (dpa). (K) Measurements of Rcan2 protein expression after standardization to beta-actin expression at the indicated time points. (L) Representative brightfield image of non-transgenic sibling 6 h after heat-shock stimulation. (M) Representative mCherry fluorescence image of non-transgenic sibling 6 h after heat-shock stimulation. (N) Representative brightfield image of transgenic Tg[hsp70:rcan2-mCherry] sibling 6 h after heat-shock stimulation. (O) Representative mCherry fluorescence image of transgenic Tg[hsp70:rcan2-mCherry] sibling 6 h after heat-shock stimulation. (P–R) Representative amputated caudal fin of heat-shocked AB non-transgenic (non-Tg) fish at time 0 days post amputation (dpa) (P) and regenerated 56 dpa (Q) together with its representative fluorescence image for mCherry (R). (S–U) Representative caudal fins of Tg[hsp70:rcan2-mCherry] transgenic fish (rcan2-Tg) at time 0 dpa (S) and regenerated 56 dpa (T) together with its representative fluorescence image for mCherry (U). (V) Fin-to-body length ratios of non-transgenic and Tg[hsp70:rcan2-mCherry] fish at the indicated time points during one-heat-shock per day regimen. (W) Wild-type 48 h post fertilization (hpf) larva (a). Sequence of rcan2 gene in exon 4. (b) The sgRNA target site is indicated by the orange box. Of the total number of injected embryos (45) assessed, only 9% showed wild-type sequence. (X) 48 hpf larva of large deletion group (a). The majority of targeted embryos (64%) harbored a similar large deletion: indicated by the gap in sequence. The downstream sequence also showed scrambled nucleotide order (b). (Y) 48 hpf larva of small deletion group (a). Over a quarter of the targeted embryos (27%) harbored similar small deletions as well as scrambled downstream sequence (b). (Z) Representative mCherry fluorescent image of CRISPR rescue expression of rcan2*-mCherry mRNA (a) and CRISPR-targeted rcan2 allele from an rcan2* mRNA-mCherry-expressing embryo (b). rcan2*-mCherry mRNA has mutated wobble-position nucleotides of codons to impair interaction with the sgRNA to continue sgRNA-mediated disruption of the rcan2 alleles while maintaining the integrity of the transgenic Rcan2 protein. (AA) Representative images of control (a) and CRISPR-targeted rcan2 KO (b) juvenile fish 76 days post fertilization. (BB) Caudal fin-to-body length ratios between control AB and rcan2 KO juvenile fish. Scale bars equal 20 μm (I), 100 μm (L–NO, W–Z), 1 mm (P–U) and 500 μm (AA). For qRT-PCR experiments from fin buds, each experiment was repeated at least 3 time (N = 3) and each repeat contained duplicate or triplicate samples. Each sample contained 80+ fin bud isolations (A, B). For the FLIM measurements of embryos, we made 3 measurements per tissue in each fin bud. At least 3 embryos were measured per group per experiment, and each experiment was repeated 3 times. Each data point represents a measurement (C). For FLIM measurements from adult fins, we measured 3 locations in the mesenchyme of each unamputated adult caudal fin: one at each of the 2 lobe tips, and one in the midline between the 2 lobes at the midline of the fin. Each data point represents 1 measurement (G). For qRT-PCR and western blots experiments with adult caudal fins, each point represents a separate sample group from an isolation of the distal tips with 2 segments from 30+ fish (H, I). For fin bud and adult fin growth measurements, each experiment was repeated 3 times. The adult fin measurements in S5V contained 10+ fish per group each time. The measurements in S5BB, each point represents 1 fish. P values represent statistical analysis by Student’s two-tailed t test. P values ≥0.05 are designated as “not significant” (NS). Numerical data used in this figure are included in S12 Data. https://doi.org/10.1371/journal.pbio.3002565.s019 (TIF) S6 Fig. CRISPR kcnk5b KO strategy, targeting efficiency of endogenous locus and Kcnk5b mutants. (A) Wild-type 48 hours post fertilization (hpf) larva (a), and sequence of kcnk5b gene in exon 1 (b). Of 100 embryos targeted, 12% do not show gene defects. (B) 48 hpf embryo (a) harboring a large deletion in the kcnk5b exon 1 (b), and 44% harbor large deletions. (C) 48 hpf embryo (a) harboring small deletions and sequence changes (b), and 44% harbor such small deletions. The sgRNA target sites are indicated by the gray box labeled “target.” (D, E) Representative embryo with kcnk5b CRISPR knockout (kcnk5b KO) brightfield (D) and lack of green fluorescence (E) of CRISPR targeted embryos. (F, G) Representative embryo kcnk5b KO brightfield (F) and GFP from control GFP mRNA. (H, I) Representative embryo kcnk5b KO rescued with expression of kcnk5b*-GFP mRNA, brightfield (H) and green fluorescence (I). kcnk5b*-GFP mRNA has mutated wobble-position nucleotides of codons to impair interaction with the sgRNA to continue sgRNA-mediated disruption of the kcnk5b alleles while maintaining the integrity of the transgenic Kcnk5b protein. (J) Representative disrupted kcnk5b allele sequence after CRISPR-targeting and kcnk5b*-GFP-expressing embryo. (K, L) Wild-type, uninjected embryo for GFP (K) and absence of mCherry (L) fluorescence. (M, N) Representative transgenic embryo for kcnk5bS345-GFP (M) and rcan2-mCherry (N) mRNAs. (O, P) Representative transgenic embryo for kcnk5bS345A-GFP mutant (O) and rcan2-mCherry (P) mRNAs. Scale bars equal 100 μm (A–C, D–I, K–P). https://doi.org/10.1371/journal.pbio.3002565.s020 (TIF) S7 Fig. Development of embryos incubated with Vinpocetine, Dibucane, and DiSBAC2(3). (A) Representative image of a 48 hpf AB non-transgenic fish after 1 heat shock at 32 hpf and treated only with the solvents ethanol (EtOH) and DSMO for 12 h starting at 36 hpf. (B) Representative image of 48 hpf AB non-transgenic fish after 1 heat shock at 32 hpf and treated with 10 μm Vinpocetine (Vin) and 40 μm Dibuciane (Dib) for 12 h starting at 36 hpf. (C) Representative image of a 48 hpf Tg[hsp70:kcnk5b-GFP] fish heat-shocked once at 32 hpf to induce expression of kcnk5b-GFP and treated for 12 h with the solvents EtOH and DMSO at 36 hpf. (D) Representative image of a 48 hpf Tg[hsp70:kcnk5b-GFP] fish heat-shocked once at 32 hpf to induce expression of kcnk5b-GFP and treated with 10 μm Vinpocetine (Vin) and 40 μm Dibuciane (Dib) for 12 h starting at 36 hpf. (E–J) Embryos incubated in 10 μm DiSBAC2(3) dye for 3 h show dye penetration in the embryos. mCherry fluorescence of embryo treated with EtOH and DMSO before DiSBAC2(3) incubation (E). Brighfield image of embryo treated with EtOH and DMSO after DiSBAC2(3) incubation (F). Fluorescence of DiSBAC2(3) in embryo treated with 10 μm Vin and 40 μm Dib before DiSBAC incubation (I). mCherry fluorescence of embryo treated with Vin and Dib before DiSBAC2(3) incubation (H). Brighfield image of embryo treated with Vin and Dib after DiSBAC2(3) incubation (I). Fluorescence of DiSBAC2(3) in embryo treated with Vin and Dib after DiSBAC2(3) incubation (J). https://doi.org/10.1371/journal.pbio.3002565.s021 (TIF) S8 Fig. Effects of drug treatments on culture cells and embryos. (A) FLIM-FRET measurements of intracellular cAMP levels. Forskolin used as positive control for cAMP production. (B) FLIM-FRET measurements of intracellular cGMP levels. SNAP used as positive control for cGMP production. (C) Patch-clamp experiment measuring the K+ leak from HEK293T cells transfected either with Kcnk5b-GFP (blue) or a mutated Kcnk5bMut-GFP (red) that displayed almost no channel activity. (D) Diagram for T- and L-type Ca2+ channel inhibition by verapamil or mibefradil. (E) qRT-PCR of SHH expression in HEK293 cells after treatment with verapamil at the indicated concentrations. (F) qRT-PCR of SHH expression in HEK293 cells after treatment with Mibefradil at the indicated concentrations. (G) Brightfield images of HEK293 cells transfected with CMV-mCherry and treated with DMSO (a) or transfected with CMV-Kcnk5b-mCherry and treated with DMSO (b) or with verapamil, a L-/T-channel inhibitor, at 50 μm (c), 100 μm (d), 140 μm (e), 200 μm (f). (H) Fluorescence of HEK293 cells transfected with CMV-mCherry and treated with DMSO (a), or transfected with CMV-Kcnk5b-mCherry and treated with DMSO (b) or with verapamil at 50 μm (c), 100 μm (d), 140 μm (e), 200 μm (f). (I) Brightfield images of HEK293 cells transfected with CMV-mCherry and treated with DMSO (a) or transfected with CMV-Kcnk5-mCherry and treated with DMSO (b) or with Mibefradil, a L-/T-channel inhibitor, at 3.5 μm (c), 5 μm (d). (J) Fluorescent images of HEK293 transfected with CMV-mCherry and treated with DMSO (a) or transfected with CMV-Kcnk5b-mCherry and treated with DMSO (b) or with Mibefradil at 3.5 μm (c) 5 μm (d). (K) Brightfield images of HEK293 cells transfected with CMV-GFP and treated with DMSO (a) or transfected with CMV-Kcnk5b-GFP and treated with DMSO (b) or with the IP3 receptor inhibitor 2-APB at 30 μm (c), 75 μm (d), 105 μm (e), 150 μm (f). (L) Fluorescence images of HEK293 cells transfected with CMV-GFP and treated with DMSO (a) or transfected with CMV-Kcnk5b-GFP and treated with DMSO (b) or the IP3 receptor inhibitor 2-APB at at 30 μm (c), 75 μm (d), 105 μm (e), 150 μm (f). Each experiment was repeated at least 3 times (N = 3) and each repeat contained duplicate or triplicate samples. P values represent statistical analysis by Student’s two-tailed t test. P values ≥0.05 are designated as “not significant” (NS). Scale bars equal 20 μm (B–E, M, O) and 100 μm (F–K). Numerical data used in this figure are included in S13 Data. https://doi.org/10.1371/journal.pbio.3002565.s022 (TIF) S9 Fig. Effects of drug treatments on culture cells and embryos. (A) qRT-PCR for SHH in HEK293 cells transfected either with GFP or Kcnk5b-GFP and treated with DMSO or the CaMKII,IV inhibitor KN-62 at the indicated concentrations. (B) Brightfield images of HEK293 cells transfected with CMV-mCherry and treated with DMSO (a) or transfected with CMV-Kcnk5b-mCherry and treated with DMSO (b) or treated with KN-62 an inhibitor for CaMKII and CaMKIV at 2 μm (c) and 3 μm (d). (C) Fluorescence images of HEK293 cells transfected with CMV-mCherry and treated with DMSO (a) or transfected with CMV-Kcnk5b-mCherry and treated with DMSO (b) or treated with KN-62 an inhibitor for CaMKII and CaMKIV at 2 μm (c) and 3 μm (d). (D) Brightfield images of cells treated transfected with CMV-GFP and treated with DSMO (a) or transfected with CMV-Kcnk5b-GFP and treated with DSMO (b) or STO-609 an inhibitor of CaMKK at 0.2 μm (c), 0.5 μm (d), 1 μm (e), 1.5 μm (f). (E) Fluorescent images of cells treated transfected with CMV-GFP and treated with DSMO (a) or transfected with CMV-Kcnk5b-GFP and treated with DSMO (b) or STO-609 an inhibitor of CaMKK at 0.2 μm (c), 0.5 μm (d), 1 μm (e), 1.5 μm (f). (F) Representative 48 hpf kcnk5b transgenic Tg[hsp70:kcnk5b-GFP] embryo 12 h post heat shock and 6 h treatment in solvent concentration of DMSO. (G) Enlarge view of kcnk5b-Tg embryo (F) shows some heart edema associated with heat-shock-induced expression of kcnk5b. (H) Representative 48 hpf kcnk5b transgenic embryo 12 h post heat shock and 4 h treatment in the IP3R inhibitor 13 μm 2-APB. (I) Enlarge view of kcnk5b transgenic embryo in (H). (J) Representative 48 hpf kcnk5b transgenic embryo 12 h post heat shock and 6 h treatment in 24 μm STO. (K) Enlarge view of kcnk5b transgenic embryo in (J). (L, M) Cells transfected with CMV-camkk1b-mCherry in representative brightfield (L) and fluorescence (M) images. Each experiment was repeated at least 3 times (N ≥ 3) and each repeat contained duplicate or triplicate samples. P values represent statistical analysis by Student’s two-tailed t test. P values ≥0.05 are designated as “not significant” (NS). Scale bars equal 20 μm (D–I, K, M, N, P, Q, M, O) and 100 μm (R–W). Numerical data used in this figure are included in S14 Data. https://doi.org/10.1371/journal.pbio.3002565.s023 (TIF) Acknowledgments Template plasmid for Cas9 mRNA synthesis: pxT7-hcas9 were gifts from HF Lin and Bao Baolong’s Labs. pMD 19-T vector for sgRNA cloning was a gift from HF Lin. We also thank the Molecular Imaging Core Facility (MICF) and the Molecular and Cell Biology Core Facility (MCBCF) at the School of Life Science and Technology, ShanghaiTech University for expertise and technical support.

Kamat, Aditya;Tran, Ngat T.;Sharda, Mohak;Sontakke, Neha;Le, Tung B. K.;Badrinarayanan, Anjana

doi: 10.1371/journal.pbio.3002540pmid: 38466718

Introduction Information regarding biological processes is primarily encoded in the genomes of living organisms. Functional modalities are further enhanced via modulation of the epigenetic states of DNA [1–4]. For example, methylation of DNA occurs in organisms across all domains of life. In bacteria, physiological DNA methylations at N5-meC, N4-meC, and N6-meA nucleotide positions play diverse functions ranging from defensive roles involving immunity against invasive genetic elements to regulatory roles in cell cycle and transcriptional control [2,5]. In contrast to these physiological modifications, certain types of methylations can be aberrant. DNA methylations on O6-meG and O4-meT positions act as a potent source of mutagenesis, while N3-meA, N1-meA, and N3-meC block DNA replication and transcription [6–9]. Bacteria cope with this threat of aberrant methylations via eliciting 2 DNA damage responses (DDRs). The first response is the well-characterized and ubiquitously conserved SOS response that consists of 2 regulatory units, RecA and LexA [10]. The transcriptional repressor LexA occupies lexA boxes present in the promoter regions of SOS genes and inhibits their expression. Upon DNA damage, an activated RecA nucleoprotein filament triggers the autocleavage of LexA leading to de-repression of the SOS response [11–13]. Three distinct features are hallmarks of this response: (a) It is induced under all forms of DNA damage and is not specific to methylation damage alone [14]; (b) the response has leaky expression even in the absence of damage [15,16]; and (c) the response can result in mutagenesis due to expression of translesion synthesis polymerases [17]. Thus, SOS response induction is a trade-off between its essentiality and its cost. The SOS response is the primary response of bacteria to DNA damage. However, recent studies indicate that the SOS response is often complemented by SOS-independent DNA damage response pathways [18]. The PafBC response was discovered in Mycobacterium tuberculosis and is essential for inducing a majority of the genes under mitomycin C exposure [19,20]. Induction of this response relies on the PafBC heterodimer which is suggested to exhibit an ssDNA-dependent activation similar to the SOS response [21]. The DriD response of Caulobacter also exhibits a similar mode of activation [22]. The inhibition of cell division by didA (belonging to the DriD regulon) under DNA damage is well characterized [23]. However, recent studies also suggest that DriD may play a role beyond DDR pathway activation [24]. Both DriD and PafBC belong to the WYL-dependent family of transcription factors that is widely conserved across bacteria [25]. These pathways seem to be induced prominently by DNA double strand breaks but their role in the repair of methylation DNA damage remains to be uncovered [23,26]. A temporally delayed DDR subsequent to the SOS response, known as the adaptive response to methylation damage (or the Ada response) has also been reported in Escherichia coli [27–29]. This response exhibits key distinguishing features compared to the SOS response: (a) It is activated independent of the SOS response and is specific to methylation damage only [30]; (b) the response is adaptive; i.e., exposure to a sublethal dose of methylation damage encodes for memory that allows cells to adapt and survive formerly lethal levels of DNA methylation damage [27,30]; (c) the response exhibits bi-stability [31], a subpopulation of cells do not induce the response even under continuous exposure to methylation damage; and (d) repair via this response is non-mutagenic [27,32]. The master regulator of this DDR is the methyltransferase EcAda, comprising of an AdaA domain in its N-terminus (which associates with “A” and “B” DNA boxes located upstream of EcAda-regulated promoters) and a C-terminal methyltransferase domain [28]. EcAda expression is tightly regulated, with only 0–2 molecules of Ada present in cells in the absence of damage [31]. Posttranslational modification (PTM) via methylation of a conserved cysteine in N-Ada is crucial for activating EcAda as a transcription factor [33–35]. Once activated, EcAda drives its own expression as well as expression of genes involved in direct repair of methylation lesions. How prevalent is the risk of methylation damage, which could have resulted in the evolution of a methylation-specific DDR? Instances of encountering methylation DNA damage are not infrequent, and bacteria often face methylation stress. Intrinsic factors, such as insidious metabolic byproducts (e.g., lipid peroxidation) pose a prominent risk for DNA methylation damage [9,28]. Environmental factors including halocarbons and methylating agents produced by bacteria as instruments of inter-microbial warfare also contribute to this damage [36,37]. Additionally, infected mammalian cells subject invading bacteria to methylation stress [38]. More recently, studies have reported instances where physiological DNA methyltransferases, which are otherwise innocuous, erroneously inflict methylation damage on genomic DNA [39]. Thus, given the prevalence of methylation stress, it is likely that many bacteria have evolved unique systems such as the Ada response to protect their genomes. Intriguingly, despite the simultaneous discovery of the adaptive response to methylation damage alongside the SOS response [27,40], our knowledge of this pathway is presently restricted only to the E. coli paradigm. Thus, the significance and conservation of a methylation-based PTM in the regulation of a damage-specific bacterial DDR remains unknown. Indeed, a limited number of computational studies suggest only sporadic conservation of EcAda across the bacterial kingdom [41–43], and many bacteria, such as Caulobacter crescentus, are reported to lack an inducible adaptive response altogether [44,45]. In this work, we discover a methylation-specific DDR in Caulobacter crescentus, which showcases key features of an adaptive response. This pathway is regulated by a conserved but uncharacterized transcription factor ccna_03845 (“Cada2” henceforth). Detailed molecular characterization reveals a novel sequence-specific DNA-binding domain in this protein, as well as a methylation-based PTM required for activation of Cada2 as a transcription factor in a manner that is distinct from its E. coli counterpart. Despite the contrasting mechanistic features of EcAda and Cada2, we observe remarkable similarities in the defining characteristics of the downstream response. Phylogenetic distribution of adaptive response regulatory proteins further reveals their widespread prevalence across the bacterial kingdom, with ubiquitous conservation of key residues required for methylation-dependent activation (either in an EcAda-like or Cada2-like form). Collectively, our work highlights the importance of a transcriptional switch mediated by methylation PTM in activating an essential bacterial response to methylation DNA damage. Results A methylation-specific DNA damage response in Caulobacter We undertook a comprehensive transcriptomic approach to identify whether Caulobacter induces an SOS-independent transcriptional response specific to methylation damage. For this, we treated Caulobacter cells with agents that predominantly induce 1 specific form of DNA damage (methyl methane sulphonate (MMS), a causative agent of methylation DNA damage [46], mitomycin-C (MMC) that induces intra-strand crosslinks and mono-adducts [47], and norfloxacin that induces double-strand breaks [48]). We collected samples at 0, 20, and 40 min post damage exposure for RNA sequencing (Figs 1A and S1A). Our analysis revealed several genes that were induced in all conditions (“universal”) and a set of genes that were specifically induced under MMS treatment, but in none of the other damaging conditions (methylation-specific) (S1A Fig). Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 1. A methylation-specific DNA damage response in Caulobacter. (A) [Left] Schematic summarizing the RNA-sequencing experiment is provided. Wild-type and ΔrecA (SOS-deficient) cells were exposed to 1.5 mM MMS for 40 min. [Centre] Volcano plots representing differentially expressed genes in comparison to cells exposed to no damage for wild-type (top) and ΔrecA (bottom) cells, respectively. Genes up-regulated in wild-type cells (log2FC > 2 and–log10(p-value) > 2) are highlighted in blue, while genes highlighted in red represent genes up-regulated in wild-type but down-regulated in ΔrecA cells. [Right] Heat maps represent log2FC values for individual genes induced in wild-type and ΔrecA cells under MMS exposure. (B) Wild-type Pccna_00746-yfp reporter (schematic inset) was exposed to 1.5 mM MMS or 0.5 μg/ml MMC, respectively for 2 h. Representative cells are shown on the left (scale bar: 2 μm, cell boundaries are marked by white dotted outline, here and in all other images). Violin plots show fluorescence intensity distribution normalized to cell area from single cells (n = 300, from 3 biological replicates). The underlying data are available in S1 Data. (C) [Top] Schematic of the experimental protocol comparing induction kinetics of the Caulobacter SOS and methylation-specific response via time lapse microscopy. [Bottom] Fluorescence intensity normalized to cell area for Pccna_00746-yfp and PsidA-yfp cells over 3 h of exposure to 1.5 mM MMS. The underlying data are available in S1 Data. https://doi.org/10.1371/journal.pbio.3002540.g001 We next asked whether the expression of these genes was independent of the SOS response. We subjected recA deleted cells to MMS treatment and carried out RNA sequencing in this background. Comparing wild type to recA deletion cells showed that the methylation damage-specific genes were induced in a recA-independent manner (SOS-independent) (Fig 1A). In contrast, genes that were induced under all DNA damaging conditions in wild type background remain uninduced in recA deletion cells (SOS-dependent) (Figs 1A and S1A). We identified ccna_00745 to be the highest induced under methylation damage in the transcriptome analysis (Fig 1A). This gene is co-operonic with a second gene ccna_00746, that is also induced in a methylation DNA damage-specific manner (S1A Fig). ccna_00745 is Bioinformatically predicted to be a 2-oxoglutarate, Fe2+-dependent dioxygenase, similar to E. coli alkB, while ccna_00746 is predicted to possess an Ada-like methyltransferase domain (PF01035) [9]. To further corroborate the RNA-seq observations, we constructed a fluorescence-based readout for promoter activity of these genes (Pccna_00746-yfp). No significant expression of yfp was detected in the absence of damage (Fig 1B). We observed YFP expression from this construct only in cells treated with MMS, and not other DNA damaging agents MMC, norfloxacin and hydroxyurea (HU) (Figs 1B and S1B). We additionally utilized a second methylation damaging agent, streptozotocin (STZ), a naturally occurring antibiotic produced by Streptomyces achromogenes var. streptozoticus [36,37]. In this case as well, we observed YFP expression from the methylation damage-specific promoter (S1B Fig). These observations were distinct from those seen for a fluorescence reporter for the promoter of sidA, a known SOS-dependent gene [49], that has been reported to be induced under other types of DNA damage in a RecA-dependent manner [49,50]. Additionally, the Pccna_00746-yfp reporter exhibited methylation damage-dependent induction in cells lacking the driD transcription factor as well (S1C Fig). Thus, Caulobacter elicits a transcriptional response to methylation damage that is independent of the SOS as well as the DriD-dependent DNA damage responses. We next investigated the temporal kinetics of the methylation-specific response in relation to the SOS response. For this, we carried out time-lapse imaging of cells carrying either the Pccna_00746-yfp or the PsidA-yfp reporter after MMS exposure. We observed that expression of yfp from the methylation damage-specific promoter expression was temporally delayed when compared to the SOS reporter (Fig 1C). Finally, we tested whether the induction kinetics of this response is adaptive. For this, we first exposed our promoter fusion strain to a low (0.5 mM) dose of MMS. This led to only a modest increase in the expression of Pccna_00746-yfp (S1C Fig). Consistent with an “adaptive” response [51], these pre-treated cells were able to induce Pccna_00746-yfp significantly faster as compared to untreated cells upon exposure to a higher dose (1.5 mM) of MMS (S1C Fig). Together, the Caulobacter methylation damage-specific response showcases key features of an adaptive response to methylation damage as reported in E. coli. Based on these observations, we henceforth refer to this response as the Caulobacter adaptive response to methylation damage. Caulobacter adaptive response to methylation damage is regulated by Cada2 Given the striking similarity between the adaptive responses of Caulobacter and E. coli, we wondered whether the Caulobacter response was regulated by an EcAda-like protein [52]. While we could not identify any EcAda-like protein in Caulobacter, we observed that 3 adaptive response candidates (ccna_00746, ccna_03845, and ccna_00725) possessed an Ada-like methyltransferase domain (PF01035) in their C-terminus region (Fig 2A). Intriguingly, domains corresponding to N-Ada of EcAda protein (required for forming sequence-specific interactions with the cognate Ada promoters) were split between ccna_00746 (with A box-binding domain) and ccna_03845 (with B box-binding domain) (Fig 2A). We thus annotated these as “cada” (Caulobacter adaptive response) genes cada1, cada2, and cada3. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 2. Caulobacter adaptive response to methylation damage is regulated by Cada2. (A) Comparative analysis of high-confidence Alphafold structural models of EcAda, Cada1, Cada2, and Cada3 proteins along with their domain organizations. (B) [Left] Representative cells showing Pccna_00746-yfp reporter induction in wild type (from Fig 1B) and individual deletion strains of the cada genes under 1.5 mM MMS damage. [Right] Violin plots showing fluorescence intensity distribution normalized to cell area from single cells (n = 300, from 3 biological replicates). The underlying data are available in S1 Data. (C) Heat map of log2FC values from RNA-seq experiments for genes up-regulated in wild type, Δcada1 and Δcada2 cells following 1.5 mM MMS treatment. (D) Survival assay of individual deletions of cada genes with and without methylating agent (STZ) (5 μg/ml). https://doi.org/10.1371/journal.pbio.3002540.g002 Next, we asked whether Cada1, Cada2, or Cada3 regulated the adaptive response. We deleted all 3 cada genes individually and assessed the activity of the Pccna_00746-yfp reporter. We found that only cells lacking cada2 were unable to induce Pccna_00746-yfp under methylation damage (Fig 2B). In support, RNA-sequencing analysis revealed that other MMS-specific, RecA-independent genes were also not induced in cells lacking cada2, but were unaffected in cells lacking cada1 (Fig 2C). Thus, A-box containing Cada1 does not drive gene expression under the adaptive response, while Cada2 possessing solely the B-box binding domain is required for response activation. We next tested cell survival upon exposure to MMS as well as STZ. We found that survival was significantly compromised specifically under STZ damage in cells lacking cada2 (Figs 2D and S1E), suggesting that the Cada2-dependent adaptive response to methylation damage was an essential DDR pathway in Caulobacter. The difference in survival between the 2 methylating agents can be attributed to distinct patterns of methylation modifications induced by STZ and MMS [53]. Thus, it is possible that the extent of essentiality is nuanced and dependent on the proportions of various base modifications that a given methylating agent induces. Cada2 associates with adaptive response promoters in a sequence-specific manner To determine whether Cada2 was directly responsible for the induction of the adaptive response genes under methylation damage, we performed ChIP-sequencing experiments (Fig 3A). For this, cells expressing cada2-3x-flag from its endogenous promoter were used. The flag-tagged strain resembled survival of wild-type cells under methylation DNA damage and damage-dependent expression of Cada2 was detected via western blot (S2A and S2B Fig). Using this strain, we carried out ChIP-seq experiments in the presence or absence of methylation damage (Fig 3A). We estimated Cada2-3xFlag enrichment across the Caulobacter genome under damage using untagged wild-type cells as control. This allowed us to identify bona fide Cada2-binding sites. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 3. Cada2 associates with adaptive response promoters in a sequence-specific manner. (A) [Top left] Schematic of the ChIP-sequencing protocol. [Bottom left] Table showcasing the Cada2 regulon. Gene names and predicted functions are listed. [Top right] Normalized reads (in rpm) for Cada2-3x-Flag ChIP-seq represented ±2.5 kb around the CDS of the Cada2 regulon genes in the presence/absence of 1.5 mM MMS damage. [Bottom right] Data represented in a similar fashion for RpoC-3x-Flag ChIP-seq. (B) Binding consensus motif for Cada2 and EcAda derived from the promoters of the respective constituents of their regulon. (C) Representative cells showing induction of variants of the Pccna_00746-yfp reporter (scramble/EcAda-like) compared to wild type under 1.5 mM MMS damage. [Bottom right] Half-violin plots show fluorescence intensity distribution normalized to cell area from single cells for the reporter variants in the presence (dark) and absence of damage (light) (n = 300, from 3 biological replicates). Wild-type data are represented again from Fig 1B. The underlying data are available in S1 Data. https://doi.org/10.1371/journal.pbio.3002540.g003 Under damage, we observed Cada2 enrichment at its own promoter as well as at other promoters of the adaptive response (Figs 3A, S2C, and S2D). This autoregulation of its own expression appears to be a conserved feature between EcAda [54] and Cada2. To further assess the methylation damage-dependent transcription of these genes, we performed ChIP-seq with flag-tagged RNA polymerase subunit C (RpoC-3xflag) construct. We did not observe localization of RpoC to the cada2-associated promoters in the absence of methylation damage (Figs 3A, S2C, and S2D). Upon exposure to damage, RpoC signal could be detected at the cada2 promoter, as well as other adaptive response promoters (Figs 3A, S2C, and S2D). ChIP-seq results overlapped significantly with the Cada2-dependent up-regulated genes identified using RNA-seq experiments (approximately 83%). Superimposing the 2 datasets allowed us to describe the Cada2 regulon comprising 6 genes (Fig 3A), most of which appear to have direct repair-associated activities. We carried out MEME analysis of Cada2-bound regions to identify any sequence-specific DNA-binding motifs. This revealed 2 sequences consistent across all promoters of the Caulobacter adaptive response (Fig 3B). One of these sequences, “GCAA,” was identical to the B-box sequence motif bound by the B-box binding domain of EcAda (Fig 3B). Thus, we annotate it as the B-box motif in case of Caulobacter as well. We did not detect an A-box sequence motif (“AAT,” bound by EcAda A-box binding domain). We instead identified a second recurrent motif that was GC-rich (“CGG”). We annotated this as “X-box” sequence motif. These 2 sequence motifs were separated by a 3-base pair spacer that exhibited no recurrently conserved sequence, but seemed to be consistently AT-rich. Together, we annotated the B-box and X-box motif, separated by the AT-rich spacer as the Cada2 binding sequence motif (Fig 3B). To assess whether this sequence is required for Cada2-mediated regulation, we scrambled the complete sequence in our Pccna_00746-yfp reporter construct. We found that such a reporter was no longer induced under methylation damage (Fig 3C). We next asked whether an EcAda-like DNA-binding motif comprising of an “A-box” instead of the “X-box” DNA motif could drive promoter activity by Cada2. Here too we observed that our reporter construct carrying an A and B box DNA motif was not responsive to methylation damage (Fig 3C). This suggests that the newly identified B+X-box motif is essential for the induction of the Caulobacter adaptive response. Cada2-like proteins are widespread and encode a novel DNA-binding domain The absence of the A-box DNA motif as well as the A-box protein domain in case of Cada2 led us to hypothesize that the X-box DNA motif must likely have a cognate DNA binding region in the Cada2 protein. We thus performed a comprehensive computational search for all Cada2-like methyltransferases across the bacterial kingdom. For this, we built a hidden Markov model (HMM) [55] profile from a multiple-sequence alignment of Cada2-like proteins. As comparison, we followed the same process to identify EcAda-like [28,35] and AdaA-like [42] proteins as well. Phylogenomic distribution of these proteins across a curated, nonredundant database of diverse bacterial species showcased that they are abundant and widespread across all major bacterial phyla (Figs 4A, S3D, and S3E). Furthermore, Cada2-like proteins infrequently co-occurred with EcAda-like proteins in the same genome (approximately 7%, S3F Fig). Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 4. Cada2-like proteins are widespread and encode a novel DNA-binding domain. (A) Phylogenetic distribution at the genus level of EcAda-like (black), Cada2-like (gray), and AdaA-like (brown) proteins across the genomes of 4 bacterial phyla: proteobacteria (red), firmicutes (yellow), actinomycetes (blue), and bacteriodetes (green). Presence/absence is shown on a 16S rRNA-based phylogenetic tree of bacterial genomes. (B) [Top] MEME motif derived from the regulatory domain of 1,000 EcAda-like proteins. [Bottom] MEME motif derived from 1,000 Cada2-like proteins reveal conserved residues in the regulatory domain. (C) [Left] Representative images showing Pccna_00746-yfp reporter induction in wild type and cada2 mutant (cada2R68A and cada2R114A) background under 1.5 mM MMS damage. [Right] Half-violin plots show fluorescence intensity distribution normalized to cell area from single cells in the presence (dark) and absence of damage (light) (n = 300, from 3 biological replicates). Wild-type data are represented again from Fig 1B. The underlying data are available in S1 Data. (D) ChIP-qPCR comparing wild-type Cada2, Cada2R68A, and Cada2R114A enrichment at cada1 and cada3 promoters following exposure to 1.5 mM MMS. In all cases, flag-tagged version of protein is expressed from a xylose-inducible promoter on a high-copy replicating vector (in a Δcada2 background). Bar graphs represent mean fold enrichment in comparison to wild-type cada2 (n = 3 independent repeats, error bars represent standard error). The underlying data are available in S1 Data. https://doi.org/10.1371/journal.pbio.3002540.g004 We zoomed into the DNA-binding (regulatory) regions of EcAda and Cada2-like proteins to identify conserved and unique features. As anticipated, both classes of proteins encoded the B-box binding domain (marked by the conserved “SPHFQR” amino acid residues (Fig 4B)). It is likely that this region associates with the B-box DNA motif that can be found in both EcAda and Cada2 regulons. The 2 proteins deviated with regards to the A-box, with A-box (marked by the 4 cysteine residues) being conserved in all EcAda-like proteins (Fig 4B), but absent in all Cada2-like proteins. Instead, proximal to the putative B-box binding domain, Cada2-like proteins possessed a unique and highly conserved “RLHD” sequence domain (Fig 4B). We highlight here that the conservation of the RLHD domain is even more than that of the B-box. AlphaFold [56,57] model of the Cada2 protein predicted that this RLHD domain falls in a helix-turn-helix domain similar to the B-box binding domain (S3A Fig). In support of the significance of these putative DNA-binding regions of Cada2, mutation of the conserved arginine (usually associated with DNA binding) in both domains to an alanine residue (B-box cada2R68A and RLHD-motif cada2R114A) abrogated Pccna_00746-yfp reporter activity under methylation damage (Fig 4C). The 2 mutants were not compromised in expression and phenocopied the cada2 deletion when assessed for survival under methylation damage (S3B and S3C Fig). Promoter binding was also compromised in both mutants as seen from ChIP-qPCR experiments for Cada2 association at promoter regions of genes belonging to the Cada2 regulon (Fig 4D). We conclude that the B-box and the novel and conserved “RLHD” domains on the Cada2 protein enable it to associate with its promoters in a sequence-specific manner. Cada2 is a methylation-responsive transcription factor We noticed in the ChIP profiles that Cada2 appeared to be modestly enriched at its own promoter even in no damage conditions (Figs 3A, S2C, and S2D). This was in contrast to RNA polymerase that was observed to bind to the cada2 promoter region only in the present of methylation damage (Figs 3A, S2C, and S2D). Indeed, Caulobacter induced the adaptive response only upon exposure to DNA methylation damage (Fig 1B), and overexpression of cada2 from a xylose-inducible promoter was insufficient to induce expression from the Pccna_00746-yfp reporter in the absence of damage (Fig 5A). Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 5. Cada2 is a methylation-dependent transcription factor. (A) [Top] Schematic representing experimental protocol for cada2 overexpression analysis (see main text for details). [Bottom] Representative images and normalized fluorescence intensity for Pccna_00746-yfp expression where cada2 is overexpressed from a xylose-inducible promoter in a Δcada2 background in presence/absence of 1.5 mM MMS damage (n = 300, from 3 biological replicates). The underlying data are available in S1 Data. (B) (Top left) Alphafold structure of Cada2 highlighting the presence of the methyltransferase domain is schematized. (Top right) MEME generated from 1,000 bacterial Cada2 proteins indicate conservation of the “PCHR” motif across methyltransferase domains. (Bottom) Zoomed in mass spectrometry fragmentation spectrum of a methylated peptide detected from a cada2-flag strain under 1.5 mM MMS exposure. The peptide (inset) includes the conserved “PCHR” motif of the Cada2 methyltransferase domain. Representative spectrum from 8 fragments across 2 biological replicates is shown. Methylation modification on Cys267 was detected in 6 out of 8 fragments. (C) [Left] Representative cells showing Pccna_00746-yfp reporter induction in wild type (from Fig 1B) and cada2C267G mutant under 1.5 mM MMS damage. [Right] Violin plots show fluorescence intensity distribution normalized to cell area from single cells (n = 300, from 3 biological replicates). The underlying data are available in S1 Data. (D) Survival assay of cada2C267G mutant in the presence and absence of streptozotocin (5 and 25 μg/ml) and mitomycin C (0.5 μg/ml) damage. https://doi.org/10.1371/journal.pbio.3002540.g005 We asked whether RNA polymerase recruitment to the promoter regions by Cada2 was mediated by physical interaction between Cada2 and the RNA polymerase holoenzyme, and if this step could be methylation dependent. We thus carried out bacterial-two-hybrid interaction analysis of Cada2 against various RNA polymerase holoenzyme subunits (RpoA, RpoB, RpoC, RpoD, and RpoZ). As a positive control, we used the helicase-nuclease protein complex components AddA and AddB [58]. In the presence of methylation damage, we observed interaction signal between Cada2 and RNA polymerase subunit A (S4A Fig). How does methylation damage activate Cada2 function? Under damage, EcAda is posttranslationally methylated at conserved cysteine residues in its A-box binding and methyltransferase (“PCHR”) domains, respectively [59,60]. However, it is the methylation of Cys38 in the A-box binding domain that is required for its activation as a transcription factor via modulation of sequence-specific DNA binding affinity of EcAda [35]. While Cada2 lacks the A-box binding domain, it does possess the methyltransferase domain (in its C-terminus) (Fig 5B). Hence, we tested whether Cada2 is methylated post exposure to methylation DNA damage. Using mass spectrometry, we identified a methylation modification on a cysteine residue (Cys267, part of the “PCHR” methyltransferase domain) of Cada2 in cells treated with methylation damage, with no detectable methylation in the absence of damage (Fig 5B). We mutated the Cys267 residue to an alanine to disrupt the methylation modification. This mutant, cada2C267A, was unable to drive the induction of the Pccna_00746-yfp reporter and phenocopied a cada2 deletion under streptozotocin treatment (S4B and S4C Fig). Significantly, mutations in the promoter-binding domains of Cada2 (B-box cada2R68A and RLHD-motif cada2R114A) did not affect the ability of Cada2 to act as a methyltransferase, suggesting that this activity occurred in a DNA sequence-independent manner (S4D Fig), and that methylation likely precedes the transcriptional response regulated by methylated Cada2. To test if methylation of Cada2 was sufficient for its activation, we mutated the Cys267 residue to a glycine (cada2C267G). Such a mutation in the A-box of EcAda results in its constitutive activation as a transcription factor [35]. In contrast to the alanine mutant which abrogated cada2 activity, we found that the cada2C267G exhibited constitutive expression of the complete Cada2 regulon (S4E Fig). Consistent with this, the Pccna_00746-yfp reporter showed robust expression even in the absence of methylation damage (Fig 5C). Furthermore, cada2C267G did not display growth defects associated with the cada2 deletion or cada2C267A under methylation damage (Figs 5D and S4C). Thus, it is likely that Cada2 is activated as a regulator of the Caulobacter adaptive response via methylation of the conserved cysteine in its methyltransferase domain (PF01035). Given that the adaptive response was constitutively ON in cada2C267G cells, we asked whether such cells would perform better that wild type under methylation damage (as they would resemble cells “primed/adapted” to methylation damage). Indeed, in comparison to wild-type cells, the cada2C267G mutant had a significant growth advantage under methylation damage and appeared to have no growth defect in untreated conditions (Fig 5D). However, despite this growth advantage, sequence analysis of the “PCHR” domain in Cada2-like proteins showed that a constitutive “ON” version of this protein is not found in nature (Fig 5B, 1,000 nonredundant genomes analyzed). We thus wondered whether there could be growth conditions where such a response must remain repressed. To test this, we subjected wild-type and cada2C267G mutant cells to MMC-treatment, an unrelated DNA damage condition where Cada2 activity is not essential and would ordinarily be in an “OFF” state. In contrast to the growth advantage under methylation damage, we observed that cada2C267G cells were considerably compromised in growth in comparison to wild type when exposed to MMC (that induces mono-adducts and intra-strand crosslinks) (Fig 5D). While the molecular mechanism underlying Cada2 activation under methylation damage as well as the crosstalk in case of the cada2C267G mutant requires further investigation, these observations support the possibility that methylation-specific regulation of Cada2 is important, due to antagonistic effects associated with this response being mis-regulated in other stress conditions. Regulatory features of Cada2 are conserved across bacteria and distinct from the EcAda paradigm Taken together, the following regulatory features constitute the Caulobacter adaptive response: (a) Promoter regions of genes under this regulon carry a B-box motif (GCAA) and an X-box motif (CGG). (b) Cada2 protein N-terminus encodes B-box binding and RLHD domains that enable it to interact with promoter regions in a sequence-specific manner. (c) Cada2 protein C-terminus encodes a methyltransferase domain. Methylation of the conserved cysteine in this domain activates Cada2 as a transcription factor, likely via enabling interaction with RNA polymerase. We asked how extrapolatable these observations would be across bacterial species. For this, we collated the presence or absence of the features listed above for a set of bacteria (including closely related and unrelated) that carry either an EcAda-like or Cada2-like protein (Fig 6A). To our surprise, we found that every organism that encoded an EcAda-like protein carried all the associated regulatory features for an E. coli-like adaptive response (Fig 6A), while organisms with Cada2-like protein had regulatory features observed in case of the Caulobacter adaptive response (Fig 6A). Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 6. Regulatory features of Cada2 are conserved across bacteria and distinct from the EcAda paradigm. (A) Regulatory Ada methyltransferases such as EcAda and Cada2 are auto-regulatory; as represented in schematics, these proteins regulate activity of their own promoters (in a methylation-dependent manner). This allowed us to predict presence/absence of regulatory features in their respective promoters. The table represents conservation of regulatory features in the protein (A, B, or RLHD amino acid sequences) and in the cognate promoter sequences (A, B, or X Box DNA motif) of identified EcAda-like and Cada2-like proteins. Presence (gray) or absence of features (blank) is indicated along with % protein sequence. (B) Cada2 from Myxococcus (MyxoCada2) can drive Pccna_00746-yfp induction in Δcada2 in response to MMS damage, but cannot drive yfp expression from the EcAda promoter. Conversely, Ada from E. coli can activate expression of yfp from the EcAda promoter, but not from the Pccna_00746 promoter. The underlying data are available in S1 Data. https://doi.org/10.1371/journal.pbio.3002540.g006 The conservation of the protein and promoter sequence across Cada2-like proteins motivated us to test whether these proteins are functionally inter-changeable. For this, we expressed a Cada2-like protein from Myxococcus xanthus (Cada2myxo) in Caulobacter lacking its own cada2. Cada2myxo has overall low identity (44%) to the Caulobacter Cada2 protein; however, it shares high similarity in terms of conserved regulatory features (Fig 6A). As a control, we expressed EcAda in the same background to assess whether EcAda could complement the absence of Cada2. We found that Cada2myxo was able to fully complement the absence of Caulobacter cada2 under methylation damage, as assessed by cell survival as well as promoter activity of Pccna_00746-yfp (Figs 6B and S5A). In contrast, EcAda was not able to complement a cada2 deletion phenotype (Fig 6B). This protein, expressed in Caulobacter, was fully proficient in driving expression of yfp from a promoter carrying the EcAda DNA binding sequence in a methylation-dependent manner (Fig 6B). However, the Cada2myxo showed Cada2-specific activity and could not induce yfp expression from the EcAda reporter (Fig 6B). Together, these data illustrate the specificity, sufficiency, and necessity of the identified regulatory features in the control of the adaptive responses across bacteria. A methylation-specific DNA damage response in Caulobacter We undertook a comprehensive transcriptomic approach to identify whether Caulobacter induces an SOS-independent transcriptional response specific to methylation damage. For this, we treated Caulobacter cells with agents that predominantly induce 1 specific form of DNA damage (methyl methane sulphonate (MMS), a causative agent of methylation DNA damage [46], mitomycin-C (MMC) that induces intra-strand crosslinks and mono-adducts [47], and norfloxacin that induces double-strand breaks [48]). We collected samples at 0, 20, and 40 min post damage exposure for RNA sequencing (Figs 1A and S1A). Our analysis revealed several genes that were induced in all conditions (“universal”) and a set of genes that were specifically induced under MMS treatment, but in none of the other damaging conditions (methylation-specific) (S1A Fig). Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 1. A methylation-specific DNA damage response in Caulobacter. (A) [Left] Schematic summarizing the RNA-sequencing experiment is provided. Wild-type and ΔrecA (SOS-deficient) cells were exposed to 1.5 mM MMS for 40 min. [Centre] Volcano plots representing differentially expressed genes in comparison to cells exposed to no damage for wild-type (top) and ΔrecA (bottom) cells, respectively. Genes up-regulated in wild-type cells (log2FC > 2 and–log10(p-value) > 2) are highlighted in blue, while genes highlighted in red represent genes up-regulated in wild-type but down-regulated in ΔrecA cells. [Right] Heat maps represent log2FC values for individual genes induced in wild-type and ΔrecA cells under MMS exposure. (B) Wild-type Pccna_00746-yfp reporter (schematic inset) was exposed to 1.5 mM MMS or 0.5 μg/ml MMC, respectively for 2 h. Representative cells are shown on the left (scale bar: 2 μm, cell boundaries are marked by white dotted outline, here and in all other images). Violin plots show fluorescence intensity distribution normalized to cell area from single cells (n = 300, from 3 biological replicates). The underlying data are available in S1 Data. (C) [Top] Schematic of the experimental protocol comparing induction kinetics of the Caulobacter SOS and methylation-specific response via time lapse microscopy. [Bottom] Fluorescence intensity normalized to cell area for Pccna_00746-yfp and PsidA-yfp cells over 3 h of exposure to 1.5 mM MMS. The underlying data are available in S1 Data. https://doi.org/10.1371/journal.pbio.3002540.g001 We next asked whether the expression of these genes was independent of the SOS response. We subjected recA deleted cells to MMS treatment and carried out RNA sequencing in this background. Comparing wild type to recA deletion cells showed that the methylation damage-specific genes were induced in a recA-independent manner (SOS-independent) (Fig 1A). In contrast, genes that were induced under all DNA damaging conditions in wild type background remain uninduced in recA deletion cells (SOS-dependent) (Figs 1A and S1A). We identified ccna_00745 to be the highest induced under methylation damage in the transcriptome analysis (Fig 1A). This gene is co-operonic with a second gene ccna_00746, that is also induced in a methylation DNA damage-specific manner (S1A Fig). ccna_00745 is Bioinformatically predicted to be a 2-oxoglutarate, Fe2+-dependent dioxygenase, similar to E. coli alkB, while ccna_00746 is predicted to possess an Ada-like methyltransferase domain (PF01035) [9]. To further corroborate the RNA-seq observations, we constructed a fluorescence-based readout for promoter activity of these genes (Pccna_00746-yfp). No significant expression of yfp was detected in the absence of damage (Fig 1B). We observed YFP expression from this construct only in cells treated with MMS, and not other DNA damaging agents MMC, norfloxacin and hydroxyurea (HU) (Figs 1B and S1B). We additionally utilized a second methylation damaging agent, streptozotocin (STZ), a naturally occurring antibiotic produced by Streptomyces achromogenes var. streptozoticus [36,37]. In this case as well, we observed YFP expression from the methylation damage-specific promoter (S1B Fig). These observations were distinct from those seen for a fluorescence reporter for the promoter of sidA, a known SOS-dependent gene [49], that has been reported to be induced under other types of DNA damage in a RecA-dependent manner [49,50]. Additionally, the Pccna_00746-yfp reporter exhibited methylation damage-dependent induction in cells lacking the driD transcription factor as well (S1C Fig). Thus, Caulobacter elicits a transcriptional response to methylation damage that is independent of the SOS as well as the DriD-dependent DNA damage responses. We next investigated the temporal kinetics of the methylation-specific response in relation to the SOS response. For this, we carried out time-lapse imaging of cells carrying either the Pccna_00746-yfp or the PsidA-yfp reporter after MMS exposure. We observed that expression of yfp from the methylation damage-specific promoter expression was temporally delayed when compared to the SOS reporter (Fig 1C). Finally, we tested whether the induction kinetics of this response is adaptive. For this, we first exposed our promoter fusion strain to a low (0.5 mM) dose of MMS. This led to only a modest increase in the expression of Pccna_00746-yfp (S1C Fig). Consistent with an “adaptive” response [51], these pre-treated cells were able to induce Pccna_00746-yfp significantly faster as compared to untreated cells upon exposure to a higher dose (1.5 mM) of MMS (S1C Fig). Together, the Caulobacter methylation damage-specific response showcases key features of an adaptive response to methylation damage as reported in E. coli. Based on these observations, we henceforth refer to this response as the Caulobacter adaptive response to methylation damage. Caulobacter adaptive response to methylation damage is regulated by Cada2 Given the striking similarity between the adaptive responses of Caulobacter and E. coli, we wondered whether the Caulobacter response was regulated by an EcAda-like protein [52]. While we could not identify any EcAda-like protein in Caulobacter, we observed that 3 adaptive response candidates (ccna_00746, ccna_03845, and ccna_00725) possessed an Ada-like methyltransferase domain (PF01035) in their C-terminus region (Fig 2A). Intriguingly, domains corresponding to N-Ada of EcAda protein (required for forming sequence-specific interactions with the cognate Ada promoters) were split between ccna_00746 (with A box-binding domain) and ccna_03845 (with B box-binding domain) (Fig 2A). We thus annotated these as “cada” (Caulobacter adaptive response) genes cada1, cada2, and cada3. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 2. Caulobacter adaptive response to methylation damage is regulated by Cada2. (A) Comparative analysis of high-confidence Alphafold structural models of EcAda, Cada1, Cada2, and Cada3 proteins along with their domain organizations. (B) [Left] Representative cells showing Pccna_00746-yfp reporter induction in wild type (from Fig 1B) and individual deletion strains of the cada genes under 1.5 mM MMS damage. [Right] Violin plots showing fluorescence intensity distribution normalized to cell area from single cells (n = 300, from 3 biological replicates). The underlying data are available in S1 Data. (C) Heat map of log2FC values from RNA-seq experiments for genes up-regulated in wild type, Δcada1 and Δcada2 cells following 1.5 mM MMS treatment. (D) Survival assay of individual deletions of cada genes with and without methylating agent (STZ) (5 μg/ml). https://doi.org/10.1371/journal.pbio.3002540.g002 Next, we asked whether Cada1, Cada2, or Cada3 regulated the adaptive response. We deleted all 3 cada genes individually and assessed the activity of the Pccna_00746-yfp reporter. We found that only cells lacking cada2 were unable to induce Pccna_00746-yfp under methylation damage (Fig 2B). In support, RNA-sequencing analysis revealed that other MMS-specific, RecA-independent genes were also not induced in cells lacking cada2, but were unaffected in cells lacking cada1 (Fig 2C). Thus, A-box containing Cada1 does not drive gene expression under the adaptive response, while Cada2 possessing solely the B-box binding domain is required for response activation. We next tested cell survival upon exposure to MMS as well as STZ. We found that survival was significantly compromised specifically under STZ damage in cells lacking cada2 (Figs 2D and S1E), suggesting that the Cada2-dependent adaptive response to methylation damage was an essential DDR pathway in Caulobacter. The difference in survival between the 2 methylating agents can be attributed to distinct patterns of methylation modifications induced by STZ and MMS [53]. Thus, it is possible that the extent of essentiality is nuanced and dependent on the proportions of various base modifications that a given methylating agent induces. Cada2 associates with adaptive response promoters in a sequence-specific manner To determine whether Cada2 was directly responsible for the induction of the adaptive response genes under methylation damage, we performed ChIP-sequencing experiments (Fig 3A). For this, cells expressing cada2-3x-flag from its endogenous promoter were used. The flag-tagged strain resembled survival of wild-type cells under methylation DNA damage and damage-dependent expression of Cada2 was detected via western blot (S2A and S2B Fig). Using this strain, we carried out ChIP-seq experiments in the presence or absence of methylation damage (Fig 3A). We estimated Cada2-3xFlag enrichment across the Caulobacter genome under damage using untagged wild-type cells as control. This allowed us to identify bona fide Cada2-binding sites. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 3. Cada2 associates with adaptive response promoters in a sequence-specific manner. (A) [Top left] Schematic of the ChIP-sequencing protocol. [Bottom left] Table showcasing the Cada2 regulon. Gene names and predicted functions are listed. [Top right] Normalized reads (in rpm) for Cada2-3x-Flag ChIP-seq represented ±2.5 kb around the CDS of the Cada2 regulon genes in the presence/absence of 1.5 mM MMS damage. [Bottom right] Data represented in a similar fashion for RpoC-3x-Flag ChIP-seq. (B) Binding consensus motif for Cada2 and EcAda derived from the promoters of the respective constituents of their regulon. (C) Representative cells showing induction of variants of the Pccna_00746-yfp reporter (scramble/EcAda-like) compared to wild type under 1.5 mM MMS damage. [Bottom right] Half-violin plots show fluorescence intensity distribution normalized to cell area from single cells for the reporter variants in the presence (dark) and absence of damage (light) (n = 300, from 3 biological replicates). Wild-type data are represented again from Fig 1B. The underlying data are available in S1 Data. https://doi.org/10.1371/journal.pbio.3002540.g003 Under damage, we observed Cada2 enrichment at its own promoter as well as at other promoters of the adaptive response (Figs 3A, S2C, and S2D). This autoregulation of its own expression appears to be a conserved feature between EcAda [54] and Cada2. To further assess the methylation damage-dependent transcription of these genes, we performed ChIP-seq with flag-tagged RNA polymerase subunit C (RpoC-3xflag) construct. We did not observe localization of RpoC to the cada2-associated promoters in the absence of methylation damage (Figs 3A, S2C, and S2D). Upon exposure to damage, RpoC signal could be detected at the cada2 promoter, as well as other adaptive response promoters (Figs 3A, S2C, and S2D). ChIP-seq results overlapped significantly with the Cada2-dependent up-regulated genes identified using RNA-seq experiments (approximately 83%). Superimposing the 2 datasets allowed us to describe the Cada2 regulon comprising 6 genes (Fig 3A), most of which appear to have direct repair-associated activities. We carried out MEME analysis of Cada2-bound regions to identify any sequence-specific DNA-binding motifs. This revealed 2 sequences consistent across all promoters of the Caulobacter adaptive response (Fig 3B). One of these sequences, “GCAA,” was identical to the B-box sequence motif bound by the B-box binding domain of EcAda (Fig 3B). Thus, we annotate it as the B-box motif in case of Caulobacter as well. We did not detect an A-box sequence motif (“AAT,” bound by EcAda A-box binding domain). We instead identified a second recurrent motif that was GC-rich (“CGG”). We annotated this as “X-box” sequence motif. These 2 sequence motifs were separated by a 3-base pair spacer that exhibited no recurrently conserved sequence, but seemed to be consistently AT-rich. Together, we annotated the B-box and X-box motif, separated by the AT-rich spacer as the Cada2 binding sequence motif (Fig 3B). To assess whether this sequence is required for Cada2-mediated regulation, we scrambled the complete sequence in our Pccna_00746-yfp reporter construct. We found that such a reporter was no longer induced under methylation damage (Fig 3C). We next asked whether an EcAda-like DNA-binding motif comprising of an “A-box” instead of the “X-box” DNA motif could drive promoter activity by Cada2. Here too we observed that our reporter construct carrying an A and B box DNA motif was not responsive to methylation damage (Fig 3C). This suggests that the newly identified B+X-box motif is essential for the induction of the Caulobacter adaptive response. Cada2-like proteins are widespread and encode a novel DNA-binding domain The absence of the A-box DNA motif as well as the A-box protein domain in case of Cada2 led us to hypothesize that the X-box DNA motif must likely have a cognate DNA binding region in the Cada2 protein. We thus performed a comprehensive computational search for all Cada2-like methyltransferases across the bacterial kingdom. For this, we built a hidden Markov model (HMM) [55] profile from a multiple-sequence alignment of Cada2-like proteins. As comparison, we followed the same process to identify EcAda-like [28,35] and AdaA-like [42] proteins as well. Phylogenomic distribution of these proteins across a curated, nonredundant database of diverse bacterial species showcased that they are abundant and widespread across all major bacterial phyla (Figs 4A, S3D, and S3E). Furthermore, Cada2-like proteins infrequently co-occurred with EcAda-like proteins in the same genome (approximately 7%, S3F Fig). Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 4. Cada2-like proteins are widespread and encode a novel DNA-binding domain. (A) Phylogenetic distribution at the genus level of EcAda-like (black), Cada2-like (gray), and AdaA-like (brown) proteins across the genomes of 4 bacterial phyla: proteobacteria (red), firmicutes (yellow), actinomycetes (blue), and bacteriodetes (green). Presence/absence is shown on a 16S rRNA-based phylogenetic tree of bacterial genomes. (B) [Top] MEME motif derived from the regulatory domain of 1,000 EcAda-like proteins. [Bottom] MEME motif derived from 1,000 Cada2-like proteins reveal conserved residues in the regulatory domain. (C) [Left] Representative images showing Pccna_00746-yfp reporter induction in wild type and cada2 mutant (cada2R68A and cada2R114A) background under 1.5 mM MMS damage. [Right] Half-violin plots show fluorescence intensity distribution normalized to cell area from single cells in the presence (dark) and absence of damage (light) (n = 300, from 3 biological replicates). Wild-type data are represented again from Fig 1B. The underlying data are available in S1 Data. (D) ChIP-qPCR comparing wild-type Cada2, Cada2R68A, and Cada2R114A enrichment at cada1 and cada3 promoters following exposure to 1.5 mM MMS. In all cases, flag-tagged version of protein is expressed from a xylose-inducible promoter on a high-copy replicating vector (in a Δcada2 background). Bar graphs represent mean fold enrichment in comparison to wild-type cada2 (n = 3 independent repeats, error bars represent standard error). The underlying data are available in S1 Data. https://doi.org/10.1371/journal.pbio.3002540.g004 We zoomed into the DNA-binding (regulatory) regions of EcAda and Cada2-like proteins to identify conserved and unique features. As anticipated, both classes of proteins encoded the B-box binding domain (marked by the conserved “SPHFQR” amino acid residues (Fig 4B)). It is likely that this region associates with the B-box DNA motif that can be found in both EcAda and Cada2 regulons. The 2 proteins deviated with regards to the A-box, with A-box (marked by the 4 cysteine residues) being conserved in all EcAda-like proteins (Fig 4B), but absent in all Cada2-like proteins. Instead, proximal to the putative B-box binding domain, Cada2-like proteins possessed a unique and highly conserved “RLHD” sequence domain (Fig 4B). We highlight here that the conservation of the RLHD domain is even more than that of the B-box. AlphaFold [56,57] model of the Cada2 protein predicted that this RLHD domain falls in a helix-turn-helix domain similar to the B-box binding domain (S3A Fig). In support of the significance of these putative DNA-binding regions of Cada2, mutation of the conserved arginine (usually associated with DNA binding) in both domains to an alanine residue (B-box cada2R68A and RLHD-motif cada2R114A) abrogated Pccna_00746-yfp reporter activity under methylation damage (Fig 4C). The 2 mutants were not compromised in expression and phenocopied the cada2 deletion when assessed for survival under methylation damage (S3B and S3C Fig). Promoter binding was also compromised in both mutants as seen from ChIP-qPCR experiments for Cada2 association at promoter regions of genes belonging to the Cada2 regulon (Fig 4D). We conclude that the B-box and the novel and conserved “RLHD” domains on the Cada2 protein enable it to associate with its promoters in a sequence-specific manner. Cada2 is a methylation-responsive transcription factor We noticed in the ChIP profiles that Cada2 appeared to be modestly enriched at its own promoter even in no damage conditions (Figs 3A, S2C, and S2D). This was in contrast to RNA polymerase that was observed to bind to the cada2 promoter region only in the present of methylation damage (Figs 3A, S2C, and S2D). Indeed, Caulobacter induced the adaptive response only upon exposure to DNA methylation damage (Fig 1B), and overexpression of cada2 from a xylose-inducible promoter was insufficient to induce expression from the Pccna_00746-yfp reporter in the absence of damage (Fig 5A). Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 5. Cada2 is a methylation-dependent transcription factor. (A) [Top] Schematic representing experimental protocol for cada2 overexpression analysis (see main text for details). [Bottom] Representative images and normalized fluorescence intensity for Pccna_00746-yfp expression where cada2 is overexpressed from a xylose-inducible promoter in a Δcada2 background in presence/absence of 1.5 mM MMS damage (n = 300, from 3 biological replicates). The underlying data are available in S1 Data. (B) (Top left) Alphafold structure of Cada2 highlighting the presence of the methyltransferase domain is schematized. (Top right) MEME generated from 1,000 bacterial Cada2 proteins indicate conservation of the “PCHR” motif across methyltransferase domains. (Bottom) Zoomed in mass spectrometry fragmentation spectrum of a methylated peptide detected from a cada2-flag strain under 1.5 mM MMS exposure. The peptide (inset) includes the conserved “PCHR” motif of the Cada2 methyltransferase domain. Representative spectrum from 8 fragments across 2 biological replicates is shown. Methylation modification on Cys267 was detected in 6 out of 8 fragments. (C) [Left] Representative cells showing Pccna_00746-yfp reporter induction in wild type (from Fig 1B) and cada2C267G mutant under 1.5 mM MMS damage. [Right] Violin plots show fluorescence intensity distribution normalized to cell area from single cells (n = 300, from 3 biological replicates). The underlying data are available in S1 Data. (D) Survival assay of cada2C267G mutant in the presence and absence of streptozotocin (5 and 25 μg/ml) and mitomycin C (0.5 μg/ml) damage. https://doi.org/10.1371/journal.pbio.3002540.g005 We asked whether RNA polymerase recruitment to the promoter regions by Cada2 was mediated by physical interaction between Cada2 and the RNA polymerase holoenzyme, and if this step could be methylation dependent. We thus carried out bacterial-two-hybrid interaction analysis of Cada2 against various RNA polymerase holoenzyme subunits (RpoA, RpoB, RpoC, RpoD, and RpoZ). As a positive control, we used the helicase-nuclease protein complex components AddA and AddB [58]. In the presence of methylation damage, we observed interaction signal between Cada2 and RNA polymerase subunit A (S4A Fig). How does methylation damage activate Cada2 function? Under damage, EcAda is posttranslationally methylated at conserved cysteine residues in its A-box binding and methyltransferase (“PCHR”) domains, respectively [59,60]. However, it is the methylation of Cys38 in the A-box binding domain that is required for its activation as a transcription factor via modulation of sequence-specific DNA binding affinity of EcAda [35]. While Cada2 lacks the A-box binding domain, it does possess the methyltransferase domain (in its C-terminus) (Fig 5B). Hence, we tested whether Cada2 is methylated post exposure to methylation DNA damage. Using mass spectrometry, we identified a methylation modification on a cysteine residue (Cys267, part of the “PCHR” methyltransferase domain) of Cada2 in cells treated with methylation damage, with no detectable methylation in the absence of damage (Fig 5B). We mutated the Cys267 residue to an alanine to disrupt the methylation modification. This mutant, cada2C267A, was unable to drive the induction of the Pccna_00746-yfp reporter and phenocopied a cada2 deletion under streptozotocin treatment (S4B and S4C Fig). Significantly, mutations in the promoter-binding domains of Cada2 (B-box cada2R68A and RLHD-motif cada2R114A) did not affect the ability of Cada2 to act as a methyltransferase, suggesting that this activity occurred in a DNA sequence-independent manner (S4D Fig), and that methylation likely precedes the transcriptional response regulated by methylated Cada2. To test if methylation of Cada2 was sufficient for its activation, we mutated the Cys267 residue to a glycine (cada2C267G). Such a mutation in the A-box of EcAda results in its constitutive activation as a transcription factor [35]. In contrast to the alanine mutant which abrogated cada2 activity, we found that the cada2C267G exhibited constitutive expression of the complete Cada2 regulon (S4E Fig). Consistent with this, the Pccna_00746-yfp reporter showed robust expression even in the absence of methylation damage (Fig 5C). Furthermore, cada2C267G did not display growth defects associated with the cada2 deletion or cada2C267A under methylation damage (Figs 5D and S4C). Thus, it is likely that Cada2 is activated as a regulator of the Caulobacter adaptive response via methylation of the conserved cysteine in its methyltransferase domain (PF01035). Given that the adaptive response was constitutively ON in cada2C267G cells, we asked whether such cells would perform better that wild type under methylation damage (as they would resemble cells “primed/adapted” to methylation damage). Indeed, in comparison to wild-type cells, the cada2C267G mutant had a significant growth advantage under methylation damage and appeared to have no growth defect in untreated conditions (Fig 5D). However, despite this growth advantage, sequence analysis of the “PCHR” domain in Cada2-like proteins showed that a constitutive “ON” version of this protein is not found in nature (Fig 5B, 1,000 nonredundant genomes analyzed). We thus wondered whether there could be growth conditions where such a response must remain repressed. To test this, we subjected wild-type and cada2C267G mutant cells to MMC-treatment, an unrelated DNA damage condition where Cada2 activity is not essential and would ordinarily be in an “OFF” state. In contrast to the growth advantage under methylation damage, we observed that cada2C267G cells were considerably compromised in growth in comparison to wild type when exposed to MMC (that induces mono-adducts and intra-strand crosslinks) (Fig 5D). While the molecular mechanism underlying Cada2 activation under methylation damage as well as the crosstalk in case of the cada2C267G mutant requires further investigation, these observations support the possibility that methylation-specific regulation of Cada2 is important, due to antagonistic effects associated with this response being mis-regulated in other stress conditions. Regulatory features of Cada2 are conserved across bacteria and distinct from the EcAda paradigm Taken together, the following regulatory features constitute the Caulobacter adaptive response: (a) Promoter regions of genes under this regulon carry a B-box motif (GCAA) and an X-box motif (CGG). (b) Cada2 protein N-terminus encodes B-box binding and RLHD domains that enable it to interact with promoter regions in a sequence-specific manner. (c) Cada2 protein C-terminus encodes a methyltransferase domain. Methylation of the conserved cysteine in this domain activates Cada2 as a transcription factor, likely via enabling interaction with RNA polymerase. We asked how extrapolatable these observations would be across bacterial species. For this, we collated the presence or absence of the features listed above for a set of bacteria (including closely related and unrelated) that carry either an EcAda-like or Cada2-like protein (Fig 6A). To our surprise, we found that every organism that encoded an EcAda-like protein carried all the associated regulatory features for an E. coli-like adaptive response (Fig 6A), while organisms with Cada2-like protein had regulatory features observed in case of the Caulobacter adaptive response (Fig 6A). Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 6. Regulatory features of Cada2 are conserved across bacteria and distinct from the EcAda paradigm. (A) Regulatory Ada methyltransferases such as EcAda and Cada2 are auto-regulatory; as represented in schematics, these proteins regulate activity of their own promoters (in a methylation-dependent manner). This allowed us to predict presence/absence of regulatory features in their respective promoters. The table represents conservation of regulatory features in the protein (A, B, or RLHD amino acid sequences) and in the cognate promoter sequences (A, B, or X Box DNA motif) of identified EcAda-like and Cada2-like proteins. Presence (gray) or absence of features (blank) is indicated along with % protein sequence. (B) Cada2 from Myxococcus (MyxoCada2) can drive Pccna_00746-yfp induction in Δcada2 in response to MMS damage, but cannot drive yfp expression from the EcAda promoter. Conversely, Ada from E. coli can activate expression of yfp from the EcAda promoter, but not from the Pccna_00746 promoter. The underlying data are available in S1 Data. https://doi.org/10.1371/journal.pbio.3002540.g006 The conservation of the protein and promoter sequence across Cada2-like proteins motivated us to test whether these proteins are functionally inter-changeable. For this, we expressed a Cada2-like protein from Myxococcus xanthus (Cada2myxo) in Caulobacter lacking its own cada2. Cada2myxo has overall low identity (44%) to the Caulobacter Cada2 protein; however, it shares high similarity in terms of conserved regulatory features (Fig 6A). As a control, we expressed EcAda in the same background to assess whether EcAda could complement the absence of Cada2. We found that Cada2myxo was able to fully complement the absence of Caulobacter cada2 under methylation damage, as assessed by cell survival as well as promoter activity of Pccna_00746-yfp (Figs 6B and S5A). In contrast, EcAda was not able to complement a cada2 deletion phenotype (Fig 6B). This protein, expressed in Caulobacter, was fully proficient in driving expression of yfp from a promoter carrying the EcAda DNA binding sequence in a methylation-dependent manner (Fig 6B). However, the Cada2myxo showed Cada2-specific activity and could not induce yfp expression from the EcAda reporter (Fig 6B). Together, these data illustrate the specificity, sufficiency, and necessity of the identified regulatory features in the control of the adaptive responses across bacteria. Discussion In this study, we identify Cada2, a novel methylation damage-specific transcription factor conserved across all major phyla of bacteria. We further implicate a critical and conserved role for PTM-based activation of this class of bacterial transcription factors, albeit with varying mechanisms of action (Cada2 versus EcAda). We hypothesize that the organizational diversity observed across regulators of the adaptive response pathways are likely an adaptation to the physiological features of bacteria. The downstream adaptive response is, however, robust to this variability. Diverse mechanisms, unifying function The mechanistic differences between Cada2 and EcAda would suggest that the response dynamics would also be dissimilar. Cada2 and EcAda use distinct DNA-binding domains to associate with their cognate promoters (B-box and “RLHD” domains versus A and B-box binding domains, respectively [35]). The 2 proteins are also activated as transcription factors in non-overlapping ways: In case of EcAda, methylation of the cysteine residue within the sequence-specific DNA-binding A-box domain activates it as a transcription factor [59,61]. In contrast, Cada2 activation occurs via PTM of the cysteine in its methyltransferase domain that is distally located from the sequence-specific DNA-bindings domain. Furthermore, unlike EcAda, unmethylated Cada2 appears to form sequence-specific interactions with its own promoter (Figs 3A and S2A). Presence of electrostatic repulsion (as in EcAda) would hinder this interaction. This would suggest that Cada2 acts via an alternate, non-electrostatic mechanism. Yet despite these contrasts, the response kinetics are invariant between the 2 organisms. The adaptive response of Caulobacter is tightly regulated, showing low levels of expression (Ada OFF) in the absence of methylation DNA damage. Upon exposure to methylating agents, Caulobacter cells also elicit a delayed adaptive response, subsequent to the SOS response. Similar to the E. coli Ada response [31], the Caulobacter response exhibits bi-stability with a subpopulation of cells resembling expression of Ada OFF cells even in the presence of DNA methylation damage. Interestingly, the OFF population is abolished in a strain overexpressing Cada2, mirroring observations made in case of EcAda [31] (Figs 1B and 5A). The factors influencing the differences in the mechanisms of action of the response regulators, while retaining the defining features of the response itself are an important and exciting avenue for future investigations. Requirement for a methylation damage-specific response in bacteria How pervasive is the adaptive response to methylation damage? Our computational analysis reveals that the regulatory Ada methyltransferase is widely conserved, but it appears to have diverse domain organizations (S3D and S3E Fig). Given the apparent modularity of these domains (S3D and S3E Fig), it is tempting to speculate that such diverse organizations arise as a result of domain-shuffling from a common ancestor protein [62–64]. Despite their prevalence, multiple regulators rarely occur on the same genome. A case in point is the comparison between EcAda and Cada2. We were able to detect only approximately 7% of genomes possessing both proteins. The genomes that encoded both EcAda and Cada2 displayed varied and unique regulatory circuits, including autoregulation as well as cross-regulation (S3F Fig). What drives the observed diversity? Comparison of EcAda and Cada2 might provide a compelling hypothesis. Both EcAda and Cada2 share the B-box motif in their cognate promoters. However, instead of the AT-rich A-box motif observed in the promoters of EcAda regulon genes, Cada2 binds to a GC-rich X-box motif. This correlates well with the GC-content of the genomes that encode these proteins. The GC-content of the E. coli genome is approximately 50% (low), while the GC-content of the Caulobacter genome is approximately 67% (high). This hypothesis is consistent with previous studies that have also suggested the interdependence of genome GC-content and presence/absence of specific DNA repair pathways [65–68]. Thus, it is likely that distinguishing physiological characteristics of bacteria warrant organizational adaptation and domain reorganizations of regulatory Ada methyltransferases. However, regardless of the diversity observed at a regulator level, the adaptive response-like pathways seem to retain certain characteristics as emphasized previously. In this context, we highlight the PTM-based transcriptional switch required for activation and regulation of the response. It is possible that costs (e.g., Fig 5D) or fitness advantages associated with this response serve as constraining forces that render this system robust to variations in regulator organization, and yet drive the retention of key regulatory features. Together, the conservation of bacterial adaptive response mechanisms, albeit in diverse regulatory forms, underscores the fundamental requirement for a dedicated methylation-specific DNA damage response. Diverse mechanisms, unifying function The mechanistic differences between Cada2 and EcAda would suggest that the response dynamics would also be dissimilar. Cada2 and EcAda use distinct DNA-binding domains to associate with their cognate promoters (B-box and “RLHD” domains versus A and B-box binding domains, respectively [35]). The 2 proteins are also activated as transcription factors in non-overlapping ways: In case of EcAda, methylation of the cysteine residue within the sequence-specific DNA-binding A-box domain activates it as a transcription factor [59,61]. In contrast, Cada2 activation occurs via PTM of the cysteine in its methyltransferase domain that is distally located from the sequence-specific DNA-bindings domain. Furthermore, unlike EcAda, unmethylated Cada2 appears to form sequence-specific interactions with its own promoter (Figs 3A and S2A). Presence of electrostatic repulsion (as in EcAda) would hinder this interaction. This would suggest that Cada2 acts via an alternate, non-electrostatic mechanism. Yet despite these contrasts, the response kinetics are invariant between the 2 organisms. The adaptive response of Caulobacter is tightly regulated, showing low levels of expression (Ada OFF) in the absence of methylation DNA damage. Upon exposure to methylating agents, Caulobacter cells also elicit a delayed adaptive response, subsequent to the SOS response. Similar to the E. coli Ada response [31], the Caulobacter response exhibits bi-stability with a subpopulation of cells resembling expression of Ada OFF cells even in the presence of DNA methylation damage. Interestingly, the OFF population is abolished in a strain overexpressing Cada2, mirroring observations made in case of EcAda [31] (Figs 1B and 5A). The factors influencing the differences in the mechanisms of action of the response regulators, while retaining the defining features of the response itself are an important and exciting avenue for future investigations. Requirement for a methylation damage-specific response in bacteria How pervasive is the adaptive response to methylation damage? Our computational analysis reveals that the regulatory Ada methyltransferase is widely conserved, but it appears to have diverse domain organizations (S3D and S3E Fig). Given the apparent modularity of these domains (S3D and S3E Fig), it is tempting to speculate that such diverse organizations arise as a result of domain-shuffling from a common ancestor protein [62–64]. Despite their prevalence, multiple regulators rarely occur on the same genome. A case in point is the comparison between EcAda and Cada2. We were able to detect only approximately 7% of genomes possessing both proteins. The genomes that encoded both EcAda and Cada2 displayed varied and unique regulatory circuits, including autoregulation as well as cross-regulation (S3F Fig). What drives the observed diversity? Comparison of EcAda and Cada2 might provide a compelling hypothesis. Both EcAda and Cada2 share the B-box motif in their cognate promoters. However, instead of the AT-rich A-box motif observed in the promoters of EcAda regulon genes, Cada2 binds to a GC-rich X-box motif. This correlates well with the GC-content of the genomes that encode these proteins. The GC-content of the E. coli genome is approximately 50% (low), while the GC-content of the Caulobacter genome is approximately 67% (high). This hypothesis is consistent with previous studies that have also suggested the interdependence of genome GC-content and presence/absence of specific DNA repair pathways [65–68]. Thus, it is likely that distinguishing physiological characteristics of bacteria warrant organizational adaptation and domain reorganizations of regulatory Ada methyltransferases. However, regardless of the diversity observed at a regulator level, the adaptive response-like pathways seem to retain certain characteristics as emphasized previously. In this context, we highlight the PTM-based transcriptional switch required for activation and regulation of the response. It is possible that costs (e.g., Fig 5D) or fitness advantages associated with this response serve as constraining forces that render this system robust to variations in regulator organization, and yet drive the retention of key regulatory features. Together, the conservation of bacterial adaptive response mechanisms, albeit in diverse regulatory forms, underscores the fundamental requirement for a dedicated methylation-specific DNA damage response. Materials and methods Bacterial strains and growth conditions All strains, plasmids, and oligos used in this study are listed in S1–S3 Tables, respectively. Caulobacter crescentus cells were grown at 30°C in peptone yeast extract (PYE) media (0.2% peptone, 0.1% yeast extract, and 0.06% MgSO4) and supplemented with antibiotics or inducers, as required, at suitable concentrations. For induction of cada2 expression, 0.3% xylose was introduced into the cultures, unless otherwise stated. Fluorescence microscopy and image analysis Imaging was performed on an epifluorescence, wide-field microscope (Eclipse Ti-2E, Nikon) with a 60×/1.4 NA oil immersion objective and a motorized stage. pE-4000 (CoolLED) was used as the LED excitation source. Exposure times for all yfp samples (λ = 490nm) was 500 ms and exposure was maintained at 50% of LED power. Images were captured using a Hamamatsu Orca Flash 4.0 camera. Focus was ensured via an infrared-based Perfect Focusing System (Nikon). Samples were prepared as detailed in [69]. For time course microscopy, 1 ml of cultures were aliquoted at various time points, pelleted and resuspended in appropriate volume of growth media, and 2 μl of the resuspended cells were spotted on 1% agarose (Invitrogen ultrapure) pads and imaged. For time lapse microscopy, 2 μl of culture were spotted on 1.5% low-melting GTG agarose pads supplemented with PYE media and 1.5 mM MMS damage. Throughout the duration of the time lapse, samples were grown in an OkoLab incubation chamber maintained at 30°C and imaged at regular intervals. Cell segmentation and fluorescence intensity analysis was carried out via Oufti software [70]. RNA-sequencing Sample collection. Overnight cultures of Caulobacter cells were back-diluted to 0.025 OD600 and grown at 30°C for 3 h. When OD600 of culture was approximately 0.1, DNA damage (MMS (1.5 mM)/MMC (0.25 μg/ml)/norfloxacin (8 μg/ml)) was introduced. Samples from 4 biological replicates were collected for no damage conditions and 2 biological replicates for samples in the presence of damage, and 2 ml of Caulobacter cells were harvested and spun at 10,000 g for 5 min. Supernatant was then discarded and the cell pellets were snap frozen with liquid nitrogen and stored at –80°C until the samples were further processed. Cells were harvested at 0, 20, and 40 min post DNA damage exposure. RNA-extraction. RNA was extracted from the cell pellets according to a previous study [58]. Cells were lysed using 400 μl of pre-heated Trizol in a thermomixer at 65°C, 2000 rpm. The lysed cells were transferred to –80°C for a minimum of 30 min and were then centrifuged at 4°C. The supernatant was carefully transferred to 100% ethanol of equivalent volume. RNA was then extracted via the protocol mentioned in the Direct-zol RNA MiniPrep (Zymo, Cat. no. R2052) kit. The extracted RNA was then subjected to DNAse treatment in order to remove any genomic DNA from the sample. Total RNA from the samples was purified using the RNA Clean & Concentrator-25 (Zymo, Cat. no. R1018) kit. Integrity of the RNA was tested using Bioanalyzer instrument. mRNA was isolated from the total sample using the RiboMinus kit (Thermo, Cat. no. K155004) and submitted for RNA-seq at the NCBS next generation sequencing facility. Details regarding kits used for RNA library preparation and the sequencing platform used for the RNA-seq experiment are summarized in S4 Table. RNA-sequencing analysis Raw reads for the sequencing results were obtained as fastq files. The reference genome sequence (.fna) and annotation (.gff) files for the Caulobacter crescentus NA1000 strain (accession number: NC_011916.1) were downloaded from the NCBI file transfer protocol (ftp) website (“ftp.ncbi.nlm.nih.gov”). The raw read quality was checked using the FastQC software (version 0.11.5). Burrows–Wheeler Aligner (BWA) (version 0.7.17-r1188) was used to index the reference genome. Reads with raw read quality > = 20 were aligned to the indexed genome using the BWA aln -q option. Samtools (version 0.1.7) was used to filter out multi-mapped reads. Bedtools (version 2.26.0) was used to calculate the reads count per gene using the annotation file (.bed) [71]. The normalization and differential gene expression analysis for the samples with replicates were carried out using edgeR [72]. Samples were grouped together with their cognate biological replicates. To estimate differential gene expression, a quasi-likelihood F-test was applied comparing normalized read values from no damage samples to samples exposed to DNA damage. For cada2C267G data, samples were normalized by calculating Reads per Kilobase per Million mapped reads (RPKM) along with a corresponding wild-type control. Fold change for cada2C267G samples was calculated by comparing RPKM values with the wild-type control (in the absence of damage). A gene was defined as up-regulated based on 2 thresholds; in a certain condition if the log2 fold change (from control) of the gene is greater than or equal to 1 (i.e., the gene content has doubled in this condition) and if the false discovery rate across replicates was less than 0.05. Similarly, a gene was down-regulated in that condition if the gene’s log2 fold change from control was less than −1 and the false discovery rate across replicates is less than 0.05. DGE analysis was done using Google Colab with R version 4.2.3 (2023-03-15). Western blotting Western blotting was performed as described in a previous study [73]. Briefly, overnight cultures in PYE media was back-diluted to 0.025 OD600 and grown at 30°C for 3 h. At approximately 0.1OD600, 1.5 mM MMS was added to the culture; 0, 2, and 4 h post introduction of DNA damage, cells corresponding to 0.6 OD600 were harvested. The culture was pelleted following which supernatant was discarded. The pelleted cells were resuspended in 200 μl of lysis buffer [500 mM Tris-HCL (pH 6.8), 8% SDS, 40% glycerol, 2-Mercaptoethanol, and sufficient amount of bromophenol blue] and heated at 95°C for 3 min. This was followed by 2 rounds of 30 s vortex spins, each followed by short spins. The resuspended cells were again heated to 95°C post which samples were loaded on to a 10% SDS-PAGE gel and subjected to electrophoresis. The resolved protein bands were then transferred on to a PVDF membrane and probed with anti-flag antibodies for estimation of Cada2 levels. The same samples were also probed with anti-RpoA antibodies in order to verify uniform loading of samples. The blots were developed using SuperSignal West PICO PLUS chemiluminescent substrate. Mass spectrometry Sample collection. Cells overexpressing flag-tagged wild-type cada2 or its mutants (cada2R68A, cada2R114A) from a xylose-inducible promoter in a Δcada2 background were used. Overnight cultures grown in PYE media containing gentamycin was back-diluted to 0.025 OD600 and grown at 30°C for 3 h. At approximately 0.1OD600, 1.5 mM MMS and 0.3% xylose were added to the culture and culture volume corresponding to 0.6 OD600 was collected after 4 h of treatment. The culture was pelleted following which supernatant was discarded. The pelleted cells were resuspended in 200 μl of lysis buffer [500 mM Tris-HCL (pH 6.8), 8% SDS, 40% glycerol, 2-Mercaptoethanol, and sufficient amount of bromophenol blue] and heated at 95°C for 3 min. This was followed by 30 s vortex spin, followed by a short spin. The above 2 steps were repeated twice. The resuspended cells were again heated to 95°C post which samples were loaded on to a 10% SDS-PAGE gel and subjected to electrophoresis. The gel was appropriately resolved and stained with Coomassie dye. Bands corresponding to the size of Cada2-3x-Flag (approximately 37 kDa) were excised from the gel, resuspended in milliQ water, and submitted to the NCBS mass spectrometry facility for further processing. Sample preparation and mass spectrometry. The excised gel fragments were first washed thrice with LC-MS water post which the supernatant water was discarded. Next, the gel pieces were chopped followed by destaining with 1:1 100 mM Triethyl ammonium bicarbonate (TEAB) with 100% acetonitrile (ACN). Post destaining, the supernatant was discarded. Approximately 200 μl of 100% ACN was then added and the sample was kept at room temperature till the gel pieces shrunk, following which the supernatant was discarded. The previous step of ACN addition was repeated for a second time. The gel pieces were allowed to dry for a few minutes at room temperature. Next, the gel was suspended in a minimum of 500 ng of Trypsin with 100 to 200 μl of 100 mM TEAB. The samples were then kept for overnight trypsin digestion at 37°C, and 100 μl of 100% ACN with 0.1% formic acid was then introduced to the digested sample and the digested sample were sonicated for 5 min. The supernatant was then collected into a fresh tube. The post-digestion step was repeated for a second time. The collected supernatant was then dried using a SpeedVac vacuum concentrator and reconstituted in 0.1% formic acid. This was followed by desalting post which samples were injected into Orbitrap Fusion Tribrid Mass Spectrometer coupled to a Thermo EASY nanoLC 1200 chromatographic system for mass spectrometry. The results were analyzed using the PeakStudio 8.0 for identifying peptide fragments and their associated PTMs (if any). Chromatin immunoprecipitation sequencing (ChIP-Seq) Fifty ml PYE was sub-inoculated with overnight Caulobacter crescentus cultures to OD600 of 0.025, cultures were grown at 30°C for another 3 h with shaking at 250 rpm until reaching OD600 of 0.1. MMS (final concentration of 1.5 mM) was then added, and the cultures were outgrowth for another 2 or 4 h before fixation with formaldehyde. Controls of no MMS treatment were also included. Cells were fixed with formaldehyde (final concentration of 1%) at room temperature for 30 min, then quenched with 0.125 M glycine for another 15 min at room temperature. Cells were washed 3 times with 1× PBS (pH 7.4) and resuspended in 1 ml of lysis buffer [20 mM K-HEPES (pH 7.9), 50 mM KCl, 10% glycerol, and Roche EDTA-free protease inhibitors]. ChIP-seq was performed as described previously in Tran and colleagues [74]. Briefly, the cell suspension was sonicated on ice using a Soniprep 150 probe-type sonicator (11 cycles, 15s on, 15s off, at setting 8) to shear the chromatin to below 1 kb, and the cell debris was cleared by centrifugation (20 min at 13,000 rpm at 4°C). The supernatant was then transferred to a new 2 ml tube and the buffer conditions were adjusted to 10 mM Tris-HCl (pH 8), 150 mM NaCl, and 0.1% NP-40. Fifty microliters of the supernatant were transferred to a separate tube for control (the input fraction) and stored at −20°C. Meanwhile, antibodies-coupled beads were washed off storage buffers before being added to the above supernatant. α-M2 FLAG antibodies coupled to sepharose beads (Merck, United Kingdom) were employed for ChIP-seq of Cada2-FLAG and RpoC-FLAG. Briefly, 100 μl α-FLAG beads were washed off the storage buffer by repeated centrifugation and resuspension in IPP150 buffer [10 mM Tris-HCl (pH 8), 150 mM NaCl, and 0.1% NP-40]. Beads were then introduced to the cleared supernatant and incubated with gentle shaking at 4°C overnight. Beads were then washed 5 times at 4°C for 2 min each with 1 ml of IPP150 buffer, then twice at 4°C for 2 min each in 1× TE buffer [10 mM Tris-HCl (pH 8) and 1 mM EDTA]. Protein-DNA complexes were then eluted twice from the beads by incubating the beads first with 150 μl of the elution buffer [50 mM Tris-HCl (pH 8), 10 mM EDTA, and 1% SDS] at 65°C for 15 min, then with 100 μl of 1X TE buffer + 1% SDS for another 15 min at 65°C. The supernatant (the ChIP fraction) was then separated from the beads and further incubated at 65°C overnight to completely reverse the crosslink. The input fraction was also de-crosslinked by incubation with 200 μl of 1X TE buffer + 1% SDS at 65°C overnight. DNA from the ChIP and input fraction were then purified using the PCR purification kit (Qiagen) according to the manufacturer’s instruction, then eluted out in 40 μl water. Purified DNA was then made into libraries suitable for Illumina sequencing using the NEXT Ultra II library preparation kit (NEB, UK). ChIP libraries were sequenced on the Illumina Hiseq 2500 at the Tufts University Genomics facility. ChIP-seq analysis Raw reads for the ChIP-seq results were obtained as fastq files. BWA was used in order to align the reads to the Caulobacter genome similar to analysis of RNA-seq data. The aligned reads in.bam format were converted into a .bed file format using the bamtobed option of Bedtools. Coverage at each nucleotide position was calculated using the Bedtools genomecov command. Subsequently, ChIP-seq profiles from these coverage files were plotted using a custom python script. Pearson correlation between ChIP-seq profiles for cada2 flag and rpoC flag was calculated using Numpy [75]. Peakzilla was used in order to identify bona fide enrichment peaks [76], and.bed files from cada2 3xflag or rpoC 3xflag ChIP-seq under MMS exposure were compared to untagged wild type cells treated under identical conditions for peak-calling. ChIP and quantitative PCR analysis For quantitative-PCR (q-PCR) analysis of cada1 and cada3 promoter enrichment in ChIP experiments, purified DNA from both input and ChIP fractions was diluted 1:4 in water and 1 μl was used for qPCR using a SYBR Green JumpStart Taq ReadyMix (CAT S4438, Merck, UK) and a BioRad CFX96 instrument. Fold enrichments were calculated using the comparative Ct method (ΔΔCt) and represent the relative abundance of cada1 and cada3 promoter DNA compared to rpoD DNA as a negative control. All fold enrichment values represent the average of 3 biological replicates. The following oligos were used in qPCR reactions: cada1 (fw, TAGGACGCGACTGCTGA, rv, TCCTTTCGTGAGGAGACCA) cada3 (fw, ATCGCCCGCATGGAATAC, rv, AGAAGGAAGCTACTACCGGAT) rpoD (fw, TCAGGCCAAGAAGGAAATGG, rv, GCCTTCATCAGGCCGATATT). Survival assay Overnight cultures of Caulobacter strains were back-diluted to OD600 0.1. The cultures were incubated at 30°C for 3 h. All cultures were then normalized to 0.3 OD600 and serially diluted in 10-fold increments (10−1–10−8), and 6 μl of each dilution was spotted on PYE plates containing appropriate chemicals (DNA damaging agents/antibiotics/inducer). The plates were incubated at 30°C for 48 h post which survival of the respective strains was determined by counting the number of spots. Computational predictions of protein structures via AlphaFold Three-dimensional structural predictions for EcAda, Cada1, Cada2, and Cada3 were made by ColabFold (AlphaFold2 coupled with MMSeq2 hosted on Google Colaboratory) [56,57]. Primary sequence of each protein was used as query sequence. Default parameters were used for other settings. Structural models shown in this paper are models with the highest pLDDT score. Computational analysis of protein conservation All “complete” and “latest” (assembly_summary.txt; as of January 2017) genome information files for approximately 6,000 bacteria were downloaded from the NCBI ftp website using in-house scripts for whole-genome sequences (.fna), protein coding nucleotide sequences (.fna), RNA sequences (.fna), and protein sequences (.faa). All the organisms were assigned respective phylum based on the KEGG classification (https://www.genome.jp/kegg/genome.html; as of May 2018). For identification of Ada domains and Ada variants, initial blastp was run using each of the 4 protein domain sequences of EcAda from Escherichia coli MG1655 as the query sequence against the UniprotKB database with an E-value cutoff of 0.0001. The 4 domains of Ada include A-box, B-box, RNAse-like, and repair. The top 1,000 full-length sequence hits were downloaded from UniProt for all the 4 domains. A domain multiple sequence alignment (MSA) was made using phmmer–A option with the top 1,000 hits as the sequence database and E. coli domain sequences as the query. An hmm profile was built using the hmmbuild command for the MSA obtained in the previous step. To find domain homologs, hmmsearch command with an E-value cutoff of 0.0001 was used with the hmm profile as the query against a database of 5,973 bacterial genome sequences. These homolog searches were done using HMMER package v3.3. Different Ada variants were identified based on the assignment of the different combinations of Ada domains to the same protein. A phylogenetic species tree was constructed using 16S rRNA sequences. One sequence per genome was extracted from a multi-fasta file. An MSA was built using muscle v3.8.31 with default options, followed by alignment trimming using BMGE v1.12. Using IQTREE v1.6.5, a maximum likelihood-based phylogeny was built with the best model chosen using ModelFinder (-m MF option) against 285 other models. Branch supports were assessed using both 1,000 ultrafast bootstrap approximations (-bb 1,000 –bnni option) and SH-like approximate likelihood ratio test (-alrt 1,000 option). Final tree for visualization was pruned to contain 4 major phyla—Proteobacteria, Actinobacteria, Firmicutes, and Bacteroidetes. Online tool Itol [77] was used for visualization and EcAda and Cada2 presence/absence was overlaid on the phylogeny. For identification of essential regulatory motifs of Cada2 and EcAda, protein blast was run on the coding sequences of EcAda or Cada2 against the UniprotKB database with an E-value cutoff of 10. The top 1,000 full-length sequence hits were downloaded from UniProt for both proteins and were submitted to the MEME discovery program in order to identify conserved features. Bacterial strains and growth conditions All strains, plasmids, and oligos used in this study are listed in S1–S3 Tables, respectively. Caulobacter crescentus cells were grown at 30°C in peptone yeast extract (PYE) media (0.2% peptone, 0.1% yeast extract, and 0.06% MgSO4) and supplemented with antibiotics or inducers, as required, at suitable concentrations. For induction of cada2 expression, 0.3% xylose was introduced into the cultures, unless otherwise stated. Fluorescence microscopy and image analysis Imaging was performed on an epifluorescence, wide-field microscope (Eclipse Ti-2E, Nikon) with a 60×/1.4 NA oil immersion objective and a motorized stage. pE-4000 (CoolLED) was used as the LED excitation source. Exposure times for all yfp samples (λ = 490nm) was 500 ms and exposure was maintained at 50% of LED power. Images were captured using a Hamamatsu Orca Flash 4.0 camera. Focus was ensured via an infrared-based Perfect Focusing System (Nikon). Samples were prepared as detailed in [69]. For time course microscopy, 1 ml of cultures were aliquoted at various time points, pelleted and resuspended in appropriate volume of growth media, and 2 μl of the resuspended cells were spotted on 1% agarose (Invitrogen ultrapure) pads and imaged. For time lapse microscopy, 2 μl of culture were spotted on 1.5% low-melting GTG agarose pads supplemented with PYE media and 1.5 mM MMS damage. Throughout the duration of the time lapse, samples were grown in an OkoLab incubation chamber maintained at 30°C and imaged at regular intervals. Cell segmentation and fluorescence intensity analysis was carried out via Oufti software [70]. RNA-sequencing Sample collection. Overnight cultures of Caulobacter cells were back-diluted to 0.025 OD600 and grown at 30°C for 3 h. When OD600 of culture was approximately 0.1, DNA damage (MMS (1.5 mM)/MMC (0.25 μg/ml)/norfloxacin (8 μg/ml)) was introduced. Samples from 4 biological replicates were collected for no damage conditions and 2 biological replicates for samples in the presence of damage, and 2 ml of Caulobacter cells were harvested and spun at 10,000 g for 5 min. Supernatant was then discarded and the cell pellets were snap frozen with liquid nitrogen and stored at –80°C until the samples were further processed. Cells were harvested at 0, 20, and 40 min post DNA damage exposure. RNA-extraction. RNA was extracted from the cell pellets according to a previous study [58]. Cells were lysed using 400 μl of pre-heated Trizol in a thermomixer at 65°C, 2000 rpm. The lysed cells were transferred to –80°C for a minimum of 30 min and were then centrifuged at 4°C. The supernatant was carefully transferred to 100% ethanol of equivalent volume. RNA was then extracted via the protocol mentioned in the Direct-zol RNA MiniPrep (Zymo, Cat. no. R2052) kit. The extracted RNA was then subjected to DNAse treatment in order to remove any genomic DNA from the sample. Total RNA from the samples was purified using the RNA Clean & Concentrator-25 (Zymo, Cat. no. R1018) kit. Integrity of the RNA was tested using Bioanalyzer instrument. mRNA was isolated from the total sample using the RiboMinus kit (Thermo, Cat. no. K155004) and submitted for RNA-seq at the NCBS next generation sequencing facility. Details regarding kits used for RNA library preparation and the sequencing platform used for the RNA-seq experiment are summarized in S4 Table. Sample collection. Overnight cultures of Caulobacter cells were back-diluted to 0.025 OD600 and grown at 30°C for 3 h. When OD600 of culture was approximately 0.1, DNA damage (MMS (1.5 mM)/MMC (0.25 μg/ml)/norfloxacin (8 μg/ml)) was introduced. Samples from 4 biological replicates were collected for no damage conditions and 2 biological replicates for samples in the presence of damage, and 2 ml of Caulobacter cells were harvested and spun at 10,000 g for 5 min. Supernatant was then discarded and the cell pellets were snap frozen with liquid nitrogen and stored at –80°C until the samples were further processed. Cells were harvested at 0, 20, and 40 min post DNA damage exposure. RNA-extraction. RNA was extracted from the cell pellets according to a previous study [58]. Cells were lysed using 400 μl of pre-heated Trizol in a thermomixer at 65°C, 2000 rpm. The lysed cells were transferred to –80°C for a minimum of 30 min and were then centrifuged at 4°C. The supernatant was carefully transferred to 100% ethanol of equivalent volume. RNA was then extracted via the protocol mentioned in the Direct-zol RNA MiniPrep (Zymo, Cat. no. R2052) kit. The extracted RNA was then subjected to DNAse treatment in order to remove any genomic DNA from the sample. Total RNA from the samples was purified using the RNA Clean & Concentrator-25 (Zymo, Cat. no. R1018) kit. Integrity of the RNA was tested using Bioanalyzer instrument. mRNA was isolated from the total sample using the RiboMinus kit (Thermo, Cat. no. K155004) and submitted for RNA-seq at the NCBS next generation sequencing facility. Details regarding kits used for RNA library preparation and the sequencing platform used for the RNA-seq experiment are summarized in S4 Table. RNA-sequencing analysis Raw reads for the sequencing results were obtained as fastq files. The reference genome sequence (.fna) and annotation (.gff) files for the Caulobacter crescentus NA1000 strain (accession number: NC_011916.1) were downloaded from the NCBI file transfer protocol (ftp) website (“ftp.ncbi.nlm.nih.gov”). The raw read quality was checked using the FastQC software (version 0.11.5). Burrows–Wheeler Aligner (BWA) (version 0.7.17-r1188) was used to index the reference genome. Reads with raw read quality > = 20 were aligned to the indexed genome using the BWA aln -q option. Samtools (version 0.1.7) was used to filter out multi-mapped reads. Bedtools (version 2.26.0) was used to calculate the reads count per gene using the annotation file (.bed) [71]. The normalization and differential gene expression analysis for the samples with replicates were carried out using edgeR [72]. Samples were grouped together with their cognate biological replicates. To estimate differential gene expression, a quasi-likelihood F-test was applied comparing normalized read values from no damage samples to samples exposed to DNA damage. For cada2C267G data, samples were normalized by calculating Reads per Kilobase per Million mapped reads (RPKM) along with a corresponding wild-type control. Fold change for cada2C267G samples was calculated by comparing RPKM values with the wild-type control (in the absence of damage). A gene was defined as up-regulated based on 2 thresholds; in a certain condition if the log2 fold change (from control) of the gene is greater than or equal to 1 (i.e., the gene content has doubled in this condition) and if the false discovery rate across replicates was less than 0.05. Similarly, a gene was down-regulated in that condition if the gene’s log2 fold change from control was less than −1 and the false discovery rate across replicates is less than 0.05. DGE analysis was done using Google Colab with R version 4.2.3 (2023-03-15). Western blotting Western blotting was performed as described in a previous study [73]. Briefly, overnight cultures in PYE media was back-diluted to 0.025 OD600 and grown at 30°C for 3 h. At approximately 0.1OD600, 1.5 mM MMS was added to the culture; 0, 2, and 4 h post introduction of DNA damage, cells corresponding to 0.6 OD600 were harvested. The culture was pelleted following which supernatant was discarded. The pelleted cells were resuspended in 200 μl of lysis buffer [500 mM Tris-HCL (pH 6.8), 8% SDS, 40% glycerol, 2-Mercaptoethanol, and sufficient amount of bromophenol blue] and heated at 95°C for 3 min. This was followed by 2 rounds of 30 s vortex spins, each followed by short spins. The resuspended cells were again heated to 95°C post which samples were loaded on to a 10% SDS-PAGE gel and subjected to electrophoresis. The resolved protein bands were then transferred on to a PVDF membrane and probed with anti-flag antibodies for estimation of Cada2 levels. The same samples were also probed with anti-RpoA antibodies in order to verify uniform loading of samples. The blots were developed using SuperSignal West PICO PLUS chemiluminescent substrate. Mass spectrometry Sample collection. Cells overexpressing flag-tagged wild-type cada2 or its mutants (cada2R68A, cada2R114A) from a xylose-inducible promoter in a Δcada2 background were used. Overnight cultures grown in PYE media containing gentamycin was back-diluted to 0.025 OD600 and grown at 30°C for 3 h. At approximately 0.1OD600, 1.5 mM MMS and 0.3% xylose were added to the culture and culture volume corresponding to 0.6 OD600 was collected after 4 h of treatment. The culture was pelleted following which supernatant was discarded. The pelleted cells were resuspended in 200 μl of lysis buffer [500 mM Tris-HCL (pH 6.8), 8% SDS, 40% glycerol, 2-Mercaptoethanol, and sufficient amount of bromophenol blue] and heated at 95°C for 3 min. This was followed by 30 s vortex spin, followed by a short spin. The above 2 steps were repeated twice. The resuspended cells were again heated to 95°C post which samples were loaded on to a 10% SDS-PAGE gel and subjected to electrophoresis. The gel was appropriately resolved and stained with Coomassie dye. Bands corresponding to the size of Cada2-3x-Flag (approximately 37 kDa) were excised from the gel, resuspended in milliQ water, and submitted to the NCBS mass spectrometry facility for further processing. Sample preparation and mass spectrometry. The excised gel fragments were first washed thrice with LC-MS water post which the supernatant water was discarded. Next, the gel pieces were chopped followed by destaining with 1:1 100 mM Triethyl ammonium bicarbonate (TEAB) with 100% acetonitrile (ACN). Post destaining, the supernatant was discarded. Approximately 200 μl of 100% ACN was then added and the sample was kept at room temperature till the gel pieces shrunk, following which the supernatant was discarded. The previous step of ACN addition was repeated for a second time. The gel pieces were allowed to dry for a few minutes at room temperature. Next, the gel was suspended in a minimum of 500 ng of Trypsin with 100 to 200 μl of 100 mM TEAB. The samples were then kept for overnight trypsin digestion at 37°C, and 100 μl of 100% ACN with 0.1% formic acid was then introduced to the digested sample and the digested sample were sonicated for 5 min. The supernatant was then collected into a fresh tube. The post-digestion step was repeated for a second time. The collected supernatant was then dried using a SpeedVac vacuum concentrator and reconstituted in 0.1% formic acid. This was followed by desalting post which samples were injected into Orbitrap Fusion Tribrid Mass Spectrometer coupled to a Thermo EASY nanoLC 1200 chromatographic system for mass spectrometry. The results were analyzed using the PeakStudio 8.0 for identifying peptide fragments and their associated PTMs (if any). Sample collection. Cells overexpressing flag-tagged wild-type cada2 or its mutants (cada2R68A, cada2R114A) from a xylose-inducible promoter in a Δcada2 background were used. Overnight cultures grown in PYE media containing gentamycin was back-diluted to 0.025 OD600 and grown at 30°C for 3 h. At approximately 0.1OD600, 1.5 mM MMS and 0.3% xylose were added to the culture and culture volume corresponding to 0.6 OD600 was collected after 4 h of treatment. The culture was pelleted following which supernatant was discarded. The pelleted cells were resuspended in 200 μl of lysis buffer [500 mM Tris-HCL (pH 6.8), 8% SDS, 40% glycerol, 2-Mercaptoethanol, and sufficient amount of bromophenol blue] and heated at 95°C for 3 min. This was followed by 30 s vortex spin, followed by a short spin. The above 2 steps were repeated twice. The resuspended cells were again heated to 95°C post which samples were loaded on to a 10% SDS-PAGE gel and subjected to electrophoresis. The gel was appropriately resolved and stained with Coomassie dye. Bands corresponding to the size of Cada2-3x-Flag (approximately 37 kDa) were excised from the gel, resuspended in milliQ water, and submitted to the NCBS mass spectrometry facility for further processing. Sample preparation and mass spectrometry. The excised gel fragments were first washed thrice with LC-MS water post which the supernatant water was discarded. Next, the gel pieces were chopped followed by destaining with 1:1 100 mM Triethyl ammonium bicarbonate (TEAB) with 100% acetonitrile (ACN). Post destaining, the supernatant was discarded. Approximately 200 μl of 100% ACN was then added and the sample was kept at room temperature till the gel pieces shrunk, following which the supernatant was discarded. The previous step of ACN addition was repeated for a second time. The gel pieces were allowed to dry for a few minutes at room temperature. Next, the gel was suspended in a minimum of 500 ng of Trypsin with 100 to 200 μl of 100 mM TEAB. The samples were then kept for overnight trypsin digestion at 37°C, and 100 μl of 100% ACN with 0.1% formic acid was then introduced to the digested sample and the digested sample were sonicated for 5 min. The supernatant was then collected into a fresh tube. The post-digestion step was repeated for a second time. The collected supernatant was then dried using a SpeedVac vacuum concentrator and reconstituted in 0.1% formic acid. This was followed by desalting post which samples were injected into Orbitrap Fusion Tribrid Mass Spectrometer coupled to a Thermo EASY nanoLC 1200 chromatographic system for mass spectrometry. The results were analyzed using the PeakStudio 8.0 for identifying peptide fragments and their associated PTMs (if any). Chromatin immunoprecipitation sequencing (ChIP-Seq) Fifty ml PYE was sub-inoculated with overnight Caulobacter crescentus cultures to OD600 of 0.025, cultures were grown at 30°C for another 3 h with shaking at 250 rpm until reaching OD600 of 0.1. MMS (final concentration of 1.5 mM) was then added, and the cultures were outgrowth for another 2 or 4 h before fixation with formaldehyde. Controls of no MMS treatment were also included. Cells were fixed with formaldehyde (final concentration of 1%) at room temperature for 30 min, then quenched with 0.125 M glycine for another 15 min at room temperature. Cells were washed 3 times with 1× PBS (pH 7.4) and resuspended in 1 ml of lysis buffer [20 mM K-HEPES (pH 7.9), 50 mM KCl, 10% glycerol, and Roche EDTA-free protease inhibitors]. ChIP-seq was performed as described previously in Tran and colleagues [74]. Briefly, the cell suspension was sonicated on ice using a Soniprep 150 probe-type sonicator (11 cycles, 15s on, 15s off, at setting 8) to shear the chromatin to below 1 kb, and the cell debris was cleared by centrifugation (20 min at 13,000 rpm at 4°C). The supernatant was then transferred to a new 2 ml tube and the buffer conditions were adjusted to 10 mM Tris-HCl (pH 8), 150 mM NaCl, and 0.1% NP-40. Fifty microliters of the supernatant were transferred to a separate tube for control (the input fraction) and stored at −20°C. Meanwhile, antibodies-coupled beads were washed off storage buffers before being added to the above supernatant. α-M2 FLAG antibodies coupled to sepharose beads (Merck, United Kingdom) were employed for ChIP-seq of Cada2-FLAG and RpoC-FLAG. Briefly, 100 μl α-FLAG beads were washed off the storage buffer by repeated centrifugation and resuspension in IPP150 buffer [10 mM Tris-HCl (pH 8), 150 mM NaCl, and 0.1% NP-40]. Beads were then introduced to the cleared supernatant and incubated with gentle shaking at 4°C overnight. Beads were then washed 5 times at 4°C for 2 min each with 1 ml of IPP150 buffer, then twice at 4°C for 2 min each in 1× TE buffer [10 mM Tris-HCl (pH 8) and 1 mM EDTA]. Protein-DNA complexes were then eluted twice from the beads by incubating the beads first with 150 μl of the elution buffer [50 mM Tris-HCl (pH 8), 10 mM EDTA, and 1% SDS] at 65°C for 15 min, then with 100 μl of 1X TE buffer + 1% SDS for another 15 min at 65°C. The supernatant (the ChIP fraction) was then separated from the beads and further incubated at 65°C overnight to completely reverse the crosslink. The input fraction was also de-crosslinked by incubation with 200 μl of 1X TE buffer + 1% SDS at 65°C overnight. DNA from the ChIP and input fraction were then purified using the PCR purification kit (Qiagen) according to the manufacturer’s instruction, then eluted out in 40 μl water. Purified DNA was then made into libraries suitable for Illumina sequencing using the NEXT Ultra II library preparation kit (NEB, UK). ChIP libraries were sequenced on the Illumina Hiseq 2500 at the Tufts University Genomics facility. ChIP-seq analysis Raw reads for the ChIP-seq results were obtained as fastq files. BWA was used in order to align the reads to the Caulobacter genome similar to analysis of RNA-seq data. The aligned reads in.bam format were converted into a .bed file format using the bamtobed option of Bedtools. Coverage at each nucleotide position was calculated using the Bedtools genomecov command. Subsequently, ChIP-seq profiles from these coverage files were plotted using a custom python script. Pearson correlation between ChIP-seq profiles for cada2 flag and rpoC flag was calculated using Numpy [75]. Peakzilla was used in order to identify bona fide enrichment peaks [76], and.bed files from cada2 3xflag or rpoC 3xflag ChIP-seq under MMS exposure were compared to untagged wild type cells treated under identical conditions for peak-calling. ChIP and quantitative PCR analysis For quantitative-PCR (q-PCR) analysis of cada1 and cada3 promoter enrichment in ChIP experiments, purified DNA from both input and ChIP fractions was diluted 1:4 in water and 1 μl was used for qPCR using a SYBR Green JumpStart Taq ReadyMix (CAT S4438, Merck, UK) and a BioRad CFX96 instrument. Fold enrichments were calculated using the comparative Ct method (ΔΔCt) and represent the relative abundance of cada1 and cada3 promoter DNA compared to rpoD DNA as a negative control. All fold enrichment values represent the average of 3 biological replicates. The following oligos were used in qPCR reactions: cada1 (fw, TAGGACGCGACTGCTGA, rv, TCCTTTCGTGAGGAGACCA) cada3 (fw, ATCGCCCGCATGGAATAC, rv, AGAAGGAAGCTACTACCGGAT) rpoD (fw, TCAGGCCAAGAAGGAAATGG, rv, GCCTTCATCAGGCCGATATT). Survival assay Overnight cultures of Caulobacter strains were back-diluted to OD600 0.1. The cultures were incubated at 30°C for 3 h. All cultures were then normalized to 0.3 OD600 and serially diluted in 10-fold increments (10−1–10−8), and 6 μl of each dilution was spotted on PYE plates containing appropriate chemicals (DNA damaging agents/antibiotics/inducer). The plates were incubated at 30°C for 48 h post which survival of the respective strains was determined by counting the number of spots. Computational predictions of protein structures via AlphaFold Three-dimensional structural predictions for EcAda, Cada1, Cada2, and Cada3 were made by ColabFold (AlphaFold2 coupled with MMSeq2 hosted on Google Colaboratory) [56,57]. Primary sequence of each protein was used as query sequence. Default parameters were used for other settings. Structural models shown in this paper are models with the highest pLDDT score. Computational analysis of protein conservation All “complete” and “latest” (assembly_summary.txt; as of January 2017) genome information files for approximately 6,000 bacteria were downloaded from the NCBI ftp website using in-house scripts for whole-genome sequences (.fna), protein coding nucleotide sequences (.fna), RNA sequences (.fna), and protein sequences (.faa). All the organisms were assigned respective phylum based on the KEGG classification (https://www.genome.jp/kegg/genome.html; as of May 2018). For identification of Ada domains and Ada variants, initial blastp was run using each of the 4 protein domain sequences of EcAda from Escherichia coli MG1655 as the query sequence against the UniprotKB database with an E-value cutoff of 0.0001. The 4 domains of Ada include A-box, B-box, RNAse-like, and repair. The top 1,000 full-length sequence hits were downloaded from UniProt for all the 4 domains. A domain multiple sequence alignment (MSA) was made using phmmer–A option with the top 1,000 hits as the sequence database and E. coli domain sequences as the query. An hmm profile was built using the hmmbuild command for the MSA obtained in the previous step. To find domain homologs, hmmsearch command with an E-value cutoff of 0.0001 was used with the hmm profile as the query against a database of 5,973 bacterial genome sequences. These homolog searches were done using HMMER package v3.3. Different Ada variants were identified based on the assignment of the different combinations of Ada domains to the same protein. A phylogenetic species tree was constructed using 16S rRNA sequences. One sequence per genome was extracted from a multi-fasta file. An MSA was built using muscle v3.8.31 with default options, followed by alignment trimming using BMGE v1.12. Using IQTREE v1.6.5, a maximum likelihood-based phylogeny was built with the best model chosen using ModelFinder (-m MF option) against 285 other models. Branch supports were assessed using both 1,000 ultrafast bootstrap approximations (-bb 1,000 –bnni option) and SH-like approximate likelihood ratio test (-alrt 1,000 option). Final tree for visualization was pruned to contain 4 major phyla—Proteobacteria, Actinobacteria, Firmicutes, and Bacteroidetes. Online tool Itol [77] was used for visualization and EcAda and Cada2 presence/absence was overlaid on the phylogeny. For identification of essential regulatory motifs of Cada2 and EcAda, protein blast was run on the coding sequences of EcAda or Cada2 against the UniprotKB database with an E-value cutoff of 10. The top 1,000 full-length sequence hits were downloaded from UniProt for both proteins and were submitted to the MEME discovery program in order to identify conserved features. Supporting information S1 Fig. A methylation-specific DNA damage response in Caulobacter. (A) Heat map of log2FC values for genes up-regulated in wild-type cells following MMS, MMC, or Norfloxacin treatment. (B) [Left] Representative cells showing Pccna_00746-yfp reporter induction upon exposure to STZ, MMC, norfloxacin, and HU. [Right] Violin plots show fluorescence intensity distribution normalized to cell area from single cells (n = 300, from 3 biological replicates). The underlying data are available in S1 Data. (C) [Left] Representative cells showing Pccna_00746-yfp reporter induction in ΔdriD background under 1.5 mM MMS damage. [Right] Violin plots showing fluorescence intensity distribution normalized to cell area from single cells (n = 300, from 3 biological replicates). The underlying data are available in S1 Data. (D) [Top] Schematic of the experimental protocol for testing the adaptive property of the Caulobacter methylation-specific damage response. Cultures of Pccna_00746-yfp cells were exposed to a sublethal dose of 0.5 mM either MMS (adapted) or no MMS (non-adapted). The cells were subsequently exposed to a higher dose of 1.5 mM MMS on agarose pads supplemented with PYE medium. [Bottom] Normalized fluorescence intensity kinetics was measured via time lapse microscopy over 3 h of lethal MMS exposure. Dotted lines (in dark) indicate mean time trace of induction kinetics while the shaded region (in light) indicates the standard deviation of all time traces for the respective conditions (here and for all other time lapse data) (n = 25). The underlying data are available in S1 Data. (E) Survival assay of individual deletions of cada genes with and without MMS exposure (1.5 mM). https://doi.org/10.1371/journal.pbio.3002540.s001 (TIF) S2 Fig. Cada2 associates with adaptive response promoters in a sequence-specific manner. (A) Survival assay of cada2-3x-flag strain in the presence or absence of STZ damage (5 μg/ml). (B) Western blot showing Cada2-3x-flag levels at 0, 2, and 4 h after 1.5 mM MMS exposure. As a loading control, RpoA is probed. (C) ChIP-seq profiles for wild type (control), Cada2-3x-Flag, or RpoC-3x-Flag ±2.5 kb around cada2 CDS before (no damage) and after (+damage) exposure to 1.5 mM MMS. (D) Zoomed-in ChIP-seq profiles for Cada2-3x-Flag and RpoC-3x-Flag ±2.5 kb around cada2 CDS before and after exposure to 1.5 mM MMS. https://doi.org/10.1371/journal.pbio.3002540.s002 (TIF) S3 Fig. Cada2-like proteins are widespread and encode a novel DNA-binding domain. (A) (Top) Domain organization and predicted Alphafold structure of Cada2 and EcAda is indicated. (Bottom) Closeup of the EcAda and Cada2 regulatory domain reveals that the Cada2 B-box binding domain (possessing the “SPFHQR” amino acid sequence) and the newly identified sequence-specific binding domain of Cada2 (possessing the “RLHD” amino acid sequence) are part of a helix-turn-helix domain similar to the B-box binding domain of EcAda. The position of these conserved motifs are highlighted and labeled as a ball-and-stick in the overall ribbon representation of the models. (B) Western blot of flag-tagged Cada2 mutants (Cada2R68A and Cada2R114A) overexpressed in a Δcada2 strain from a xylose-inducible promoter treated with MMS damage. (C) Survival assay of cada2R68A and cada2R114A with or without STZ damage. (D) [Left] Prevalence of EcAda-like, Cada2-like, and AdaA-like proteins estimated from a curated, nonredundant database of bacterial genomes is analyzed at the genus level. Numbers represent the presence of these proteins for the major bacterial clades. [Right] domain organization of the respective adaptive response regulatory proteins analyzed here. (E) [Left] As (D) for a curated, nonredundant database of bacterial genomes at the species level. [Right] Prevalence of other adaptive response methyltransferases estimated from a curated, nonredundant database of bacterial genomes is analyzed at the species level. (F) Table represents presence and absence of EcAda-like and Cada2-like proteins and their co-occurrence. In the instances of Cada2-EcAda co-occurrence, potential regulatory circuits predicted via identifying Cada2 and EcAda binding motif in their cognate promoters are shown. https://doi.org/10.1371/journal.pbio.3002540.s003 (TIF) S4 Fig. Cada2 is a methylation-dependent transcription factor. (A) Bacterial two-hybrid assay to test interaction between Cada2 and RNA polymerase subunits. T18-Cada2 was tested for interaction with T25-RNA polymerase holoenzyme subunits with and without 1.5 mM MMS. The presence of red colonies indicated positive interaction. As a positive control AddA (T18) and AddB (T25) are used, and empty vectors (T18 and T25) are used as negative control. Representative images from 2 independent repeats are shown. (B) [Left] Representative cells showing Pccna_00746-yfp reporter induction in wild type and cada2267A mutant background under 1.5 mM MMS damage. [Right] Violin plots showing fluorescence intensity distribution normalized to cell area from single cells in the presence (dark) and absence of damage (light) (n = 200, from 2 biological replicates). Wild-type data are represented again from Fig 1B. The underlying data are available in S1 Data. (C) Survival assay of the cada2 mutants (cada2C267A or cada2C267G) with and without STZ damage (5 μg/ml). (D) [Above] Table representing peptides corresponding to the Cada2 methyltransferase domain bearing the “PCHR” motif as identified via mass spectrometry. In the absence of MMS, peptides corresponding to wild-type Cada2-flag are unmethylated. Upon exposure to MMS damage, peptides corresponding to wild type as well as mutant Cada2-3x-Flag exhibit methylation at the Cys267 residue. [Below] Representative mass spectrometry fragmentation patterns for peptides in the above table. In case of wild type (no damage) representative spectrum from 3 fragments across 2 biological replicates is shown. No methylation modification on Cys267 was detected. In case of Cada2R68A representative spectrum from 4 fragments across 2 biological replicates is shown. Methylation modification on Cys267 was detected in 3 out of 4 fragments. In case of Cada2R114A representative spectrum from 7 fragments across 2 biological replicates is shown. Methylation modification on Cys267 was detected in 5 out of 7 fragments. (E) Heat maps represent log2FC values for Cada2 regulon genes in wild type and cada2C267G cells under MMS exposure. https://doi.org/10.1371/journal.pbio.3002540.s004 (TIF) S5 Fig. Regulatory features of Cada2 are conserved across bacteria and distinct from the EcAda paradigm. (A) Survival assay of Δcada2 strain overexpressing cada2caulo or cada2myxo (from xylose-inducible promoter) under methylation damage (5 μg/ml STZ). Survival of these strains was compared to a control strain comprising of an empty vector in a Δcada2 background. https://doi.org/10.1371/journal.pbio.3002540.s005 (TIF) S1 Table. Strains used in present study. https://doi.org/10.1371/journal.pbio.3002540.s006 (DOCX) S2 Table. Plasmids used in present study. https://doi.org/10.1371/journal.pbio.3002540.s007 (DOCX) S3 Table. Oligos used in present study. https://doi.org/10.1371/journal.pbio.3002540.s008 (DOCX) S4 Table. Kits used for preparing RNA libraries and sequencing platform used for the RNA-seq experiment. https://doi.org/10.1371/journal.pbio.3002540.s009 (DOCX) S1 Data. Numerical data underlying graphs and plots shown in main and supplementary figures. https://doi.org/10.1371/journal.pbio.3002540.s010 (XLSX) S1 Raw Images. Uncropped images for western blot in S2B and S3B Figs. https://doi.org/10.1371/journal.pbio.3002540.s011 (PDF) Acknowledgments We thank Dr. Asha Joseph for contribution to RNA-sequencing experiments and Kaustav Mitra for carrying out initial bacterial-two-hybrid experiments. We are grateful to members of the AB lab for helpful comments and discussions. We acknowledge support from NCBS Mass Spectrometry facility and the Next Generations Sequencing facility for key experiments.

Johnson, Christopher J.;Razy-Krajka, Florian;Zeng, Fan;Piekarz, Katarzyna M.;Biliya, Shweta;Rothbächer, Ute;Stolfi, Alberto

doi: 10.1371/journal.pbio.3002555pmid: 38478577

Introduction Tunicates, the sister group to the vertebrates, comprise a diverse group of marine non-vertebrate chordates [1,2]. Most tunicate species are classified in the order Ascidiacea, commonly known as ascidians [3], although phylogenetic evidence suggests this is not a monophyletic group within Tunicata [4–6]. The majority of ascidians have a biphasic life cycle that alternates between a swimming larva and a sessile adult. The larva functions exclusively to disperse the species, not feeding until it has found a suitable location on which to settle and trigger metamorphosis [7]. Recent work has started to reveal the cellular and molecular basis of larval settlement and metamorphosis. Key to the process of settlement and metamorphosis are the papillae, which comprise a set of 3 anterior sensory/adhesive organs in the laboratory model species of the genus Ciona and a majority of other ascidian genera as well (Fig 1) [8–11]. The papillae are composed of a few different cell types that have been characterized by both electron and fluorescence microscopy [9,12–14]. Several cells appear to secrete the “glue” or bioadhesive material required for the attachment of the larva to the substrate, termed “collocytes” [9,15]. Other cells are clearly neuronal (4 ciliated neurons per papilla) [9] and are required to trigger the onset of metamorphosis [16], which was also recently shown to depend on mechanical stimulation of the papillae [17]. Finally, at the very center of each papilla are 4 “Axial Columnar Cells” (ACCs), which have been suggested to possess chemosensory and contractile properties [11,18,19]. Although they have been called papilla “sensory cells” or “neurons,” they are not innervated and have little structural and molecular overlap with the other 2 cell types. Furthermore, single-cell RNA sequencing (scRNAseq) revealed that they do not express genes typically associated with neuronal function [20]. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 1. Development of the papillae of Ciona. (A) Diagram showing the early cell lineages that give rise to the papillae. The papillae invariantly derive from Foxc+ cells in the anterior neural plate, more specifically the anterior daughter cells of “Row 6” of the neural plate, which activate Foxg downstream of Foxc. Foxg is also activated in the posterior daughter cells of “Row 5,” which go on to give rise to part of the OSP. Numbers in each cell indicate their invariant identity according to the Conklin cell lineage nomenclature. Black bars indicate sibling cells born from the same mother cell. (B) Diagram of what is currently known about the later lineage and fates of the Foxg+ “Anterior Row 6” cells shown in panel A. As the cells divide mediolaterally, some cells up-regulate Sp6/7/8 and down-regulate Foxg (gray cells). Those cells that maintain Foxg expression turn on Islet and coalesce as 3 clusters of cells (pink with green outline): 1 medial, more ventral cluster, and 2 left/right, more dorsal clusters. Later, these 3 clusters organize the territory into the 3 protruding papillae of the larva, which contains several cell types described in detail by TEM [9]. Dashed cell outlines indicate uncertain number/provenance of cells. A-P: anterior-posterior. D-V: dorsal-ventral. Lineages and gene networks are based mostly on: [21,22,86,87]. OSP, oral siphon primordium; TEM, transmission electron microscopy. https://doi.org/10.1371/journal.pbio.3002555.g001 In Ciona, previous work had established that the 3 papillae likely arise from 3 clusters of Foxg+/Islet+ cells arranged roughly as a triangle—2 dorsal clusters (left and right) and single ventral cluster [21,22]. Although Foxg is initially activated in an entire row of cells at the very anterior of the neural plate, Sp6/7/8 (also known as Zfp220 or Buttonhead) is required to refine this swath of expression down to 3 “spots” of Foxg, which is required for expression of Islet in these cell clusters (Fig 1) [21]. MEK/ERK (e.g., MAPK) signaling also appears to play an important role in this refinement, as treatment with the MEK inhibitor U0126 results in a “U”-shaped band of Islet expression instead of 3 discrete foci (Fig 1) [22]. Similarly, BMP inhibition also causes a similar “U-shape” swath of Foxg/Islet expression, resulting in a single protrusion instead of the normal 3, termed the “cyrano” phenotype [23,24]. However, it has not been shown how these early specification events connect to the final cell type diversity and arrangement of the papillae. Here, we describe novel genetic markers and reporter constructs that allowed us to visualize each of the different cell type of the papillae and follow their development upon various molecular perturbations targeting specific transcription factors or signaling pathways. We show that different transcription factors contribute to the specification of the different cell types and that cell-cell signaling in the FGF/MAPK and Delta/Notch pathways are crucial for patterning and arranging these cells in the 3 papillae. Altering papilla development in different ways contributes to different processes of post-settlement larval body plan rearrangements, revealing the complex molecular and cellular underpinning of tunicate larval metamorphosis. Methods Ciona handling Ciona robusta (intestinalis Type A) were shipped from San Diego (M-REP), while Ciona intestinalis (Type B) were shipped from Roscoff Biological Station, France. Eggs were fertilized in vitro, dechorionated, and electroporated following established protocols [25–27]. Staging at different temperatures was estimated based on the published C. robusta developmental table from TUNICANATO [28]. Unc-76 tags were used as a default for fluorescent proteins (FPs) for optimal cell labeling as previously described [29], which excludes the FPs from the nucleus and ensures transport down axons. Typically, 40 to 100 μg of untagged or Unc-76-tagged FP plasmids and 10 to 35 μg of histone (H2B) fusion FP plasmids were used per 700 μl of electroporation solution. For CRISPR, typically 35 to 40 μg of Cas9 plasmid and 25 to 40 μg of each gRNA plasmid was used per 700 μl of electroporation solution, except when validating single-chain guide RNAs (sgRNAs) (see further below). For Sp6/7/8, Pou4, and Foxg, the 2 sgRNAs validated for each gene (S3 Fig) were used in combination. For Villin, all 3 validated sgRNAs were used in combination, while Tuba3 only had 1 validated sgRNA. For Islet, most experiments used only the Islet.2 sgRNA, unless otherwise specified. Precise electroporation mixes for given perturbation experiments and controls are specified in the S1 File. C. robusta embryos were raised at 20 °C and C. intestinalis embryos were raised at 18 °C, unless otherwise specified. For U0126 treatment, U0126 stock solution resuspended in DMSO was diluted to 10 μm final concentration in artificial seawater prior to transferring embryos at stage 16 (approximately 7.5 hpf). Negative control embryos were transferred to seawater with the equivalent volume of DMSO vehicle. For DMH1 treatment, concentrated stock solution was diluted to 2.5 μm final concentration in artificial seawater prior to transferring embryos at stage 10 (4 hpf), as previously established [23]. Fixation, staining, imaging, scoring, and statistical analyses Embryos and larvae were fixed for fluorescent protein imaging in MEM-FA fixation solution (3.7% formaldehyde, 0.1 M MOPS (pH 7.4), 0.5 M NaCl, 1 mM EGTA, 2 mM MgSO4, 0.1% Triton-X100), rinsed in 1× PBS, 0.4% Triton-X100, 50 mM NH4Cl, and 1× PBS, 0.1% Triton-X100. For mRNA in situ hybridization, embryos/larvae were fixed in MEM-PFA fixation solution (4% paraformaldehyde, 0.1 M MOPS (pH 7.4), 0.5 M NaCl, 1 mM EGTA, 2 mM MgSO4, 0.05% Tween-20) and in situ hybridization was carried out as previously described [30,31]. All probe template sequences are shown in the S1 File. Immunolabeling of Flag-tag (DYKDDDDK), β-galactosidase, mCherry (alone or in conjunction with mRNA in situ hybridization) was carried out as previously described [32], on embryos/larvae using mouse anti-DYKDDDDK Tag (Thermo Fisher catalog number MA1-91878, 1:1,000), mouse anti-β-gal (Promega catalog number Z3781, 1:1,000), and rabbit anti-mCherry (BioVision, accession number ACY24904, 1:500) primary antibodies. Specimens were mounted in 50% glycerol/1X PBS/2% DABCO mounting solution on slides with double-sided tape spacing between the slide and coverslip and imaged on Leica DM IL LED or DMI8 inverted epifluorescence microscopes, with maximum Z projection processing and cell measurements performed in LAS X. PNA staining was carried out on 4% PFA fixed larvae, using Tris-buffered saline (pH 8.0) supplemented with 5 mM CaCl2 and 0.1% Triton X-100 (TBS-T). Unspecific background was blocked by 3% BSA in TBS-T for 2 h at room temperature. Biotinylated peanut agglutinin (PNA; Vector Laboratories, B-1075) was diluted in BSA-TBS-T to a final concentration of 25 μg/ml and applied to the specimen overnight at 4 °C. After several washes in TBS-T over 2 h, larvae were incubated for 1 h in fluorescent streptavidin (Vector Laboratories, SA-5006) diluted 1:300 in BSA-TBS-T at room temperature. PNA stainings were mounted in Vectashield (Vector Laboratories, H-1000-10) and imaged using a Leica SP5 II confocal scanning microscope. Stacks were acquired sequentially and z-projected. Images were analyzed with ImageJ (Version 1.52 h). Only Foxc>H2B::mCherry+/lacZ+ embryos, larvae, and juveniles were scored, unless otherwise noted in results or figure and legend. For tests of proportion between 2 groups where there were 2 outcomes, Fisher’s exact test was used, while for tests of proportion between more than 2 groups/outcomes, chi-square test was performed. All results of tests of proportions shown in S4 Data. For continuous variable measurements, see “Quantitative image analyses” subsection below. CRISPR/Cas9 sgRNA design and validation The Cas9 [33] and Cas9::Geminin-Nterminus [34] protein-coding sequences have been described before. sgRNAs were designed using the CRISPOR website [35](crispor.tefor.net). Those sgRNAs with high Doench ‘16 score, high MIT specificity score, and not spanning known SNPs were selected for testing. Validation of sgRNAs was performed by co-electroporation 25 μg of Eef1a>Cas9 or Eef1a>Cas9::Geminin-Nterminus and 75 μg of the sgRNA plasmid, per 700 μl of total electroporation volume. Genomic DNA was extracted from pooled larvae electroporated with a single sgRNA, using the QIAamp DNA micro kit (Qiagen). PCR products spanning each sgRNA target site were amplified from the corresponding genomic DNA, with primers designed so that the amplicon was to be 150 to 450 bp in size. Amplicons were purified by QIAquick PCR purification kit (Qiagen) and submitted for Amplicon-EZ Illumina-based sequencing by Azenta/Genewiz (New Jersey, United States of America), which returned mutagenesis rates and indel plots. CRISPR “rescue” cDNAs for Islet, Foxg, and Sp6/7/8 were designed with silent (i.e., synonymous) point mutations disrupting our sgRNA targets sites and/or their PAMs (see S1 File). In the case of the Islet.2 sgRNA, this one binds to a sequence at an intron/exon boundary and therefore no mutation was needed for the rescue cDNA. RNA sequencing and analysis The scRNAseq data from Cao and colleagues were re-analyzed in Seurat [36]. Combined larva stage data was clustered and plotted using 30 dimensions (S1A Fig). Clusters 3 and 33 were determined to contain papilla cell types and were re-clustered separately, also using 30 dimensions (S1B Fig). Differential gene expression plots (S1C Fig) were explored to find candidate papilla cell type markers, which appeared to be enriched in subclusters 8 and 9 (S1 Data). Some were then confirmed by in situ hybridization (S 1D Fig) and/or reporter plasmids. All code and Seurat files can be downloaded from: https://osf.io/sc7pr/. An alternative filtering and clustering approach (https://osf.io/dbv42) used in parallel to find specifically papilla neurons (PNs) resulted in a different TSNE plot (https://osf.io/6cg4h). From this, clusters 8, 10, and 18 were selected based on known papilla cell type markers and re-clustered, which led to the identification of a new subcluster “10” enriched for both ACC and PN markers. ACC markers (cluster “J”) identified in Sharma and colleagues were subtracted from subcluster 10 markers to generate a list of potential PN-specific markers (https://osf.io/7xqp2). Bulk RNA integrity numbers were determined using the Agilent Bioanalyzer RNA 6000 Nano kit and used as a QC measure. All samples with RINs over 7 were used for library preparation. mRNA was enriched using the NEBNext Poly(A) mRNA isolation module and Illumina compatible libraries were prepared using the NEBNext Ultra II RNA directional library preparation kit. QC on the libraries was performed on the Agilent Bioanalyzer 2100 and concentrations were determined fluorometrically. The libraries were then pooled and sequenced on the NovaSeq 6000 with an SP Flow Cell to get PE100bp reads. The RNA-seq raw files were analyzed in Galaxy hub (usegalaxy.org) [37]. Firstly, the raw fastq files were inspected using FastQC Read Quality Reports (Galaxy Version 0.73+galaxy0) and MultiQC (Galaxy Version 1.11+galaxy0). The reads were then filtered and trimmed with Cutadapt (Galaxy Version 4.0+galaxy0). The minimum read length was set to 20 and the reads that did not meet the quality cutoff of 20 were discarded. Then, FastQC and MultiQC were used again to assess the resulting files after filtering and trimming. Next, the technical replicates were combined and used as the input to the mapping tool (RNA STAR, Galaxy Version 2.7.8a+galaxy0, length of the SA pre-indexing string of 12), together with the custom “KY21” version of the Ciona reference genome sequence and gene models (“Kyoto 2021”, obtained from the Ghost Database; http://ghost.zool.kyoto-u.ac.jp/download_ht.html) [38]. The counts were generated using featureCounts (Galaxy Version 2.0.1+galaxy2; minimum mapping quality per gene was set to 10). Lastly, the differential gene expression analysis (S2 Data) was performed with DESeq2 (Galaxy Version 2.11.40.7+galaxy1). KY21 gene models were linked to KyotoHoya (“KH”) version gene models using the Ciona Gene Model Converter application https://github.com/katarzynampiekarz/ciona_gene_model_converter. Raw sequencing reads available from the SRA database under accession PRJNA949791. Analysis code and files can be found at: https://osf.io/wzrdk/. Quantitative image analyses Larvae subjected to papilla-specific knockout of Islet, Villin, or Tuba3 (using Foxc>Cas9, see S1 File for detailed electroporation recipes) and negative control larvae were fixed at 17 hpf, 20 °C and mounted as above. Islet intron 1 + -473/-9>Unc-76::GFP+ or CryBG>Unc-76::GFP+ cells were imaged using a K3M camera mounted on a Leica DMI8 inverted epifluorescence microscope and the greatest distance between the apical and the basal extremities of each GFP+ papilla (not individual cells) in LAS X, based on visible GFP fluorescence in the ACCs at a given focal plane (see example images with superimposed lines and measurements in S8D Fig). Sometimes, 2 or more papillae were GFP+ in the same larva. In these cases, each papilla was measured independently. Individual papilla length measurements are listed in S3 Data. For analysis of Islet perturbations on Villin reporter expression, fluorescence from Villin -1978/-1>Unc-76::GFP and Foxc>H2B::mCherry reporters were acquired as above but with fixed illumination intensity and exposure times (50 ms for GFP, 100 ms for mCherry). Mean fluorescence intensities in both channels (GFP, mCherry) were measured in mCherry+ areas corresponding to the papillae, and ratio of mean values (GFP/mCherry) was calculated. Ciona handling Ciona robusta (intestinalis Type A) were shipped from San Diego (M-REP), while Ciona intestinalis (Type B) were shipped from Roscoff Biological Station, France. Eggs were fertilized in vitro, dechorionated, and electroporated following established protocols [25–27]. Staging at different temperatures was estimated based on the published C. robusta developmental table from TUNICANATO [28]. Unc-76 tags were used as a default for fluorescent proteins (FPs) for optimal cell labeling as previously described [29], which excludes the FPs from the nucleus and ensures transport down axons. Typically, 40 to 100 μg of untagged or Unc-76-tagged FP plasmids and 10 to 35 μg of histone (H2B) fusion FP plasmids were used per 700 μl of electroporation solution. For CRISPR, typically 35 to 40 μg of Cas9 plasmid and 25 to 40 μg of each gRNA plasmid was used per 700 μl of electroporation solution, except when validating single-chain guide RNAs (sgRNAs) (see further below). For Sp6/7/8, Pou4, and Foxg, the 2 sgRNAs validated for each gene (S3 Fig) were used in combination. For Villin, all 3 validated sgRNAs were used in combination, while Tuba3 only had 1 validated sgRNA. For Islet, most experiments used only the Islet.2 sgRNA, unless otherwise specified. Precise electroporation mixes for given perturbation experiments and controls are specified in the S1 File. C. robusta embryos were raised at 20 °C and C. intestinalis embryos were raised at 18 °C, unless otherwise specified. For U0126 treatment, U0126 stock solution resuspended in DMSO was diluted to 10 μm final concentration in artificial seawater prior to transferring embryos at stage 16 (approximately 7.5 hpf). Negative control embryos were transferred to seawater with the equivalent volume of DMSO vehicle. For DMH1 treatment, concentrated stock solution was diluted to 2.5 μm final concentration in artificial seawater prior to transferring embryos at stage 10 (4 hpf), as previously established [23]. Fixation, staining, imaging, scoring, and statistical analyses Embryos and larvae were fixed for fluorescent protein imaging in MEM-FA fixation solution (3.7% formaldehyde, 0.1 M MOPS (pH 7.4), 0.5 M NaCl, 1 mM EGTA, 2 mM MgSO4, 0.1% Triton-X100), rinsed in 1× PBS, 0.4% Triton-X100, 50 mM NH4Cl, and 1× PBS, 0.1% Triton-X100. For mRNA in situ hybridization, embryos/larvae were fixed in MEM-PFA fixation solution (4% paraformaldehyde, 0.1 M MOPS (pH 7.4), 0.5 M NaCl, 1 mM EGTA, 2 mM MgSO4, 0.05% Tween-20) and in situ hybridization was carried out as previously described [30,31]. All probe template sequences are shown in the S1 File. Immunolabeling of Flag-tag (DYKDDDDK), β-galactosidase, mCherry (alone or in conjunction with mRNA in situ hybridization) was carried out as previously described [32], on embryos/larvae using mouse anti-DYKDDDDK Tag (Thermo Fisher catalog number MA1-91878, 1:1,000), mouse anti-β-gal (Promega catalog number Z3781, 1:1,000), and rabbit anti-mCherry (BioVision, accession number ACY24904, 1:500) primary antibodies. Specimens were mounted in 50% glycerol/1X PBS/2% DABCO mounting solution on slides with double-sided tape spacing between the slide and coverslip and imaged on Leica DM IL LED or DMI8 inverted epifluorescence microscopes, with maximum Z projection processing and cell measurements performed in LAS X. PNA staining was carried out on 4% PFA fixed larvae, using Tris-buffered saline (pH 8.0) supplemented with 5 mM CaCl2 and 0.1% Triton X-100 (TBS-T). Unspecific background was blocked by 3% BSA in TBS-T for 2 h at room temperature. Biotinylated peanut agglutinin (PNA; Vector Laboratories, B-1075) was diluted in BSA-TBS-T to a final concentration of 25 μg/ml and applied to the specimen overnight at 4 °C. After several washes in TBS-T over 2 h, larvae were incubated for 1 h in fluorescent streptavidin (Vector Laboratories, SA-5006) diluted 1:300 in BSA-TBS-T at room temperature. PNA stainings were mounted in Vectashield (Vector Laboratories, H-1000-10) and imaged using a Leica SP5 II confocal scanning microscope. Stacks were acquired sequentially and z-projected. Images were analyzed with ImageJ (Version 1.52 h). Only Foxc>H2B::mCherry+/lacZ+ embryos, larvae, and juveniles were scored, unless otherwise noted in results or figure and legend. For tests of proportion between 2 groups where there were 2 outcomes, Fisher’s exact test was used, while for tests of proportion between more than 2 groups/outcomes, chi-square test was performed. All results of tests of proportions shown in S4 Data. For continuous variable measurements, see “Quantitative image analyses” subsection below. CRISPR/Cas9 sgRNA design and validation The Cas9 [33] and Cas9::Geminin-Nterminus [34] protein-coding sequences have been described before. sgRNAs were designed using the CRISPOR website [35](crispor.tefor.net). Those sgRNAs with high Doench ‘16 score, high MIT specificity score, and not spanning known SNPs were selected for testing. Validation of sgRNAs was performed by co-electroporation 25 μg of Eef1a>Cas9 or Eef1a>Cas9::Geminin-Nterminus and 75 μg of the sgRNA plasmid, per 700 μl of total electroporation volume. Genomic DNA was extracted from pooled larvae electroporated with a single sgRNA, using the QIAamp DNA micro kit (Qiagen). PCR products spanning each sgRNA target site were amplified from the corresponding genomic DNA, with primers designed so that the amplicon was to be 150 to 450 bp in size. Amplicons were purified by QIAquick PCR purification kit (Qiagen) and submitted for Amplicon-EZ Illumina-based sequencing by Azenta/Genewiz (New Jersey, United States of America), which returned mutagenesis rates and indel plots. CRISPR “rescue” cDNAs for Islet, Foxg, and Sp6/7/8 were designed with silent (i.e., synonymous) point mutations disrupting our sgRNA targets sites and/or their PAMs (see S1 File). In the case of the Islet.2 sgRNA, this one binds to a sequence at an intron/exon boundary and therefore no mutation was needed for the rescue cDNA. RNA sequencing and analysis The scRNAseq data from Cao and colleagues were re-analyzed in Seurat [36]. Combined larva stage data was clustered and plotted using 30 dimensions (S1A Fig). Clusters 3 and 33 were determined to contain papilla cell types and were re-clustered separately, also using 30 dimensions (S1B Fig). Differential gene expression plots (S1C Fig) were explored to find candidate papilla cell type markers, which appeared to be enriched in subclusters 8 and 9 (S1 Data). Some were then confirmed by in situ hybridization (S 1D Fig) and/or reporter plasmids. All code and Seurat files can be downloaded from: https://osf.io/sc7pr/. An alternative filtering and clustering approach (https://osf.io/dbv42) used in parallel to find specifically papilla neurons (PNs) resulted in a different TSNE plot (https://osf.io/6cg4h). From this, clusters 8, 10, and 18 were selected based on known papilla cell type markers and re-clustered, which led to the identification of a new subcluster “10” enriched for both ACC and PN markers. ACC markers (cluster “J”) identified in Sharma and colleagues were subtracted from subcluster 10 markers to generate a list of potential PN-specific markers (https://osf.io/7xqp2). Bulk RNA integrity numbers were determined using the Agilent Bioanalyzer RNA 6000 Nano kit and used as a QC measure. All samples with RINs over 7 were used for library preparation. mRNA was enriched using the NEBNext Poly(A) mRNA isolation module and Illumina compatible libraries were prepared using the NEBNext Ultra II RNA directional library preparation kit. QC on the libraries was performed on the Agilent Bioanalyzer 2100 and concentrations were determined fluorometrically. The libraries were then pooled and sequenced on the NovaSeq 6000 with an SP Flow Cell to get PE100bp reads. The RNA-seq raw files were analyzed in Galaxy hub (usegalaxy.org) [37]. Firstly, the raw fastq files were inspected using FastQC Read Quality Reports (Galaxy Version 0.73+galaxy0) and MultiQC (Galaxy Version 1.11+galaxy0). The reads were then filtered and trimmed with Cutadapt (Galaxy Version 4.0+galaxy0). The minimum read length was set to 20 and the reads that did not meet the quality cutoff of 20 were discarded. Then, FastQC and MultiQC were used again to assess the resulting files after filtering and trimming. Next, the technical replicates were combined and used as the input to the mapping tool (RNA STAR, Galaxy Version 2.7.8a+galaxy0, length of the SA pre-indexing string of 12), together with the custom “KY21” version of the Ciona reference genome sequence and gene models (“Kyoto 2021”, obtained from the Ghost Database; http://ghost.zool.kyoto-u.ac.jp/download_ht.html) [38]. The counts were generated using featureCounts (Galaxy Version 2.0.1+galaxy2; minimum mapping quality per gene was set to 10). Lastly, the differential gene expression analysis (S2 Data) was performed with DESeq2 (Galaxy Version 2.11.40.7+galaxy1). KY21 gene models were linked to KyotoHoya (“KH”) version gene models using the Ciona Gene Model Converter application https://github.com/katarzynampiekarz/ciona_gene_model_converter. Raw sequencing reads available from the SRA database under accession PRJNA949791. Analysis code and files can be found at: https://osf.io/wzrdk/. Quantitative image analyses Larvae subjected to papilla-specific knockout of Islet, Villin, or Tuba3 (using Foxc>Cas9, see S1 File for detailed electroporation recipes) and negative control larvae were fixed at 17 hpf, 20 °C and mounted as above. Islet intron 1 + -473/-9>Unc-76::GFP+ or CryBG>Unc-76::GFP+ cells were imaged using a K3M camera mounted on a Leica DMI8 inverted epifluorescence microscope and the greatest distance between the apical and the basal extremities of each GFP+ papilla (not individual cells) in LAS X, based on visible GFP fluorescence in the ACCs at a given focal plane (see example images with superimposed lines and measurements in S8D Fig). Sometimes, 2 or more papillae were GFP+ in the same larva. In these cases, each papilla was measured independently. Individual papilla length measurements are listed in S3 Data. For analysis of Islet perturbations on Villin reporter expression, fluorescence from Villin -1978/-1>Unc-76::GFP and Foxc>H2B::mCherry reporters were acquired as above but with fixed illumination intensity and exposure times (50 ms for GFP, 100 ms for mCherry). Mean fluorescence intensities in both channels (GFP, mCherry) were measured in mCherry+ areas corresponding to the papillae, and ratio of mean values (GFP/mCherry) was calculated. Results Identification of novel markers and reporters for specific cell types in the papillae We searched Ciona robusta (i.e., intestinalis Type A) whole-larva scRNAseq data [39] for evidence of the cell types described by transmission electron microscopy (TEM) of the papillae [9]. While a cell cluster annotated as “Palp Sensory Cells” (PSCs) appeared enriched for known markers of ACCs like CryBG (KH.S605.3) and KH.C3.516 [20,40], genes expressed in other papilla cell types were also enriched in this cluster as well, including Sp6/7/8 (KH.C13.22) [21,22] and Pou4 (KH.C2.42) [16,23]. Re-analysis and re-clustering of these data revealed novel potential markers for different cell types in and around the papillae (S1A–S1C Fig and S1 Data). We performed in situ mRNA hybridization for several of these candidate markers in C. robusta larvae (S1D Fig). As we had hoped, they appeared to label different cells in the papilla territory. Some appeared to label cells in the center of each papilla, while others were expressed in cells surrounding or on the outermost edges of each papilla. These vastly different expression patterns supported the idea of mixed cell identities in the PSC scRNAseq cluster. To further confirm the expression patterns of these and other candidate markers, we made reporter plasmids from their upstream cis-regulatory sequences and electroporated these into Ciona embryos. None of the selected genes showed any appreciable homology to genes of known function in other organisms, but we reasoned that they might serve as useful markers for specific papilla cell types. First, the gene KH.L96.43, predicted to encode a secreted protein with TSP1 repeats and a trypsin-like serine protease domain (S1E Fig and S1 File), was expressed in cells surrounding and in between the 3 papillae (S1D Fig). This pattern was recapitulated by a KH.L96.43 reporter plasmid (“L96.43>GFP,” Fig 2A). Co-electroporation with the papilla-specific Foxg>mCherry reporter [39] showed clear, mutually exclusive expression between the 2 reporters. We propose that L96.43 marks a population of “peri-papillary” and/or “inter-papillary” cells previously identified as “basal cells” that are part of the larger papilla region but excluded from the 3 protruding, Foxg+ papillae sensu stricto [9]. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 2. Novel genetic markers label distinct cell types of the papillae. (A) GFP reporter plasmid (green) constructed using the cis-regulatory sequences from the KH.L96.43 gene labels basal cells in between and surrounding the protruding papillae labeled by Foxg reporter plasmid (pink). (B) TGFB>GFP reporter (green) labels PNs, the axons of which make contacts with BTN axons labeled by a BTN-specific Islet reporter (pink), at 23.5 h hpf, approximately corresponding to Hotta stage 30. (C) A KH.C4.78 reporter (C4.78>GFP) also labels PNs, which are also labeled by Foxg>H2B::mCherry (mCh) reporter (pink nuclei). (D) Lack of overlap between expression of C4.78>GFP (green) and a papilla-specific Islet reporter plasmid (pink nuclei) showing that PNs do not arise from Islet+ cells. (E, F) Co-electroporation of C11.360>GFP (green) with H2B::mCherry reporter plasmids (pink nuclei) indicates these cells come from Foxg-expressing cells that also express Islet. (G) C11.360>mCherry reporter (pink) labels centrally located ICs adjacent to ACCs labeled by CryBG>LacZ reporter (green). (H, I) L141.36>GFP reporter (green) labels OCs that arise from Foxg+ cells (pink nuclei) but do not express Islet (pink nuclei). (J) ICs and OCs are distinct cells as there is no overlap between C11.360 (green) and L141.36 (pink) reporter plasmid expression. (K) Ciona intestinalis (Type B) larva ICs labeled with a reporter plasmid made from the corresponding cis-regulatory sequence of the C. intestinalis Chr11.1038 gene, orthologous to C. robusta KH.C11.360. (L) C. intestinalis larva OCs labeled by a Chr7.130 reporter, corresponding to the C. robusta ortholog KH.L141.36. (M) Summary of the main marker genes and corresponding reporter plasmids used in this study to label different subsets of papilla progenitors and their derivative cell types. All GFP and mCherry reporters fused to the Unc-76 tag, unless specified (see Methods and supplement for details). Weaker Foxg -2863/-3 promoter used in panel A, all other Foxg reporters used the improved Foxg -2863/+54 sequence instead. All Islet reporters shown correspond to the Islet intron 1 + bpFOG>H2B::mCherry plasmid. White channel shows either DAPI (nuclei) and/or larva outline in brightfield, depending on the panel. All C. robusta raised at 20 °C to 18 hpf (roughly st. 28) except: panel B (23.5 hpf, ~st. 30); panels C–F (17 hpf, ~st. 27); panels H–J (20 hpf, ~st. 29). C. intestinalis raised at 18 °C to 20–22 hpf (Hotta stage 28). ACC, axial columnar cell; BTN, bipolar tail neuron; hpf, hours post-fertilization; IC, inner collocyte; OC, outer collocyte; PN, papilla neuron. https://doi.org/10.1371/journal.pbio.3002555.g002 Next, we further confirmed that the PNs are distinct from the ACCs [9]. Previously identified as a potential PN marker by in situ hybridization [41], a TGFB reporter clearly labeled PNs (Fig 2B and S2A Fig), which are distinguished as the only papilla cell types bearing an axon [9]. However, co-electroporation of TGFB reporter with an ACC-specific CryBG reporter [40] appeared to result in “cross-talk,” or cross-plasmid transvection in which a cis-regulatory element in one plasmid activates the transcription of a reporter protein-encoding gene on another, distinct co-electroporated plasmid (S2B Fig). Indeed, other PN-specific reporters tested did not cross-talk with CryBG, including the previously published Gnrh1 reporter [42], and the novel reporter KH.C4.78 (“C4.78>GFP”) (S2C and S2D Fig). KH.C4.78 encodes a predicted transmembrane protein with a single extracellular Sel1-like repeat (S1E Fig and S1 File). Interestingly, PN axons continued to extend posteriorly during the swimming phase to contact the anterior axon branches of the bipolar tail neurons (Fig 2B), which project their posterior axon branches to the very tip of the tail [43]. This hints at a potential mechanism for transducing sensory information from the papillae to the tail tip where tail retraction initiates, especially during later time points when larvae are competent to settle [44]. Double electroporation with KH.C4.78 and Foxg reporters (Fig 2C and S2D Fig) revealed that, unlike the basal cells, PNs are specified from Foxg+ cells in the papillae. However, co-electroporation with a papilla-specific Islet reporter plasmid also revealed that PNs are adjacent to, but distinct from, the central Islet+ “core” of each papilla (Fig 2D and S2D Fig). In contrast, a KH.C11.360 reporter (“C11.360>GFP/mCherry”) labeled cells that were both Foxg+ and Islet+, but were clearly not the ACCs (Fig 2E–2G and S2C Fig). The KH.C11.360 gene encodes a predicted secreted/transmembrane protein with no other recognizable domains or motifs (S1 File). The C11.360+ cells were adjacent to the ACCs but lacked the thin protrusions into the hyaline cap that are typical of the ACCs and also lacked axons typical of the PNs. Therefore, these cells appear to be collocytes, proposed to be adhesive-secreting cells responsible for attachment to the substrate during larval settlement [9]. Previous characterization of the papillae by TEM described 12 collocytes in each papilla [9], yet the C11.360 reporter appeared to only label at most 4 cells per papilla. This suggested the existence of cryptic collocyte subtypes. In fact, those same TEM images showed certain qualitative differences in cytoplasmic contents between peripheral collocytes and the more central collocytes [9]. Indeed, we identified another reporter, that of the gene KH.L141.36 (“L141.36>GFP”), that labeled Foxg+ but Islet-negative cells that are at the periphery of each papilla but that are not PNs as they do not have axons (Fig 2H and 2I, and S2D Fig). KH.L141.36 encodes a predicted transmembrane protein with at least 4 extracellular Sushi/SCR/CCP domains (S1E Fig and S1 File). Co-electroporation of L141.36 and C11.360 reporters labeled mutually exclusive groups of cells (Fig 2J and S2D Fig). We propose that these respective reporters delineate more peripheral, or “outer” collocytes (OCs) versus more central, or “inner” collocytes (ICs). Interestingly, strong KH.L141.36 reporter expression was not visible in early larvae (approximately 17 hpf) like most of the other reporters described, suggesting a later onset of activation. When using these C. robusta reporter plasmids to electroporate the closely related C. intestinalis (i.e., Type B) sourced from Roscoff, France [45], we noticed that their expression was very weak (S2E and S2F Fig). This led us to re-cloning the orthologous sequences from the C. intestinalis Type B genome [46] (S1 File). Percent identity over the alignable portions of these noncoding sequences (disregarding large gaps or insertions) was 89% for C11.360 and 66% for L141.36. Electroporation of Type B embryos with Type B-specific reporter plasmids resulted in much stronger, reliable expression (Fig 2K and 2L). This suggests relatively significant changes to the cis-regulatory sequences of these cell type-specific genes in these otherwise nearly indistinguishable cryptic species. Although we also obtained additional reporters that labeled 1 or more different papilla cell types (S2G and S2H Fig), we now had a full set of papilla cell type-specific marker genes and reporter plasmids for a deeper investigation of papilla patterning and development (Fig 2M). Finally, it is also important to note that some of these reporters are also expressed in cell types outside the papillae (e.g., CryBG in the otolith and KH.C4.78 in the descending decussating neurons of the motor ganglion). Specification of ACCs, ICs, and OCs by Islet and Sp6/7/8 combinatorial logic How are the cell types of the papillae (ACCs, ICs, OCs, and PNs) specified? In situ mRNA hybridization previously revealed partially overlapping expression territories of 3 genes encoding sequence-specific transcription factors (Fig 3A): a central domain of Islet+ cells, surrounded by a ring of cells that express both Islet and Sp6/7/8 (and Emx, though distinct from the earlier expression of Emx at neurula stages), and additional cells surrounding them expressing only Sp6/7/8 [22]. Additionally, overexpression of Islet had been previously shown to generate a single large papilla expressing the ACC reporter CryBG>GFP [22]. We therefore asked whether these transcription factors might be patterning the papillae into an ordered array of cell types (Fig 3A). Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 3. The transcription factor Islet is required for specification of ACCs and ICs. (A) Diagram depicting a partially overlapping expression patterns of Islet and Sp6/7/8, as originally shown by in situ mRNA hybridizations (Wagner and colleagues), and the correlation of these patterns with the later arrangement of ACCs, ICs, and OCs in the papillae. “Late” Emx expression in a ring of cells expressing both Islet and Sp6/7/8 appears to be distinct from earlier Emx expression in Foxg-negative cells (see text and S2 Fig for details). (B) Papilla lineage-specific CRISPR/Cas9-mediated mutagenesis of Islet using Foxc>Cas9 and a the U6>Islet.2 sgRNA plasmid shows reduction of larvae showing expression of reporters labeling ACCs and ICs, but not OCs or PNs. Results compared to a negative “control” condition using a negative control sgRNA (U6>Control, see text for details). Nuclei counterstained with DAPI (white). (C) Scoring data for larvae represented in panel B, averaged between biological duplicates. Foxc>H2B::mCherry+ larvae were scored for quantity of papillae showing visible expression of the corresponding GFP reporter plasmid. Due to mosaic uptake or retention of the plasmids after electroporation, number of papillae with GFP fluorescence is variable and rarely seen in all 3 papillae even in control larvae. Normally larvae have 3 papilla (GFP+ or not), but some mutants have more/fewer than 3. ACC/IC/OC subpanels in panel B at 20 hpf/20 °C (~st. 29), PN subpanels at 21 hpf/20 °C (~st. 29). Same applies to scoring data in Panel C. All experiments were performed in duplicate, with number of embryos ranging from 76 to 100 per condition per replicate. **** p < 0.0001 in both duplicates, as determined by chi-square test. ns = not statistically significant in at least 1 duplicate, also by chi-square test. See S4 Data for the data underlying the graphs and for statistical test details. ACC, axial columnar cell; IC, inner collocyte; OC, outer collocyte; PN, papilla neuron; sgRNA, single-chain guide RNA. https://doi.org/10.1371/journal.pbio.3002555.g003 First we asked, does Islet specify the centrally located ACCs and ICs? To test this, we turned to tissue-specific CRISPR/Cas9-mediated mutagenesis [33]. To knock out Islet in the papillae, we electroporated a previously validated sgRNA expression construct targeting its intron/exon 2 boundary (U6>Islet.2, 44% mutagenesis efficacy, S3 Fig) [47] together with Foxc>Cas9. Papilla-specific CRISPR-based knockout of Islet and resulting loss of ACC cell fate was confirmed by loss of CryBG>GFP expression, compared to negative control individuals electroporated instead with previously published U6>Control sgRNA vector [33] targeting no sequence (Fig 3B and 3C, and S4A Fig). CryBG>GFP activation was rescued by expressing an Islet cDNA that is not targeted by our sgRNAs, demonstrating that its loss is not likely due to off-target CRISPR effects (S5A Fig). Therefore, we conclude that Islet is required for the specification and differentiation of ACCs. A smaller portion of larvae completely lost expression of the IC reporter, C11.360>GFP, but expression was still substantially reduced relative to the control (Fig 3B and 3C). This difference might be due to lower sensitivity of the IC reporter to Islet knockout, or might simply reflect the lower level of mosaicism of C11.360>GFP expression observed in the control. To test whether Islet is required for other cell types of the papillae, we repeated papilla-specific Islet CRISPR knockout using our different reporters to monitor the specification or differentiation of OCs (L141.36>GFP) and PNs (TGFB>GFP). While Islet knockout altered the general morphology of the papillae (see further below), it did not cause significant loss of OC or PN reporter expression (Fig 3B and 3C, and S4A Fig). We therefore conclude that Islet is required for the specification and/or differentiation of ACCs and ICs, but not OCs or PNs. Because it was reported that an outer Emx+ “ring” of Islet+ cells in each papilla co-express Sp6/7/8 [22], we hypothesized that Sp6/7/8 might be required for a fate choice between ACCs and ICs. Corroborating the idea that these outer Islet+ cells are specified as ICs, we cloned an intronic cis-regulatory element from the Emx gene that is sufficient to drive late expression specifically in ICs (S2G Fig). This late ring of Emx expression is not to be confused with the earlier expression of Emx in Foxc+/Foxg-negative cells at the neurula stage [21], which represent a distinct lineage (Fig 1). To test the role of Sp6/7/8 in IC versus ACC fate choice, we used the papilla-specific Islet cis-regulatory element to overexpress Islet or Sp6/7/8. While Islet>Islet did not reduce expression of either reporter (Fig 4A and 4B, with overexpression confirmed by immunostaining for a Flag tag epitope fused to Islet, S2I Fig), Islet>Sp6/7/8 specifically abolished ACC reporter expression, but not that of the IC reporter (Fig 4A and 4B). In fact, IC reporter expression appeared to be slightly expanded in approximately 29% of larvae electroporated with Islet>Sp6/7/8, as assayed by perfect overlap with Islet>H2B::mCherry reporter (S5C Fig). Taken together, these results suggest that overexpression of Sp6/7/8 in the Islet+ cells of the papillae is sufficient to abolish ACC fate and might convert the cells to an IC fate instead. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 4. Specification of ACCs, ICs, and OCs by a combinatorial logic of Islet and Sp6/7/8. (A) Overexpression of Sp6/7/8 (using the Islet>Sp6/7/8 plasmid) in all Islet+ papilla cells results in loss of ACCs (assayed by expression of CryBG>Unc-76::GFP, green), but not of ICs (assayed by expression of C11.360>Unc-76::GFP, green). Islet overexpression (with Islet>Flag::Islet-rescue) does not significantly impact the specification of ACCs or ICs. Larvae at 20 hpf/20 °C (~st. 29). (B) Scoring data showing presence or absence of ICs or ACCs in Foxc>H2B::mCherry+ larvae, as represented in panel A. Experiments were performed in duplicate with 99 or 100 larvae in each duplicate. (C) Cell type specification assayed by reporter plasmid expression (green) in larvae subjected to various Islet and/or Sp6/7/8 perturbation conditions (see main text for details). For ICs and ACCs, the “control” condition is negative control CRISPR (U6>Control), while for OCs it is Foxc>lacZ. Overexpression ACC/IC subpanels are at 18.5 hpf/20 °C (~st. 28), all CRISPR and OC panels at 20 hpf/20 °C (~st. 29). (D) Scoring data for most larvae represented in panel C. Foxc>H2B::mCherry+ larvae were scored for cell type-specific GFP reporter expression that was “heterogeneous” (mixed on/off GFP expression, with all “wild type” patterns of expression falling under this category), “predominant” (ectopic/supernumerary GFP+ cells), “sparse” (reduced frequency/intensity of GFP expression), or “absent” (no GFP visible). (E) IC or ACC reporter (C11.360>Unc-76::GFP, CryBG>Unc-76::GFP) expression scored in Foxc>H2B::mCherry+ larvae represented in top 2 panels of right-most column in C. Experiment was performed and scored in duplicate, with number of larvae in each duplicate ranging from 100 to 105. (F) OC-specific reporter (L141.36>Unc-76::GFP) expression scored in Foxc>H2B::mCherry+ larvae represented by the bottom/right-most subpanel in panel B. Scoring strategy same as in Fig 3. Asterisk denotes when a duplicate of the negative control condition was also used for plots in Fig 3, as multiple CRISPR experiments were performed in parallel. All experiments were performed in duplicate, with number of embryos ranging from 76 to 100 per duplicate. Foxc>Cas9 used for all CRISPR/Cas9 experiments. The Islet cis-regulatory sequence used (panels A and B) was always Islet intron 1 + -473/-9. For overexpression conditions, Foxc>lacZ or Islet>LacZ were used to normalize total amount of DNA (see S1 File for detailed electroporation recipes). All error bars indicate upper and lower limits. **** p < 0.0001 in both duplicates as determined by Fisher’s exact test (panels B and E) or chi-square test (panel F). ns = not statistically significant in at least 1 duplicate. See S4 Data for the data underlying the graphs and statistical test details. ACC, axial columnar cell; IC, inner collocyte; OC, outer collocyte. https://doi.org/10.1371/journal.pbio.3002555.g004 To further show that the combination of Islet and Sp6/7/8 is sufficient to specify IC cell fate, we used the Foxc promoter to drive expression of Islet, Sp6/7/8, or a combination of both in the entire papilla territory. Foxc>Islet alone strongly promoted ACC reporter expression, as previously reported [22], but resulted in more scattered IC reporter expression (Fig 4C and 4D). In contrast, co-electroporation of Foxc>Islet and Foxc>Sp6/7/8 resulted more often in a large, single papilla expressing predominantly the IC reporter, not the ACC reporter (Fig 4C and 4D). Finally, we performed papilla-specific CRISPR knockout of Sp6/7/8, following the same strategy for Islet detailed above, using a combination of 2 new sgRNAs that were designed and validated (S3A Fig). Indeed, CRISPR/Cas9-mediated mutagenesis of Sp6/7/8 in the papilla territory resulted in loss of IC cell fate, as assayed by expression of C11.360>GFP (Fig 4C and 4E). Reduced IC reporter expression was also observed using either individual Sp6/7/8 sgRNAs independently (S4B Fig), and was overcome by Foxc promoter-driven overexpression of an Sp6/7/8 rescue cDNA (S5B Fig), suggesting this effect was not due to CRISPR off-targeting. In contrast, the same perturbation did not diminish the expression of the ACC reporter (Fig 4C and 4E). We noticed that Foxc>Sp6/7/8 alone resulted in a large proportion of larvae lacking either ACC or IC reporter expression (Fig 4D). This suggested the possibility that Sp6/7/8 alone might be promoting another papilla cell fate. Indeed, we found that Sp6/7/8 knockout by CRISPR abolishes the expression of the OC reporter (L141.36>GFP), while Foxc>Sp6/7/8 expands it slightly (Fig 4C, 4D and 4F). In contrast, Foxc>Islet alone or in combination with Foxc>Sp6/7/8 suppressed OC reporter expression (Fig 4C and 4D), while Islet knockout did not affect it, as shown further above (Fig 3B and 3C, and S4A Fig). Taken together, these results suggest that a combinatorial transcriptional logic underlies papilla cell fate choices between ACCs (Islet alone), ICs (Islet + Sp6/7/8), and OCs (Sp6/7/8 alone). Identifying the adhesive-secreting cells of the papillae Previous data revealed PNA staining as a marker for glue-secreting cell granules, the adhesive papillary cap, and adhesive prints left by larvae on the substrate [9,15]. The delineation of 2 collocyte populations opened the question of whether both (ICs and OCs) are equally PNA-positive. To answer this question, we performed PNA stainings on larvae expressing IC or OC reporter plasmids (Fig 5A and 5B). Interestingly, ICs contained PNA-stained granules only at the very apical tip, on top of which the strongest PNA staining is seen extracellularly (Fig 5A and S6 Fig and S1 Movie), while the majority of PNA-stained intracellular granules were not within the ICs at this stage (Fig 5A). Consistently, the OCs were the main cells showing PNA-stained granules located within the papillae (Fig 5B and S6 Fig and S2 Movie). This distribution of PNA staining corresponds to the distribution of granules previously identified by high-pressure freezing electron microscopy [9], in which collocytes located in the central core of the papilla contain granules mostly at their apical end. Indeed, in cross-sections, granules were most abundant inside the papillary body, likely in cells identified here as OCs. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 5. Both types of collocytes contribute to production of adhesive material. (A) PNA-stained granules (pink) are seen in the hyaline cap and the apical tip of ICs (left panel, white arrow) in a C. intestinalis larva labeled by the C. intestinalis C11.360>Unc-76::GFP reporter (green). PNA-stained granules are also seen in cells not labeled by the IC reporter (right subpanel, hollow arrowhead), suggesting they are localized in a different cell type. Left and right subpanels are from different focal planes of the same papilla. (B) OCs labeled with C. robusta L141.36>Unc-76::GFP (green) in a C. robusta larva, with PNA-stained granules (pink) in both apical and basal positions within the cell (white arrows). DAPI in blue. (C) PNA staining (pink) in C. robusta upon overexpression of Sp6/7/8 alone, showing reduction of IC specification as assayed by C11.360>Unc-76::GFP expression (green). Weak PNA staining and GFP expression are still visible in some papillae (solid arrow), but not others (open arrowhead). (D) PNA staining (pink) and C11.360>Unc-76::GFP expression (green) in C. robusta upon overexpression of both Islet and Sp6/7/8, showing expansion of IC fate in a single large papilla (arrow). PNA staining is similarly expanded over the entire IC cluster, confirming that ICs produce the adhesive glue. Foxc>lacZ expression (β-galactosidase immunostaining) shown in blue in both C and D. (E) Scoring of larvae represented in panels C and D, averaged across duplicates. Weak PNA staining is observed upon partial suppression of IC fate, but strong PNA staining is seen upon expansion of supernumerary ICs, confirming that this cell type is one of the major contributors of PNA-positive adhesive glue. Total larvae (duplicate 1) or β-galactosidase+ larvae (duplicate 2) were scored. **** p < 0.0001 in both duplicates, as determined by chi-square test. See S4 Data for sample size, statistical test details, and for the data underlying the graphs. C. intestinalis raised to 20–22 hpf at 18 °C (~st. 28), C. robusta raised to 20 hpf at 20 °C (~st. 29). See Supplemental Movies for full confocal stacks and S6 Fig for single-channel images. IC, inner collocyte; OC, outer collocyte; PNA, peanut agglutinin. https://doi.org/10.1371/journal.pbio.3002555.g005 To further investigate the contributions of both ICs and OCs to glue secretion, we performed PNA staining on larvae in distinct perturbation conditions. Namely, we electroporated larvae with Foxc>Sp6/7/8, which was shown above to suppress IC specification, or with Foxc>Islet and Foxc>Sp6/7/8 combined, which was shown to convert most of the papilla territory into ICs. Although Foxc>Sp6/7/8 eliminated most IC reporter expression, PNA staining was still weakly present (Fig 5C and 5E), likely due to continued presence of OCs. In contrast, Foxc>Islet + Foxc>Sp6/7/8 resulted in a single enlarged papilla with supernumerary ICs, and the entire papilla was often covered by strong PNA staining (Fig 5D and 5E). Taken together, these results suggest that both ICs and OCs contribute to the production of adhesive material, but that the ICs (or their progenitors) are likely the more important contributors. Specification of PNs and OCs from cells that have down-regulated Foxg With the specification of ACCs/ICs/OCs explained in large part due to overlapping expression domains of Islet and Sp6/7/8, the precise developmental origins of the PNs and OCs still remained elusive. While it has become clear that the Islet+ cells at the core of each papilla give rise to ACCs and ICs, Papilla-specific CRISPR knockout of Islet did not abolish PNs or OCs, as shown above (Fig 3). This suggested they do not arise from these core Islet+ cells, consistent with their more lateral positions as shown previously by TEM [9]. Furthermore, recently published in situ hybridization data showing presumptive Pou4-expressing PN precursors surrounding Islet-expressing cells at late tailbud stage [23]. Indeed, co-electroporation of Islet reporter and PN- or OC-specific reporter plasmids clearly showed PNs and OCs immediately adjacent to, but distinct from, Islet+ cells (Fig 2D and 2I, and S2D Fig). Might PNs and OCs be arising from the cells in which Foxg is down-regulated (likely via repression by Sp6/7/8) and that do not go on to express Islet (Fig 6A) [21–23]? To test this, we used the MEK (MAPK kinase) inhibitor U0126 to expand Islet expression as previously done (Fig 6A) [22]. While treatment with 10 μm U0126 at 7.5 hpf (between stages 16 and 17, or late neurula and early tailbud) predictably expanded Islet reporter expression, it also eliminated expression of the PN reporter C4.78>GFP, as well as that of the OC reporter L141.36>GFP (Fig 6B and 6C). These results suggest that Foxg+ papilla cells that maintain Foxg expression go on to express Islet and give rise to ACCs and ICs, while the cells that activate Sp6/7/8 and down-regulate Foxg in response to MAPK signaling go on to give rise to OCs and PNs instead. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 6. Specification of PNs and OCs from Islet-negative cells by MAPK and Notch pathways. (A) Diagram showing effect of MAPK inhibition with the pharmacological MEK inhibitor U0126, based on findings from Wagner and colleagues and Liu and Satou. Inhibition of FGF/MAPK results in expansion of Foxg and Islet from 3 discrete foci to a large “U-shaped” swath, transforming 3 papillae into a single, enlarged papilla (similar results reported with BMP inhibition by Roure and colleagues). (B) The 10 μm U0126 treatment at 7.5 hpf/20 °C (st. 16) results in loss of PNs (assayed by C4.78>Unc-76::GFP at 17 hpf/20 °C, ~st. 27, green) upon expansion of Islet+ cells (pink nuclei), relative to DMSO alone. (C) The same treatment results in loss of OCs (L141.36>Unc-76::GFP at either 17 or 20 hpf/20 °C, ~st. 27–29 green) upon expansion of Islet+ cells (pink nuclei). Both U0126 experiments were performed in duplicate, with 100 larvae per condition per duplicate. (D) Two-color, whole-mount mRNA in situ hybridization for Foxg (green in merged image) and Ascl.a (KH. L9.13, pink). (E) Larva electroporated with Ascl.a>Unc-76::GFP labeling several papilla cells including PNs. (F) Myt1>mNeonGreen labeling PNs and other neurons including CENs. (G) Two-color in situ hybridization of Foxg (green) and Pou4 (pink), the latter labeling adjacent PNs and possibly RTENs. (H) Lineage-specific CRISPR/Cas9-mediated mutagenesis of Pou4 results in loss of PN reporter expression (C4.78>Unc-76::GFP, green). Asterisk denoted background/leaky expression in mesenchyme. Larvae at 17 hpf/20 °C (~st. 27). (I) Scoring of Foxc>H2B::mCherry+ larvae represented in panel H and in Sp6/7/8 CRISPR mutagenesis condition showing Sp6/7/8 does not appear to play a major role in PN reporter expression like Pou4. Experiments repeated in duplicate with 100 larvae in each. (J) Inhibition of Delta/Notch signaling using Foxc>SUH-DBM results in reduced expression of OC reporter (L141.36>Unc-76::GFP, green) at 21 hpf/20 °C (~st. 29). Experiment was repeated in duplicate with 100 larvae in each. (K) Notch inhibition also results in concomitant expansion of supernumerary PNs at 17 hpf/20 °C (~st. 27, labeled by C4.78>Unc-76::GFP, green) relative to Foxc>lacZ control. Experiment was repeated in duplicate, with 42 to 50 larvae in each. (L) Summary diagram and model of effects of Delta/Notch inhibition on PN/OC fate choice in Islet-negative (but formerly Foxg+) papilla progenitor cells. All Islet reporters are the Islet intron 1 + bpFOG>H2B::mCherry. All error bars indicate upper and lower limits. **** p < 0.0001, *** p = 0.0003 in both duplicates as determined by Fisher’s exact test, ns = not statistically significant in at least 1 duplicate. See S4 Data for the data underlying the graphs and for statistical test details. CEN, caudal epidermal neuron; OC, outer collocyte; PN, papilla neuron; RTEN, rostral trunk epidermal neuron. https://doi.org/10.1371/journal.pbio.3002555.g006 PNs are specified by common peripheral neuron regulators Previous papilla-specific TALEN knockout of the neuronal transcription factor-encoding gene Pou4 successfully eliminated PNs and the larva’s tail resorption response to mechanical stimuli [16]. Pou4 has been previously implicated in a Myt1-dependent regulatory cascade that specifies the caudal epidermal neurons (CENs) of the tail, from neurogenic midline cells expressing the proneural bHLH transcription factor Ascl.a (KH.L9.13, sometimes called Ascl2 or Ascl.b previously) [23,48–51]. To precisely visualize the neurogenic cells of the papillae, we performed double (two-color) mRNA in situ hybridization for Ascl.a and Foxg at the mid-tailbud stage. Indeed, Ascl.a expression was seen broadly in the papilla territory surrounding the 3 Foxg+ cell clusters (Fig 6D). This was confirmed by an Ascl.a fluorescent protein reporter plasmid that labeled a broad set of papilla territory cells, including PNs and their axons (Fig 6E). Furthermore, a previously published Myt1 reporter [52] was also found to be expressed in the PNs (Fig 6F). Double in situ of Pou4 and Foxg revealed Pou4+ cells surrounding each Foxg+ cluster, corroborating a recent report [23] (Fig 6G). It was not immediately clear which Pou4+ cells were PN precursors and which were nearby rostral trunk epidermal neuron (RTEN) precursors. Based on our images and those of the most recent study [23], we propose that there are initially 2 Pou4+ cells per papilla, later dividing to give rise to the 4 PNs per papilla as previously described [9]. This would mirror the development of the epidermal neurons of the tail, in which neurons are born side-by-side as pairs after a final cell division by a committed mother cell [49]. Papilla-specific CRISPR knockout of Pou4 with a combination of 2 newly validated sgRNAs (S3D Fig) recapitulated the loss of PN differentiation by the previously published TALEN knockout [16], as assayed by C4.78 and TGFB reporter expression (Fig 6H and 6I, and S4 Fig). In contrast, Pou4 knockout had no effect on the specification of ACCs or OCs, suggesting Pou4 function is specific for PN fate in the papillae (S4 Fig). Taken together, these results suggest that PNs are specified from interspersed neurogenic progenitors that are carved out by MAPK signaling. Interestingly, CRISPR knockout of Sp6/7/8 did not substantially affect PN specification (Fig 6I). This suggests that even though Sp6/7/8 down-regulates Foxg in these cells [21], it does not appear to be required for their neurogenic potential. Notch signaling regulates the fate choice between PNs and OCs Because both OCs and PNs appeared to arise from Foxg-down-regulating, Islet-negative cells, we sought to test whether an additional regulatory step is required for the fate choice between these 2 cell types. In the neurogenic midline territory of the tail epidermis, lateral inhibition by Delta/Notch signaling regulates the final number and spacing of CENs [48,49,53]. Delta/Notch limits the expression of Myt1, which in turn activates Pou4 expression. In the tail epidermis, the major ligand involved is the putative Delta like non-canonical Notch ligand homolog (encoded by gene KH.L50.6), which is also expressed in alternating pattern in the papillae (S7A Fig). We therefore decided to test whether a similar mechanism in controlling the number of PNs and OCs surrounding each papilla. To test the requirement of Delta/Notch, we overexpressed a DNA-binding mutant of the Notch co-factor RBPJ/SUH (SUH-DBM) [54]. Indeed, electroporation with Foxc>SUH-DBM resulted in loss of OC reporter expression (Fig 6J), and concomitant expansion of PN reporter expression (Fig 6K). We conclude that Delta/Notch signaling regulates PN versus OC fate choice in neurogenic progenitor cells surrounding each presumptive papilla, with Notch delimiting the specification of supernumerary neurons, thus allowing OCs to form (Fig 6L). In contrast, SUH-DBM had minimal effect on IC/ACC fate choice (S7B Fig), suggesting Delta/Notch might only regulate neurogenesis, and not cell fate choice in general, in the papilla territory. This common origin of PNs and OCs is also supported by the recent finding that the latter appear to have basal bodies like the PNs, but without the accompanying sensory cilia [9]. Interestingly, papilla-specific knockout of Foxg resulted in moderate loss of PN reporter expression (TGFB>GFP), and very little effect on the OC reporter (S4A Fig). This suggests differing requirement for Foxg in different cell type-specific branches of the papilla regulatory network, despite all these cell types arising from cells that initially express Foxg. Regulation of papilla morphogenesis by Islet It was previously shown that Foxg or Islet overexpression induces the formation of a single enlarged “megapapilla,” in which all cells are substantially elongated relative to the rest of the epidermis [21,22]. We have shown above that this appears to be driven by expansion of ACCs and/or ICs, which are atypically elongated in the apical-basal direction and form apical protrusions and microvilli. Islet is sufficient for apical-basal elongation of epidermal cells [22], and morpholino-knockdown of Foxg (which is upstream of Islet) also impairs proper papilla morphogenesis [21]. We asked if Islet is required for papilla morphogenesis, using papilla-specific CRISPR knockout of Islet. Knocking out Islet in the papilla territory impaired the formation of the typically “pointy-shaped” papillae, resulting instead in blunt cells with flat, broader apical surfaces and reduced cell length along the apical-basal axis (Fig 7A and 7B, S8C and S8D Fig). This result suggested that transcriptional targets downstream of Islet might be regulating the distinct cell shape of ACCs/ICs. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 7. Islet is also required for papilla morphogenesis. (A) Papilla shape is shortened and blunt at the apical end upon tissue-specific CRISPR/Cas9-mediated mutagenesis of Islet. Embryos were electroporated with Islet intron 1 + -473/-9>Unc-76::GFP and Foxc>Cas9. Islet CRISPR was performed using U6>Islet.2 sgRNA plasmid and the negative control used U6>Control. Larvae were imaged at 20 hpf/20 °C (~st. 28). Right: Scoring of percentage of GFP+ larvae classified as having normal “protruding” or blunt papillae, as represented to the left. Experiment was performed and scored in duplicate, using 2 different GFP fusions: Unc-76::GFP and DcxΔC::GFP [88]. Replicate 1: n = 100 for either condition; replicate 2: n = 55 for either condition **** p < 0.0001 in both duplicates as determined by Fisher’s exact test. (B) Quantification of papilla cell (Islet intron 1 +-473/-9>Unc-76::GFP+) lengths along apical-basal axis in negative control and Islet CRISPR larvae at 18 hpf/20 °C (~st. 28). Both Islet.1 and Islet.2 sgRNAs used in combination. Statistical significance tested by unpaired t test (two-tailed). See S8 Fig for duplicate experiment. (C) In situ mRNA hybridization of Villin, showing expression in Foxg+/Islet+ central papilla cells at 10 hpf/20 °C (st. 21, left) and at larval stage (~st. 27, right). (D) Villin -1978/-1>Unc-76::GFP showing expression in the papilla territory of electroporated larvae (~st. 28), strongest in the central cells. (E) Villin -1978/-1>Unc-76::GFP in st. 28 larvae is down-regulated by tissue-specific CRISPR/Cas9 mutagenesis of Islet (Foxc>Cas9 + U6>Islet.1 + U6>Islet.2, see text for details). (F) Quantification of effect of Islet CRISPR (as in panel E) on Villin -1978/-1>Unc 76::GFP/Foxc>H2B::mCherry mean fluorescence intensity ratios in ROIs defined by the mCherry+ nuclei (see Methods for details). Significance determined by Mann–Whitney test (two-tailed). (G) Villin reporter is up-regulated in st. 28 larvae by overexpressing Islet (Foxc>Islet, see text for details). (H) GFP/mCherry ratio quantification done in identical manner as in F, but comparing Islet overexpression (as in panel G) and control lacZ larvae. (I) Quantification of ACC lengths measured in negative control and papilla-specific Villin CRISPR larvae at 17 hpf/20 °C (~st. 27). Significance tested by unpaired t test (two-tailed). Although no statistically significant difference between control and CRISPR larvae was observed in this replicate, average ACC length was significantly shorter in the CRISPR condition in an additional replicate (S8 Fig). (J) mRNA in situ hybridization for Tuba3, showing enrichment in the central cells of the papillae in st. 27 larvae. (K) Tuba3>Unc-76::GFP reporter plasmid is broadly expressed in the papillae of st. 27 larvae but stronger in central cells. (L) Papilla-specific CRISPR knockout of Tuba3 does not result in decrease of average ACC apical-basal cell length compared to negative control CRISPR using U6>Control sgRNA instead. Significance tested by unpaired t test (two-tailed). ns = not significant. All large bars indicate medians and smaller bars indicate interquartile ranges. See S3 and S4 Data for the data underlying the graphs and for statistical test details. ACC, axial columnar cell; ROI, region of interest; sgRNA, single-chain guide RNA. https://doi.org/10.1371/journal.pbio.3002555.g007 To identify potential candidate effectors of morphogenesis downstream of Islet, we used bulk RNAseq to measure differential gene expression between different Islet perturbation conditions. We compared “negative control” embryos to (1) embryos in which Islet was overexpressed in the whole territory using the Foxc promoter (Foxc>Islet); and (2) embryos in which Islet was knocked out specifically in the papilla lineage by CRISPR/Cas9. For this, we designed an additional sgRNA targeting the first exon of Islet, to be used in combination with the already published sgRNA to generate larger deletions. This new sgRNA vector, which we named U6>Islet.1, resulted in a mutagenesis efficacy of 20% (S3B Fig). Whole embryos from each condition were collected at 12 hpf (Islet conditions) at 20 °C in biological triplicate. RNA was extracted from pooled embryos in each sample, and RNAseq libraries were prepared from poly(A)-selected RNAs and sequenced by Illumina NovaSeq. This bulk RNAseq approach revealed that Islet overexpression results in the up-regulation of several ACC markers from previous scRNAseq analysis (S2 Data) [20]. With Islet overexpression, this included ACC markers previously validated by mRNA in situ hybridization or reporter gene expression, such as CryBG (KH.S605.3) and Atp2a (KH.L116.40). Many ACC markers were conspicuously absent, but this may be due to the relatively early time point (12 hpf, late tailbud stage), well before hatching and ACC differentiation. This was a deliberate choice, as we were focused on papilla morphogenesis, which begins around this stage [22]. One resulting candidate Islet target revealed by RNAseq was Astl-related (KH.C9.850), and its expression in the Islet+ cells of the papillae was confirmed by in situ hybridization (S8A Fig). Indeed, Islet knockout by CRISPR eliminated Astl-related reporter expression, supporting our approach to identifying new targets of Islet (S8B Fig). Furthermore, the top up-regulated gene by Islet overexpression (and 17th most down-regulated by Islet CRISPR) was KY21.Chr10.318, which encodes a Fibrillin-related (Fbn) protein. This gene was previously shown to be specifically expressed in the central Islet+ cells by in situ hybridization [23], further validating our approach to identifying putative Islet target genes. One particularly interesting ACC-specific candidate that was among the genes most highly up-regulated by Islet overexpression was Villin (KH.C9.512), an ortholog of the Villin family of genes encoding effectors of actin regulators [55]. The apical extensions of the ACCs are highly enriched for actin filaments and microtubules [9], suggesting that cytoskeletal modulation may be important for the extended length of these cells relative to surrounding cells. We confirmed the expression of Villin in the papillae by in situ hybridization and reporter plasmids (Fig 7C and 7D). In the Islet CRISPR condition, Villin was the top down-regulated gene by Islet CRISPR knockout as well. Villin reporter expression was reduced in intensity but not completely lost upon knockout of Islet by CRISPR (Fig 7E and 7F), yet was dramatically up-regulated by Islet overexpression (Fig 7G and 7H). This suggests partially redundant activation of Villin by another factor, likely at earlier developmental stages (e.g., by Foxc or Foxg), and that Islet might be required for its sustained expression specifically in the central cells of the papilla throughout morphogenesis. This is consistent with the weak but broad expression of Villin>GFP in the entire papilla territory (Fig 7D), and the fact that papilla territory cells are already more elongated than epidermal cells in other parts of the embryo even at earlier stages [22]. To test whether Villin is required for proper morphogenesis of Islet+ cells in the papilla, we performed tissue-specific CRISPR knockout using a combination of 3 validated sgRNAs spanning most of the coding sequence (S3E Fig). Because the functionally important “headpiece” domain is encoded by the last exon, we combined an sgRNA targeting this exon with 2 sgRNAs targeting more upstream exons. In one batch of Villin CRISPR larvae, ACCs were not significantly shorter on average along the apical-basal axis than in control larvae (Fig 7I). However, average ACC length was significantly shorter in CRISPR larvae than control larvae in a duplicate experiment (S8E Fig). This contrast between replicates was found to be entirely due to variability between batches of control larvae, not the Villin CRISPR larvae (S8F Fig). Because the ACCs have been shown to dynamically extend or contract in length, possibly in response to external stimuli [56,57], we suspect that these differences in average length in different batches of control animals are due to as of yet unidentified environmental conditions. In addition to actin regulators, we searched our list of putative Islet targets for microtubule components and regulators, since microtubule bundles were reported in the apical protrusions of the ACCs in Distaplia occidentalis [58]. We identified a gene encoding a divergent Tubulin alpha monomer (Tuba3, KH.C3.736) as one such potential target. Enrichment of Tuba3 expression in the central papillae was confirmed by in situ hybridization (Fig 7J) and a Tuba3 reporter plasmid (Fig 7K). However, papilla-specific CRISPR knockout of Tuba3 did not result in significantly shorter ACCs either (Fig 7L). Taken together, these results suggest that Islet is required for proper papilla morphogenesis, and that this may be due to its role in activating the expression of numerous effector genes. However, knocking out individual candidate effector genes like Villin or Tuba3 has not yet revealed a key role for any one of these putative downstream targets. An investigation into the cell and molecular basis of larval settlement and metamorphosis With our different CRISPR knockouts affecting different cell types of the papillae, we asked how these different perturbations might affect larval metamorphosis. Only the involvement of the PNs in triggering metamorphosis has been demonstrated [16,17], but it is not yet known how the regulatory networks and cell types of the papillae affect different processes during metamorphosis. We performed papilla-specific CRISPR as above using the Foxc>Cas9 vector, targeting the 4 different transcription factors we have shown to be involved in patterning the cell types of the papillae: Pou4, Islet, Foxg, and Sp6/7/8. We assayed tail retraction and body rotation at the last stage of metamorphosis [28] (Fig 8A and 8B), as these are 2 processes that can be uncoupled in certain genetic perturbations or naturally occurring mutants [59]. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 8. Genetic perturbations of metamorphosis. (A) Ciona robusta juvenile undergoing metamorphosis, showing the retracted tail and rotated anterior-posterior body axis (dashed lines). PNs in the former papilla (now substrate attachment stolon, or holdfast) labeled by TGFB>Unc-76::GFP (green). Animal counterstained with DAPI (blue). (B) Scoring of Foxc>H2B::mCherry+ individuals showing tail retraction and/or body rotation at 48 hpf/20 °C in various papilla territory-specific (using Foxc>Cas9) CRISPR-based gene knockouts. Experiments were performed and scored in duplicate and percentages averaged, except for Foxg CRISPR for which a third replicate was performed (see S9 Fig). Number scored individuals in each replicate indicated underneath. “Tailed juveniles” have undergone body rotation but not tail retraction, whereas normally body rotation follows tail retraction. The sgRNA plasmids used for each condition were as follows- Control: U6>Control; Pou4: U6>Pou4.3.21 + U6>Pou4.4.106; Islet: U6>Islet.2; Foxg: U6>Foxg.1.116 + U6>Foxg.5.419; Sp6/7/8: U6>Sp6/7/8.4.29 + U6>Sp6/7/8.8.117. (C) Plot showing lack of any discernable metamorphosis defect after eliminating ACCs using Islet intron 1 + bpFOG>Sp6/7/8 (images not shown). Only Islet intron 1 + bpFOG>H2B::mCherry+ individuals were scored. Experiment was performed and scored in duplicate and averaged (n = 100 each duplicate). ACC specification was scored using the CryBG>Unc-76::GFP reporter. (D) Example of “tailed juveniles” at 47 hpf/20 °C compared to a larva in which no tail retraction or body rotation has occurred, elicited by tissue-specific Foxg CRISPR (Foxc>Cas9 + U6>Foxg.1.116 + U6>Foxg5.419). See S9 Fig for scoring. All error bars denote upper and lower limits. **** p < 0.0001, *** p < 0.0015 in both duplicates, ns = not significant in at least 1 duplicate, as determined by chi-square test comparing to the control conditions. See S4 Data for the data underlying the graphs and for statistical test details. ACC, axial columnar cell; PN, papilla neuron; sgRNA, single-chain guide RNA. https://doi.org/10.1371/journal.pbio.3002555.g008 Knockout of Pou4 recapitulated recent published results on this transcription factor [16]. Namely, both tail retraction and body rotation were blocked in the vast majority of individuals. This suggests that proper specification and/or differentiation of PNs by Pou4 is crucial for the ability of the larva to trigger the onset of metamorphosis. In contrast, Islet knockout did not affect tail retraction, but body rotation appeared somewhat impaired. This suggested that ACCs/ICs are not required for tail retraction, but might play a role in regulating body rotation downstream of it. Eliminating ACCs using Islet>Sp6/7/8 had no effect on either tail retraction or body rotation (Fig 8C), confirming that ACCs are not required for metamorphosis, but that perhaps certain Islet targets might specifically regulate body rotation. Unsurprisingly, Foxg knockout modestly impaired both tail retraction and body rotation (Fig 8B and 8D, and S9 Fig), but also resulted in a noticeable fraction (approximately 19% on average) of “tailed juveniles” in which body rotation begins even in the absence of tail retraction. This unusual effect was seen even when repeating the experiment independently a third time, revealing consistent uncoupling of these 2 processes upon Foxg knockout (S9 Fig). Finally, Sp6/7/8 CRISPR did not substantially alter either tail retraction or body rotation. Taken together, these results paint a more complex picture of regulation of metamorphosis by the papillae. Our findings suggest that different cell types of the papillae might play distinct roles in the regulation of metamorphosis, perhaps interacting with one another to regulate different steps, or that certain transcription factors might be required for the expression of key rate-limiting components of these different processes. Further work will be required to disentangle these different cellular and genetic factors, which we hope will be aided by our cell type-specific reporters and CRISPR reagents. Identification of novel markers and reporters for specific cell types in the papillae We searched Ciona robusta (i.e., intestinalis Type A) whole-larva scRNAseq data [39] for evidence of the cell types described by transmission electron microscopy (TEM) of the papillae [9]. While a cell cluster annotated as “Palp Sensory Cells” (PSCs) appeared enriched for known markers of ACCs like CryBG (KH.S605.3) and KH.C3.516 [20,40], genes expressed in other papilla cell types were also enriched in this cluster as well, including Sp6/7/8 (KH.C13.22) [21,22] and Pou4 (KH.C2.42) [16,23]. Re-analysis and re-clustering of these data revealed novel potential markers for different cell types in and around the papillae (S1A–S1C Fig and S1 Data). We performed in situ mRNA hybridization for several of these candidate markers in C. robusta larvae (S1D Fig). As we had hoped, they appeared to label different cells in the papilla territory. Some appeared to label cells in the center of each papilla, while others were expressed in cells surrounding or on the outermost edges of each papilla. These vastly different expression patterns supported the idea of mixed cell identities in the PSC scRNAseq cluster. To further confirm the expression patterns of these and other candidate markers, we made reporter plasmids from their upstream cis-regulatory sequences and electroporated these into Ciona embryos. None of the selected genes showed any appreciable homology to genes of known function in other organisms, but we reasoned that they might serve as useful markers for specific papilla cell types. First, the gene KH.L96.43, predicted to encode a secreted protein with TSP1 repeats and a trypsin-like serine protease domain (S1E Fig and S1 File), was expressed in cells surrounding and in between the 3 papillae (S1D Fig). This pattern was recapitulated by a KH.L96.43 reporter plasmid (“L96.43>GFP,” Fig 2A). Co-electroporation with the papilla-specific Foxg>mCherry reporter [39] showed clear, mutually exclusive expression between the 2 reporters. We propose that L96.43 marks a population of “peri-papillary” and/or “inter-papillary” cells previously identified as “basal cells” that are part of the larger papilla region but excluded from the 3 protruding, Foxg+ papillae sensu stricto [9]. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 2. Novel genetic markers label distinct cell types of the papillae. (A) GFP reporter plasmid (green) constructed using the cis-regulatory sequences from the KH.L96.43 gene labels basal cells in between and surrounding the protruding papillae labeled by Foxg reporter plasmid (pink). (B) TGFB>GFP reporter (green) labels PNs, the axons of which make contacts with BTN axons labeled by a BTN-specific Islet reporter (pink), at 23.5 h hpf, approximately corresponding to Hotta stage 30. (C) A KH.C4.78 reporter (C4.78>GFP) also labels PNs, which are also labeled by Foxg>H2B::mCherry (mCh) reporter (pink nuclei). (D) Lack of overlap between expression of C4.78>GFP (green) and a papilla-specific Islet reporter plasmid (pink nuclei) showing that PNs do not arise from Islet+ cells. (E, F) Co-electroporation of C11.360>GFP (green) with H2B::mCherry reporter plasmids (pink nuclei) indicates these cells come from Foxg-expressing cells that also express Islet. (G) C11.360>mCherry reporter (pink) labels centrally located ICs adjacent to ACCs labeled by CryBG>LacZ reporter (green). (H, I) L141.36>GFP reporter (green) labels OCs that arise from Foxg+ cells (pink nuclei) but do not express Islet (pink nuclei). (J) ICs and OCs are distinct cells as there is no overlap between C11.360 (green) and L141.36 (pink) reporter plasmid expression. (K) Ciona intestinalis (Type B) larva ICs labeled with a reporter plasmid made from the corresponding cis-regulatory sequence of the C. intestinalis Chr11.1038 gene, orthologous to C. robusta KH.C11.360. (L) C. intestinalis larva OCs labeled by a Chr7.130 reporter, corresponding to the C. robusta ortholog KH.L141.36. (M) Summary of the main marker genes and corresponding reporter plasmids used in this study to label different subsets of papilla progenitors and their derivative cell types. All GFP and mCherry reporters fused to the Unc-76 tag, unless specified (see Methods and supplement for details). Weaker Foxg -2863/-3 promoter used in panel A, all other Foxg reporters used the improved Foxg -2863/+54 sequence instead. All Islet reporters shown correspond to the Islet intron 1 + bpFOG>H2B::mCherry plasmid. White channel shows either DAPI (nuclei) and/or larva outline in brightfield, depending on the panel. All C. robusta raised at 20 °C to 18 hpf (roughly st. 28) except: panel B (23.5 hpf, ~st. 30); panels C–F (17 hpf, ~st. 27); panels H–J (20 hpf, ~st. 29). C. intestinalis raised at 18 °C to 20–22 hpf (Hotta stage 28). ACC, axial columnar cell; BTN, bipolar tail neuron; hpf, hours post-fertilization; IC, inner collocyte; OC, outer collocyte; PN, papilla neuron. https://doi.org/10.1371/journal.pbio.3002555.g002 Next, we further confirmed that the PNs are distinct from the ACCs [9]. Previously identified as a potential PN marker by in situ hybridization [41], a TGFB reporter clearly labeled PNs (Fig 2B and S2A Fig), which are distinguished as the only papilla cell types bearing an axon [9]. However, co-electroporation of TGFB reporter with an ACC-specific CryBG reporter [40] appeared to result in “cross-talk,” or cross-plasmid transvection in which a cis-regulatory element in one plasmid activates the transcription of a reporter protein-encoding gene on another, distinct co-electroporated plasmid (S2B Fig). Indeed, other PN-specific reporters tested did not cross-talk with CryBG, including the previously published Gnrh1 reporter [42], and the novel reporter KH.C4.78 (“C4.78>GFP”) (S2C and S2D Fig). KH.C4.78 encodes a predicted transmembrane protein with a single extracellular Sel1-like repeat (S1E Fig and S1 File). Interestingly, PN axons continued to extend posteriorly during the swimming phase to contact the anterior axon branches of the bipolar tail neurons (Fig 2B), which project their posterior axon branches to the very tip of the tail [43]. This hints at a potential mechanism for transducing sensory information from the papillae to the tail tip where tail retraction initiates, especially during later time points when larvae are competent to settle [44]. Double electroporation with KH.C4.78 and Foxg reporters (Fig 2C and S2D Fig) revealed that, unlike the basal cells, PNs are specified from Foxg+ cells in the papillae. However, co-electroporation with a papilla-specific Islet reporter plasmid also revealed that PNs are adjacent to, but distinct from, the central Islet+ “core” of each papilla (Fig 2D and S2D Fig). In contrast, a KH.C11.360 reporter (“C11.360>GFP/mCherry”) labeled cells that were both Foxg+ and Islet+, but were clearly not the ACCs (Fig 2E–2G and S2C Fig). The KH.C11.360 gene encodes a predicted secreted/transmembrane protein with no other recognizable domains or motifs (S1 File). The C11.360+ cells were adjacent to the ACCs but lacked the thin protrusions into the hyaline cap that are typical of the ACCs and also lacked axons typical of the PNs. Therefore, these cells appear to be collocytes, proposed to be adhesive-secreting cells responsible for attachment to the substrate during larval settlement [9]. Previous characterization of the papillae by TEM described 12 collocytes in each papilla [9], yet the C11.360 reporter appeared to only label at most 4 cells per papilla. This suggested the existence of cryptic collocyte subtypes. In fact, those same TEM images showed certain qualitative differences in cytoplasmic contents between peripheral collocytes and the more central collocytes [9]. Indeed, we identified another reporter, that of the gene KH.L141.36 (“L141.36>GFP”), that labeled Foxg+ but Islet-negative cells that are at the periphery of each papilla but that are not PNs as they do not have axons (Fig 2H and 2I, and S2D Fig). KH.L141.36 encodes a predicted transmembrane protein with at least 4 extracellular Sushi/SCR/CCP domains (S1E Fig and S1 File). Co-electroporation of L141.36 and C11.360 reporters labeled mutually exclusive groups of cells (Fig 2J and S2D Fig). We propose that these respective reporters delineate more peripheral, or “outer” collocytes (OCs) versus more central, or “inner” collocytes (ICs). Interestingly, strong KH.L141.36 reporter expression was not visible in early larvae (approximately 17 hpf) like most of the other reporters described, suggesting a later onset of activation. When using these C. robusta reporter plasmids to electroporate the closely related C. intestinalis (i.e., Type B) sourced from Roscoff, France [45], we noticed that their expression was very weak (S2E and S2F Fig). This led us to re-cloning the orthologous sequences from the C. intestinalis Type B genome [46] (S1 File). Percent identity over the alignable portions of these noncoding sequences (disregarding large gaps or insertions) was 89% for C11.360 and 66% for L141.36. Electroporation of Type B embryos with Type B-specific reporter plasmids resulted in much stronger, reliable expression (Fig 2K and 2L). This suggests relatively significant changes to the cis-regulatory sequences of these cell type-specific genes in these otherwise nearly indistinguishable cryptic species. Although we also obtained additional reporters that labeled 1 or more different papilla cell types (S2G and S2H Fig), we now had a full set of papilla cell type-specific marker genes and reporter plasmids for a deeper investigation of papilla patterning and development (Fig 2M). Finally, it is also important to note that some of these reporters are also expressed in cell types outside the papillae (e.g., CryBG in the otolith and KH.C4.78 in the descending decussating neurons of the motor ganglion). Specification of ACCs, ICs, and OCs by Islet and Sp6/7/8 combinatorial logic How are the cell types of the papillae (ACCs, ICs, OCs, and PNs) specified? In situ mRNA hybridization previously revealed partially overlapping expression territories of 3 genes encoding sequence-specific transcription factors (Fig 3A): a central domain of Islet+ cells, surrounded by a ring of cells that express both Islet and Sp6/7/8 (and Emx, though distinct from the earlier expression of Emx at neurula stages), and additional cells surrounding them expressing only Sp6/7/8 [22]. Additionally, overexpression of Islet had been previously shown to generate a single large papilla expressing the ACC reporter CryBG>GFP [22]. We therefore asked whether these transcription factors might be patterning the papillae into an ordered array of cell types (Fig 3A). Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 3. The transcription factor Islet is required for specification of ACCs and ICs. (A) Diagram depicting a partially overlapping expression patterns of Islet and Sp6/7/8, as originally shown by in situ mRNA hybridizations (Wagner and colleagues), and the correlation of these patterns with the later arrangement of ACCs, ICs, and OCs in the papillae. “Late” Emx expression in a ring of cells expressing both Islet and Sp6/7/8 appears to be distinct from earlier Emx expression in Foxg-negative cells (see text and S2 Fig for details). (B) Papilla lineage-specific CRISPR/Cas9-mediated mutagenesis of Islet using Foxc>Cas9 and a the U6>Islet.2 sgRNA plasmid shows reduction of larvae showing expression of reporters labeling ACCs and ICs, but not OCs or PNs. Results compared to a negative “control” condition using a negative control sgRNA (U6>Control, see text for details). Nuclei counterstained with DAPI (white). (C) Scoring data for larvae represented in panel B, averaged between biological duplicates. Foxc>H2B::mCherry+ larvae were scored for quantity of papillae showing visible expression of the corresponding GFP reporter plasmid. Due to mosaic uptake or retention of the plasmids after electroporation, number of papillae with GFP fluorescence is variable and rarely seen in all 3 papillae even in control larvae. Normally larvae have 3 papilla (GFP+ or not), but some mutants have more/fewer than 3. ACC/IC/OC subpanels in panel B at 20 hpf/20 °C (~st. 29), PN subpanels at 21 hpf/20 °C (~st. 29). Same applies to scoring data in Panel C. All experiments were performed in duplicate, with number of embryos ranging from 76 to 100 per condition per replicate. **** p < 0.0001 in both duplicates, as determined by chi-square test. ns = not statistically significant in at least 1 duplicate, also by chi-square test. See S4 Data for the data underlying the graphs and for statistical test details. ACC, axial columnar cell; IC, inner collocyte; OC, outer collocyte; PN, papilla neuron; sgRNA, single-chain guide RNA. https://doi.org/10.1371/journal.pbio.3002555.g003 First we asked, does Islet specify the centrally located ACCs and ICs? To test this, we turned to tissue-specific CRISPR/Cas9-mediated mutagenesis [33]. To knock out Islet in the papillae, we electroporated a previously validated sgRNA expression construct targeting its intron/exon 2 boundary (U6>Islet.2, 44% mutagenesis efficacy, S3 Fig) [47] together with Foxc>Cas9. Papilla-specific CRISPR-based knockout of Islet and resulting loss of ACC cell fate was confirmed by loss of CryBG>GFP expression, compared to negative control individuals electroporated instead with previously published U6>Control sgRNA vector [33] targeting no sequence (Fig 3B and 3C, and S4A Fig). CryBG>GFP activation was rescued by expressing an Islet cDNA that is not targeted by our sgRNAs, demonstrating that its loss is not likely due to off-target CRISPR effects (S5A Fig). Therefore, we conclude that Islet is required for the specification and differentiation of ACCs. A smaller portion of larvae completely lost expression of the IC reporter, C11.360>GFP, but expression was still substantially reduced relative to the control (Fig 3B and 3C). This difference might be due to lower sensitivity of the IC reporter to Islet knockout, or might simply reflect the lower level of mosaicism of C11.360>GFP expression observed in the control. To test whether Islet is required for other cell types of the papillae, we repeated papilla-specific Islet CRISPR knockout using our different reporters to monitor the specification or differentiation of OCs (L141.36>GFP) and PNs (TGFB>GFP). While Islet knockout altered the general morphology of the papillae (see further below), it did not cause significant loss of OC or PN reporter expression (Fig 3B and 3C, and S4A Fig). We therefore conclude that Islet is required for the specification and/or differentiation of ACCs and ICs, but not OCs or PNs. Because it was reported that an outer Emx+ “ring” of Islet+ cells in each papilla co-express Sp6/7/8 [22], we hypothesized that Sp6/7/8 might be required for a fate choice between ACCs and ICs. Corroborating the idea that these outer Islet+ cells are specified as ICs, we cloned an intronic cis-regulatory element from the Emx gene that is sufficient to drive late expression specifically in ICs (S2G Fig). This late ring of Emx expression is not to be confused with the earlier expression of Emx in Foxc+/Foxg-negative cells at the neurula stage [21], which represent a distinct lineage (Fig 1). To test the role of Sp6/7/8 in IC versus ACC fate choice, we used the papilla-specific Islet cis-regulatory element to overexpress Islet or Sp6/7/8. While Islet>Islet did not reduce expression of either reporter (Fig 4A and 4B, with overexpression confirmed by immunostaining for a Flag tag epitope fused to Islet, S2I Fig), Islet>Sp6/7/8 specifically abolished ACC reporter expression, but not that of the IC reporter (Fig 4A and 4B). In fact, IC reporter expression appeared to be slightly expanded in approximately 29% of larvae electroporated with Islet>Sp6/7/8, as assayed by perfect overlap with Islet>H2B::mCherry reporter (S5C Fig). Taken together, these results suggest that overexpression of Sp6/7/8 in the Islet+ cells of the papillae is sufficient to abolish ACC fate and might convert the cells to an IC fate instead. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 4. Specification of ACCs, ICs, and OCs by a combinatorial logic of Islet and Sp6/7/8. (A) Overexpression of Sp6/7/8 (using the Islet>Sp6/7/8 plasmid) in all Islet+ papilla cells results in loss of ACCs (assayed by expression of CryBG>Unc-76::GFP, green), but not of ICs (assayed by expression of C11.360>Unc-76::GFP, green). Islet overexpression (with Islet>Flag::Islet-rescue) does not significantly impact the specification of ACCs or ICs. Larvae at 20 hpf/20 °C (~st. 29). (B) Scoring data showing presence or absence of ICs or ACCs in Foxc>H2B::mCherry+ larvae, as represented in panel A. Experiments were performed in duplicate with 99 or 100 larvae in each duplicate. (C) Cell type specification assayed by reporter plasmid expression (green) in larvae subjected to various Islet and/or Sp6/7/8 perturbation conditions (see main text for details). For ICs and ACCs, the “control” condition is negative control CRISPR (U6>Control), while for OCs it is Foxc>lacZ. Overexpression ACC/IC subpanels are at 18.5 hpf/20 °C (~st. 28), all CRISPR and OC panels at 20 hpf/20 °C (~st. 29). (D) Scoring data for most larvae represented in panel C. Foxc>H2B::mCherry+ larvae were scored for cell type-specific GFP reporter expression that was “heterogeneous” (mixed on/off GFP expression, with all “wild type” patterns of expression falling under this category), “predominant” (ectopic/supernumerary GFP+ cells), “sparse” (reduced frequency/intensity of GFP expression), or “absent” (no GFP visible). (E) IC or ACC reporter (C11.360>Unc-76::GFP, CryBG>Unc-76::GFP) expression scored in Foxc>H2B::mCherry+ larvae represented in top 2 panels of right-most column in C. Experiment was performed and scored in duplicate, with number of larvae in each duplicate ranging from 100 to 105. (F) OC-specific reporter (L141.36>Unc-76::GFP) expression scored in Foxc>H2B::mCherry+ larvae represented by the bottom/right-most subpanel in panel B. Scoring strategy same as in Fig 3. Asterisk denotes when a duplicate of the negative control condition was also used for plots in Fig 3, as multiple CRISPR experiments were performed in parallel. All experiments were performed in duplicate, with number of embryos ranging from 76 to 100 per duplicate. Foxc>Cas9 used for all CRISPR/Cas9 experiments. The Islet cis-regulatory sequence used (panels A and B) was always Islet intron 1 + -473/-9. For overexpression conditions, Foxc>lacZ or Islet>LacZ were used to normalize total amount of DNA (see S1 File for detailed electroporation recipes). All error bars indicate upper and lower limits. **** p < 0.0001 in both duplicates as determined by Fisher’s exact test (panels B and E) or chi-square test (panel F). ns = not statistically significant in at least 1 duplicate. See S4 Data for the data underlying the graphs and statistical test details. ACC, axial columnar cell; IC, inner collocyte; OC, outer collocyte. https://doi.org/10.1371/journal.pbio.3002555.g004 To further show that the combination of Islet and Sp6/7/8 is sufficient to specify IC cell fate, we used the Foxc promoter to drive expression of Islet, Sp6/7/8, or a combination of both in the entire papilla territory. Foxc>Islet alone strongly promoted ACC reporter expression, as previously reported [22], but resulted in more scattered IC reporter expression (Fig 4C and 4D). In contrast, co-electroporation of Foxc>Islet and Foxc>Sp6/7/8 resulted more often in a large, single papilla expressing predominantly the IC reporter, not the ACC reporter (Fig 4C and 4D). Finally, we performed papilla-specific CRISPR knockout of Sp6/7/8, following the same strategy for Islet detailed above, using a combination of 2 new sgRNAs that were designed and validated (S3A Fig). Indeed, CRISPR/Cas9-mediated mutagenesis of Sp6/7/8 in the papilla territory resulted in loss of IC cell fate, as assayed by expression of C11.360>GFP (Fig 4C and 4E). Reduced IC reporter expression was also observed using either individual Sp6/7/8 sgRNAs independently (S4B Fig), and was overcome by Foxc promoter-driven overexpression of an Sp6/7/8 rescue cDNA (S5B Fig), suggesting this effect was not due to CRISPR off-targeting. In contrast, the same perturbation did not diminish the expression of the ACC reporter (Fig 4C and 4E). We noticed that Foxc>Sp6/7/8 alone resulted in a large proportion of larvae lacking either ACC or IC reporter expression (Fig 4D). This suggested the possibility that Sp6/7/8 alone might be promoting another papilla cell fate. Indeed, we found that Sp6/7/8 knockout by CRISPR abolishes the expression of the OC reporter (L141.36>GFP), while Foxc>Sp6/7/8 expands it slightly (Fig 4C, 4D and 4F). In contrast, Foxc>Islet alone or in combination with Foxc>Sp6/7/8 suppressed OC reporter expression (Fig 4C and 4D), while Islet knockout did not affect it, as shown further above (Fig 3B and 3C, and S4A Fig). Taken together, these results suggest that a combinatorial transcriptional logic underlies papilla cell fate choices between ACCs (Islet alone), ICs (Islet + Sp6/7/8), and OCs (Sp6/7/8 alone). Identifying the adhesive-secreting cells of the papillae Previous data revealed PNA staining as a marker for glue-secreting cell granules, the adhesive papillary cap, and adhesive prints left by larvae on the substrate [9,15]. The delineation of 2 collocyte populations opened the question of whether both (ICs and OCs) are equally PNA-positive. To answer this question, we performed PNA stainings on larvae expressing IC or OC reporter plasmids (Fig 5A and 5B). Interestingly, ICs contained PNA-stained granules only at the very apical tip, on top of which the strongest PNA staining is seen extracellularly (Fig 5A and S6 Fig and S1 Movie), while the majority of PNA-stained intracellular granules were not within the ICs at this stage (Fig 5A). Consistently, the OCs were the main cells showing PNA-stained granules located within the papillae (Fig 5B and S6 Fig and S2 Movie). This distribution of PNA staining corresponds to the distribution of granules previously identified by high-pressure freezing electron microscopy [9], in which collocytes located in the central core of the papilla contain granules mostly at their apical end. Indeed, in cross-sections, granules were most abundant inside the papillary body, likely in cells identified here as OCs. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 5. Both types of collocytes contribute to production of adhesive material. (A) PNA-stained granules (pink) are seen in the hyaline cap and the apical tip of ICs (left panel, white arrow) in a C. intestinalis larva labeled by the C. intestinalis C11.360>Unc-76::GFP reporter (green). PNA-stained granules are also seen in cells not labeled by the IC reporter (right subpanel, hollow arrowhead), suggesting they are localized in a different cell type. Left and right subpanels are from different focal planes of the same papilla. (B) OCs labeled with C. robusta L141.36>Unc-76::GFP (green) in a C. robusta larva, with PNA-stained granules (pink) in both apical and basal positions within the cell (white arrows). DAPI in blue. (C) PNA staining (pink) in C. robusta upon overexpression of Sp6/7/8 alone, showing reduction of IC specification as assayed by C11.360>Unc-76::GFP expression (green). Weak PNA staining and GFP expression are still visible in some papillae (solid arrow), but not others (open arrowhead). (D) PNA staining (pink) and C11.360>Unc-76::GFP expression (green) in C. robusta upon overexpression of both Islet and Sp6/7/8, showing expansion of IC fate in a single large papilla (arrow). PNA staining is similarly expanded over the entire IC cluster, confirming that ICs produce the adhesive glue. Foxc>lacZ expression (β-galactosidase immunostaining) shown in blue in both C and D. (E) Scoring of larvae represented in panels C and D, averaged across duplicates. Weak PNA staining is observed upon partial suppression of IC fate, but strong PNA staining is seen upon expansion of supernumerary ICs, confirming that this cell type is one of the major contributors of PNA-positive adhesive glue. Total larvae (duplicate 1) or β-galactosidase+ larvae (duplicate 2) were scored. **** p < 0.0001 in both duplicates, as determined by chi-square test. See S4 Data for sample size, statistical test details, and for the data underlying the graphs. C. intestinalis raised to 20–22 hpf at 18 °C (~st. 28), C. robusta raised to 20 hpf at 20 °C (~st. 29). See Supplemental Movies for full confocal stacks and S6 Fig for single-channel images. IC, inner collocyte; OC, outer collocyte; PNA, peanut agglutinin. https://doi.org/10.1371/journal.pbio.3002555.g005 To further investigate the contributions of both ICs and OCs to glue secretion, we performed PNA staining on larvae in distinct perturbation conditions. Namely, we electroporated larvae with Foxc>Sp6/7/8, which was shown above to suppress IC specification, or with Foxc>Islet and Foxc>Sp6/7/8 combined, which was shown to convert most of the papilla territory into ICs. Although Foxc>Sp6/7/8 eliminated most IC reporter expression, PNA staining was still weakly present (Fig 5C and 5E), likely due to continued presence of OCs. In contrast, Foxc>Islet + Foxc>Sp6/7/8 resulted in a single enlarged papilla with supernumerary ICs, and the entire papilla was often covered by strong PNA staining (Fig 5D and 5E). Taken together, these results suggest that both ICs and OCs contribute to the production of adhesive material, but that the ICs (or their progenitors) are likely the more important contributors. Specification of PNs and OCs from cells that have down-regulated Foxg With the specification of ACCs/ICs/OCs explained in large part due to overlapping expression domains of Islet and Sp6/7/8, the precise developmental origins of the PNs and OCs still remained elusive. While it has become clear that the Islet+ cells at the core of each papilla give rise to ACCs and ICs, Papilla-specific CRISPR knockout of Islet did not abolish PNs or OCs, as shown above (Fig 3). This suggested they do not arise from these core Islet+ cells, consistent with their more lateral positions as shown previously by TEM [9]. Furthermore, recently published in situ hybridization data showing presumptive Pou4-expressing PN precursors surrounding Islet-expressing cells at late tailbud stage [23]. Indeed, co-electroporation of Islet reporter and PN- or OC-specific reporter plasmids clearly showed PNs and OCs immediately adjacent to, but distinct from, Islet+ cells (Fig 2D and 2I, and S2D Fig). Might PNs and OCs be arising from the cells in which Foxg is down-regulated (likely via repression by Sp6/7/8) and that do not go on to express Islet (Fig 6A) [21–23]? To test this, we used the MEK (MAPK kinase) inhibitor U0126 to expand Islet expression as previously done (Fig 6A) [22]. While treatment with 10 μm U0126 at 7.5 hpf (between stages 16 and 17, or late neurula and early tailbud) predictably expanded Islet reporter expression, it also eliminated expression of the PN reporter C4.78>GFP, as well as that of the OC reporter L141.36>GFP (Fig 6B and 6C). These results suggest that Foxg+ papilla cells that maintain Foxg expression go on to express Islet and give rise to ACCs and ICs, while the cells that activate Sp6/7/8 and down-regulate Foxg in response to MAPK signaling go on to give rise to OCs and PNs instead. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 6. Specification of PNs and OCs from Islet-negative cells by MAPK and Notch pathways. (A) Diagram showing effect of MAPK inhibition with the pharmacological MEK inhibitor U0126, based on findings from Wagner and colleagues and Liu and Satou. Inhibition of FGF/MAPK results in expansion of Foxg and Islet from 3 discrete foci to a large “U-shaped” swath, transforming 3 papillae into a single, enlarged papilla (similar results reported with BMP inhibition by Roure and colleagues). (B) The 10 μm U0126 treatment at 7.5 hpf/20 °C (st. 16) results in loss of PNs (assayed by C4.78>Unc-76::GFP at 17 hpf/20 °C, ~st. 27, green) upon expansion of Islet+ cells (pink nuclei), relative to DMSO alone. (C) The same treatment results in loss of OCs (L141.36>Unc-76::GFP at either 17 or 20 hpf/20 °C, ~st. 27–29 green) upon expansion of Islet+ cells (pink nuclei). Both U0126 experiments were performed in duplicate, with 100 larvae per condition per duplicate. (D) Two-color, whole-mount mRNA in situ hybridization for Foxg (green in merged image) and Ascl.a (KH. L9.13, pink). (E) Larva electroporated with Ascl.a>Unc-76::GFP labeling several papilla cells including PNs. (F) Myt1>mNeonGreen labeling PNs and other neurons including CENs. (G) Two-color in situ hybridization of Foxg (green) and Pou4 (pink), the latter labeling adjacent PNs and possibly RTENs. (H) Lineage-specific CRISPR/Cas9-mediated mutagenesis of Pou4 results in loss of PN reporter expression (C4.78>Unc-76::GFP, green). Asterisk denoted background/leaky expression in mesenchyme. Larvae at 17 hpf/20 °C (~st. 27). (I) Scoring of Foxc>H2B::mCherry+ larvae represented in panel H and in Sp6/7/8 CRISPR mutagenesis condition showing Sp6/7/8 does not appear to play a major role in PN reporter expression like Pou4. Experiments repeated in duplicate with 100 larvae in each. (J) Inhibition of Delta/Notch signaling using Foxc>SUH-DBM results in reduced expression of OC reporter (L141.36>Unc-76::GFP, green) at 21 hpf/20 °C (~st. 29). Experiment was repeated in duplicate with 100 larvae in each. (K) Notch inhibition also results in concomitant expansion of supernumerary PNs at 17 hpf/20 °C (~st. 27, labeled by C4.78>Unc-76::GFP, green) relative to Foxc>lacZ control. Experiment was repeated in duplicate, with 42 to 50 larvae in each. (L) Summary diagram and model of effects of Delta/Notch inhibition on PN/OC fate choice in Islet-negative (but formerly Foxg+) papilla progenitor cells. All Islet reporters are the Islet intron 1 + bpFOG>H2B::mCherry. All error bars indicate upper and lower limits. **** p < 0.0001, *** p = 0.0003 in both duplicates as determined by Fisher’s exact test, ns = not statistically significant in at least 1 duplicate. See S4 Data for the data underlying the graphs and for statistical test details. CEN, caudal epidermal neuron; OC, outer collocyte; PN, papilla neuron; RTEN, rostral trunk epidermal neuron. https://doi.org/10.1371/journal.pbio.3002555.g006 PNs are specified by common peripheral neuron regulators Previous papilla-specific TALEN knockout of the neuronal transcription factor-encoding gene Pou4 successfully eliminated PNs and the larva’s tail resorption response to mechanical stimuli [16]. Pou4 has been previously implicated in a Myt1-dependent regulatory cascade that specifies the caudal epidermal neurons (CENs) of the tail, from neurogenic midline cells expressing the proneural bHLH transcription factor Ascl.a (KH.L9.13, sometimes called Ascl2 or Ascl.b previously) [23,48–51]. To precisely visualize the neurogenic cells of the papillae, we performed double (two-color) mRNA in situ hybridization for Ascl.a and Foxg at the mid-tailbud stage. Indeed, Ascl.a expression was seen broadly in the papilla territory surrounding the 3 Foxg+ cell clusters (Fig 6D). This was confirmed by an Ascl.a fluorescent protein reporter plasmid that labeled a broad set of papilla territory cells, including PNs and their axons (Fig 6E). Furthermore, a previously published Myt1 reporter [52] was also found to be expressed in the PNs (Fig 6F). Double in situ of Pou4 and Foxg revealed Pou4+ cells surrounding each Foxg+ cluster, corroborating a recent report [23] (Fig 6G). It was not immediately clear which Pou4+ cells were PN precursors and which were nearby rostral trunk epidermal neuron (RTEN) precursors. Based on our images and those of the most recent study [23], we propose that there are initially 2 Pou4+ cells per papilla, later dividing to give rise to the 4 PNs per papilla as previously described [9]. This would mirror the development of the epidermal neurons of the tail, in which neurons are born side-by-side as pairs after a final cell division by a committed mother cell [49]. Papilla-specific CRISPR knockout of Pou4 with a combination of 2 newly validated sgRNAs (S3D Fig) recapitulated the loss of PN differentiation by the previously published TALEN knockout [16], as assayed by C4.78 and TGFB reporter expression (Fig 6H and 6I, and S4 Fig). In contrast, Pou4 knockout had no effect on the specification of ACCs or OCs, suggesting Pou4 function is specific for PN fate in the papillae (S4 Fig). Taken together, these results suggest that PNs are specified from interspersed neurogenic progenitors that are carved out by MAPK signaling. Interestingly, CRISPR knockout of Sp6/7/8 did not substantially affect PN specification (Fig 6I). This suggests that even though Sp6/7/8 down-regulates Foxg in these cells [21], it does not appear to be required for their neurogenic potential. Notch signaling regulates the fate choice between PNs and OCs Because both OCs and PNs appeared to arise from Foxg-down-regulating, Islet-negative cells, we sought to test whether an additional regulatory step is required for the fate choice between these 2 cell types. In the neurogenic midline territory of the tail epidermis, lateral inhibition by Delta/Notch signaling regulates the final number and spacing of CENs [48,49,53]. Delta/Notch limits the expression of Myt1, which in turn activates Pou4 expression. In the tail epidermis, the major ligand involved is the putative Delta like non-canonical Notch ligand homolog (encoded by gene KH.L50.6), which is also expressed in alternating pattern in the papillae (S7A Fig). We therefore decided to test whether a similar mechanism in controlling the number of PNs and OCs surrounding each papilla. To test the requirement of Delta/Notch, we overexpressed a DNA-binding mutant of the Notch co-factor RBPJ/SUH (SUH-DBM) [54]. Indeed, electroporation with Foxc>SUH-DBM resulted in loss of OC reporter expression (Fig 6J), and concomitant expansion of PN reporter expression (Fig 6K). We conclude that Delta/Notch signaling regulates PN versus OC fate choice in neurogenic progenitor cells surrounding each presumptive papilla, with Notch delimiting the specification of supernumerary neurons, thus allowing OCs to form (Fig 6L). In contrast, SUH-DBM had minimal effect on IC/ACC fate choice (S7B Fig), suggesting Delta/Notch might only regulate neurogenesis, and not cell fate choice in general, in the papilla territory. This common origin of PNs and OCs is also supported by the recent finding that the latter appear to have basal bodies like the PNs, but without the accompanying sensory cilia [9]. Interestingly, papilla-specific knockout of Foxg resulted in moderate loss of PN reporter expression (TGFB>GFP), and very little effect on the OC reporter (S4A Fig). This suggests differing requirement for Foxg in different cell type-specific branches of the papilla regulatory network, despite all these cell types arising from cells that initially express Foxg. Regulation of papilla morphogenesis by Islet It was previously shown that Foxg or Islet overexpression induces the formation of a single enlarged “megapapilla,” in which all cells are substantially elongated relative to the rest of the epidermis [21,22]. We have shown above that this appears to be driven by expansion of ACCs and/or ICs, which are atypically elongated in the apical-basal direction and form apical protrusions and microvilli. Islet is sufficient for apical-basal elongation of epidermal cells [22], and morpholino-knockdown of Foxg (which is upstream of Islet) also impairs proper papilla morphogenesis [21]. We asked if Islet is required for papilla morphogenesis, using papilla-specific CRISPR knockout of Islet. Knocking out Islet in the papilla territory impaired the formation of the typically “pointy-shaped” papillae, resulting instead in blunt cells with flat, broader apical surfaces and reduced cell length along the apical-basal axis (Fig 7A and 7B, S8C and S8D Fig). This result suggested that transcriptional targets downstream of Islet might be regulating the distinct cell shape of ACCs/ICs. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 7. Islet is also required for papilla morphogenesis. (A) Papilla shape is shortened and blunt at the apical end upon tissue-specific CRISPR/Cas9-mediated mutagenesis of Islet. Embryos were electroporated with Islet intron 1 + -473/-9>Unc-76::GFP and Foxc>Cas9. Islet CRISPR was performed using U6>Islet.2 sgRNA plasmid and the negative control used U6>Control. Larvae were imaged at 20 hpf/20 °C (~st. 28). Right: Scoring of percentage of GFP+ larvae classified as having normal “protruding” or blunt papillae, as represented to the left. Experiment was performed and scored in duplicate, using 2 different GFP fusions: Unc-76::GFP and DcxΔC::GFP [88]. Replicate 1: n = 100 for either condition; replicate 2: n = 55 for either condition **** p < 0.0001 in both duplicates as determined by Fisher’s exact test. (B) Quantification of papilla cell (Islet intron 1 +-473/-9>Unc-76::GFP+) lengths along apical-basal axis in negative control and Islet CRISPR larvae at 18 hpf/20 °C (~st. 28). Both Islet.1 and Islet.2 sgRNAs used in combination. Statistical significance tested by unpaired t test (two-tailed). See S8 Fig for duplicate experiment. (C) In situ mRNA hybridization of Villin, showing expression in Foxg+/Islet+ central papilla cells at 10 hpf/20 °C (st. 21, left) and at larval stage (~st. 27, right). (D) Villin -1978/-1>Unc-76::GFP showing expression in the papilla territory of electroporated larvae (~st. 28), strongest in the central cells. (E) Villin -1978/-1>Unc-76::GFP in st. 28 larvae is down-regulated by tissue-specific CRISPR/Cas9 mutagenesis of Islet (Foxc>Cas9 + U6>Islet.1 + U6>Islet.2, see text for details). (F) Quantification of effect of Islet CRISPR (as in panel E) on Villin -1978/-1>Unc 76::GFP/Foxc>H2B::mCherry mean fluorescence intensity ratios in ROIs defined by the mCherry+ nuclei (see Methods for details). Significance determined by Mann–Whitney test (two-tailed). (G) Villin reporter is up-regulated in st. 28 larvae by overexpressing Islet (Foxc>Islet, see text for details). (H) GFP/mCherry ratio quantification done in identical manner as in F, but comparing Islet overexpression (as in panel G) and control lacZ larvae. (I) Quantification of ACC lengths measured in negative control and papilla-specific Villin CRISPR larvae at 17 hpf/20 °C (~st. 27). Significance tested by unpaired t test (two-tailed). Although no statistically significant difference between control and CRISPR larvae was observed in this replicate, average ACC length was significantly shorter in the CRISPR condition in an additional replicate (S8 Fig). (J) mRNA in situ hybridization for Tuba3, showing enrichment in the central cells of the papillae in st. 27 larvae. (K) Tuba3>Unc-76::GFP reporter plasmid is broadly expressed in the papillae of st. 27 larvae but stronger in central cells. (L) Papilla-specific CRISPR knockout of Tuba3 does not result in decrease of average ACC apical-basal cell length compared to negative control CRISPR using U6>Control sgRNA instead. Significance tested by unpaired t test (two-tailed). ns = not significant. All large bars indicate medians and smaller bars indicate interquartile ranges. See S3 and S4 Data for the data underlying the graphs and for statistical test details. ACC, axial columnar cell; ROI, region of interest; sgRNA, single-chain guide RNA. https://doi.org/10.1371/journal.pbio.3002555.g007 To identify potential candidate effectors of morphogenesis downstream of Islet, we used bulk RNAseq to measure differential gene expression between different Islet perturbation conditions. We compared “negative control” embryos to (1) embryos in which Islet was overexpressed in the whole territory using the Foxc promoter (Foxc>Islet); and (2) embryos in which Islet was knocked out specifically in the papilla lineage by CRISPR/Cas9. For this, we designed an additional sgRNA targeting the first exon of Islet, to be used in combination with the already published sgRNA to generate larger deletions. This new sgRNA vector, which we named U6>Islet.1, resulted in a mutagenesis efficacy of 20% (S3B Fig). Whole embryos from each condition were collected at 12 hpf (Islet conditions) at 20 °C in biological triplicate. RNA was extracted from pooled embryos in each sample, and RNAseq libraries were prepared from poly(A)-selected RNAs and sequenced by Illumina NovaSeq. This bulk RNAseq approach revealed that Islet overexpression results in the up-regulation of several ACC markers from previous scRNAseq analysis (S2 Data) [20]. With Islet overexpression, this included ACC markers previously validated by mRNA in situ hybridization or reporter gene expression, such as CryBG (KH.S605.3) and Atp2a (KH.L116.40). Many ACC markers were conspicuously absent, but this may be due to the relatively early time point (12 hpf, late tailbud stage), well before hatching and ACC differentiation. This was a deliberate choice, as we were focused on papilla morphogenesis, which begins around this stage [22]. One resulting candidate Islet target revealed by RNAseq was Astl-related (KH.C9.850), and its expression in the Islet+ cells of the papillae was confirmed by in situ hybridization (S8A Fig). Indeed, Islet knockout by CRISPR eliminated Astl-related reporter expression, supporting our approach to identifying new targets of Islet (S8B Fig). Furthermore, the top up-regulated gene by Islet overexpression (and 17th most down-regulated by Islet CRISPR) was KY21.Chr10.318, which encodes a Fibrillin-related (Fbn) protein. This gene was previously shown to be specifically expressed in the central Islet+ cells by in situ hybridization [23], further validating our approach to identifying putative Islet target genes. One particularly interesting ACC-specific candidate that was among the genes most highly up-regulated by Islet overexpression was Villin (KH.C9.512), an ortholog of the Villin family of genes encoding effectors of actin regulators [55]. The apical extensions of the ACCs are highly enriched for actin filaments and microtubules [9], suggesting that cytoskeletal modulation may be important for the extended length of these cells relative to surrounding cells. We confirmed the expression of Villin in the papillae by in situ hybridization and reporter plasmids (Fig 7C and 7D). In the Islet CRISPR condition, Villin was the top down-regulated gene by Islet CRISPR knockout as well. Villin reporter expression was reduced in intensity but not completely lost upon knockout of Islet by CRISPR (Fig 7E and 7F), yet was dramatically up-regulated by Islet overexpression (Fig 7G and 7H). This suggests partially redundant activation of Villin by another factor, likely at earlier developmental stages (e.g., by Foxc or Foxg), and that Islet might be required for its sustained expression specifically in the central cells of the papilla throughout morphogenesis. This is consistent with the weak but broad expression of Villin>GFP in the entire papilla territory (Fig 7D), and the fact that papilla territory cells are already more elongated than epidermal cells in other parts of the embryo even at earlier stages [22]. To test whether Villin is required for proper morphogenesis of Islet+ cells in the papilla, we performed tissue-specific CRISPR knockout using a combination of 3 validated sgRNAs spanning most of the coding sequence (S3E Fig). Because the functionally important “headpiece” domain is encoded by the last exon, we combined an sgRNA targeting this exon with 2 sgRNAs targeting more upstream exons. In one batch of Villin CRISPR larvae, ACCs were not significantly shorter on average along the apical-basal axis than in control larvae (Fig 7I). However, average ACC length was significantly shorter in CRISPR larvae than control larvae in a duplicate experiment (S8E Fig). This contrast between replicates was found to be entirely due to variability between batches of control larvae, not the Villin CRISPR larvae (S8F Fig). Because the ACCs have been shown to dynamically extend or contract in length, possibly in response to external stimuli [56,57], we suspect that these differences in average length in different batches of control animals are due to as of yet unidentified environmental conditions. In addition to actin regulators, we searched our list of putative Islet targets for microtubule components and regulators, since microtubule bundles were reported in the apical protrusions of the ACCs in Distaplia occidentalis [58]. We identified a gene encoding a divergent Tubulin alpha monomer (Tuba3, KH.C3.736) as one such potential target. Enrichment of Tuba3 expression in the central papillae was confirmed by in situ hybridization (Fig 7J) and a Tuba3 reporter plasmid (Fig 7K). However, papilla-specific CRISPR knockout of Tuba3 did not result in significantly shorter ACCs either (Fig 7L). Taken together, these results suggest that Islet is required for proper papilla morphogenesis, and that this may be due to its role in activating the expression of numerous effector genes. However, knocking out individual candidate effector genes like Villin or Tuba3 has not yet revealed a key role for any one of these putative downstream targets. An investigation into the cell and molecular basis of larval settlement and metamorphosis With our different CRISPR knockouts affecting different cell types of the papillae, we asked how these different perturbations might affect larval metamorphosis. Only the involvement of the PNs in triggering metamorphosis has been demonstrated [16,17], but it is not yet known how the regulatory networks and cell types of the papillae affect different processes during metamorphosis. We performed papilla-specific CRISPR as above using the Foxc>Cas9 vector, targeting the 4 different transcription factors we have shown to be involved in patterning the cell types of the papillae: Pou4, Islet, Foxg, and Sp6/7/8. We assayed tail retraction and body rotation at the last stage of metamorphosis [28] (Fig 8A and 8B), as these are 2 processes that can be uncoupled in certain genetic perturbations or naturally occurring mutants [59]. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 8. Genetic perturbations of metamorphosis. (A) Ciona robusta juvenile undergoing metamorphosis, showing the retracted tail and rotated anterior-posterior body axis (dashed lines). PNs in the former papilla (now substrate attachment stolon, or holdfast) labeled by TGFB>Unc-76::GFP (green). Animal counterstained with DAPI (blue). (B) Scoring of Foxc>H2B::mCherry+ individuals showing tail retraction and/or body rotation at 48 hpf/20 °C in various papilla territory-specific (using Foxc>Cas9) CRISPR-based gene knockouts. Experiments were performed and scored in duplicate and percentages averaged, except for Foxg CRISPR for which a third replicate was performed (see S9 Fig). Number scored individuals in each replicate indicated underneath. “Tailed juveniles” have undergone body rotation but not tail retraction, whereas normally body rotation follows tail retraction. The sgRNA plasmids used for each condition were as follows- Control: U6>Control; Pou4: U6>Pou4.3.21 + U6>Pou4.4.106; Islet: U6>Islet.2; Foxg: U6>Foxg.1.116 + U6>Foxg.5.419; Sp6/7/8: U6>Sp6/7/8.4.29 + U6>Sp6/7/8.8.117. (C) Plot showing lack of any discernable metamorphosis defect after eliminating ACCs using Islet intron 1 + bpFOG>Sp6/7/8 (images not shown). Only Islet intron 1 + bpFOG>H2B::mCherry+ individuals were scored. Experiment was performed and scored in duplicate and averaged (n = 100 each duplicate). ACC specification was scored using the CryBG>Unc-76::GFP reporter. (D) Example of “tailed juveniles” at 47 hpf/20 °C compared to a larva in which no tail retraction or body rotation has occurred, elicited by tissue-specific Foxg CRISPR (Foxc>Cas9 + U6>Foxg.1.116 + U6>Foxg5.419). See S9 Fig for scoring. All error bars denote upper and lower limits. **** p < 0.0001, *** p < 0.0015 in both duplicates, ns = not significant in at least 1 duplicate, as determined by chi-square test comparing to the control conditions. See S4 Data for the data underlying the graphs and for statistical test details. ACC, axial columnar cell; PN, papilla neuron; sgRNA, single-chain guide RNA. https://doi.org/10.1371/journal.pbio.3002555.g008 Knockout of Pou4 recapitulated recent published results on this transcription factor [16]. Namely, both tail retraction and body rotation were blocked in the vast majority of individuals. This suggests that proper specification and/or differentiation of PNs by Pou4 is crucial for the ability of the larva to trigger the onset of metamorphosis. In contrast, Islet knockout did not affect tail retraction, but body rotation appeared somewhat impaired. This suggested that ACCs/ICs are not required for tail retraction, but might play a role in regulating body rotation downstream of it. Eliminating ACCs using Islet>Sp6/7/8 had no effect on either tail retraction or body rotation (Fig 8C), confirming that ACCs are not required for metamorphosis, but that perhaps certain Islet targets might specifically regulate body rotation. Unsurprisingly, Foxg knockout modestly impaired both tail retraction and body rotation (Fig 8B and 8D, and S9 Fig), but also resulted in a noticeable fraction (approximately 19% on average) of “tailed juveniles” in which body rotation begins even in the absence of tail retraction. This unusual effect was seen even when repeating the experiment independently a third time, revealing consistent uncoupling of these 2 processes upon Foxg knockout (S9 Fig). Finally, Sp6/7/8 CRISPR did not substantially alter either tail retraction or body rotation. Taken together, these results paint a more complex picture of regulation of metamorphosis by the papillae. Our findings suggest that different cell types of the papillae might play distinct roles in the regulation of metamorphosis, perhaps interacting with one another to regulate different steps, or that certain transcription factors might be required for the expression of key rate-limiting components of these different processes. Further work will be required to disentangle these different cellular and genetic factors, which we hope will be aided by our cell type-specific reporters and CRISPR reagents. Discussion Sensory systems are crucial for interactions between organisms and their environment. The concentration of sensory functions in the head is thought to have played a central role in vertebrate evolution, leading to a more active behavior emerging from early filter-feeding chordate ancestors [60–62]. The peripheral components of the sensory systems in vertebrates arise from 2 physically close but distinct ectodermal cell populations, the cranial sensory placodes and the neural crest [63]. Cranial sensory placodes are characterized by their common ontogenetic origin from a crescent-shaped region surrounding the anterior neural plate. Our understanding of the evolutionary origins of structures long presented as vertebrate novelties has benefited from an increasing number of comparative studies with tunicates. Several discrete populations of peripheral sensory cells originating from distinct ectodermal regions in tunicates have respectively been linked to neural crest and cranial placodes, among them the sensory adhesive papillae [9,64–67]. Our results have confirmed the existence of molecularly distinct cell types in the Ciona papillae and the developmental pathways that specify them (summarized in Fig 9). Using CRISPR/Cas9-mediated mutagenesis, we have shown that different transcription factors are required for their specification, differentiation, and morphogenesis. Namely, ACCs and ICs are specified from Foxg+/Islet+ cells at the center of each of the 3 papillae, while OCs and PNs are specified from interleaved Islet-negative cells that nonetheless derive from initially Foxg+ cells. While Sp6/7/8 specifies IC versus ACC fate among Islet+ cells, Delta/Notch signaling suppresses PN fate and promotes OC fate among Islet-negative cells. While there appear to be 2 molecularly distinct collocyte subtypes (OCs and ICs), both contain granules that are stained by PNA, and therefore both are likely to be involved in glue production. Where they differ might be in the timing of glue production and/or secretion, as they showed distinct subcellular localization of PNA+ granules, and PNA production was previously shown to start very early [9]. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 9. Summary diagram. (A) Updated diagram of the development of the anterior descendants of Row 6 in the neural plate to show the proposed patterns of MAPK and Delta/Notch signaling that set up the 3 Foxg+ clusters and interleaved Foxg-negative neurogenic cells. (B) Diagram proposing the contributions of Foxg+ and Foxg-negative cells to later patterns of transcription factors that specify the different cell types found in each papilla, which is in turn is repeated 3 times, thanks to the process shown in panel A. (C) Papilla development shown as cell lineages, with dashed lines indicating uncertain cell divisions and lineage history. MAPK regulates binary fate choices, promoting Foxg expression in the papillae proper at first but later suppressing it (and Islet). Lastly, Delta-Notch signaling promotes OC fate and limits PN specification through lateral inhibition. Cell type numbers based on [9]. (D) Provisional gene regulatory network diagram of the transcription factors involved in specification and differentiation of the different papilla cell types. Arrowheads indicate activating gene expression or promoting cell fate, while blunt ends indicate repression of gene expression of cell fate. Solid lines indicate regulatory links (direct or indirect) that are supported by the current data and literature. Dashed lines indicate regulatory links that have not been tested, or need to be investigated in more detail. A-P, anterior-posterior; D-V, dorsal-ventral; OC, outer collocyte; OSP, oral siphon primordium; PN, papilla neuron. https://doi.org/10.1371/journal.pbio.3002555.g009 Our results also demonstrate a clear distinction between CryBG+ ACCs and Pou4+ PNs. Previously, these cells types have been confused and only recently distinguished by TEM and different molecular markers [9]. Here we show that, while both arise from Foxc+/Foxg+ cells, ACCs are not specified by Pou4, and PNs are not specified by Islet. However, because Pou4 can activate Foxg expression in a proposed feedback loop [68], overexpression of Pou4 might result in ectopic activation of ACC markers via ectopic Foxg and Islet activation. There are still unanswered questions that we hope future work will address: (1) How do the 3 “spots” of Foxg+/Islet+ cells form in an invariant manner? Ephrin-Eph signaling is often responsible for suppression of FGF/MAPK signaling in alternating cells in Ciona embryos, via asymmetric inheritance/activation of p120 RasGAP [69,70]. This is also true in the earlier patterning of the papilla territory, where EphrinA.d suppresses FGF/MAPK to promote Foxg activation [21]. Curiously, later expression of EphrinA.d in the lineage appears to be stronger in medial Foxg+ cells than in lateral cells [21]. This distribution would suffice to result in the alternating ON/OFF pattern of MAPK activation at the tailbud stage that results in the 3 foci of Foxg/Islet expression. Thus, it may be informative to test the ongoing functions of Ephrin-Eph signaling in this lineage throughout development. Interestingly, we did not observe substantial loss of PNs in Sp6/7/8 CRISPR larvae, even though Sp6/7/8 down-regulates Foxg in between the 3 “spots” [21] and is necessary for OC reporter expression (Fig 4F). It is possible that FGF/MAPK suppresses Islet in parallel, and that loss of Sp6/7/8 is not sufficient to expand Islet expression and ACC/IC progenitor fate (at the expense of PN/OC progenitors) in the same manner that the MEK inhibitor U0126 does. (2) How are PNs specified adjacent to the Islet+ cells? Since Delta/Notch signaling is involved in PN versus OC fate, we propose that there is something that biases Notch signaling to be activated preferentially in those cells not touching the Islet+ cells. This could be due to cell-autonomous activation of Notch signaling in the Islet+ cells, which in turn would allow for suppression of Notch in adjacent cells fated to become PNs. A recent study showed Pou4 expansion with concomitant “U”-shaped expansion of Islet in larvae with a single, enlarged papilla when inhibiting BMP signaling [23]. However, our expansion of Islet with treatment of U0126 (based on experiments from Wagner and colleagues) suggests the opposite, the elimination of Pou4+ PNs. Why the discrepancy? One possibility is that inhibiting BMP results in specification of supernumerary RTEN-like neurons from adjacent epidermis, not PNs. While Pou4 is expressed in all epidermal neurons including PNs and RTENs, our preferred PN marker KH.C4.78 is not expressed in RTENs. However, we did not observe either loss or expansion of C4.78>GFP in larvae with enlarged, single papillae resulting from treatment with the BMP inhibitor DMH1 (S10 Fig). This suggests that inhibition of FGF/MEK and BMP have slightly different effects on patterning and neurogenesis in the papillae. Further work will be needed to resolve these and other intriguing nuances. (3) What activates the expression of Sp6/7/8 in Islet+ cells, ultimately promoting IC specification? We do not yet know the exact mitotic history of the ACCs/ICs. How do the initially four Islet+ cells divide, and which daughter cells give rise to ACCs versus ICs? Are ACCs/ICs specified in an invariant manner, or is there some variability? Finally, what allows the “creeping” activation of Sp6/7/8 in the outer ring of cells that likely become the ICs? Is this due to additional asymmetric FGF/MAPK activation downstream of Ephrin-Eph? Or could this be due to some other signaling pathway? Is there an inductive signal from adjacent cells, for instance PNs or common PN/OC progenitors? (4) How do the different papilla cell types regulate metamorphosis? We noticed some uncoupling of tail resorption and body rotation upon targeting different transcription factors for deletion in the papillae (Fig 8). This was most apparent in the Foxg knockout, in which a substantial portion of individuals displayed the “tailed juvenile” phenotype in which body rotation proceeds even in the absence of tail resorption. From the Pou4 knockout, it is clear that PNs are upstream of both tail resorption and body rotation, but the partial uncoupling seen with the other manipulations were particularly intriguing. This uncoupling has been reported before in Cellulose synthase mutants, which results in similar tailed juveniles [71]. Additionally, perturbation of Gonadotropin-releasing hormone (GnRH) or the prohormone convertase enzyme (PC2) necessary for its processing similarly blocks tail resorption but not body rotation and further adult organ growth [72]. Thus, it is possible that while Pou4 disrupts PN specification altogether, Foxg might be more specifically required for GnRH or other neuropeptide expression/processing in the PNs. Supporting this idea, the Foxg CRISPR did not disrupt PN specification (as assayed by TGFB>GFP reporter expression) as robustly as did Pou4 CRISPR. Alternatively, this uncoupling may also be a result of the different roles of Foxg in regulating different papilla cell types, which may be unequally and variably affected due to CRISPR knockout mosaicism in F0. Finally, the appearance of juveniles with resorbed tails but no further body rotation in the Islet CRISPR condition suggests a crucial role for the “core” cells of the papilla (ACCs/ICs) in metamorphosis downstream of PNs. However, body rotation was not affected by eliminating either ICs (Sp6/7/8 CRISPR, Fig 8B) or ACCs (Islet>Sp6/7/8, Fig 8C), suggesting Islet is required for the expression of a “body rotation” factor independently of IC/ACC specification. Clearly, more work will be required to understand the contributions of different cell types, and potentially different molecular pathways in the same cell type, towards either activation or suppression of specific body plan rearrangement processes in tunicate larval metamorphosis. (5) Are the tunicate larval papillae homologous to vertebrate cement glands and/or olfactory placodes? The papillae have often been compared to the cement glands of fish and amphibian larvae, which are transient adhesive organs secreting sticky mucus [73]. Even though they are innervated by trigeminal fibers, the secreting cells from the cement gland differentiate from a surface ectoderm region anterior to the oral ectoderm and the panplacodal domain [74]. Therefore, they are usually not considered placodal derivatives. Despite their variability in size, number, and location, head adhesive organs are proposed to be homologous across vertebrate species based on their shared expression of Pitx1/2 and BMP4 genes, innervation by trigeminal fibers, and inhibiting mechanism of swimming behavior [73,74]. While recent papers have revealed an important role for BMP signaling in pattering the tunicate papilla territory [23,24], suggesting a potential evolutionary connection, additional work on the molecular basis of the papillary glue in tunicates will be required to answer questions of homology between these adhesive organs. Our identification of molecular signatures for both collocyte subtypes in the papillae of Ciona provides a starting point for future investigations and may allow for broader evolutionary comparisons between chordates and other bilaterians. If one considers the vertebrate cement gland and the trigeminal neurons that innervate it as a functional unit but no homology to the tunicate papillae can be established, they might represent an interesting case of evolutionary convergence. Besides adhesion, a key role of the papilla is to act as an organ for bimodal mechano- and chemo-sensation regulating larval settlement and metamorphosis [75]. It has been proposed that the papilla territory may be a homolog of the olfactory placode, based on the expression of several regulators investigated in this study including Sp6/7/8, Islet, Foxc, and Foxg [18,21]. The roles of Foxg in specifying the papilla territory may also be linked evolutionarily to the functions of Foxg1 in vertebrate olfactory development [76,77]. Intriguingly, the specification of sensory PNs and probable adhesive-secreting OCs from shared Ascl.a+ progenitors (as revealed by Notch pathway inhibition experiments, Fig 6) suggests a close regulatory link between sensory/adhesive functions in Ciona papillae. In vertebrates, Ascl1 not only regulates sensory cell specification in developing olfactory epithelium and taste buds [78–80] but is also required for intestinal secretory cells [81,82]. Additionally, Ascl3 is expressed in vertebrate salivary gland duct progenitors, which are also highly enriched for orthologs of other tunicate papilla regulators like Foxc, Sp6/7/8, and Islet [83,84]. Thus, while the papillae of Ciona might represent a tunicate-specific evolutionary novelty, overlaying unrelated networks for sensory neurons and adhesive-secreting cells together in a single embryonic territory, it is also possible that they rely on a shared sensory/exocrine program that might have deeper evolutionary origins instead [85]. Supporting information S1 Fig. Finding papilla cell type-specific markers in single-cell RNAseq data. (A) Cell clusters based from reanalysis and re-clustering of whole-larva single-cell RNA sequencing (scRNAseq) data from Cao and colleagues (see S1 Data). Dashed red box indicated clusters 3 and 33, which appeared to correspond to several papilla cell types. (B) Cells from clusters 3 and 33 from plot A set aside and re-clustered. (C) Differential expression plots showing examples of candidate papilla cell type marker genes mapped onto clusters in B. (D) Fluorescent, whole-mount in situ mRNA hybridization (green) for certain genes plotted in C, labeling different cells in the papillae of Ciona robusta (intestinalis Type A) hatched larvae. (E) Protein domain prediction diagrams for select cell type-specific marker proteins generated by SMART [89]. Unless specifically named, genes are indicated by KyotoHoya (KH) ID numbers (e.g., KH.L96.43). All larvae were fixed at 18 h post-fertilization (hpf), 20 °C, except for C11.360 and C2.1013 (18.5 hpf). Blue counterstain is DAPI. https://doi.org/10.1371/journal.pbio.3002555.s001 (TIF) S2 Fig. Additional marker genes and reporter plasmids expressed in papillae. (A) TGFB>Unc-76::GFP reporter (green) is not co-expressed in the same cells as the Islet intron 1 + -473/-9>mCherry reporter (pink) at 20.5 hpf (~st. 29). (B) Cross-talk between CryBG>Unc-76::GFP and TGFB>Unc-76::mCherry reporter plasmids at 16 hpf (~st. 26), showing aberrant co-expression in ACCs and/or PNs only when co-electroporated. (C) Mutually exclusive expression of CryBG>lacZ in ACCs (cyan), Gnrh1>Unc-76::GFP in PNs (yellow), and C11.360>Unc-76::mCherry in ICs (magenta), with DAPI counterstained in gray. This larva is the same as in main Fig 2G, with an additional channel and different false coloring. (D) Images from Fig 2 with mCherry and GFP channels displayed separately. (E) C. intestinalis (Type B) larva electroporated with C. robusta C11.360>Unc-76::GFP reporter plasmid, showing specific but weak expression. (F) C. intestinalis (Type B) larva electroporated with C. robusta L96.43>Unc-76::GFP reporter plasmid, also showing weak expression. Papillae in panels E and F outlined by dashed lines. (G) Reporter plasmid containing the first intronic region of Emx drives expression in ICs at 20 hpf (~st. 29), likely corresponding to the “ring” of late Emx expression in Islet+ cells reported in Wagner and colleagues and distinct from earlier Emx expression in the papilla lineage as described in Liu and Satou. (H) C14.116>Unc-76::mCherry reporter expressed in central cells (ACCs+ICs, pink) and basal cells around the 3 papillae at 20.5 hpf (~st. 29). (I) Immunostaining for the Flag epitope tag fused to the Islet-rescue protein used for Islet>Islet experiments in Fig 4. Flag immunostaining in green and Foxc>H2B::mCherry in pink in merged image. Larvae fixed at 19 hpf (~st. 28). DAPI in gray. ACCs, axial columnar cells; PN, papilla neuron. https://doi.org/10.1371/journal.pbio.3002555.s002 (TIF) S3 Fig. Validation of sgRNAs for CRISPR/Cas9-mediated mutagenesis. Gene loci diagrams for the 4 transcription factor-encoding genes investigated in this study: Sp6/7/8, Foxg, Islet, and Pou4. Plots underneath each gene show validation by Illumina sequencing (“Next-generation sequencing” or NGS) of amplicons, performed as “Amplicon-EZ” service by Azenta. Mutagenesis efficacies are calculated by this service, and histograms of mapped reads show specificity of indels elicited by each sgRNA. Negative control amplicons are amplified from samples that were electroporated with no sgRNA, U6>Control sgRNA, or sgRNAs targeting unrelated amplicon regions. Note different y axis scales for each plot. Asterisks in Villin exon 5 and Tuba3 amplicon plots indicate naturally occurring indels. Precise calculation of mutagenesis efficacy for Villin.5.105 and Tuba3.3.24 sgRNAs was not given due to these natural indels. https://doi.org/10.1371/journal.pbio.3002555.s003 (TIF) S4 Fig. Effect of various CRISPR knockouts on specification of ACCs, PNs, and OCs. (A) Scoring of effect of papilla-specific CRISPR knockout of Foxg or Pou4 on specification of ACCs and PNs. Embryos were electroporated with Foxc>H2B::mCherry, Foxc>Cas9, CryBG>Unc-76::GFP (ACC reporter), TGFB>Unc-76::GFP (PN reporter), or L141.36>Unc-76::GFP (OC reporter), and gene-specific sgRNA combinations (see below for specific combinations). All were performed in duplicate and scores averaged, but some replicates and conditions are represented in Figs 3 and 4 also. Total embryos ranged between 76 and 100 per condition per replicate. Specific sgRNAs used: Foxg: U6>Foxg.1.116 + U6>Foxg.5.419; Pou4: U6>Pou4.3.21 + U6>Pou4.4.106; Sp6/7/8: U6>Sp6/7/8.4.29 + U6>Sp6/7/8.8.117; Islet: U6>Islet.2; Control: U6>Control. (B) Foxg, Pou4, and Sp6/7/8 sgRNAs were also tested alone (as opposed to pairs in combination) using reporter assays as in Figs 3 and 4. Those sgRNAs used further are highlighted in blue font. Additional sgRNAs abandoned due to low efficacy indicated in black font. For all plots, only larvae showing Foxc>H2B::mCherry expression in the papillae were scored. See S4 Data for the data underlying the graphs and for statistical test details. https://doi.org/10.1371/journal.pbio.3002555.s004 (TIF) S5 Fig. CRISPR rescues and possible expansion of ICs in Islet>Sp6/7/8. (A) CryBG>GFP expression in Islet (left) and Foxg (right) CRISPR larvae is rescued by co-electroporation with Islet intron 1 + bpFOG>Flag::Islet-rescue or Foxg>Foxg-rescue constructs, respectively, thanks to silent point mutations disrupting the sgRNA target binding sites. (B) Expression of C11.360>GFP is rescued in Sp6/7/8 CRISPR larvae upon co-electroporation with an Islet intron 1 + bpFOG>Sp6/7/8-rescue construct. (C) Example of expanded IC reporter (C11.360>Unc-76::GFP, green) in larvae (20 hpf/20 °C, ~st. 29) electroporated with Islet intron 1 + -473/-9>Sp6/7/8, as determined by perfect overlap with the Islet intron 1 + -473/-9>H2B::mCherry reporter (pink). See text for more details. See S1 File for exact sequences and detailed electroporation recipes. See S4 Data for the data underlying the graphs and for statistical test details. https://doi.org/10.1371/journal.pbio.3002555.s005 (TIF) S6 Fig. Single-channel images of PNA-stained larvae shown in Fig 5. https://doi.org/10.1371/journal.pbio.3002555.s006 (TIF) S7 Fig. Delta/Notch components and lack of SUH-DBM effect on ACC/IC fate choice. (A) Fluorescent whole-mount in situ mRNA hybridization for Delta like non-canonical Notch ligand (KH.L50.6) in a stage 23 embryo, marking epidermal sensory neurons including papilla neurons (arrows) and caudal epidermal neurons (CENs). (B) No significant difference in expression of either CryBG>mCherry or C11.360>GFP in larvae at 19 hpf at 20 °C (~st. 28) electroporated with the Delta-Notch pathway-inhibiting Foxc>SUH-DBM. Right: representative panels showing expression of both reporters in control and SUH-DBM-expressing larvae. See S4 Data for the data underlying the graphs and statistical test details. https://doi.org/10.1371/journal.pbio.3002555.s007 (TIF) S8 Fig. Investigating the regulation of papilla morphogenesis by Islet and its putative transcriptional targets. (A) In situ mRNA hybridization (ISH) showing expression of Astl-related (green) specifically in the Islet+ cells of the papillae (labeled by Islet intron 1 + -473/-9>mCherry, pink nuclei). (B) Tissue-specific CRISPR/Cas-mediated mutagenesis of Islet results in loss of Astl-related>Unc-76::GFP reporter expression in ACCs/ICs (green). Foxc>Cas9 was used to restrict CRISPR/Cas9 to the papilla territory (labeled by Foxc>H2B::mCherry, pink nuclei). Asterisk denotes residual reporter expression in cells outside the papilla territory. Right: Scoring of larvae represented in panel B, following criteria used for Fig 3. **** p < 0.0001 using chi-square test. (C) Second duplicate of Islet CRISPR experiment in Fig 7B. Statistical significance was determined by unpaired t test (two-tailed). (D) Representative images of control and Islet CRISPR larvae used for measurements in Fig 7B, with example of apical-basal cell length measurements. (E) Both duplicates of Villin CRISPR experiments side by side. Statistical significance was calculated using Mann–Whitney test (two-tailed) for replicate 1 and unpaired t test (two-tailed) for replicate 2. ns = not significant. (F) Comparison of ACC lengths measured in control larvae from the 2 duplicate Villin CRISPR experiments, showing statistically significant difference in average ACC lengths between different batches of larvae, calculated using Mann–Whitney test (two-tailed). See S3 and S4 Data for the data underlying the graphs and for statistical test details. https://doi.org/10.1371/journal.pbio.3002555.s008 (TIF) S9 Fig. Third replicate of Foxg CRISPR effects on metamorphosis. Scoring of Foxc>H2B::mCherry+ individuals as represented in Fig 7D, in third replicate of data in Fig 7B. See S1 File for detailed plasmid electroporation recipes. **** p < 0.0001 calculated by Fisher’s exact test. See S4 Data for the data underlying the graphs and for statistical test details. https://doi.org/10.1371/journal.pbio.3002555.s009 (TIF) S10 Fig. Pharmacological inhibition of BMP signaling and papilla neuron specification. Larva co-electroporated with Islet intron 1 + -473/-9>mCherry (pink) and PN reporter C4.78>Unc-76::GFP (green) showing lack of substantial loss or expansion of PNs in larvae treated with the BMP pathway inhibitor DMH1, in spite of expanded Islet reporter expression and a single, enlarged papilla. Left: scoring of PN reporter expression in Islet>mCherry+ DMSO (negative control) and DMH1-treated larvae. All larvae raised to 19 hpf at 20 °C (~st. 28). See S4 Data for the data underlying the graphs and for statistical test details. https://doi.org/10.1371/journal.pbio.3002555.s010 (TIF) S1 Data. Differential gene expression table of re-analyzed whole larva single-cell RNAseq data from Cao and colleagues (new clusters 3 and 33 subclustered into 11 subclusters). https://doi.org/10.1371/journal.pbio.3002555.s011 (XLSX) S2 Data. Differential gene expression table of bulk RNAseq analysis of Islet overexpression or Islet CRISPR embryos vs. control embryos. https://doi.org/10.1371/journal.pbio.3002555.s012 (XLSX) S3 Data. Papilla length measurements in different CRISPR and control larvae. https://doi.org/10.1371/journal.pbio.3002555.s013 (XLSX) S4 Data. Summary of statistical tests of proportions (i.e., scoring). https://doi.org/10.1371/journal.pbio.3002555.s014 (XLSX) S1 Movie. Confocal stack represented by Fig 5A. https://doi.org/10.1371/journal.pbio.3002555.s015 (AVI) S2 Movie. Confocal stack represented by Fig 5B. https://doi.org/10.1371/journal.pbio.3002555.s016 (AVI) S1 File. All relevant DNA and protein sequences used in this study (supplemental sequence file). https://doi.org/10.1371/journal.pbio.3002555.s017 (DOCX) Acknowledgments We thank members of the labs at Georgia Tech and Innsbruck for critical feedback and support. We thank Susanne Gibboney, Tanner Shearer, Alex Gurgis, Lindsey Cohen, Akhil Kulkarni, and Eduardo Gigante for technical assistance.

Bekdash, Ramsey;Yoshida, Kazushige;Nair, Manoj S.;Qiu, Lauren;Ahdout, Johnathan;Tsai, Hsiang-Yi;Uryu, Kunihiro;Soni, Rajesh K.;Huang, Yaoxing;Ho, David D.;Yazawa, Masayuki

doi: 10.1371/journal.pbio.3002522pmid: 38483887

Introduction The Coronavirus Disease 2019 (COVID-19) pandemic caused by the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) [1–3] has affected approximately 800 million people and counting in the world. More than 6 million people have passed away due to the viral infection. Because the mutagenesis rate in SARS-CoV-2 genes such as Spike is high [4–6], it is a challenge to develop sustainable approaches for prevention and treatment. While several new vaccines and drug candidates have become available, the number of COVID-19 infections and deaths are still increasing, and new variants are being reported [7–10]. Therefore, it would be ideal if we could target coronavirus genes that are highly conserved in SARS-CoV-1 and SARS-CoV-2 variants, as well as other human coronaviruses. Human coronaviruses such as SARS-CoV-2 and Middle East Respiratory Syndrome Coronavirus (MERS-CoV) express an Envelope (E) protein that forms an ion channel essential for viral function called a viroporin [11–17]. E protein is known to induce cellular toxicity via a number of different molecules and signaling pathways [12,16,18–23] and E protein is also thought to be involved in the Spike protein protection and maturation in host cells [20,24]. Compared to the other molecules, E protein is highly conserved among coronaviruses: SARS-CoV-2 E (2E) protein’s 75 amino acid residues have high homology with SARS-CoV-1 E protein (approximately 96%), with identical amino (N)-terminus, transmembrane, and pore structures while their carboxyl (C)-terminus is slightly different [11,25–31] (Fig 1A). Previous studies reported that deficiency of SARS-CoV-1 E gene significantly reduced viral propagation [12], suggesting that 2E may also play essential roles in viral function and can be a potential therapeutic target for COVID-19 and future variants [10]. Therefore, in this study we seek to develop screening platforms and applied them to identify drug candidates against 2E. In addition, we examined whether our therapeutic approach targeting 2E can be applicable to the E protein of other human coronaviruses. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 1. Testing the amino-terminal fragment of SARS-CoV-2 Envelope named MY18 on lysosomal pH imaging in mammalian cells. (A) Alignment of Envelope of SARS-CoV (SARS1-E) and SARS-CoV-2 (SARS2-E), showing the difference in amino acids (red), the targeted amino-terminal region named MY18 (red underlined), and putative transmembrane region (black underlined, 18–39). (B) Representative confocal fluorescent and bright field images of NIH 3T3 cells loaded with DND-189, a lysosomal pH green fluorescent dye, and transfected with MY18 peptide construct, empty vector (mock) and SARS2-E fused with mKate2 red fluorescent protein (2E-mKate2). Scale bar, 5 μm. (C) Relative fluorescent intensity of DND-189 dye in NIH 3T3 cells transfected using mock (n = 40) or 2E-mKate2 plasmid without (-, n = 36) and with MY18 plasmid (+MY18, n = 30). One-way ANOVA with Tukey’s multiple comparisons test (**** P < 0.0001; n.s. not significant). (D) The sequences of cell-penetrating peptide candidates, 6-Arg (6Arg), TAT and Penetratin (Pen), for MY18 peptide uptake in mammalian cells. (E) Relative fluorescent intensity of DND-189 dye in NIH 3T3 cells incubated with MY18 (10 μM, n = 93), 6Arg-MY18 (0.1 μM, n = 40; 1, n = 41; 10, n = 106), TAT-MY18 (0.1 μM, n = 27; 1, n = 20; 10, n = 92), and Pen-MY18 peptides (0.1 μM, n = 27; 1, n = 32; 10, n = 68) with 2E-mKate2 plasmid transfection. Mock (n = 116) and non-treated 2E-mKate2 (-, n = 139) were also tested as their controls. In 1 μM condition, TAT-MY18 peptide is not different from mock (blue highlighted) while 6Arg and Pen are significantly lower than mock. One-way ANOVA with Dunnett’s multiple comparisons test (**** P < 0.0001; *** P < 0.001; ** P < 0.01; n.s. not significant, compared to mock). The data underlying this figure can be found in S1 Data. All the graphs in the figure are mean ± SD. NIH, National Institute of Health; SARS‑CoV‑2, Severe Acute Respiratory Syndrome Coronavirus 2. https://doi.org/10.1371/journal.pbio.3002522.g001 Results Previous studies have reported oligomeric structures and molecular interactions in SARS-CoV-1 E and 2E [25,26,32]. Following the results, we hypothesized that the highly conserved N-terminal region is crucial for oligomerization and that the N-terminal fragment might be able to disrupt 2E protein oligomerization and function because the N-termini might interact with each other in the protein oligomers. Because our previous study demonstrates that the overexpression of 2E affects proton homeostasis in intracellular organelles such as Golgi apparatus and lysosomes in mammalian cells [19], we examined the effect of the N-terminal fragment, which is named MY18 (18 amino acids, MYSFVSEETGTLIVNSVL), on 2E function using DND-189 pH fluorescent dye and MY18 plasmid transfection in mammalian cells. DND-189-based pH fluorescent imaging shows that MY18 co-overexpression in 2E-expressing mammalian cells significantly restores DND-189 fluorescence to the normal level (Fig 1B and 1C). These results encouraged further investigation in using MY18 as a peptide that inhibits 2E. Next, to apply MY18 as an exogenous synthetic peptide, we tested 3 cell-penetrating amino acid motifs: an arginine repeat, TAT and Penetratin [33] (Fig 1D). The DND-189-based pH imaging demonstrates that the TAT version of MY18 is the most promising cell-penetrating peptide among the cell-penetrating peptide candidates (Fig 1E). While DND-189-based pH fluorescent imaging is useful as a drug screening platform of live mammalian cells, the dynamic range of the dye is somewhat limited, the standard deviation of fluorescent readout is relatively large, and its throughput is not as high as that of an assay for screening. These limitations may give rise to difficulties in further optimizing the MY18 peptide using molecular biological approach with mutagenesis. To develop a higher-throughput screening platform, we overexpressed 2E in mammalian cells and explored other reliable and quantitative readouts. Interestingly, global proteomics results showed increases of various key signaling molecules such as JUN/AP-1 (Fig 2A). A follow-up experiment using quantitative RT-PCR (qPCR) confirmed that the expression of most of the genes including JUN transcript are significantly increased in 2E-expressing cells compared to mock (Figs 2B and S1). The transcript expression of a JUN-related molecule, NFATC4/NFAT3, is also up-regulated significantly (Fig 2C), though the increase of NFATC4/NFAT3 protein (approximately 120%) did not reach to significance in statistical analysis of the global proteomics results with the standard false discovery rate (FDR, 0.05, Fig 2A). Following these results, we decided to apply the NFAT response element of the human IL-2 gene where NFAT and JUN/AP-1 interact [34]. To obtain precise readouts, we used a dual luciferase reporter system containing NFAT Firefly luciferase reporter and pRL-TK-Renilla luciferase reporter as the transfection control using HSV TK, herpes simplex virus thymidine kinase, promoter [35] (Fig 2D). The luciferase assay result obtained using plasmid DNA co-transfection shows that 2E overexpression significantly increases the Firefly luciferase activity in mammalian cells and that MY18 co-expression significantly suppresses the effect of 2E on the NFAT/AP-1 pathway, though it does not fully restore it to mock levels (Fig 2E). These results suggest that MY18 is not sufficient to prevent 2E from altering the NFAT/AP-1 pathway completely, though we observed that MY18 restores proton homeostasis in DND-189-based pH fluorescent imaging (Fig 1). Looking to optimize MY18, we next examined whether deletion or extension of MY18 might improve the effect on interrupting 2E function using the luciferase reporter. We found that none of the constructs significantly improved the effect, as the majority reduced efficacy (Fig 2F). Next, we tested a variety of MY18 mutant constructs using mutagenesis, co-transfection and luciferase assay, and found that the substitution of glutamate to aspartate at seventh and eighth residues (EE7-8DD or 2ED) significantly improved MY18 (Fig 2G). We combined the mutant 2ED and TAT cell-penetrating motif (Fig 1D and 1E) and called TAT-MY18-2ED “iPep-SARS2-E” (inhibitory Peptide against SARS-CoV-2 Envelope). In addition, we confirmed that iPep-SARS2-E had the same rescuing effect on lysosomal pH phenotype in 2E-transfected mammalian cells as the plasmid construct did (S2A Fig). Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 2. Mutagenesis of MY18 peptide to develop iPep-SARS2-E. (A) Volcano plots of global proteomics of HEK 293S cells transfected with SARS2-E fused to mKate2 (2E-mKate2). Red plots demonstrate significant increases in 2E-mKate2 compared to empty vector (mock) (FDR: 0.05). (B and C) The expressions of JUN/AP-1 (B) and NFATC4 transcripts (C) significantly increased in HEK 293S cells transfected to 2E-mKate2 compared to mock. Unpaired Student’s t test was used (** P < 0.01; * P < 0.05, n = 3–4). (D) Schematic representation of dual luminescence reporter system using 4-repeated NFAT response element (RE) of human IL-2 gene and Firefly luciferase gene (NFAT-FLuc) and herpes simplex virus thymidine kinase (HSV TK) promoter-driven Renilla luciferase gene (TK-RLuc) as transfection control. (E) Relative NFAT-FLuc/TK-RLuc activity in HEK 293T cells transfected using empty vector (mock, n = 9) or 2E-mKate2 plasmid without (-, n = 9) and with MY18 plasmid (+MY18, n = 12). One-way ANOVA with Tukey’s multiple comparisons test was used (**** P < 0.0001). (F) Relative NFAT-FLuc/TK-RLuc activity in HEK 293T cells transfected using empty vector (mock, n = 9) or 2E-mKate2 plasmid without (-, n = 6) and with various sizes of SARS2-E amino-terminal constructs including MY18 (each, n = 6, inset). One-way ANOVA with Dunnett’s multiple comparisons test was used (**** P < 0.0001; ** P < 0.01; * P < 0.05; n.s., not significant, compared to MY18). (G) Relative NFAT-FLuc/TK-RLuc activity in HEK 293T cells transfected using empty vector (mock, n = 3) or 2E-mKate2 plasmid without (-, n = 3) and with MY18 mutants (each, n = 3). One-way ANOVA with Dunnett’s multiple comparisons test was used (**** P < 0.0001; ** P < 0.01; * P < 0.05; n.s., not significant, compared to MY18 wild-type, WT, n = 9). S-T switch, Ser and Thr replaced each other. The inset demonstrates the amino acid sequence of MY18-2ED (EE7-8DD, underlined) mutant, which significantly improves the inhibitory effect on SARS2-E-mediated NFAT-JUN/AP-1 activation. The data underlying this figure can be found in S1 Data. All the graphs in the figure are mean ± SD. FDR, false discovery rate; HEK, human embryonic kidney. https://doi.org/10.1371/journal.pbio.3002522.g002 To characterize iPep-SARS2-E, we first produced a monoclonal antibody against the N-terminal region of 2E. Following previous studies of viral envelope topology, we believed that the N-terminus may be in the extracellular region of SARS-CoV-2 [25,36]. To produce the monoclonal antibody, we used a synthetic peptide composed of the first 18 amino acids of E protein (i.e., MY18) with keyhole limpet haemocyanin (KLH) as the antigen. Western blotting result reveals that a hybridoma produces a 2E monoclonal antibody (2E-N; clone, N2A5E8) that can recognize 2E proteins expressed in a mammalian heterologous expression system (S2B Fig). Next, we conducted an ELISA using the antibody to compare the affinity of our 2E-N antibody to our 2ED and wild-type peptides. We confirmed that the 2ED mutation reduces the binding affinity of the 2E antibody, which was produced using the wild-type MY18 peptide as the antigen (S2C Fig). Next, we used ELISA to examine how stable iPep-SARS2-E is in phosphate-buffered saline (PBS) at 37°C. We did not observe obvious peptide degradation in 24 h, though the peptide might become unstable after 48 h because the standard deviation became larger compared to earlier time points (S2D Fig). Using an apoptosis/necrosis assay with flow cytometry, we confirmed that iPep-SARS2-E does not cause cellular toxicity in mammalian cells in vitro (S2E–S2G Fig). To investigate the interaction of MY18-2ED and 2E protein biochemically, we first conducted immunoprecipitation using 6xHis-tagged MY18-2ED (His-MY18) and 2E-YFP with Ni column. The western blotting using anti-GFP antibody, which can recognize 2E-YFP protein band, demonstrates that Ni column with His-MY18 can pull down 2E-YFP proteins (Fig 3A), suggesting an interaction between MY18-2ED and 2E protein, though there was no obvious difference in MY18-2E protein interaction between the wild-type and 2ED peptide constructs, according to the immunoprecipitation result (S2H Fig). To confirm this in situ, we co-expressed 2E-mKate2 with His-MY18 in NIH 3T3 cells using lipofection and anti-His tag antibody conjugated to Alexa Fluor 488 to examine whether MY18 peptide interacts 2E protein directly. The fluorescent imaging result reveals that His-MY18 are co-localized with 2E proteins in the mammalian cells (Fig 3B). To further investigate the molecular mechanism underlying the inhibitory effect of MY18-2ED, we next used our established electrophysiological recording [19] to examine whether MY18-2ED co-transfected with 2E has a direct effect on 2E channel activity. The result demonstrates that MY18-2ED significantly reduced 2E channel current in HEK cells (Fig 3C and 3D), suggesting that MY18-2ED may inhibit 2E function directly. Together, the results suggest that MY18-2ED might be integrated into 2E protein oligomers and inhibit 2E function. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 3. Characterization of iPep-SARS2-E. (A) Immunoprecipitation of 2E protein using Ni column and HEK 293T cells transfected using 2E-YFP with or without 6xHis-MY18-2ED (His-MY18). Anti-GFP antibody was used to blot 2E-YFP protein bands. (B) Representative epi-fluorescence image of NIH 3T3 cells co-transfected with 2E-mKate2 (red) and His-MY18 plasmids. Anti-His tag antibody conjugated to Alexa Fluor 488 (green) and Hoechst 33258 dye (blue, nucleus) were used after cell fixation. White arrowheads, co-localization of red and green fluorescence. Scale bar, 5 μm. (C) Representative SARS2-E currents in 2E-PM-expressing HEK 293S cells mock-transfect and co-transfected with MY18-2ED peptide construct. (D) MY18-2ED significantly blocked 2E currents. Student’s t test was used (*** P < 0.001). (E) Representative immunoblot images of HEK 293T cells transfected with SARS2-E fused with YFP (2E-YFP) and 48 h treated with 10 μM TAT-MY18-2ED or MY18-WT peptides (negative control). Anti-GFP (for 2E-YFP, top) and GAPDH antibodies (as loading control, bottom) were used. Putative aggregates of 2E-YFP proteins (&) were observed even though Urea-based lysis buffer was used for the sample preparation. #, nonspecific bands around 40 and 50 kDa according to the YFP blotting (S3B Fig). The whole blotting image of GAPDH is shown in S3A Fig. (F and G) Quantification of 2E-YFP monomeric form (F) and aggregates (G) of HEK 293T cells transfected with 2E-YFP and treated using TAT-MY18-2ED (n = 4) or MY18-WT peptides (n = 4). Student’s t test was used (*** P < 0.001). (H) Representative immunoblot images of HEK 293T cells transfected with YFP plasmid and 48 h treated with TAT-MY18-2ED (10 μM, n = 3) and MY18-WT peptides (10 μM, n = 3, a negative control) and non-treated cells (n = 3, another negative control). Anti-GFP (for YFP, top) and GAPDH antibodies (as loading control, bottom) were used. The whole blotting images and quantification are shown in S3B and S3C Fig, respectively. (I) Schematic representation of the molecular mechanism underlying the effect of MY18 peptide on 2E protein and lysosomal function. The images are prepared using BioRender: Top, 2E induces deacidification in lysosome (Fig 1B and 1C); bottom, MY18 peptides binds 2E proteins (Fig 3A and 3B), resulting in 2E inhibition (Fig 3C and 3D) and restored lysosomal activity (Figs 1B, 1C and S2A) and 2E protein reduction (Fig 3E–3G). The data underlying this figure can be found in S1 Data. All the graphs in the figure are mean ± SD. HEK, human embryonic kidney; NIH, National Institute of Health. https://doi.org/10.1371/journal.pbio.3002522.g003 To examine the effect of MY18-2ED on 2E protein, we incubated mammalian cells with iPep-SARS2-E and transfected them with 2E-YFP construct. Interestingly, iPep-SARS2-E significantly reduced 2E-YFP protein expression in mammalian cells. While we observed both monomeric and aggregate bands of 2E proteins, both forms were significantly decreased in the treated cells (Figs 3E–3G and S3A). As a control experiment, we used YFP-transfected cells with peptide treatment and did not observe any effect of iPep-SARS2-E on YFP protein expression (Figs 3H, S3B and S3C). The results indicate that iPep-SARS2-E does not affect lipofection but reduces 2E expression in mammalian cells. Based on the results obtained in the series of experiments, the suggested molecular mechanism underlying the inhibitory effect of iPep-SARS2-E is as follows (Fig 3I): the peptides might interact with 2E proteins (Fig 3A and 3B) and inhibit 2E channel function (Fig 3C and 3D), resulting in lysosomal pH restoration (Figs 1B–1E and S2A) and 2E protein degradation (Fig 3E–3G). To examine the kinetics of peptide penetration into mammalian cells, we conjugated a fluorescent probe, Alexa Fluor 594, to the N- or C-terminus of iPep-SARS2-E. Fluorescent imaging in situ reveals that the C-terminal fused version exhibited faster cell-penetration than the N-terminal version in NIH 3T3 cells. We found that most cells can uptake the peptides in 2 h (S4A and S4B Fig). This result suggests that the N-terminal conjugation of Alexa Fluor 594 might slightly affect the function of the TAT motif in the peptide. To examine the off kinetics, we next treated cells with iPep-SARS2-E fluorescent peptide (C-terminal version, TAT-MY18-2ED-Alexa Fluor 594) for 24 h, washed out the culture medium, and then monitored the red fluorescence (S4C Fig). The in situ imaging suggests that the peptide is not reduced or degraded for 96 h; however, the peptide might become unstable and aggregated after 72 h at 37°C in the live cells, since larger fluorescent puncta were observed compared to earlier time points (S4D Fig). In addition, we tested the C-terminal version in another mammalian cell line, Vero-E6, which has been commonly used in virological studies using SARS-CoV-2. We confirmed that the peptides can penetrate into these cells as well (S5A and S5B Fig). To examine the effect of iPep-SARS2-E on SARS-CoV-2 infection, as a proof-of-concept experiment in vitro, we conducted a cytopathic assay using a mammalian cell line, Vero-E6, and SARS-CoV-2 (WA1 strain, Fig 4A). We used the wild-type MY18 peptide (non-TAT version, MY18-WT) as a negative control in the assay because MY18-WT does not have any cell-penetrating motifs or effect on 2E in the pH imaging in NIH 3T3 cells (Fig 1E). The cytopathic assay result demonstrates that iPep-SARS2-E significantly inhibits viral toxicity in vitro (Fig 4B: IC50 of iPep-SARS2-E, approximately 400 nM). Following these results, we next conducted time-course experiments to elucidate the mechanism underlying the inhibitory effect of iPep-SARS2-E on viral function (Fig 4C). The qPCR result demonstrates that there is no difference in SARS-CoV-2 nucleocapsid (N) gene expression, suggesting no effect of iPep-SARS2-E on virus transcription and entry (Fig 4D). On the other hand, there is a significant difference in SARS-CoV-2 N gene detection between the culture supernatants of PBS-treated control and iPep-SARS2-E-treated cells sampled at 24 h post-infection (Fig 4E), demonstrating a significant reduction of virus release from iPep-SARS2-E-treated cells. Importantly, we found that iPep-SARS2-E could significantly restore the expression of JUN/AP-1 (Fig 4F), which we had used as a reporter of SARS2-E cellular toxicity in this study to optimize the MY18 peptide series (Fig 2). Electron microscopy reveals that viral particles can be observed in large vacuoles of PBS-treated cells (Fig 4G and 4H), which could be deacidified and disrupted lysosomes, according to a previous study [37]. However, no vacuoles containing multiple viral particles were found in iPep-SARS2-E-treated cells, while small particles were found in the endoplasmic reticulum and nuclear envelope (Figs 4I and S6A). This suggests that the particles might be virions, though it is not clear whether the virions are mature in iPep-SARS2-E-treated cells. Therefore, we next examined infectivity of these intracellular particles from iPep-SARS2-E-treated cells treated via an endpoint titration assay used to quantitate intracellular virus particles (S6B Fig) [38–40]. We found that there was a modest but significant (P < 0.0001) reduction in virus titers of iPep-SARS2-E-treated cell samples compared to PBS-treated control cells, but that the intracellular particles of iPep-SARS2-E-treated cell samples are still infectious (S6C Fig). Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 4. iPep-SARS2-E in vitro test. (A) Experimental design for the iPep-SARS2-E (TAT-MY18-2ED) test using SARS-CoV-2 WA1 virus (MOI, 0.10) and Vero-E6 cells in vitro. (B) Inhibition of iPep-SARS2-E in the cytopathic effect of SARS-CoV-2 WA1 virus on Vero-E6 cells. MY18-WT (non-TAT) was used as a control. (C) Design of time-course experiment using SARS-CoV-2 WA1 virus (MOI, 0.10) and Vero-E6 cells in vitro. (D) Time-course qPCR of SARS2-CoV-2 nucleocapsid (SARS2 N) expression of PBS- and iPep-SARS2-E (10 μM)-treated Vero-E6 cells. The expression of N gene was normalized to a house-keeping gene, GAPDH. Student’s t test was used at each time point (n.s., not significant). (E) qPCR of SARS2 N expression of PBS- and iPep-SARS2-E (10 μM)-treated Vero-E6 cell culture supernatant (sup) at 24-h post-infection. Student’s t test was used (**** P < 0.0001). (F) qPCR of JUN/AP-1 expression of PBS- and iPep-SARS2-E (10 μM)-treated Vero-E6 cells comparing to non-infected cells. The expression of JUN was normalized to GAPDH. One-way ANOVA with Tukey’s multiple comparisons test was used (* P < 0.05; n.s., not significant). (G) Representative electron microscopic (EM) image of PBS-treated Vero-E6 cells at 24 h post-infection. Scale bar, 500 nm. (H) Higher magnification of the EM image of PBS-treated Vero-E6 cells at 24 h post-infection (a box shown in Fig 4G). Scale bar, 100 nm. (I) Representative EM image of iPep-SARS2-E-treated Vero-E6 cells at 24 h post-infection. Arrowheads, virus-like particles accumulated in the endoplasmic reticulum. The other EM image is also shown in S6A Fig. Scale bar, 500 nm. (J) Representative confocal fluorescent images of Vero-E6 cells treated with PBS or iPep-SARS2-E at 24 h post-infection. SARS2 N antibody (red) and Hoechst 33258 dye (blue, for nucleus) were used with antibodies of subcellular organelle markers (green): BiP for endoplasmic reticulum (ER), ERGIC-53 for ER Golgi inter compartment (ERGIC), and LAMP1 for lysosome. Scale bar, 5 μm. The data underlying this figure can be found in S1 Data. All the graphs in the figure are mean ± SD. EM, electron microscopic; ER, endoplasmic reticulum; ERGIC, ER-Golgi inter-compartment; PBS, phosphate-buffered saline; qPCR, quantitative polymerase chain reaction; SARS‑CoV‑2, Severe Acute Respiratory Syndrome Coronavirus 2. https://doi.org/10.1371/journal.pbio.3002522.g004 E protein is thought to be involved in Spike protein protection and maturation [20,24]. Therefore, we examined the effect of iPep-SARS2-E on Spike protein expression. The western blotting result demonstrates significant reductions of Spike protein expression (S6D Fig). Immunocytochemistry confirms that, in iPep-SARS2-E-treated cells, N proteins are co-localized to BiP/GRP78 (a marker of endoplasmic reticulum, or ER), and to ERGIC-53 (a marker of ER-Golgi inter-compartment, or ERGIC); in PBS-treated cells, N proteins were more highly and broadly expressed, and co-localized to LAMP1 (a lysosomal marker) (Fig 4J). Following these results, we conducted in vitro experiments using iPep-SARS2-E at a later time-point to examine its effect further. qPCR shows that there is a significant decrease in SARS-CoV-2 N and E gene expression in iPep-SARS2-E-treated Vero-E6 cells at 48 h post-infection compared to the PBS-treated control cells (S6E and S6F Fig). In addition, iPep-SARS2-E significantly restores the expression of JUN/AP-1 (S6G Fig). Importantly, detection of SARS-CoV-2 genes in the culture supernatant harvested at 48 h post-infection is significantly reduced in iPep-SARS2-E-treated cells compared to PBS-treated control cells (S6H Fig), demonstrating that iPep-SARS2-E suppresses virus release. Immunocytochemistry shows that the majority of infected Vero-E6 cells became round and apoptotic-like in the PBS-treated group, while iPep-SARS2-E-treated cells exhibited moderate expression of viral N and normal cellular morphology (S6I Fig). This is consistent with the results of the cytopathic assay (Fig 4B). Next, to validate the inhibitory effect of iPep-SARS2-E further, we conducted a preclinical experiment in vivo using iPep-SARS2-E intravenous (i.v.) injection to Balb/c mice infected with a mouse-adapted strain of SARS-CoV-2 (MA10, [41–43]). First, we conducted i.v. injection of the C-terminal fluorescent version of iPep-SARS2-E (TAT-MY18-2ED-Alexa Fluor 594) to mice and confirmed that the peptide can permeate and is detectable in mouse lung tissues 2 h after administration (Fig 5A). Following this result, we next injected iPep-SARS2-E post-infection. The mice were sacrificed 4 days after MA10 viral infection, and their lungs were harvested for viral titer, qPCR, and histology (Fig 5B). The results show no difference in body weight between the groups, but a significant reduction of viral propagation in iPep-SARS2-E-treated mouse lungs compared to the control (Fig 5C and 5D). qPCR also confirms the inhibitory effect of iPep-SARS2-E on the viral propagation in vivo (Fig 5E). Lung histology reveals no immune infiltration or alveolar damage in iPep-SARS2-E-treated mouse lungs, while minimal interstitial infiltrates with patchy lymphoid aggregates and protein accumulation were observed in the non-treated control group (Fig 5F). Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 5. iPep-SARS2-E in vivo preclinical study. (A) Representative fluorescent and bright field images of lung tissues isolated from mice administrated i.v. with PBS or Alexa594-conjugated iPep-SARS2-E peptide (TAT-MY18-2ED-A594, 300 μM, 2 h and 24 h). After isolating the tissues, the samples were washing using PBS 3 times, and the fluorescent and bright field images were taken by a fluorescent stereoscope. Scale bar, 2 mm. (B) Experimental design using the iPep-SARS2-E i.v. injection in vivo. (C) There is no difference in body weight between PBS control (n = 11) and iPep-SARS2-E-treated mice (n = 7). (D) There is a significant reduction of lung viral titer in iPep-SARS2-E-treated mice compared to the control. Student’s t test was used (* P < 0.05; n.s., not significant). Median TCID is normalized to lung wet weight (g) measured before tissue homogenization to isolate the virus. (E) iPep-SARS2-E significantly reduced the transcript expression of SARS2 N in MA10-infected Balb/c mouse lung tissues (PBS, n = 11; iPep-SARS2-E, n = 7, normalized to mouse Gapdh expression). Student’s t test was used (** P < 0.01). (F) Representative images of hematoxylin and eosin (HE) staining of mouse lung tissues of PBS control and iPep-SARS2-E-treated mice at 4 days post-infection. Protein accumulation (x) and immune cells (arrowheads) are indicated. Scale bar, 50 μm. The data underlying this figure can be found in S1 Data. All the graphs in the figure are mean ± SD. HE, hematoxylin and eosin; PBS, phosphate-buffered saline; TCID, tissue culture infection dose. https://doi.org/10.1371/journal.pbio.3002522.g005 To examine whether iPep-SARS2-E could be applied for prevention of SARS-CoV-2 infection, we conducted another experimental series in vivo using iPep-SARS2-E intranasal administration to Balb/c mice infected with MA10 virus. First, we administrated the C-terminal fluorescent version of iPep-SARS2-E to mice intranasally and confirmed that the peptide can permeate and is detectable in mouse nasal tissues 2 h after administration (S7A and S7B Fig). Next, we conducted a safety study in vivo (S7C Fig) using intranasal administration to Balb/c mice to examine the effect of iPep-SARS2-E on body weight and inflammation markers, comparing to non-treated and PBS-administrated groups. We did not find any significant differences in body weight, Cxcl12 and C5a among iPep-SARS2-E-, PBS-, and non-treated groups (S7D–S7F Fig). Following the results using intranasal administration, we applied iPep-SARS2-E intranasal administration to Balb/c mice infected with MA10 virus (S7G Fig). We found that iPep-SARS2-E significantly prevents the body weight loss and suppresses 2E protein expression in infected mouse lungs in vivo (S7H–S7K Fig). These results using in vivo experiments demonstrate that SARS2-E inhibition can be a novel strategy to prevent SARS-CoV-2 toxicity and propagation in vivo. To validate iPep-SARS2-E further, we continued to conduct control experiments, preparing an additional negative control and experimental conditions. When we examined the effect of deletion on MY18 constructs using the luciferase reporter, we had found that none of the constructs significantly improved the effect: the majority did not reduce its efficacy significantly, but deletion at the 5th and 17th/18th residues significantly reduced the inhibitory effect of MY18 on 2E-mediated NFAT/AP-1 pathway alteration (Fig 2F). Following the results, we introduced Ala-substitution, targeting these amino acids of MY18 (S8A Fig). NFAT/AP-1 Luc assay and DND-189 imaging results demonstrate that the mutagenesis significantly reduced the inhibitory effect of MY18, although NFAT/AP-1 Luc assay result suggests that it was not sufficient to fully counteract the effect of MY18 (S8B and S8C Fig). After further mutagenesis, we found that the addition of an L12A substitution was able to negate the inhibitory effect of MY18 against 2E. We confirm using qPCR, that, like PBS, the mutant peptide cannot reduce SARS-CoV-2 N expression in Vero-E6 cell culture supernatant (S8D Fig). In the following experiments, we used this mutant peptide as a new negative control. The qPCR results using Vero-E6 cell culture supernatant and lysate samples at 24 h post-infection with SARS-CoV-2 (Fig 4C–4E) suggest that iPep-SARS2-E may not have any effect on the virus entry. However, this time-course transcription profiling may not be sufficient as a readout of the virus entry. To address this concern, we conducted an experiment using pseudo virus containing SARS-CoV-2 Spike, Membrane, E proteins, and YFP reporter with the negative control peptide and iPep-SARS2-E because pseudo virus is useful to examine the effect of drug candidates on the virus entry [44]. The pseudo virus infection resulted in no difference in YFP-positive cell number between the negative control peptide- and iPep-SARS2-E-treated cells, revealing no effect of iPep-SARS2-E on the virus entry (S8E and S8F Fig). To validate the inhibitory effect of iPep-SARS2-E peptide further, we conducted in vitro experiments using human pluripotent stem cell-derived branching lung organoids and WA1 virus (Fig 6A and 6B). We found that there is a significant reduction in SARS-CoV-2 N transcript of the culture supernatant at 24 h post-infection between the negative-control peptide- and iPep-SARS2-E-treated organoids but not in the organoid lysate (Fig 6C and 6D), suggesting that virus egress is blocked by iPep-SARS2-E. Immunocytochemistry allows us to observe higher expression of SARS-CoV-2 N in the negative control peptide-treated organoids compared to iPep-SARS2-E-treated organoids (Fig 6E). The results using the lung organoids are consistent with the results using Vero-E6 cells at 24 h post-infection (Fig 4). Our results in monolayer cells and in human 3D lung organoids reveal a new strategy to prevent SARS-CoV-2 propagation using iPep-SARS2-E, which inhibits 2E activity and virus egress. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 6. iPep-SARS2-E and its negative control peptide test using in vitro human lung organoid model. (A) Experimental design for the iPep-SARS2-E test using SARS-CoV-2 WA1 virus (MOI, 0.10) and human pluripotent stem cell-derived branching lung organoids in vitro. The image is prepared using BioRender. (B) Representative phase contrast images of human pluripotent stem cell-derived branching lung organoids. Scale bar, 50 μm. (C and D) qPCR of SARS2 N expression of the negative control mutant peptide (neg. Ctrl, S8 Fig)- and iPep-SARS2-E (10 μM)-treated organoids culture supernatant (sup, C) and cells (D) at 24 h post-infection. Student’s t test was used (* P < 0.05; n.s., not significant). (E) Representative merged section images of confocal fluorescence and bright field of human lung organoids treated with neg. Ctrl or iPep-SARS2-E at 24 h post-infection. SARS2 N antibody (red, arrowheads), EpCAM antibody (green), and Hoechst dye (blue, for nucleus) were used. Scale bar, 20 μm. The data underlying this figure can be found in S1 Data. All the graphs in the figure are mean ± SD. qPCR, quantitative polymerase chain reaction; SARS‑CoV‑2, Severe Acute Respiratory Syndrome Coronavirus 2. https://doi.org/10.1371/journal.pbio.3002522.g006 Next, we conducted another in vivo mouse study (S8G Fig) using a single intranasal administration to Balb/c mice to examine the effect of iPep-SARS2-E and the negative control peptide on body weight, SARS-CoV-2 titer, and N transcript in the mouse lungs. We found that iPep-SARS2-E could significantly prevent their body weight loss while the negative control did not (S8H Fig). The viral titer and qPCR results confirm the inhibitory effect of iPep-SARS2-E in vivo (S8I and S8J Fig). These in vivo results demonstrate that SARS2-E inhibition using a single intranasal dose of iPep-SARS2-E can be useful to prevent SARS-CoV-2 toxicity and propagation in vivo. Next, we hypothesized that this peptide design and strategy could be customized and applied to other human coronaviruses because coronavirus E proteins are highly conserved (Figs 7A and S9). Therefore, following iPep-SARS2-E development, we designed inhibitory peptide constructs against E proteins from each of the other human coronaviruses: MERS-CoV, HCoV-NL63, -OC43, -HKU1, and -229E (Fig 7B). First, we found that the overexpression of MERS-CoV and HCoV-NL63 E proteins significantly increased NFAT/AP-1 luciferase reporter activity in mammalian cells while the E proteins of HCoV-OC43, -HKU1, and -229E do not have an effect on NFAT/AP-1 pathway (Fig 7C). Following these results, we focused our testing to MER-CoV and HCoV-NL63 MY18 WT peptide constructs using the NFAT/AP-1 luciferase reporter assay. The reporter assay results demonstrate that MERS-CoV and HCoV-NL63 MY18 WT could significantly reduce the effect of each E protein on NFAT/AP-1 reporter while substitution of Glu/E to Asp/D or Asp/D to Glu/E does not improve the MY18 constructs for MERS-CoV and HCoV-NL63, respectively (Fig 7D and 7E). To improve each MY18 further, we conducted mutagenesis of MERS-CoV MY18 and HCoV-NL63 MY18 constructs. We found that HCoV-NL63 MY18 2DE and N9D and MERS-CoV MY18 R8H are the best to inhibit the effect of HCoV-NL63 and MERS E proteins on NFAT/AP-1 pathway in mammalian cells, respectively (Fig 7D and 7E). Next, using DND-189 pH imaging, we found that the overexpression of MERS-CoV, HCoV-NL63, and -HKU1 E proteins significantly reduced DND-189 fluorescence in mammalian cells while the E proteins of HCoV-OC43 and -229E do not have a significant effect on lysosomal proton homeostasis (Fig 7F). Following these results, we focused on testing MY18 constructs on MERS-CoV, HCoV-NL63, and -HKU1 and found that MERS-CoV MY18 R8H, HCoV-NL63 MY18 2DE and N9D, HCoV-HKU1 MY18 D8E peptide constructs could significantly rescue the phenotypes in proton homeostasis in mammalian cells caused by the respective E proteins (Fig 7F). The results of this experiment reveal that the MY18 peptides can be applicable for not only SARS-CoV-2 but also some of the other human coronaviruses such as MERS-CoV, HCoV-NL63, and HCoV-HKU1, demonstrating that E protein can more broadly be a potential therapeutic target for human coronaviruses. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 7. MY18 peptide application for other human coronaviruses. (A) Alignment of the N-terminal region of human coronavirus Envelope (E) of SARS-CoV-2, MERS-CoV, HCoV-229E, HCoV-NL63, HCoV-OC43, and HCoV-HKU1. Glu (red), Asp (blue), and other amino acid residues (purple) are targeted for further mutagenesis to customize MY18 inhibitory peptides for each viral E. (B) MY18 peptide design for each human coronavirus E of SARS-CoV-2, MERS-CoV, HCoV-229E, HCoV-NL63, HCoV-OC43, and HCoV-HKU1. Glu and Asp are replaced (red), following iPep-SARS2-E. (C) Relative NFAT-FLuc/TK-RLuc activity in HEK 293T cells transfected using mock (n = 17), SARS2-E (n = 10), and the other coronavirus E (each, n = 8). One-way ANOVA with Dunnett’s multiple comparisons test was used (**** P < 0.0001; * P < 0.05; n.s., not significant, compared to mock). (D) Relative NFAT-FLuc/TK-RLuc activity in HEK 293T cells transfected using mock (n = 9), NL63-E without (n = 12), and with NL63-MY18 constructs (WT, n = 12; 2DE, n = 3; the other mutants, n = 6). One-way ANOVA with Dunnett’s multiple comparisons test was used (**** P < 0.0001; *** P < 0.001; n.s., not significant, compared to mock). (E) Relative NFAT-FLuc/TK-RLuc activity in HEK 293T cells transfected using mock (n = 9), MERS-E without (n = 12), and with MERS-MY18 constructs (WT, n = 12; E7D, n = 3; the other mutants, n = 6). One-way ANOVA with Dunnett’s multiple comparisons test was used (**** P < 0.0001; ** P < 0.01; n.s., not significant, compared to mock). (F) Relative fluorescent intensity of DND-189 dye in NIH 3T3 cells transfected with mock (n = 78) and the other human coronavirus E-mKate2 constructs: OC43 (n = 74), 229E (n = 66), MERS (n = 63), NL63 (n = 65), and HKU1 (n = 78). Each MY18 mutant construct was also tested: MERS-MY18 (R8H, n = 36), NL63-MY18 (2DE and N9D, n = 37), and HKU1-MY18 (D8E, n = 31). One-way ANOVA with Dunnett’s multiple comparisons test was used (**** P < 0.0001; n.s., not significant, compared to mock). The data underlying this figure can be found in S1 Data. All the graphs in the figure are mean ± SD. HEK, human embryonic kidney; MERS-CoV, Middle East Respiratory Syndrome Coronavirus; SARS‑CoV‑2, Severe Acute Respiratory Syndrome Coronavirus 2. https://doi.org/10.1371/journal.pbio.3002522.g007 Discussion We applied a pH fluorescent dye, DND-189, to identify TAT-MY18 as a therapeutic candidate against E proteins (Fig 1). However, uptake of the dye could be dependent on cell viability and intracellular organelle function because DND-189 is a single green fluorescent dye. Therefore, the fluorescent change might not be due to a direct effect of E protein, but a secondary effect. In addition to the DND-189 dye, we applied the NFAT/AP-1 luciferase reporter assay as a higher throughput screening platform to optimize the MY18 peptide using molecular biological approach (Fig 2). However, NFAT/AP-1 alteration might be a secondary consequence induced by other molecular phenotypes of E protein overexpression in mammalian cells. Therefore, immunoprecipitation, immunocytochemistry, and electrophysiological recording were useful to validate the inhibitory effect of the MY18 peptides on E protein function (Fig 3). Control safety experiments in vitro and in vivo demonstrate that there is no obvious toxicity of iPep-SARS2-E in vitro and in vivo (S2E–S2G and S7C–S7F Figs). iPep-SARS2-E was shown to be efficacious against SARS-CoV-2 both in vitro and in vivo (Figs 4–6 and S6–S8). The cytopathic assay result demonstrates that, MY18-WT, which was validated as a negative control in DND-189 imaging using NIH 3T3 cells (Fig 1E), also had a moderate inhibitory effect on the virus in Vero-E6 cells (Fig 4B), although no cell-penetrating motif is present. This suggests that the non-TAT version might be spontaneously taken up into Vero-E6 cells by endocytosis. Alternatively, we generated and validated another negative control peptide using mutagenesis (Figs 6 and S8). iPep-SARS2-E significantly reduces 2E-YFP protein expression in mammalian cells while having no effect on YFP and GAPDH protein (Figs 3E–3H and S3A–S3C). Altered lysosomal pH due to 2E may result in a protective effect that allows for viral proteins to be sequestered from proteolysis and promote viral assembly and release as reported in the other viruses [19,37]. Therefore, it is possible that iPep-SARS2-E-induced 2E inhibition may restore lysosomal pH, allowing lysosomes to resume normal activity and resulting in proteolysis of 2E and Spike proteins (Figs 3E–3G and S6D). Importantly, the in vitro study suggests that iPep-SARS2-E may inhibit virus egress, since the viral transcript detection was significantly lower in iPep-SARS2-E-treated cell culture supernatant than in the PBS-treated control (Fig 4E). On the other hand, we found that there is no difference in SARS-CoV-2 N transcript expression and pseudo-virus infection between iPep-SARS2-E and the controls (Figs 4D, S8E and S8F), suggesting no effect of iPep-SARS2-E on virus entry and transcription. The cytopathic assay at later time-point suggests that the intracellular particles are still infectious in iPep-SARS2-E-treated cells (S6B and S6C Fig), though the stability of iPep-SARS2-E remains unclear over 48 h (S2D and S4D Fig). The western blotting result demonstrates that iPep-SARS2-E significantly reduced Spike protein expression in WA1-infected cells (S6D Fig). The results are consistent with the previous study reporting that E protein is involved in Spike protein expression and protection [20,24]. Together, these results suggest the molecular mechanism underlying the inhibitory effect of iPep-SARS2-E as follows: the peptides interact with 2E proteins and inhibit 2E channel function, resulting in lysosomal pH restoration and reduction of 2E- protective effect, which is crucial for the viral proteins, such as Spike, to be sequestered from proteolysis and promote virus egress. E protein has been considered as a pharmaceutical target [45–50]. Compared to small molecules inhibiting E protein, iPep-SARS2-E might be more customizable to each human coronavirus E protein (Fig 7) and their future variants because it is peptide-based, though further optimization of MY18 peptides is required for each variant. The difference of this amino acid residue size (i.e., Glu versus Asp) might be crucial for human coronavirus E oligomerization. 2DE mutation could be applied for HCoV-NL63 version with N9D while ED mutation has a negative effect on the MERS-CoV version of MY18 peptide (Fig 7D and 7E). In our previous study for a bacterial lactate-binding protein LldR, the E103D mutant significantly increases the affinity and specificity of LldR to lactate [51]. Though E is a different molecule from LldR, E/D and D/E substitutions could be a useful strategy for peptide and protein engineering. In summary, we established and applied 2 screening platforms using the lysosomal pH fluorescent imaging and the NFAT/AP-1 luminescence reporter system with our expertise in proteomics, imaging and human cellular modeling [52–54], to identify novel therapeutic candidates against human coronavirus E protein. Our inhibitory peptides can be customizable and applicable for targeting E protein of not only SARS-CoV-2 but also the other coronaviruses, such as MERS-CoV, HCoV-NL63, and HCoV-HKU1, demonstrating more broadly that E protein is a potential therapeutic target for human coronaviruses. Due to the unique effects on virus egress, iPep-SARS2-E could prove to be a promising therapeutic candidate against SARS-CoV-2, particularly if combined with other therapeutic candidates such as protease and polymerase inhibitors for a synergistic effect. Methods Ethical statement Columbia University is a National Institute of Health (NIH) Office of Laboratory Animal Welfare assured institution (ID#, D16-00003/A3007-01), which has complied with the NIH Public Health Service Policy and adhered to the standards in the guide for the care and use of laboratory animals. The animal study is approved by Columbia University Institutional Animal Care and Use Committee (protocol #, AC-AABP2571). This virus study using SARS-CoV-2 is approved by Columbia University ADARC committee and BSL3 facility committee. Cell culture Human embryonic kidney (HEK) 293S cells (ATCC, Cat#CRL-3022) were cultured in Dulbecco’s Modified Eagle Media Nutrient Mixture F-12 (DMEM/F-12, Thermo Fisher Gibco # 11320033) and HEK 293T cells (ATCC, Cat#CRL-3216), NIH 3T3 cells (ATCC, Cat#CRL-1658), and Vero-E6 cells (Cat# CRL-1586) were cultured in and DMEM (Thermo Fisher/Gibco #10313021). Both media were supplemented with GlutaMax-I and penicillin/streptomycin (PS) and 10% fetal bovine serum (FBS, not heat-inactivated, HyClone, #SH30071.03, Thermo Fisher) under normoxia (20% O2, 5% CO2, at 37°C). The cell lines were passaged using trypsin-EDTA (0.25%, Thermo Fisher, # 25200–056) every 2 to 3 days. Human branching lung organoids were prepared using our normal human-induced pluripotent stem cell lines [52] with an established differentiation medium kit (STEMCELL Technologies, #100-0195/0528). Molecular biology constructs Plasmid DNA constructs were generated using standard methods with restriction enzymes (New England BioLabs), DNA ligase (MightyMix, TaKaRa Bio/Clontech), and polymerase chain reaction (PCR) with Phusion polymerase (Thermo Fisher). Construct inserts for these experiments were synthesized (Integrated DNA Technologies, IDT) and subcloned into pcDNA3 vector (Life Technologies). Mock transfections were performed by using pcDNA3 empty vector. Imaging experiments and pH measurements For the imaging experiments in Figs 1 and 7, NIH 3T3 cells were plated at a cell density of 2.5 × 105 cells/ml. The following day, the cells were transfected using Lipofectamine 3000 reagents (Thermo Fisher, # L3000001), and 4 μg of plasmid were mixed into 125 μl of serum-free OptiMEM with 5 μl of P3000 reagent. This was then added to another 125 μl of serum-free OptiMEM containing 7.5 μl of Lipofectamine 3000. Plasmid/P3000-lipofectamine complex was incubated for 15 min at room temperature, and then added to the plate. The medium was replaced 20 to 24 h after transfection, and 1 μM of Lysosensor Green DND-189 (Thermo Fisher, L7535) was added. The cells were incubated for 30 min in a 37°C, 5% CO2 incubator. The medium was replaced one final time prior to imaging. The live cell imaging was conducted on a customized/automated fluorescence microscope (Ti-E, Nikon) using an environmental chamber (TOKAI HIT) and the culture medium to maintain normal cell culture conditions (37°C, 5% CO2, 20% O2). Transfection efficiency was estimated by counting cells that showed mKate2 red fluorescence and was typically between 25 and 35%. Fluorescence quantification and analysis was performed with ImageJ software and Prism 7/8/9 (GraphPad). Representative images were also gathered on a Leica DMi8 confocal microscope. The following peptides (>95% purity) were synthesized by Thermo Fisher/Pierce Custom Peptide team: MY18: MYSFVSEETGTLIVNSVL 6-Arg (Arg)-MY18: RRRRRR-MYSFVSEETGTLIVNSVL TAT-MY18: GRKKRRQRRRPPQ-MYSFVSEETGTLIVNSVL Penetratin (Pen)-MY18: RQIKIWFQNRRMKWKK-MYSFVSEETGTLIVNSVL Global quantitative proteomics by mass spectrometry For global quantitative proteomics of HEK 293S transfected using Lipofectamine 2000 (Invitrogen 11668027) with pcDNA3-SARS-CoV-2 Envelope (WT)-mKate2, tandem mass tag (TMT)-based quantitative proteomics was used. In brief, frozen cells were lysed by bead-beating in 9 M urea and 200 mM EPPS (pH 8.5), supplemented with protease and phosphatase inhibitors. Samples were reduced with 5 mM tris(2-carboxyethyl)phosphine and alkylated with 10 mM iodoacetamide that was quenched with 10 mM Dithiothreitol. A total of 100 μg of protein was chloroform−methanol precipitated. Protein was reconstituted in 200 mM EPPS (pH 8.5) and digested by Lys-C overnight and trypsin for 6 h, both at a 1:50 protease-to-peptide ratio. Digested peptides were quantified using a Nanodrop at 280 nm and 50 μg of peptide from each sample were labeled with 400 μg TMT reagent using 10-plex TMT kit [53]. TMT labels were checked, 0.5 μg of each sample was pooled, desalted and analyzed by short synchronous precursor selection (SPS) MS3 method, and using normalization factor, samples were bulk mixed at 1:1 across all channels and 500 μg of the bulk mixed sample was used for total proteome analysis. Mixed TMT-labeled samples were vacuum centrifuged and desalted with C18 Sep-Pak (100 mg) solid-phase extraction column. The desalted sample was fractionated using BPRP chromatography. Peptides were subjected to a 50 min linear gradient from 5 to 42% acetonitrile in 10 mM ammonium bicarbonate (pH 8) at a flow rate of 0.6 ml/min over Water X-bridge C18 column (3.5 μm particles, 4.6 mm ID, and 250 mm in length). The peptide mixture was fractionated into a total of 96 fractions, which were consolidated into 28 fractions. Fractions were subsequently acidified with 1% formic acid, and vacuum centrifuged to near dryness and desalted via SDB-RP StageTip. For total proteome analysis, 28 desalted fractions were dissolved in 10 μl of 3% acetonitrile/0.1% formic acid injected using SPS-MS3. The UltiMate 3000 UHPLC system (Thermo Fisher) and EASY-Spray PepMap RSLC C18 50 cm × 75 μm ID column (Thermo Fisher) coupled with Orbitrap Fusion (Thermo Fisher) were used to separate fractioned peptides with a 5% to 30% acetonitrile gradient in 0.1% formic acid over 45 min at a flow rate of 250 nl/min. After each gradient, the column was washed with 90% buffer B for 10 min and re-equilibrated with 98% buffer A (0.1% formic acid, 100% HPLC-grade water) for 40 min. The full MS spectra was acquired in the Orbitrap Fusion Tribrid Mass Spectrometer (Thermo Fisher) at a resolution of 120,000. The 10 most intense MS1 ions were selected for MS2 analysis. The isolation width was set at 0.7 Da and isolated precursors were fragmented by CID at normalized collision energy (NCE) of 35% and analyzed in the ion trap using “turbo” scan speed. Following the acquisition of each MS2 spectrum, a SPS-MS3 scan was collected on the top 10 most intense ions in the MS2 spectrum. SPS-MS3 precursors were fragmented by higher energy collision-induced dissociation at an NCE of 65% and analyzed using the Orbitrap. Raw mass spectrometric data were analyzed using Proteome Discoverer 2.4 to perform database search and TMT reporter ions quantification. TMT tags on lysine residues and peptide N termini (+229.163 Da) and the carbamidomethylation of cysteine residues (+57.021 Da) was set as static modifications, while the oxidation of methionine residues (+15.995 Da), deamidation (+0.984) on asparagine and glutamine were set as a variable modification. Data were searched against a UniProt human with peptide-spectrum match and protein-level at 1% FDR. The signal-to-noise (S/N) measurements of each protein were normalized so that the sum of the signal for all proteins in each channel was equivalent to account for equal protein loading. Results obtained from PD2.4 were further analyzed using Perseus statistical package [54] which is part of the MaxQuant distribution. Significantly changed protein abundance was determined by ANOVA with P < 0.05 (permutation-based FDR correction). Pathway analysis was performed using ingenuity IPA (Qiagen). Luciferase assay HEK 293T cells were plated at 0.5 × 105 cells/well in 24-well plates (Corning) coated with poly-ornithine (Sigma-Aldrich). The following day, the cells were transfected with DNAs encoding the Envelope protein of interest, an NFAT Firefly Luc (NFAT-FLuc) reporter (using 4× NFAT site from human IL-2 gene), and pRL-TK-Renilla Luc reporter (TK-RLuc, transfection control reporter using HSV TK, herpes simplex virus thymidine kinase, promoter) using Lipofectamine 2000 reagents. The standard transfection ratio of Envelope protein: NFAT-FLuc: TK-RLuc was as follows: 0.3 μg: 0.3 μg: 0.03 μg. Peptides of interest were added to this mixture at an amount of 0.1 μg. The cells were then incubated overnight at 37°C in a CO2 incubator. The following day, the cells were treated with 1 μM Phorbol 12-myristate 13-acetate (Sigma-Aldrich, P1585) for 8 h at 37°C in a CO2 incubator. Luc activity levels were then assayed using Dual Luciferase assay kit and Veritas 96-well luminometer (Promega, E1910) following the manufacturer’s instructions. Following the luciferase results, TAT-MY18-2ED peptide (>95% purity) was synthesized by Thermo Fisher/Pierce Custom Peptide team: TAT-MY18-2ED: GRKKRRQRRRPPQ-MYSFVSDDTGTLIVNSVL (M.W. = 3,662.2416). Western blotting Western blot experiments were conducted using standard method. In brief, HEK 293T cells were plated in a 6-well dish at a 1 × 106 cells/well density. Cell samples receiving iPep-SARS2-E peptide (10 μm) were treated with the peptide for 24 h prior to transfection, and the peptides were refreshed during transfection. For transfection, 4 μg of the plasmid were mixed into 125 μl of serum-free OptiMEM, and 5 μl of Lipofectamine 2000 added to another 125 μl of serum-free OptiMEM. The 2 pots were combined and incubated at room temperature for 15 min. The next day, cells were lysed in cell lysis buffer (10×, Cell Signaling Technology, #9803) with 1% protease inhibitor cocktail (Sigma-Aldrich/Millipore-Sigma). Infected Vero-E6 cells were also treated and inactivated using the same lysis buffer. Samples were denatured using 2× SDS sample buffer (8 M urea, 40 mM Tris-Cl (pH 6.8), 2% SDS, 10% 2-mercaptoethanol) and boiled at 95°C for 5 min. SDS-polyacrylamide gel electrophoresis was performed using home-made gels with Tris-Glycine (Bio-Rad) containing 10% Acrylamide-Bis (Fisher Scientific), 4% to 20% gradient or 16% pre-made gels (Novex, Thermo Fisher), which were then transferred to polyvinylidene difluoride (PVDF) membranes. Primary antibodies to anti-Spike S2 (R&D, MAB10557100, 1/5,000 dilution), anti-GFP (MBL#598, 1/8,000 dilution), and anti-GAPDH (rabbit recombinant monoclonal antibody, ab181602, 1:10,000 dilution, Abcam). Secondary antibody α-mouse (Thermo Fisher, #31430, 1/8,000 dilution) or α-rabbit (Thermo Fisher, #31460, 1/8,000 dilution) with 5% skim milk in Tris-buffered saline with 0.1% Tween 20 (TBS-T) was used for blocking of the PVDF membranes. Pierce enhanced chemiluminescence (ECL) western blotting substrate (Thermo Fisher, #32209) was used for the chemiluminescent reaction with films or ChemiDoc MP Imaging system (Bio-Rad). Monoclonal antibody production Immunization of animals: The emulsion was prepared by mixing synthesized SARS2-E N-term peptide (Thermo Fisher) in 50% Dimethylsulfoxide (DMSO) in PBS and the adjuvant complete freund (BD, #263810) evenly to make final peptide concentration of 1 mg/ml. Eight-week-old WKY/NCrl female rat purchased from Charles River Laboratories (Massachusetts, United States of America) was injected intramuscularly at the right and left tail base with 100 μl each of emulsion (total 200 μg peptide/rat) under our animal protocol (AC-AABC3508). Three times booster injections were done using the same method except at days 14, 17, and 20 where adjuvant incomplete Freund (BD, #963910) was used instead of complete Freund. Lymphocyte harvest: As a partner cell, mouse myeloma cell line NS-1 (P3/NS1/1-Ag4.1, Sigma-Aldrich, # 85011427-1VL) was used. NS-1 cells were maintained with NS-1 medium (DMEM with 20% FBS, 1× GlutaMax supplement and 1× PS; Gibco #10313–021, HyClone #SH30071.03, Gibco #35050–061 and Gibco #15140–122, respectively) at 37°C, 5% CO2. Two days after the last antigen injection, the rat was euthanized, and the medial iliac lymph node was taken aseptically and placed into 1 ml of NS-1 medium and cut to small pieces using sterile surgical blades. Following gentle pipetting to separate lymphocyte cells from other tissues, the lymphocyte cells were strained using a 100 μm pore cell strainer (Falcon, #352360). After counting, lymphocytes were frozen in NS-1 medium with 10% DMSO and kept in liquid nitrogen for storage. Cell fusion: Lymphocytes were initiated the day before fusion and cultured in NS-1 medium. Fifteen million lymphocytes and 30 million NS-1 cells were mixed and spun down at 500xg for 2 min. After washing with HBSS (Gibco, #14175095), the cell pellet was resuspended in the fusion medium (0.3 M mannitol with 0.1 mM CaCl2 and MgCl2). The mixture was then put into the fusion chamber and fused using an electrofusion method (NAPA GENE, #ECFG21, for align; 30V for 20 s. For fusion; 350 V for 30 μsec 3 times with 0.5 s interval). The mixture was then collected from the fusion chamber and spun down at 500xg for 2 min. The fusion pellet was then resuspended to first culture medium (fresh NS-1 medium mixed with equal volume NS-1 cultured conditioned medium with 1× Hymax Hybridoma Fusion & Cloning Supplement (Antibody Research Corporation, Missouri, USA, #113004)) and plated on a 96-well plate. Hybridomas were selected by HAT sectioning culture for 2 weeks with NS-1 medium with 1× HAT media supplement (Sigma-Aldrich, #H0262) and 1× Hymax supplement, followed by HT maintenance culture with NS-1 medium with 1× HT media supplement (Sigma-Aldrich, #H0137) and 1× Hymax solution. Screening: Primary screening was done by immunocytochemistry using 2E-YFP expressing plasmid transfected NIH 3T3 cells. The cells were plated on a Nunc Lab-Tek II chamber slide (Thermo Fisher, #154534) at a density of 15,000 cells/well and transfected with plasmid using Lipofectamine 3000 (Invitrogen, #L3000001) following the manufacturer’s protocol. Twenty hours later, the cells were fixed using 4% paraformaldehyde in PBS. After 3 times wash with PBS, the cells were incubated with hybridoma culture supernatant with 0.5% NP-40 for 1 h at room temperature. After 3 times wash with PBS, the cells were incubated with anti-rat IgG secondary antibody conjugated with Alexa Fluor 594 (1:2,000 dilution, Abcam, #ab150160) for 30 min at room temperature. Positive clones were expanded to 24-well plate. After reaching 80% confluency, western blot was done as a secondary screening. For western blot preparation, SARS2-E-YFP plasmid was transfected into HEK 293T cells in a 10 cm dish using Lipofectamine 2000 following manufacturer’s protocol. Twenty-four hours later, the cells were lysed using lysis buffer (Cell Signaling, #9803S) and spun down to collect the supernatant. The supernatant was mixed with same volume of the 2× SDS-urea sample buffer and boiled at 95°C for 5 min. Following the SDS page using 10% Bis-acrylamide gel, the protein was transferred to a PVDF membrane by electroblotting. Following 1 h blocking by 5% milk in TBS-T, the membrane was cut longitudinally into 0.5 cm wide strip. The strips were incubated with hybridoma culture supernatant for 1 h at room temperature. After 3 times washing with TBS-T, the strips were incubated with anti-rat IgG secondary antibody conjugated with HRP (Invitrogen, #31470) for 30 min at room temperature. After 3 times washing, development was done using ECL solution following the manufacture’s protocol. The anti-GFP antibody was used as a positive control. If positive, single-cell cloning was done using a limiting dilution method and another round of immunocytochemistry and western blotting was completed to confirm the result. Following immunocytochemistry and western blotting confirmation for the single cloned hybridomas, positive clones were passaged with gradually reduced concentration of Hymax supplement (1/2, 1/4, 1/10, and finally without Hymax) until the hybridomas could be maintained by NS-1 medium with 1× HT solution. Hybridoma isotyping was done using rat isotyping kit (Bio-Rad, #RMT1). Antibody purification: 50 ml of hybridoma culture supernatant was collected followed by filtration using a 0.22 μm filter. The culture supernatant was mixed with an equal volume of 20 mM Na-phosphate solution (pH 7.0), with additions of NaCl (150 mM, final conc.) and Tween-20 (0.02%, final conc.) as well, and 0.25 ml of Protein-G agarose (Thermo Fisher, #20399) was added and gently rocked for 1 h at room temperature. Protein-G agarose was packed in a column using an open column method and washed with 20 ml of 20 mM Na-phosphate solution (pH 7.0); 100 mM Glycine-Cl solution (pH 2.7) was used as elution buffer and eluted solution was immediately neutralized by 1/10 volume of 1 M Tris-HCl solution (pH 8.0). The neutralized antibody solution was concentrated, and buffer changed to 20 mM Na-phosphate with 0.25 M NaCl solution using an Amicon Ultra-4 filter (size of 3K, Millipore, #UFC800396). Antibody concentration was measured using a rat IgG ELISA kit (Abcam, #ab189578). Electrophysiological recording SARS2-E-PM-mKate2 construct was co-expressed with pcDNA3-MY18-2ED or pcDNA3 empty vector in HEK 293S cells as described previously [19]. Whole-cell patch-clamp recordings of the transfected mKate2-positive cells were conducted using a MultiClamp 700B patch-clamp amplifier (Molecular Devices) and an inverted microscope equipped with differential interface optics (Nikon, Ti-U). The glass pipettes were prepared using borosilicate glass (Sutter Instrument, BF150-110-10) using a micropipette puller (Sutter Instrument, Model P-97). Voltage-clamp measurements were conducted using an extracellular solution consisting of normal Tyrode solution containing 140 mM NaCl, 5.4 mM KCl, 1 mM MgCl2, 10 mM glucose, 1.8 mM CaCl2, and 10 mM HEPES (pH 7.4 with NaOH at 25°C) using the pipette solution: 120 mM K D-gluconate, 25 mM KCl, 4 mM MgATP, 2 mM NaGTP, 4 mM Na2-phospho-creatin, 10 mM EGTA, 1 mM CaCl2, and 10 mM HEPES (pH 7.4 with KCl at 25°C). The recordings were conducted using the extracellular solution warmed at 37°C. The patch-clamp data was acquired and analyzed using pClamp 10 and Clampfit 10.4 (Molecular Devices). Immunoprecipitation pcDNA3-SARS2-E-YFP with pcDNA3-6xHis-MY18-2ED or pcDNA3 empty vector were co-transfected to HEK 293T cells using Lipofectamine 2000. Twenty-four hours after the transfection, the cells were washed with PBS once and then treated with cell lysis buffer (Cell Signaling Technology). Standard purification method for His-tagged construct with Ni Sepharose 6FF (Sigma-Aldrich/Cytiva, #17-5318-01) was conducted using our established method as described previously [51]. Peptide permeability assay The peptides (>95% purity) were synthesized by Thermo Fisher/Pierce Custom Peptide team. Peptides were resolved in water and stored at a 2.5 mM stock concentration. For permeability assay, NIH 3T3 and Vero-E6 cells were plated on 35 mm dishes at a density of 1 × 105 cells/ml. ON kinetics: peptides were added to cell culture medium at a concentration of 10 μM and incubated under normoxia conditions until time points for imaging. OFF kinetics: peptides were added to the cell medium at a concentration of 10 μM and incubated under normoxia conditions. After 24 h, the medium was replaced (no new peptide added). Cells were then imaged at determined time points. For all imaging, cell media was replaced with Tyrode’s solution (140 mM NaCl, 5.4 mM KCl, 1 mM MgCl2, 10 mM glucose, 1.8 mM CaCl2, and 10 mM HEPES, pH buffered to 7.4 with NaOH). Two types of Alexa Fluor 594-conjugated MY18-2ED peptides: Alexa Fluor 594-TAT-MY18-2ED-: Alexa Fluor 594-[C]G-GRKKRRQRRRPPQ-MYSFVSDDTGTLIVNSVL TAT-MY18-2ED-Alexa Fluor 594: GRKKRRQRRRPPQ-MYSFVSDDTGTLIVNSVL-L[C]-Alexa Fluor 594 Stability assay of peptides Peptide was synthesized by Thermo Scientific/Pierce and resolved in water at a 2.5 mM stock concentration and aliquoted for storage at −80°C. Peptides were thawed on ice and diluted to 40 μg/ml using PBS. Samples were incubated at 37°C. Baseline levels were measured using freshly thawed peptide diluted by 37°C pre-warmed PBS. Samples were transferred to an EIA/RIA plate (Corning, #3591) at 50 μl/well and incubated overnight for coating at 4°C. The next day, the plate was washed using 100 μl/well of PBS-T 3 times. The plate was blocked for 1 h at room temperature using 3% bovine serum albumin (BSA, Sigma-Aldrich) in PBS-T with 250 rpm shaking. Following this, the plate was washed 3 times using PBS-T. The primary antibody solution was then added at 50 μl/well (N2A5E8, 0.45 μg/ml) for 1 h at room temperature with 250 rpm shaking followed by 3 times wash with PBS-T. The anti-rat IgG secondary antibody HRP conjugated solution (1:10,000) was then added for 1 h at room temperature with 250 rpm shaking followed by 3 times wash with PBS-T. The development step was done using TMB solution (Thermo Fisher, #34021) following the manufacturer’s manual. OD600 value was detected using a SpectraMax iD3 Plate Reader (Molecular Devices). Statistical analysis was done using GraphPad Prism software. This procedure was done in biological triplicate. Apoptosis/necrosis assay using peptides Jurkat cells (ATCC, #TIB-152, clone E6-1) were cultured and treated for 48 h with 10 μM of iPep-SARS2-E peptide (TAT-MY18-2ED) or for 3 h with 10 μM (S)-(+)-camptothecin (positive control group, Sigma-Aldrich, #C9911). After treatment, the cells were collected for an Annexin-V assay (Thermo Fisher/Invitrogen, #V13241) following the manufacturer’s manual. The data was collected by a ZE5 Cell Analyzer (Bio-Rad Laboratories) and analyzed using FlowJo software (BD Biosciences). This procedure was done in biological triplicate. Viral infection in vitro and sample processing Cytopathic assay using WA10 virus (MOI, 0.10) in Vero-E6 cells was conducted as done in the previous study using standard methods [5]. Immunocytochemistry was conducted using a standard method using fixation solution containing 4% paraformaldehyde (Electron Microscopy Sciences) and 2% sucrose (Sigma-Aldrich) in PBS, blocking/permeability solution (2% BSA and 0.25% NP-40, Sigma-Aldrich, in PBS) and antibodies to ERGIC/p58 (Sigma-Aldrich, E1031), LAMP1 (Abcam, ab24170), BiP/GPR78 (Abcam, ab21685), and SARS-CoV-2 nucleocapsid (Thermo Fisher, PIMA17404). Goat anti-mouse IgG Alexa Fluor 594 antibody (Abcam, ab150116) and Goat anti-rabbit IgG Alexa Fluor 488 antibody (Abcam, ab150077) were used. For electron microscopy, infected Vero-E6 cells were fixed using 2% paraformaldehyde, 2% glutaraldehyde (Electron Microscopy Sciences), and 2 mM CaCl2 (Sigma-Aldrich) in 100 mM cacodylate buffer (pH 7.4, Electron Microscopy Sciences). For human branching lung organoids, EpCAM antibody (Abcam. ab223582) was used. Quantitative RT-PCR RNA samples of Vero-E6 cells and mouse lung tissues were prepared using TRIzol Plus RNA Purification kit and PureLink DNase set (Thermo Fisher) while RNeasy Mini kit and RNase-Free DNase set (Qiagen) was used for HEK 293 cells. cDNA was synthesized using the SuperScript III First-Strand Synthesis System for RT–PCR (Thermo Fisher). FAST or Power SYBR Green PCR Master Mix and QuantStudio 3/7 real time PCR systems (Thermo Fisher) with StepOne software (version 2.3, Life Technologies) or CFX Opus 96 Real-Time PCR system (Bio-Rad) were used for qPCR using the below primer sets. SARS-CoV-2 qPCR N forward primer: CTCTTGTAGATCTGTTCTCTAAACGAAC SARS-CoV-2 qPCR N reverse primer: GGTCCACCAAACGTAATGCG SARS-CoV-2 qPCR E forward primer: CTCATTCGTTTCGGAAGAGACAG SARS-CoV-2 qPCR E reverse primer: AGACCAGAAGATCAGGAACTCTAG Mouse Gapdh qPCR forward primer: CTTCACCACCATGGAGAAGG Mouse Gapdh qPCR reverse primer: TGAAGTCGCAGGAGACAACC Monkey qPCR GAPDH (for Vero-E6) forward primer: GAAGGTGAAGGTCGGAGTCAAC Monkey qPCR GAPDH (for Vero-E6) reverse primer: TCGTTGTCATACCAGGAAATGAGC Human NFATC4 qPCR forward primer: CTTCTCCGATGCCTCTGACG Human NFATC4 qPCR reverse primer: CGGGGCTTGGACCATACAG Human JUN/AP-1 qPCR forward primer: ACTCGGACCTTCTCACGTC Human JUN/AP-1 qPCR reverse primer: GGTCGGTGTAGTGGTGATGT Human GAPDH qPCR forward primer: GATGACATCAAGAAGGTGGTGA Human GAPDH qPCR reverse primer: GTCTACATGGCAACTGTGAGGA Viral infection in vivo and sample processing Balb/c mice (8 to 11 weeks old, both male and female, Charles River) were infected intranasally with 5 × 104 PFU of SARS-CoV-2 (MA10) in a final volume of 50 μl (a single dose) with either mock (PBS), the negative control peptide, or iPep-SARS2-E treatment, following isoflurane sedation. After viral infection, mice were monitored daily for body weight, temperature, and foods. Mice showing >20% loss of their initial body weight were defined as reaching experimental endpoint and humanely euthanized before day 4. The peptides were provided intravenously (i.v., 2 mM, 150 μl in PBS, pH 7.0 adjusted with NaOH, a single dose), following a previous peptide-related study [9]. In case of a single dose of intranasal administration, the peptides, iPep-SARS2-E and the negative control mutant peptide, were used together with the infection under isoflurane sedation (2.5 mM, 50 μl in PBS, pH 7.0 adjusted with NaOH). The lung tissue samples were collected at the endpoint (4 days post-infection) for RNA preparation, lung histology, and/or lung viral titer using standard methods [9] as well as our optimized method of SARS2-E protein blotting described as the next section. Lung tissue western blotting To inactivate the virus, MA10-infected mouse lung was incubated in 0.5% SDS in PBS for 1 h at room temperature. After being mashed by a plastic masher, the sample mixture was spun down to collect the supernatant. The supernatant was mixed with same volume of 2× SDS-urea sample buffer and boiled at 95°C for 5 min. Following the SDS page using 20% Bis-acrylamide gel, the protein was transferred to a PVDF membrane by electroblotting. Following 1 h blocking by 5% skim milk in TBS-T, the membrane was incubated with the primary antibody solution (N2A5E8, 0.2 μg/ml in TBS-T) for overnight at 4°C. After 3 times washing with TBS-T, the membrane was incubated with anti-rat IgG secondary antibody conjugated with HRP for 1 h at room temperature. After 3 times washing, development was done using ECL solution following the manufacture’s protocol. Fluorescent stereoscopic imaging of mouse tissues iPep-SARS2-E peptide conjugated with Alexa594 at the C-terminus end (TAT-MY18-2ED-A594, 10 μM, 20 μl) or PBS was provided intranasally under the Isoflurane anesthesia. Two hours later, the mice were euthanized using CO2. The skull was cut with sagittal section in the middle, followed by taking out the nasal septum. After 3 times wash in PBS, imaging for the lateral side of nasal cavity was conducted on a fluorescent stereoscope (Leica, M165 FC). Regarding i.v. injection method, iPep-SARS2-E group mice were injected with TAT-MY18-2ED-A594 peptide (300 μM, 50 μl) via tail vein before 24 h or 2 h of sacrifice. The control mouse was injected with PBS before 2 h of sacrifice. After sacrifice using CO2 and cervical dislocation, whole body perfusion with 15 ml of PBS was conducted to wash out the peptides in their bloods. Harvested tissues, such as lungs, were briefly washed by PBS and imaging was conducted on the fluorescent stereoscope. Cytokine/Inflammation array Cytokine inflammation panel was done in accordance with the manufacturer’s protocol for the mouse cytokine array kit panel A (R&D Systems, Cat# ARY006). Serum samples used were collected from blood heat-inactivated at 65°C for 30 min (to inactivate any possible viruses) and spun down at 15,000 rpm for 10 min. Approximately 50 μl of blood serum samples were used for the assay. Cxcl12, C5a, MCSF, and CD54 were detected in the array using the denatured blood samples. Pseudo-virus design and production Following a previous study [44], our pseudo viruses were designed and produced using HEK 293T cells transfected with pCMV-SARS-CoV-2 Spike (delta variant) and pcDNA3-M-P2A-E, pCMV-dR8.2 dvpr packaging plasmid (Addgene, #8455) and YFP reporter expressing plasmid, which was generated using standard PCR and LV-Cre-SD (#12105, no longer available at Addgene) with EcoRI and XhoI cloning sites to subclone YFP. The virus was filtered using a 0.45 μm syringe and then added to Vero-E6 cells (MOI, approximately 0.05). The YFP imaging was conducted 48 h post-infection using an epi-fluorescent microscopy (Nikon, TS100F). Statistics and reproducibility The statistics used for every figure have been indicated in the corresponding figure legends. The Student’s t test (paired and unpaired) was conducted with the t test functions in Microsoft Excel software. The Student’s t test was two-tailed. The one-way ANOVA with Tukey’s, Sidak’s, Bonferroni’s, or Dunnett’s post hoc multiple comparison analysis was conducted with the GraphPad Prism 6/7/8/9 software. All the data meet the assumptions of the statistical tests. All the samples used in this study were biological repeats, not technical repeats. All experiments, except ELISA for the peptide-antibody binding assay (S2C Fig), were conducted using at least 2 independent experimental materials/cohorts to reproduce similar results. No samples were excluded from the analysis in this study. All the graphs in the figures are mean ± SD. Ethical statement Columbia University is a National Institute of Health (NIH) Office of Laboratory Animal Welfare assured institution (ID#, D16-00003/A3007-01), which has complied with the NIH Public Health Service Policy and adhered to the standards in the guide for the care and use of laboratory animals. The animal study is approved by Columbia University Institutional Animal Care and Use Committee (protocol #, AC-AABP2571). This virus study using SARS-CoV-2 is approved by Columbia University ADARC committee and BSL3 facility committee. Cell culture Human embryonic kidney (HEK) 293S cells (ATCC, Cat#CRL-3022) were cultured in Dulbecco’s Modified Eagle Media Nutrient Mixture F-12 (DMEM/F-12, Thermo Fisher Gibco # 11320033) and HEK 293T cells (ATCC, Cat#CRL-3216), NIH 3T3 cells (ATCC, Cat#CRL-1658), and Vero-E6 cells (Cat# CRL-1586) were cultured in and DMEM (Thermo Fisher/Gibco #10313021). Both media were supplemented with GlutaMax-I and penicillin/streptomycin (PS) and 10% fetal bovine serum (FBS, not heat-inactivated, HyClone, #SH30071.03, Thermo Fisher) under normoxia (20% O2, 5% CO2, at 37°C). The cell lines were passaged using trypsin-EDTA (0.25%, Thermo Fisher, # 25200–056) every 2 to 3 days. Human branching lung organoids were prepared using our normal human-induced pluripotent stem cell lines [52] with an established differentiation medium kit (STEMCELL Technologies, #100-0195/0528). Molecular biology constructs Plasmid DNA constructs were generated using standard methods with restriction enzymes (New England BioLabs), DNA ligase (MightyMix, TaKaRa Bio/Clontech), and polymerase chain reaction (PCR) with Phusion polymerase (Thermo Fisher). Construct inserts for these experiments were synthesized (Integrated DNA Technologies, IDT) and subcloned into pcDNA3 vector (Life Technologies). Mock transfections were performed by using pcDNA3 empty vector. Imaging experiments and pH measurements For the imaging experiments in Figs 1 and 7, NIH 3T3 cells were plated at a cell density of 2.5 × 105 cells/ml. The following day, the cells were transfected using Lipofectamine 3000 reagents (Thermo Fisher, # L3000001), and 4 μg of plasmid were mixed into 125 μl of serum-free OptiMEM with 5 μl of P3000 reagent. This was then added to another 125 μl of serum-free OptiMEM containing 7.5 μl of Lipofectamine 3000. Plasmid/P3000-lipofectamine complex was incubated for 15 min at room temperature, and then added to the plate. The medium was replaced 20 to 24 h after transfection, and 1 μM of Lysosensor Green DND-189 (Thermo Fisher, L7535) was added. The cells were incubated for 30 min in a 37°C, 5% CO2 incubator. The medium was replaced one final time prior to imaging. The live cell imaging was conducted on a customized/automated fluorescence microscope (Ti-E, Nikon) using an environmental chamber (TOKAI HIT) and the culture medium to maintain normal cell culture conditions (37°C, 5% CO2, 20% O2). Transfection efficiency was estimated by counting cells that showed mKate2 red fluorescence and was typically between 25 and 35%. Fluorescence quantification and analysis was performed with ImageJ software and Prism 7/8/9 (GraphPad). Representative images were also gathered on a Leica DMi8 confocal microscope. The following peptides (>95% purity) were synthesized by Thermo Fisher/Pierce Custom Peptide team: MY18: MYSFVSEETGTLIVNSVL 6-Arg (Arg)-MY18: RRRRRR-MYSFVSEETGTLIVNSVL TAT-MY18: GRKKRRQRRRPPQ-MYSFVSEETGTLIVNSVL Penetratin (Pen)-MY18: RQIKIWFQNRRMKWKK-MYSFVSEETGTLIVNSVL Global quantitative proteomics by mass spectrometry For global quantitative proteomics of HEK 293S transfected using Lipofectamine 2000 (Invitrogen 11668027) with pcDNA3-SARS-CoV-2 Envelope (WT)-mKate2, tandem mass tag (TMT)-based quantitative proteomics was used. In brief, frozen cells were lysed by bead-beating in 9 M urea and 200 mM EPPS (pH 8.5), supplemented with protease and phosphatase inhibitors. Samples were reduced with 5 mM tris(2-carboxyethyl)phosphine and alkylated with 10 mM iodoacetamide that was quenched with 10 mM Dithiothreitol. A total of 100 μg of protein was chloroform−methanol precipitated. Protein was reconstituted in 200 mM EPPS (pH 8.5) and digested by Lys-C overnight and trypsin for 6 h, both at a 1:50 protease-to-peptide ratio. Digested peptides were quantified using a Nanodrop at 280 nm and 50 μg of peptide from each sample were labeled with 400 μg TMT reagent using 10-plex TMT kit [53]. TMT labels were checked, 0.5 μg of each sample was pooled, desalted and analyzed by short synchronous precursor selection (SPS) MS3 method, and using normalization factor, samples were bulk mixed at 1:1 across all channels and 500 μg of the bulk mixed sample was used for total proteome analysis. Mixed TMT-labeled samples were vacuum centrifuged and desalted with C18 Sep-Pak (100 mg) solid-phase extraction column. The desalted sample was fractionated using BPRP chromatography. Peptides were subjected to a 50 min linear gradient from 5 to 42% acetonitrile in 10 mM ammonium bicarbonate (pH 8) at a flow rate of 0.6 ml/min over Water X-bridge C18 column (3.5 μm particles, 4.6 mm ID, and 250 mm in length). The peptide mixture was fractionated into a total of 96 fractions, which were consolidated into 28 fractions. Fractions were subsequently acidified with 1% formic acid, and vacuum centrifuged to near dryness and desalted via SDB-RP StageTip. For total proteome analysis, 28 desalted fractions were dissolved in 10 μl of 3% acetonitrile/0.1% formic acid injected using SPS-MS3. The UltiMate 3000 UHPLC system (Thermo Fisher) and EASY-Spray PepMap RSLC C18 50 cm × 75 μm ID column (Thermo Fisher) coupled with Orbitrap Fusion (Thermo Fisher) were used to separate fractioned peptides with a 5% to 30% acetonitrile gradient in 0.1% formic acid over 45 min at a flow rate of 250 nl/min. After each gradient, the column was washed with 90% buffer B for 10 min and re-equilibrated with 98% buffer A (0.1% formic acid, 100% HPLC-grade water) for 40 min. The full MS spectra was acquired in the Orbitrap Fusion Tribrid Mass Spectrometer (Thermo Fisher) at a resolution of 120,000. The 10 most intense MS1 ions were selected for MS2 analysis. The isolation width was set at 0.7 Da and isolated precursors were fragmented by CID at normalized collision energy (NCE) of 35% and analyzed in the ion trap using “turbo” scan speed. Following the acquisition of each MS2 spectrum, a SPS-MS3 scan was collected on the top 10 most intense ions in the MS2 spectrum. SPS-MS3 precursors were fragmented by higher energy collision-induced dissociation at an NCE of 65% and analyzed using the Orbitrap. Raw mass spectrometric data were analyzed using Proteome Discoverer 2.4 to perform database search and TMT reporter ions quantification. TMT tags on lysine residues and peptide N termini (+229.163 Da) and the carbamidomethylation of cysteine residues (+57.021 Da) was set as static modifications, while the oxidation of methionine residues (+15.995 Da), deamidation (+0.984) on asparagine and glutamine were set as a variable modification. Data were searched against a UniProt human with peptide-spectrum match and protein-level at 1% FDR. The signal-to-noise (S/N) measurements of each protein were normalized so that the sum of the signal for all proteins in each channel was equivalent to account for equal protein loading. Results obtained from PD2.4 were further analyzed using Perseus statistical package [54] which is part of the MaxQuant distribution. Significantly changed protein abundance was determined by ANOVA with P < 0.05 (permutation-based FDR correction). Pathway analysis was performed using ingenuity IPA (Qiagen). Luciferase assay HEK 293T cells were plated at 0.5 × 105 cells/well in 24-well plates (Corning) coated with poly-ornithine (Sigma-Aldrich). The following day, the cells were transfected with DNAs encoding the Envelope protein of interest, an NFAT Firefly Luc (NFAT-FLuc) reporter (using 4× NFAT site from human IL-2 gene), and pRL-TK-Renilla Luc reporter (TK-RLuc, transfection control reporter using HSV TK, herpes simplex virus thymidine kinase, promoter) using Lipofectamine 2000 reagents. The standard transfection ratio of Envelope protein: NFAT-FLuc: TK-RLuc was as follows: 0.3 μg: 0.3 μg: 0.03 μg. Peptides of interest were added to this mixture at an amount of 0.1 μg. The cells were then incubated overnight at 37°C in a CO2 incubator. The following day, the cells were treated with 1 μM Phorbol 12-myristate 13-acetate (Sigma-Aldrich, P1585) for 8 h at 37°C in a CO2 incubator. Luc activity levels were then assayed using Dual Luciferase assay kit and Veritas 96-well luminometer (Promega, E1910) following the manufacturer’s instructions. Following the luciferase results, TAT-MY18-2ED peptide (>95% purity) was synthesized by Thermo Fisher/Pierce Custom Peptide team: TAT-MY18-2ED: GRKKRRQRRRPPQ-MYSFVSDDTGTLIVNSVL (M.W. = 3,662.2416). Western blotting Western blot experiments were conducted using standard method. In brief, HEK 293T cells were plated in a 6-well dish at a 1 × 106 cells/well density. Cell samples receiving iPep-SARS2-E peptide (10 μm) were treated with the peptide for 24 h prior to transfection, and the peptides were refreshed during transfection. For transfection, 4 μg of the plasmid were mixed into 125 μl of serum-free OptiMEM, and 5 μl of Lipofectamine 2000 added to another 125 μl of serum-free OptiMEM. The 2 pots were combined and incubated at room temperature for 15 min. The next day, cells were lysed in cell lysis buffer (10×, Cell Signaling Technology, #9803) with 1% protease inhibitor cocktail (Sigma-Aldrich/Millipore-Sigma). Infected Vero-E6 cells were also treated and inactivated using the same lysis buffer. Samples were denatured using 2× SDS sample buffer (8 M urea, 40 mM Tris-Cl (pH 6.8), 2% SDS, 10% 2-mercaptoethanol) and boiled at 95°C for 5 min. SDS-polyacrylamide gel electrophoresis was performed using home-made gels with Tris-Glycine (Bio-Rad) containing 10% Acrylamide-Bis (Fisher Scientific), 4% to 20% gradient or 16% pre-made gels (Novex, Thermo Fisher), which were then transferred to polyvinylidene difluoride (PVDF) membranes. Primary antibodies to anti-Spike S2 (R&D, MAB10557100, 1/5,000 dilution), anti-GFP (MBL#598, 1/8,000 dilution), and anti-GAPDH (rabbit recombinant monoclonal antibody, ab181602, 1:10,000 dilution, Abcam). Secondary antibody α-mouse (Thermo Fisher, #31430, 1/8,000 dilution) or α-rabbit (Thermo Fisher, #31460, 1/8,000 dilution) with 5% skim milk in Tris-buffered saline with 0.1% Tween 20 (TBS-T) was used for blocking of the PVDF membranes. Pierce enhanced chemiluminescence (ECL) western blotting substrate (Thermo Fisher, #32209) was used for the chemiluminescent reaction with films or ChemiDoc MP Imaging system (Bio-Rad). Monoclonal antibody production Immunization of animals: The emulsion was prepared by mixing synthesized SARS2-E N-term peptide (Thermo Fisher) in 50% Dimethylsulfoxide (DMSO) in PBS and the adjuvant complete freund (BD, #263810) evenly to make final peptide concentration of 1 mg/ml. Eight-week-old WKY/NCrl female rat purchased from Charles River Laboratories (Massachusetts, United States of America) was injected intramuscularly at the right and left tail base with 100 μl each of emulsion (total 200 μg peptide/rat) under our animal protocol (AC-AABC3508). Three times booster injections were done using the same method except at days 14, 17, and 20 where adjuvant incomplete Freund (BD, #963910) was used instead of complete Freund. Lymphocyte harvest: As a partner cell, mouse myeloma cell line NS-1 (P3/NS1/1-Ag4.1, Sigma-Aldrich, # 85011427-1VL) was used. NS-1 cells were maintained with NS-1 medium (DMEM with 20% FBS, 1× GlutaMax supplement and 1× PS; Gibco #10313–021, HyClone #SH30071.03, Gibco #35050–061 and Gibco #15140–122, respectively) at 37°C, 5% CO2. Two days after the last antigen injection, the rat was euthanized, and the medial iliac lymph node was taken aseptically and placed into 1 ml of NS-1 medium and cut to small pieces using sterile surgical blades. Following gentle pipetting to separate lymphocyte cells from other tissues, the lymphocyte cells were strained using a 100 μm pore cell strainer (Falcon, #352360). After counting, lymphocytes were frozen in NS-1 medium with 10% DMSO and kept in liquid nitrogen for storage. Cell fusion: Lymphocytes were initiated the day before fusion and cultured in NS-1 medium. Fifteen million lymphocytes and 30 million NS-1 cells were mixed and spun down at 500xg for 2 min. After washing with HBSS (Gibco, #14175095), the cell pellet was resuspended in the fusion medium (0.3 M mannitol with 0.1 mM CaCl2 and MgCl2). The mixture was then put into the fusion chamber and fused using an electrofusion method (NAPA GENE, #ECFG21, for align; 30V for 20 s. For fusion; 350 V for 30 μsec 3 times with 0.5 s interval). The mixture was then collected from the fusion chamber and spun down at 500xg for 2 min. The fusion pellet was then resuspended to first culture medium (fresh NS-1 medium mixed with equal volume NS-1 cultured conditioned medium with 1× Hymax Hybridoma Fusion & Cloning Supplement (Antibody Research Corporation, Missouri, USA, #113004)) and plated on a 96-well plate. Hybridomas were selected by HAT sectioning culture for 2 weeks with NS-1 medium with 1× HAT media supplement (Sigma-Aldrich, #H0262) and 1× Hymax supplement, followed by HT maintenance culture with NS-1 medium with 1× HT media supplement (Sigma-Aldrich, #H0137) and 1× Hymax solution. Screening: Primary screening was done by immunocytochemistry using 2E-YFP expressing plasmid transfected NIH 3T3 cells. The cells were plated on a Nunc Lab-Tek II chamber slide (Thermo Fisher, #154534) at a density of 15,000 cells/well and transfected with plasmid using Lipofectamine 3000 (Invitrogen, #L3000001) following the manufacturer’s protocol. Twenty hours later, the cells were fixed using 4% paraformaldehyde in PBS. After 3 times wash with PBS, the cells were incubated with hybridoma culture supernatant with 0.5% NP-40 for 1 h at room temperature. After 3 times wash with PBS, the cells were incubated with anti-rat IgG secondary antibody conjugated with Alexa Fluor 594 (1:2,000 dilution, Abcam, #ab150160) for 30 min at room temperature. Positive clones were expanded to 24-well plate. After reaching 80% confluency, western blot was done as a secondary screening. For western blot preparation, SARS2-E-YFP plasmid was transfected into HEK 293T cells in a 10 cm dish using Lipofectamine 2000 following manufacturer’s protocol. Twenty-four hours later, the cells were lysed using lysis buffer (Cell Signaling, #9803S) and spun down to collect the supernatant. The supernatant was mixed with same volume of the 2× SDS-urea sample buffer and boiled at 95°C for 5 min. Following the SDS page using 10% Bis-acrylamide gel, the protein was transferred to a PVDF membrane by electroblotting. Following 1 h blocking by 5% milk in TBS-T, the membrane was cut longitudinally into 0.5 cm wide strip. The strips were incubated with hybridoma culture supernatant for 1 h at room temperature. After 3 times washing with TBS-T, the strips were incubated with anti-rat IgG secondary antibody conjugated with HRP (Invitrogen, #31470) for 30 min at room temperature. After 3 times washing, development was done using ECL solution following the manufacture’s protocol. The anti-GFP antibody was used as a positive control. If positive, single-cell cloning was done using a limiting dilution method and another round of immunocytochemistry and western blotting was completed to confirm the result. Following immunocytochemistry and western blotting confirmation for the single cloned hybridomas, positive clones were passaged with gradually reduced concentration of Hymax supplement (1/2, 1/4, 1/10, and finally without Hymax) until the hybridomas could be maintained by NS-1 medium with 1× HT solution. Hybridoma isotyping was done using rat isotyping kit (Bio-Rad, #RMT1). Antibody purification: 50 ml of hybridoma culture supernatant was collected followed by filtration using a 0.22 μm filter. The culture supernatant was mixed with an equal volume of 20 mM Na-phosphate solution (pH 7.0), with additions of NaCl (150 mM, final conc.) and Tween-20 (0.02%, final conc.) as well, and 0.25 ml of Protein-G agarose (Thermo Fisher, #20399) was added and gently rocked for 1 h at room temperature. Protein-G agarose was packed in a column using an open column method and washed with 20 ml of 20 mM Na-phosphate solution (pH 7.0); 100 mM Glycine-Cl solution (pH 2.7) was used as elution buffer and eluted solution was immediately neutralized by 1/10 volume of 1 M Tris-HCl solution (pH 8.0). The neutralized antibody solution was concentrated, and buffer changed to 20 mM Na-phosphate with 0.25 M NaCl solution using an Amicon Ultra-4 filter (size of 3K, Millipore, #UFC800396). Antibody concentration was measured using a rat IgG ELISA kit (Abcam, #ab189578). Electrophysiological recording SARS2-E-PM-mKate2 construct was co-expressed with pcDNA3-MY18-2ED or pcDNA3 empty vector in HEK 293S cells as described previously [19]. Whole-cell patch-clamp recordings of the transfected mKate2-positive cells were conducted using a MultiClamp 700B patch-clamp amplifier (Molecular Devices) and an inverted microscope equipped with differential interface optics (Nikon, Ti-U). The glass pipettes were prepared using borosilicate glass (Sutter Instrument, BF150-110-10) using a micropipette puller (Sutter Instrument, Model P-97). Voltage-clamp measurements were conducted using an extracellular solution consisting of normal Tyrode solution containing 140 mM NaCl, 5.4 mM KCl, 1 mM MgCl2, 10 mM glucose, 1.8 mM CaCl2, and 10 mM HEPES (pH 7.4 with NaOH at 25°C) using the pipette solution: 120 mM K D-gluconate, 25 mM KCl, 4 mM MgATP, 2 mM NaGTP, 4 mM Na2-phospho-creatin, 10 mM EGTA, 1 mM CaCl2, and 10 mM HEPES (pH 7.4 with KCl at 25°C). The recordings were conducted using the extracellular solution warmed at 37°C. The patch-clamp data was acquired and analyzed using pClamp 10 and Clampfit 10.4 (Molecular Devices). Immunoprecipitation pcDNA3-SARS2-E-YFP with pcDNA3-6xHis-MY18-2ED or pcDNA3 empty vector were co-transfected to HEK 293T cells using Lipofectamine 2000. Twenty-four hours after the transfection, the cells were washed with PBS once and then treated with cell lysis buffer (Cell Signaling Technology). Standard purification method for His-tagged construct with Ni Sepharose 6FF (Sigma-Aldrich/Cytiva, #17-5318-01) was conducted using our established method as described previously [51]. Peptide permeability assay The peptides (>95% purity) were synthesized by Thermo Fisher/Pierce Custom Peptide team. Peptides were resolved in water and stored at a 2.5 mM stock concentration. For permeability assay, NIH 3T3 and Vero-E6 cells were plated on 35 mm dishes at a density of 1 × 105 cells/ml. ON kinetics: peptides were added to cell culture medium at a concentration of 10 μM and incubated under normoxia conditions until time points for imaging. OFF kinetics: peptides were added to the cell medium at a concentration of 10 μM and incubated under normoxia conditions. After 24 h, the medium was replaced (no new peptide added). Cells were then imaged at determined time points. For all imaging, cell media was replaced with Tyrode’s solution (140 mM NaCl, 5.4 mM KCl, 1 mM MgCl2, 10 mM glucose, 1.8 mM CaCl2, and 10 mM HEPES, pH buffered to 7.4 with NaOH). Two types of Alexa Fluor 594-conjugated MY18-2ED peptides: Alexa Fluor 594-TAT-MY18-2ED-: Alexa Fluor 594-[C]G-GRKKRRQRRRPPQ-MYSFVSDDTGTLIVNSVL TAT-MY18-2ED-Alexa Fluor 594: GRKKRRQRRRPPQ-MYSFVSDDTGTLIVNSVL-L[C]-Alexa Fluor 594 Stability assay of peptides Peptide was synthesized by Thermo Scientific/Pierce and resolved in water at a 2.5 mM stock concentration and aliquoted for storage at −80°C. Peptides were thawed on ice and diluted to 40 μg/ml using PBS. Samples were incubated at 37°C. Baseline levels were measured using freshly thawed peptide diluted by 37°C pre-warmed PBS. Samples were transferred to an EIA/RIA plate (Corning, #3591) at 50 μl/well and incubated overnight for coating at 4°C. The next day, the plate was washed using 100 μl/well of PBS-T 3 times. The plate was blocked for 1 h at room temperature using 3% bovine serum albumin (BSA, Sigma-Aldrich) in PBS-T with 250 rpm shaking. Following this, the plate was washed 3 times using PBS-T. The primary antibody solution was then added at 50 μl/well (N2A5E8, 0.45 μg/ml) for 1 h at room temperature with 250 rpm shaking followed by 3 times wash with PBS-T. The anti-rat IgG secondary antibody HRP conjugated solution (1:10,000) was then added for 1 h at room temperature with 250 rpm shaking followed by 3 times wash with PBS-T. The development step was done using TMB solution (Thermo Fisher, #34021) following the manufacturer’s manual. OD600 value was detected using a SpectraMax iD3 Plate Reader (Molecular Devices). Statistical analysis was done using GraphPad Prism software. This procedure was done in biological triplicate. Apoptosis/necrosis assay using peptides Jurkat cells (ATCC, #TIB-152, clone E6-1) were cultured and treated for 48 h with 10 μM of iPep-SARS2-E peptide (TAT-MY18-2ED) or for 3 h with 10 μM (S)-(+)-camptothecin (positive control group, Sigma-Aldrich, #C9911). After treatment, the cells were collected for an Annexin-V assay (Thermo Fisher/Invitrogen, #V13241) following the manufacturer’s manual. The data was collected by a ZE5 Cell Analyzer (Bio-Rad Laboratories) and analyzed using FlowJo software (BD Biosciences). This procedure was done in biological triplicate. Viral infection in vitro and sample processing Cytopathic assay using WA10 virus (MOI, 0.10) in Vero-E6 cells was conducted as done in the previous study using standard methods [5]. Immunocytochemistry was conducted using a standard method using fixation solution containing 4% paraformaldehyde (Electron Microscopy Sciences) and 2% sucrose (Sigma-Aldrich) in PBS, blocking/permeability solution (2% BSA and 0.25% NP-40, Sigma-Aldrich, in PBS) and antibodies to ERGIC/p58 (Sigma-Aldrich, E1031), LAMP1 (Abcam, ab24170), BiP/GPR78 (Abcam, ab21685), and SARS-CoV-2 nucleocapsid (Thermo Fisher, PIMA17404). Goat anti-mouse IgG Alexa Fluor 594 antibody (Abcam, ab150116) and Goat anti-rabbit IgG Alexa Fluor 488 antibody (Abcam, ab150077) were used. For electron microscopy, infected Vero-E6 cells were fixed using 2% paraformaldehyde, 2% glutaraldehyde (Electron Microscopy Sciences), and 2 mM CaCl2 (Sigma-Aldrich) in 100 mM cacodylate buffer (pH 7.4, Electron Microscopy Sciences). For human branching lung organoids, EpCAM antibody (Abcam. ab223582) was used. Quantitative RT-PCR RNA samples of Vero-E6 cells and mouse lung tissues were prepared using TRIzol Plus RNA Purification kit and PureLink DNase set (Thermo Fisher) while RNeasy Mini kit and RNase-Free DNase set (Qiagen) was used for HEK 293 cells. cDNA was synthesized using the SuperScript III First-Strand Synthesis System for RT–PCR (Thermo Fisher). FAST or Power SYBR Green PCR Master Mix and QuantStudio 3/7 real time PCR systems (Thermo Fisher) with StepOne software (version 2.3, Life Technologies) or CFX Opus 96 Real-Time PCR system (Bio-Rad) were used for qPCR using the below primer sets. SARS-CoV-2 qPCR N forward primer: CTCTTGTAGATCTGTTCTCTAAACGAAC SARS-CoV-2 qPCR N reverse primer: GGTCCACCAAACGTAATGCG SARS-CoV-2 qPCR E forward primer: CTCATTCGTTTCGGAAGAGACAG SARS-CoV-2 qPCR E reverse primer: AGACCAGAAGATCAGGAACTCTAG Mouse Gapdh qPCR forward primer: CTTCACCACCATGGAGAAGG Mouse Gapdh qPCR reverse primer: TGAAGTCGCAGGAGACAACC Monkey qPCR GAPDH (for Vero-E6) forward primer: GAAGGTGAAGGTCGGAGTCAAC Monkey qPCR GAPDH (for Vero-E6) reverse primer: TCGTTGTCATACCAGGAAATGAGC Human NFATC4 qPCR forward primer: CTTCTCCGATGCCTCTGACG Human NFATC4 qPCR reverse primer: CGGGGCTTGGACCATACAG Human JUN/AP-1 qPCR forward primer: ACTCGGACCTTCTCACGTC Human JUN/AP-1 qPCR reverse primer: GGTCGGTGTAGTGGTGATGT Human GAPDH qPCR forward primer: GATGACATCAAGAAGGTGGTGA Human GAPDH qPCR reverse primer: GTCTACATGGCAACTGTGAGGA Viral infection in vivo and sample processing Balb/c mice (8 to 11 weeks old, both male and female, Charles River) were infected intranasally with 5 × 104 PFU of SARS-CoV-2 (MA10) in a final volume of 50 μl (a single dose) with either mock (PBS), the negative control peptide, or iPep-SARS2-E treatment, following isoflurane sedation. After viral infection, mice were monitored daily for body weight, temperature, and foods. Mice showing >20% loss of their initial body weight were defined as reaching experimental endpoint and humanely euthanized before day 4. The peptides were provided intravenously (i.v., 2 mM, 150 μl in PBS, pH 7.0 adjusted with NaOH, a single dose), following a previous peptide-related study [9]. In case of a single dose of intranasal administration, the peptides, iPep-SARS2-E and the negative control mutant peptide, were used together with the infection under isoflurane sedation (2.5 mM, 50 μl in PBS, pH 7.0 adjusted with NaOH). The lung tissue samples were collected at the endpoint (4 days post-infection) for RNA preparation, lung histology, and/or lung viral titer using standard methods [9] as well as our optimized method of SARS2-E protein blotting described as the next section. Lung tissue western blotting To inactivate the virus, MA10-infected mouse lung was incubated in 0.5% SDS in PBS for 1 h at room temperature. After being mashed by a plastic masher, the sample mixture was spun down to collect the supernatant. The supernatant was mixed with same volume of 2× SDS-urea sample buffer and boiled at 95°C for 5 min. Following the SDS page using 20% Bis-acrylamide gel, the protein was transferred to a PVDF membrane by electroblotting. Following 1 h blocking by 5% skim milk in TBS-T, the membrane was incubated with the primary antibody solution (N2A5E8, 0.2 μg/ml in TBS-T) for overnight at 4°C. After 3 times washing with TBS-T, the membrane was incubated with anti-rat IgG secondary antibody conjugated with HRP for 1 h at room temperature. After 3 times washing, development was done using ECL solution following the manufacture’s protocol. Fluorescent stereoscopic imaging of mouse tissues iPep-SARS2-E peptide conjugated with Alexa594 at the C-terminus end (TAT-MY18-2ED-A594, 10 μM, 20 μl) or PBS was provided intranasally under the Isoflurane anesthesia. Two hours later, the mice were euthanized using CO2. The skull was cut with sagittal section in the middle, followed by taking out the nasal septum. After 3 times wash in PBS, imaging for the lateral side of nasal cavity was conducted on a fluorescent stereoscope (Leica, M165 FC). Regarding i.v. injection method, iPep-SARS2-E group mice were injected with TAT-MY18-2ED-A594 peptide (300 μM, 50 μl) via tail vein before 24 h or 2 h of sacrifice. The control mouse was injected with PBS before 2 h of sacrifice. After sacrifice using CO2 and cervical dislocation, whole body perfusion with 15 ml of PBS was conducted to wash out the peptides in their bloods. Harvested tissues, such as lungs, were briefly washed by PBS and imaging was conducted on the fluorescent stereoscope. Cytokine/Inflammation array Cytokine inflammation panel was done in accordance with the manufacturer’s protocol for the mouse cytokine array kit panel A (R&D Systems, Cat# ARY006). Serum samples used were collected from blood heat-inactivated at 65°C for 30 min (to inactivate any possible viruses) and spun down at 15,000 rpm for 10 min. Approximately 50 μl of blood serum samples were used for the assay. Cxcl12, C5a, MCSF, and CD54 were detected in the array using the denatured blood samples. Pseudo-virus design and production Following a previous study [44], our pseudo viruses were designed and produced using HEK 293T cells transfected with pCMV-SARS-CoV-2 Spike (delta variant) and pcDNA3-M-P2A-E, pCMV-dR8.2 dvpr packaging plasmid (Addgene, #8455) and YFP reporter expressing plasmid, which was generated using standard PCR and LV-Cre-SD (#12105, no longer available at Addgene) with EcoRI and XhoI cloning sites to subclone YFP. The virus was filtered using a 0.45 μm syringe and then added to Vero-E6 cells (MOI, approximately 0.05). The YFP imaging was conducted 48 h post-infection using an epi-fluorescent microscopy (Nikon, TS100F). Statistics and reproducibility The statistics used for every figure have been indicated in the corresponding figure legends. The Student’s t test (paired and unpaired) was conducted with the t test functions in Microsoft Excel software. The Student’s t test was two-tailed. The one-way ANOVA with Tukey’s, Sidak’s, Bonferroni’s, or Dunnett’s post hoc multiple comparison analysis was conducted with the GraphPad Prism 6/7/8/9 software. All the data meet the assumptions of the statistical tests. All the samples used in this study were biological repeats, not technical repeats. All experiments, except ELISA for the peptide-antibody binding assay (S2C Fig), were conducted using at least 2 independent experimental materials/cohorts to reproduce similar results. No samples were excluded from the analysis in this study. All the graphs in the figures are mean ± SD. Supporting information S1 Fig. The effect of SARS2-E overexpression on mammalian transcripts. Following the global proteomics results shown in Fig 2A, the expression of the gene transcripts was examined using qPCR. The expressions of HSPA6 (A), DAGLB (B), IP6K2 (C), AGAP3 (D), and RELB transcripts (E) significantly increased in HEK 293S cells transfected to 2E-mKate2 compared to mock. The expression of the other genes, TNC (F), CALU (G), PKLR (H), NOLC1 (I), and ATF3 (J), did not significantly increase in the transfected HEK 293S cells though all the gene transcriptions slightly increased. Unpaired Student’s t test was used (*** P < 0.001; ** P < 0.01; * P < 0.05; n.s., not significant, n = 6). The data underlying this figure can be found in S1 Data. All the graphs in the figure are mean ± SD. https://doi.org/10.1371/journal.pbio.3002522.s001 (PDF) S2 Fig. Characterization of iPep-SARS2-E in situ and in vitro. (A) Relative fluorescent intensity of DND-189 dye in NIH 3T3 cells transfected using mock (n = 32) or 2E-mKate2 plasmid without (-, n = 27) and with TAT-MY18-2ED peptide 24-h incubation (10 μM, n = 50). One-way ANOVA with Tukey’s multiple comparisons test (**** P < 0.0001; n.s. not significant). (B) Representative immunoblot images of HEK 293T cells transfected with SARS2-E fused with YFP (2E-YFP). Anti-SARS2-E (2E-N, clone, N2A5E8, left), GFP (right), and GAPDH antibodies (as loading control, bottom) were used. The rat monoclonal antibody (2E-N mAb, clone N2A5E8) was produced using MY18 peptide conjugated to keyhole limpet haemocyanin (KLH) as the antigen. (C) ELISA assay for the comparison of binding capacity of the anti-SARS2-E monoclonal antibody to wild-type (WT) and 2ED (EE7-8DD) mutant TAT-MY18 peptides. (D) The stability test of iPep-SARS2-E (TAT-MY18-2ED) using the same ELISA assay with anti-SARS2-E antibody. The peptide was incubated at 37°C in phosphate-buffered solution (PBS). (E–G) The toxicity test of iPep-SARS2-E (TAT-MY18-2ED, 10 μM, 48 h) using Jurkat cells and flowcytometry with apoptosis/necrosis assay. Healthy cells (E, %), apoptotic cells (F), and dead/necrotic cells (G) were counted. Camptothecin (10 μM, 3 h) was used as a positive control. One-way ANOVA with Dunnett’s multiple comparisons test was used (**** P < 0.0001; n.s. not significant, compared to non-treated). (H) Immunoprecipitation of 2E protein using Ni column and HEK 293T cells transfected using 2E-YFP with 6xHis-MY18-2ED (2ED) or 6xHis-MY18 wild-type constructs (WT). Anti-GFP antibody was used to blot 2E-YFP protein bands. The data underlying this figure can be found in S1 Data. All the graphs in the figure, except S2C Fig, are mean ± SD. S2C Fig uses single-sample datasets. https://doi.org/10.1371/journal.pbio.3002522.s002 (PDF) S3 Fig. The effect of iPep-SARS2-E on SARS2-E expression. (A) Representative GAPDH immunoblot image of HEK 293T cells transfected with SARS2-E fused with YFP (2E-YFP) and treated with 10 μM TAT-MY18-2ED or MY18-WT (negative control). The image between 30 and 40 kDa is used in Fig 3E. (B) Representative immunoblot images of HEK 293T cells transfected with YFP plasmid and treated with 10 μM TAT-MY18-2ED or MY18-WT (negative control) for 48 h. Anti-GFP (for YFP, top) and GAPDH antibodies (as loading control, bottom) were used. The short- and long-exposure film images are shown. #, nonspecific bands around 40 and 50 kDa are found in the cell lysate. The band images are used in Fig 3H. (C) Quantification of YFP protein expression of HEK 293T cells transfected using YFP plasmid non-treated (n = 3) and treated with TAT-MY18-2ED (n = 3) or MY18-WT peptides (n = 3). One-way ANOVA with Tukey’s multiple comparisons test was used (n.s., not significant). The data underlying this figure can be found in S1 Data. The graph in the figure is mean ± SD. https://doi.org/10.1371/journal.pbio.3002522.s003 (PDF) S4 Fig. Permeability of iPep-SARS2-E. (A) Representative fluorescent and bright field images of time-course cell-penetrating test using Alexa Fluor 594(A594)-conjugated iPep-SARS2-E peptides, A594-TAT-MY18-2ED (amino-terminal conjugation, N-term, 10 μM, bottom), and TAT-MY18-2ED-A594 (carboxyl-terminal, C-term, 10 μM, top) in NIH 3T3 cells after the incubation started. Scale bar, 50 μm. (B) Quantification of red fluorescence-positive cells treated with the A594-conjugated peptides for the peptide cell-penetrating “on” kinetics (mean ± SD). The data underlying this figure can be found in S1 Data. (C) Experimental design for the peptide stability, “off” kinetics, quantification. (D) Representative fluorescent and bright field images after washout of A594-conjugated TAT-MY18-2ED peptide (C-term version) in NIH 3T3 cells. White arrowheads, fluorescent puncta. Scale bar, 50 μm. https://doi.org/10.1371/journal.pbio.3002522.s004 (PDF) S5 Fig. Cell penetration of iPep-SARS2-E peptide in Vero-E6 cells. (A) Representative red fluorescent and bright field images of Vero-E6 cells incubated with TAT-MY18-2ED-AlexaFluor594 (C-term conjugated version, 10 μM). Scale bars, 50 μm. (B) Quantification of red fluorescence-positive Vero-E6 cells treated with the AlexaFluor594-conjugated peptides (C-term conjugated version) for measuring the peptide cell-penetrating “on” kinetics (mean ± SD). The data underlying this figure can be found in S1 Data. https://doi.org/10.1371/journal.pbio.3002522.s005 (PDF) S6 Fig. iPep-SARS2-E in vitro validation. (A) Electron microscopic image of iPep-SARS2-E-treated Vero-E6 cells at 24 h post-infection. Arrowheads, small particles found in the nuclear envelope. Scale bar, 1 μm. (B) Experimental design to examine whether intracellular particles are infectious in Vero-E6 cells treated with iPep-SARS2-E. (C) Result of quantitative endpoint titration assay used to quantify intracellular virus particles of PBS- and iPep-SARS2-E (10 μM)-treated Vero-E6 cells. There is a significant reduction of infectivity in iPep-SARS2-E though still infectious. Each well (cell plating at 24 h, 2,500 harvested cells onto 4 × 104 uninfected fresh cells per a well that were seeded the night before) was scored based on infectivity compared to virus controls with zero indicating no infection and 100 indicating complete infection (CPE). Student’s t test was used (**** P < 0.0001). (D) Western blots of Spike and GAPDH proteins in PBS- and iPep-SARS2-E (10 μM)-treated Vero-E6 cells at 24 h post-infection with SARS-CoV-2 WA1 (MOI, 0.10, 24 h), suggesting the effect of iPep-SARS2-E on Spike expression and/or stability. (E–H) qPCR of SARS-CoV-2 N (E), E (F), JUN/AP-1 expression (G) of PBS (n = 6)- and iPep-SARS2-E (10 μM, n = 6)-treated Vero-E6 cells comparing to non-infected cells (n = 6) at 48 h post-infection. The expression of these genes was normalized to a house-keeping gene, GAPDH. One-way ANOVA with Tukey’s multiple comparisons test was used (**** P < 0.0001; *** P < 0.001; ** P < 0.01; n.s., not significant). (H) qPCR of SARS2-CoV-2 N expression of PBS (n = 6)- and iPep-SARS2-E (10 μM, n = 6)-treated Vero-E6 cell culture supernatant (sup) at 48 h post-infection. Student’s t test was used (**** P < 0.0001). The cDNA samples of cells and cell culture sup were prepared with TRIzol Plus RNA Purification kit, PureLink DNase set and SuperScript III and then diluted (1/5) using UltraPure distilled water for conducting qPCR. (I) Representative confocal fluorescent images of Vero-E6 cells treated with PBS or iPep-SARS2-E at 48 h post-infection. SARS-CoV-2 N antibody (red) and Hoechst 33258 dye (blue, for nucleus) were used with antibodies of subcellular organelle markers (green): BiP for endoplasmic reticulum (ER), ERGIC-53 for ER Golgi inter compartment (ERGIC), and LAMP1 for lysosome. Scale bar, 5 μm. The data underlying this figure can be found in S1 Data. All the graphs in the figure are mean ± SD. https://doi.org/10.1371/journal.pbio.3002522.s006 (PDF) S7 Fig. iPep-SARS2-E in vivo test using intranasal administration. (A) Schematic representative of mouse intranasal administration of iPep-SARS2-E. Red box demonstrates the nasal tissue region harvested for the following fluorescent imaging. The image is from BioRender software. (B) Representative fluorescent and bright field images of nasal tissues isolated from mice administrated intranasally with PBS or Alexa594-conjugated iPep-SARS2-E peptide (TAT-MY18-2ED-A594, 10 μM, 2 h). After isolating the tissues, the samples were washing using PBS 3 times, and the fluorescent and bright field images were taken by a fluorescent stereoscope. Scale bar, 1 mm. (C) Experimental design for the iPep-SARS2-E safety test in vivo. (D) There is no significant difference the effects on body weight among iPep-SARS2-E-treated (n = 6), non-treated (n = 5), and PBS-treated Balb/c mouse groups (n = 5). One-way ANOVA with Tukey’s multiple comparisons was used at each day. (E, F) There were no significant differences in Cxcl12 (E) and C5a (F) among iPep-SARS2-E-treated (n = 5), non-treated (n = 4), and PBS-treated mice (n = 4). One-way ANOVA with Tukey’s multiple comparisons was used (n.s., not significant). (G) Experimental design for the iPep-SARS2-E test in vivo using intranasal administration. (H) iPep-SARS2-E prevents body weight loss in SARS-CoV-2 MA10-infected Balb/c mice (5.0 × 10^4 PFU/mouse). Student’s t test was used at each day (** P < 0.01; * P < 0.05). (I, J) Representative immunoblots of SARS-CoV-2 E (2E, I) and mouse Gapdh proteins (J) in SARS-CoV-2 MA10-infected mouse lung tissues with PBS or iPep-SARS2-E treatment. (K) iPep-SARS2-E peptides significantly reduced the protein expression of 2E in MA10-infected Balb/c mouse lung tissues (PBS, n = 4; iPep-SARS2-E, n = 4). Student’s t test was used (* P < 0.05). The data underlying this figure can be found in S1 Data. All the graphs in the figure are mean ± SD. https://doi.org/10.1371/journal.pbio.3002522.s007 (PDF) S8 Fig. iPep-SARS2-E and negative control peptide test in vitro and in vivo. (A) Alignment of MY18 WT and mutant candidate sequences (MT, targeted amino acids, underline), following the result using the MY18 deletion constructs (Fig 2F) and mutagenesis. (B) Testing iPep-SARS2-E negative control (neg. Ctrl) mutant constructs using NFAT/AP-1 assay in mock- or 2E-transfected HEK 293T cells. One-way ANOVA with Tukey’s multiple comparisons test (**** P < 0.0001; * P < 0.05; n.s., not significant). (C) Relative fluorescent intensity of DND-189 dye in NIH 3T3 cells transfected using mock or 2E-mKate2 plasmid without (-) and with the neg. Ctrl peptide constructs. One-way ANOVA with Tukey’s multiple comparisons test (**** P < 0.0001; *** P < 0.001; n.s., not significant). (D) qPCR of SARS2 N expression of the neg. Ctrl mutant peptide-, iPep-SARS2-E-, and PBS- treated Vero E6 cell culture supernatant. All the peptides (10 μM) were used overnight (approximately 18 h) and then washed before SARS-CoV-2 WA1 infection. One-way ANOVA with Tukey’s multiple comparisons test (** P < 0.01; n.s., not significant). (E) Representative images of phase contrast and yellow fluorescence of Vero-E6 cells infected with pseudo virus (MOI, 0.05) produced by SARS2 Spike, E, M, dR8.2 and YFP reporter using iPep-SARS2-E treatment (10 μM). The neg. Ctrl mutant peptide (10 μM) was used as a negative control. Scale bar, 20 μm. (F) There is no significant difference in YFP-positive cells between iPep-SARS2-E and neg. Ctrl, suggesting no effect of iPep-SARS2-E on the virus entry. Student’s t test was used (n.s., not significant). (G) Experimental design of the iPep-SARS2-E intranasal administration with the neg. Ctrl mutant peptide as a negative control in vivo. (H) Body weight changes of the mouse groups. Student’s t test was used (* P < 0.05). (I) There is a significant reduction of lung viral titer in iPep-SARS2-E-treated mice compared to the neg. Ctrl. Median tissue culture infection dose (TCID) is normalized to lung wet weight (g) measured before the tissue homogenization to isolate the virus. (J) iPep-SARS2-E significantly reduced the transcript expression of SARS-CoV-2 N in MA10-infected Balb/c mouse lung tissues. Student’s t test was used (**** P < 0.0001; ** P < 0.01). The data underlying this figure can be found in S1 Data. All the graphs in the figure are mean ± SD. https://doi.org/10.1371/journal.pbio.3002522.s008 (PDF) S9 Fig. Alignment of human coronavirus envelope proteins. Envelope protein sequence alignment of SARS-CoV-2, MERS-CoV, HCoV-229E, HCoV-NL63, HCoV-OC43, and HCoV-HKU1. CLUSTALW 2.1 multiple sequence alignment software is used to obtain the alignment. https://doi.org/10.1371/journal.pbio.3002522.s009 (PDF) S1 Data. Raw datasets of the experiments. https://doi.org/10.1371/journal.pbio.3002522.s010 (XLSX) S1 Raw Images. Western blotting original images. https://doi.org/10.1371/journal.pbio.3002522.s011 (PDF) Acknowledgments We thank N. Harrison, R. Katz, M. Rahmany, C. Y.l. Sobolevsky, M.V. Yelshanskaya, C. Aston, E. Passague, and J. Stein (Columbia University) for their helpful support and discussion; Y. Tomono and K. Yamamoto (Shigei Medical Research Institute, Japan) for helpful advice with rat monoclonal antibody production; C. Castagna, Y. Luo, S. Sozomenu, A. Matveyenko and M.H. Blumenkrantz (Columbia University) and A. Poddar (Peddie High School, NJ) for helpful assistance.

Hofstadter, William A.;Tsopurashvili, Elene;Cristea, Ileana M.

doi: 10.1371/journal.pbio.3002529pmid: 38442090

Introduction A defining feature of eukaryotic cells is the compartmentalization of cellular functions. The subcellular space of eukaryotic cells is separated into organelles, which each have unique structures and compositions that are suited for their specific cellular roles. Traditionally, organelles have been studied as isolated entities that function independently of one another. However, advances in microscopy over the past 2 decades have revealed a highly complex picture of organelles that are associated in interconnected networks, giving rise to the field of membrane contact site (MCS) research [1–5]. MCSs are a close association (approximately 15 to 30 nm) of 2 or more organelles that allow rapid inter-organelle crosstalk without inducing membrane fusion. Organelle contacts are extensively found in yeast, plant, and animal cells, with every organelle, including membrane-less organelles, forming MCSs [6–8]. The hallmark roles of these associations include facilitating non-vesicular exchange of material (e.g., ions, lipids, and metabolites), fine-tuning organelle identity and dynamics, and regulating vesicle trafficking [3]. Organelle contact formation and function is controlled by several different classes of MCS proteins, which together give each MCS its distinct character [3]. Perhaps most simply, MCS proteins can serve structural roles, acting as tethers to connect organelles, or spacers to maintain a defined distance between organelle membranes [9]. MCS proteins can also serve more functional roles, facilitating the transfer of lipids, ions, or metabolites between organelles [10,11]. Furthermore, MCS proteins can regulate the identity and activity of a given MCS by recruiting other MCS proteins to a contact, altering the activity of proteins present at MCSs, or tuning the extent of contact [12]. A single protein can also serve several of the above listed roles. In fact, most functional MCS proteins also display some tethering ability [13]. Due to their essential roles in cellular homeostasis, organelle contacts are highly regulated. As such, MCS dysregulation is a hallmark of several neurodegenerative disorders, as well as genetic and metabolic diseases [14–18]. Similarly, viral infections can trigger reorganization of organelle contacts to modulate organelle structure and function, and to establish a successful replication cycle [19–27]. Viruses are obligate parasites, indicating that they must co-opt existing cellular pathways and modulate their homeostatic functions to instead support viral replication. In doing so, viruses must also be efficient, given that they encode relatively few proteins. Therefore, to maximize this limited coding capacity, viruses commonly dysregulate MCSs, given that these represent cellular communication points that control both organelle function and transport of molecules. Diverse viruses are known to exploit MCSs to shift intracellular processes, including calcium signaling, lipid trafficking, and apoptotic and innate immune pathways, toward the benefit of the virus (Table 1). To accomplish this, viruses control MCS protein abundance, localization, posttranslational modifications, and protein–protein interactions [27–30]. In this Essay, we explore several of the core functions that MCSs can serve during viral replication, as well as providing specific examples of shared or striking virus-induced MCS modulations. Download: PPT PowerPoint slide PNG larger image TIFF original image Table 1. Diverse viruses employ MCS proteins to remodel contacts. https://doi.org/10.1371/journal.pbio.3002529.t001 MCSs as critical determinants of successful viral infections Advances in the tools used to study MCSs have revealed how diverse viruses rely on MCS regulation for efficient replication. While MCSs can serve several critical roles, a common function for MCSs during viral replication is the movement of ions and lipids such as calcium and cholesterol [20,59]. Altering the flux of these metabolites can influence every step of the viral replication cycle, including entry, genome replication, assembly, and egress. A wide range of DNA and RNA viruses depend on endocytosis for entering cells and subsequent trafficking through the endolysosomal system for release into the cellular environment, processes that rely on host MCSs [60–62]. For example, 2 RNA viruses recently involved in epidemics, Ebola virus and Middle East respiratory syndrome coronavirus (MERS), require endosome-localized MCS proteins for release into cells [24,48]. These viruses rely on the acidic environment of the lysosome to promote uncoating and release of the virion (Fig 1A). In order to properly mature into a lysosome, endosomes must maintain contact with the endoplasmic reticulum (ER) [62]. To induce this maturation, Ebola virus and MERS target the MCS protein TPC1, which promotes calcium flux [63], to regulate the endosome–ER MCS (Table 1). Suppression of TPC1 impaired the ability of Ebola and MERS virions to travel through the endolysosomal system, which was sufficient to prevent Ebola virus infection in macrophages and to restrict MERS infectivity [24,48]. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 1. MCS alterations across the viral replication cycle. (A) Schematic of Ebola virus entry, which relies on IP3R and TPC1 ion channels (cutout). (B) HRV infection promotes cholesterol flux into Golgi-derived replication organelles (ROs) to promote genome replication. This is accomplished by the MCS proteins OSBP and VAPB, which mediate the exchange of PI4P from the Golgi for cholesterol from the ER and LDs (cutout). (C) Lassa virus opens extracellular calcium channels to promote virus egress. Lassa virus proteins induce the release of calcium (Ca2+) from the ER, which stimulates STIM1 to relocalize to ER–plasma membrane MCSs, together with ORAI1, to induce extracellular calcium import (cutout). ER, endoplasmic reticulum; HRV, human rhinovirus; LD, lipid droplet; OSBP, oxysterol-binding protein; RO, replication organelle. https://doi.org/10.1371/journal.pbio.3002529.g001 Following entry, viruses must replicate their genomes and assemble all virion components. Viruses have developed many different strategies for genome replication and assembly, depending on their genome type and virion structure. For example, most DNA viruses replicate in the host cell nucleus, where they can co-opt existing host replication machinery [64]. Alternatively, all positive-stranded RNA (+RNA) viruses replicate in an elaborate cytoplasmic web of membranous compartments called replication organelles that are enriched in the viral proteins and host factors needed for efficient genome replication [65]. Organelle contacts have emerged as critical host factors that are exploited by many viruses to generate and modulate replication organelles [66]. These pathogens have developed strategies for redirecting MCS proteins to the organelle contacts close to the replication sites to promote lipid flux for replication organelle biogenesis and maintenance [20,66,67]. Several +RNA virus families, including picornaviruses and rhinoviruses, remodel MCSs at the ER–Golgi interface to redirect lipid flow (Fig 1B). In an uninfected cell, oxysterol-binding proteins (OSBPs) control cholesterol transfer to the Golgi from the ER [68]. In doing so, OSBPs also metabolize the majority of phosphatidylinositol 4-phosphate (PI4P) lipids present at the Golgi membrane. Among others, rhinoviruses rely on this cholesterol–PI4P exchange to create cholesterol-enriched viral replication compartments [50] (Fig 1B and Table 1). Broadly, this is achieved by first increasing PI4P levels at the Golgi-derived replication organelle via one of the phosphatidylinositol 4 kinases. OSBP then mediates the exchange of PI4P for cholesterol to transport cholesterol to the replication organelle. To maintain a pool of cholesterol on the ER to be transported to the replication organelle, other MCS proteins, including OSBP-like (OSBPL) proteins, mediate lipid droplet and/or endosome MCSs with the ER at sites of viral replication. Highlighting the importance of these MCS-derived structures, knock-down of OSBP1 or expression of an OSBP1 mutant that lacks the ability to bind to PI4P, prevented the replication of several human rhinoviruses (HRVs) [50]. While most pervasive in +RNA viruses, the formation of replication organelle structures is also observed in some cytoplasmic-replicating DNA viruses, such as vaccinia virus [69]. Now assembled, the maturing virions must exit the cell via a process termed egress. For egress, viruses must again interface with the plasma membrane and endomembrane systems. Therefore, many of the MCS proteins involved in entry can also be rewired to promote viral egress. For example, several hemorrhagic fever viruses, including negative-stranded RNA (−RNA) arenaviruses and filoviruses, modulate the MCS proteins ORAI1 and STIM1 for egress [54], the same proteins that DNA virus herpes simplex virus type-1 (HSV-1) relies on for entry (Table 1) [53]. For the hemorrhagic fever viruses, viral matrix proteins stimulate ER calcium release, which then triggers STIM1 to relocalize to the plasma membrane and contact with ORAI1, thereby promoting the import of extracellular calcium into the cell (Fig 1C). This external calcium supply can then support many different aspects of viral replication, including promoting membrane fusion for entry or egress [70]. During HSV-1 infection, activation of this pathway leads to TRPC1 relocation to the plasma membrane, where it can bind to the viral glycoprotein gD, found on the surface of HSV-1 virions, and facilitate viral entry [53]. Alternatively, there are many egress specific tactics. Reovirus infection promotes the formation of viral inclusions, which are membranous structures where viral replication takes place [71]. These viral inclusions then use unknown MCS proteins to transfer mature virions into modified lysosomes, termed sorting organelles, which are subsequently targeted to the plasma membrane for non-lytic egress [71]. MCSs and intrinsic immunity While many viruses have developed unique methods for usurping MCSs throughout their replication cycles, it is important to remember that the host cell is not defenseless. Once viral replication is detected within a host cell, several intrinsic immune signaling pathways have evolved mechanisms of shutting down commonly subverted MCS proteins. One way through which the host cell does this is by activating MCS proteins that are antagonistic to the needs of the virus. For example, the enveloped RNA virus influenza A requires cholesterol enrichment at the plasma membrane for virion assembly, a process that is promoted by the MCS protein NPC1 (Table 1) [45]. NPC1 functions to mobilize cholesterol out of endosomes and into the ER, where downstream MCS proteins can then transfer it to the plasma membrane (Fig 2A). To counteract this flow, the host cell employs the MCS protein ANXA6, which instead promotes the accumulation of cholesterol in endosomes (Table 1) [42]. This is a critical stage in influenza A infection, given that suppression of NPC1 activity or overexpression of ANXA6 prevents viral assembly [42,45]. Similarly, cholesterol loading can prevent the trafficking of vesicular stomatitis virus protein G to the plasma membrane [72]. Therefore, in addition to co-opting MCS functions, viruses must also subvert the antiviral activities of some MCS proteins. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 2. MCSs serve proviral and antiviral functions during infection. (A) Influenza A infection requires cholesterol enrichment at the plasma membrane for efficient replication. To reach the plasma membrane, cholesterol must first be trafficked out of endosomes and into the ER, a process which is promoted by the MCS protein NPC1 but countered by ANXA6. (B) Many viruses employ OSBP to flux cholesterol into the Golgi or Golgi-derived membranes. The antiviral protein IFITM3 functions to outcompete OSBP for binding to its partner VAPB, thus preventing cholesterol enrichment at the Golgi. (C) RIG-I recognizes cytosolic viral RNA and then localizes to ER–mitochondria MCS to activate MAVS and downstream antiviral signaling. ER, endoplasmic reticulum; MCS, membrane contact site; OSBP, oxysterol-binding protein. https://doi.org/10.1371/journal.pbio.3002529.g002 Alternatively, host cells may more directly block the function of a proviral MCS proteins. As mentioned, cholesterol redistribution is a common requirement for viral infection, and therefore, represents a broad-spectrum target for antiviral signaling pathways. This is exemplified by IFITM3, which is activated in response to diverse viral infections and exerts one of its antiviral functions by disrupting proviral MCS protein interactions [28]. When stimulated, IFITM3 outcompetes OSBP for binding to VAPA, which subsequently dysregulates cholesterol cycling (Fig 2B). The resultant accumulation of cholesterol traps the invading virions in multivesicular bodies, preventing viral replication before it starts [73]. Further indicating that this step is a common lynchpin in viral replication, antiviral signaling pathways also promote oxidation of cholesterol, forming such oxysterols as 25-hydroxycholesterol and 27-hydroxycholesterol. These chemicals exhibit broad-spectrum antiviral activities against diverse DNA, RNA, enveloped and non-enveloped viruses by altering membrane lipid compositions and thus suppressing membrane fusion [74,75]. Just as host cells have developed ways of suppressing proviral MCS functions, viruses have also developed to use MCSs to dampen antiviral signaling pathways. One common target for suppressing antiviral pathways is the mitochondria, which in addition to being the main energy producer for the cell, also acts as a central hub for many immune signaling axes. For example, RNA viruses can be detected by the host cell via the cytoplasmic sensor RIG-I, which in turn activates MAVS at ER–mitochondria MCS to ultimately induce an immune response [29,76] (Fig 2C). During hepatitis C virus (HCV) infection, the viral protein NS4A localizes specifically to ER–mitochondria MCSs to cleave MAVS and stunt this signaling pathway (Table 1) [27–29]. Alternatively, during dengue virus infection, the viral protein NS4B prevents the activation of host mitochondrial fission-factor Drp1 (DNM1L in humans), resulting in increased MCSs between mitochondria and virus-induced replication organelles (Table 1) [31]. This modulation leads to a concurrent decrease in ER–mitochondria MCSs, and thus dampens MAVS activation [31]. Conversely, viral proteins can directly suppress mitochondrial membrane potential, which is required for MAVS activation [77]. Pandemic strains of influenza A employ the viral protein PB1-F2 to form MCSs between the inner and outer mitochondria membranes, resulting in the formation of the permeability transition pore complex and subsequent membrane depolarization [40,77]. This is a particularly useful strategy for viruses that replicate quickly and do not require energy production from mitochondria. Virus-induced MCS dysregulation remodels the cellular landscape MCS proteins help to give individual organelles their identity, while also facilitating inter-organelle communication. Hence, MCS protein dysregulation during viral infection promotes many of the observed changes to the organelle landscape, as well as many of the hallmark features of particular infections. One striking example of MCS-induced organelle remodeling is during infection with human cytomegalovirus (HCMV; Fig 3A). This large DNA virus replicates over a lengthy 120-h period, during which viral proteins must accomplish the dual roles of generating enough energy for viral replication and keeping the host cell alive until virion egress. To achieve this, HCMV infection up-regulates nearly all MCS protein abundances throughout the viral replication cycle [27], concurrent with the remodeling of every subcellular organelle [26,78,79]. In particular, HCMV infection remodels ER–mitochondria MCSs to form structures termed mitochondria–ER encapsulations (MENCs) [27]. MENCs consist of a temporally stable, asymmetric cupping of mitochondria by ER tubules (Fig 3B). The MCS proteins VAPB and PTPIP51 localize to MENCs during infection and are necessary for efficient viral replication (Table 1) [27]. While the function of this contact remains unclear, MENCs are well situated to promote calcium [11,80] and lipid [81] exchange or to prevent mitochondria degradation [82] in order to promote mitochondrial activity. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 3. HCMV modulates organelle structure and function by co-opting MCSs. (A) HCMV remodels nearly every organelle during infection, including mitochondria (orange), the nucleus (blue), peroxisomes (green), the ER (light blue), and Golgi/endosomes (purple). (B) MENCs mediated by PTPIP51 and VAPB become the dominant ER–mitochondria MCSs by late stages of HCMV infection. (C) HCMV also promotes the formation of a subset of enlarged and irregularly shaped peroxisomes using the MCS proteins ACBD5 and VAPB. ER, endoplasmic reticulum; HCMV, human cytomegalovirus; MCS, membrane contact site; MENC, mitochondria–ER encapsulation. https://doi.org/10.1371/journal.pbio.3002529.g003 VAPB establishes inter-organelle tethers with many different MCS proteins, forming contacts between the ER and diverse organelles. For example, HCMV also leverages VAPB to promote ER–peroxisome contacts through an interaction with peroxisomal protein ACBD5 [27] (Fig 3C and Table 1). ACBD5 and VAPB help to coordinate the synthesis of plasmalogen lipids [83], which is initiated in peroxisomes and finalized in the ER, as well as to promote peroxisome enlargement [27]. Consequently, throughout HCMV infection, there is an increase in plasmalogen synthesis in conjunction with the formation of a subset of enlarged and irregularly shaped peroxisomes [84]. Similarly, ACBD5 has been implicated in forming enlarged peroxisomes at HCV replication organelles (Table 1) [58]. Plasmalogens are necessary for secondary envelopment during HCMV virion assembly [84] and are enriched in the envelope of HCMV virions [85]. Somewhat surprisingly, both ACBD5 overexpression and knockdown decrease HCMV replication efficiency [27]. This finding highlights the need for viruses to temporally tune MCS protein abundance across their replication cycle. While HCMV is a good example of a virus that rewires MCSs to increase the productivity of the host cell, many other viruses that replicate over shorter time frames instead employ MCSs to induce host cell death. Viruses including poliovirus, human immunodeficiency virus-1 (HIV-1), and influenza A promote host cell death by regulating the abundance of calcium channels (Table 1) [39,40,86]. All of these viruses stimulate the sudden influx of calcium from the ER into mitochondria via the MCS proteins VDACs [59]. This rapid influx of calcium results in mitochondrial collapse and subsequent activation of intrinsic apoptosis pathways. Given that this mechanism is conserved across several viral families, and that virus-induced cell death contributes to pathogenesis, blocking the interactions between viral proteins and VDACs could lead to a potent broad-spectrum antiviral response [39]. Intriguingly, many viral proteins first localize to ER–mitochondria MCSs prior to promoting mitochondria dysfunction [33,87]. This further illustrates how MCSs act as control centers for modulating many different organelle functions. Method improvements for MCS characterization during virus infection Technological advancements have allowed us to identify, quantify, and characterize many new membrane contacts with greater precision than ever before. In 1956, electron microscopy (EM) was the first technology that allowed for the visualization of MCSs [88]. Since then, advancements in EM and electron tomography (ET) have allowed researchers to directly visualize MCSs between organelles and to quantify several parameters of an organelle contact, such as distance between membranes, number of contacts, and length of contact [3,89,90]. EM has also been used to identify many hallmark features of viral infections involving MCSs, including the formation of replication organelles, such as the double membrane vesicles formed during coronavirus infections [91]. The cellular environment is intrinsically crowded and complex, which can make it difficult to identify specific structures within an EM image. To combat this, correlative light and electron microscopy (CLEM) can be used. CLEM is achieved by imaging fluorescently labeled samples with both light microscopy and EM, then integrating these 2 images to harness the specificity of light microscopy with the resolution of EM [63,92]. This technique is particularly useful during viral infections where the subcellular landscape is dramatically remodeled, which can erase common landmarks seen in uninfected cells. Furthermore, CLEM can leverage the ability of light microscopy to visualize the functions of MCSs. For example, CLEM was used to show that, during HCV infection, the viral protein NS5A colocalizes with sites of MCS-dependent cholesterol flux, which are used to form the replication organelle [44]. Alternatively, advancements in machine learning and artificial intelligence have allowed for the automated identification, segmentation, and quantification of MCSs in EM images [93,94]. While light microscopy cannot compete with the resolution of EM, this technique is much easier and more economical to perform, while also supporting the imaging of live cells. Moreover, new technologies have pushed the limits of what is possible with light microscopy, making this technique an attractive alternative. For example, advances in spectral imaging allow researchers to concurrently monitor over a dozen fluorescent probes. Spectral imaging can subsequently be leveraged for screening many different MCSs in a single sample [95]. A myriad of advances in super-resolution microscopy have also allowed for visualization of MCSs with increased temporal and spatial resolution [96,97]. In a recent study, 3D super-resolution microscopy was integrated with deep learning to explore how Zika virus proteins remodel organelles. The viral protein NS4B was shown to disrupt MCSs between the ER and mitochondria, contributing to subversion of host intrinsic immune signaling [98]. The intrinsic qualitative nature of microscopy has been complemented by advancements in quantifying MCS frequency in both fixed and live samples. For example, in a proximity ligation assay (PLA) [99], a specific fluorescent signal is generated if antibodies added to fixed cells are in close proximity (<40 nm). This allows for quantification of the localization and frequency of endogenous protein–protein interactions present at MCSs. Therefore, while light microscopy may not have the resolution to determine if 2 organelles are within the defined MCS range of 15 to 30 nm, PLA can confirm this, as well as identify the specific proteins contributing to that MCS. This powerful tool was applied during HCMV infection to confirm increased ER–peroxisome and ER–mitochondria MCSs [27]. Alternatively, split fluorescent proteins can be used to quantify MCSs in live cells [100]. In this assay, 2 plasmids are designed to each have a localization signal for a particular organelle of interest, as well as complementary halves of a split fluorescent protein such as GFP. Therefore, when these constructs are in close proximity, a quantifiable signal is generated by the reformation of the fluorescent protein. In addition to quantifying the abundance and localization of MCSs, there is significant interest in identifying the different proteins present at a given MCS to better ascertain their function. One efficient tool for answering this question is a split BioID method, termed ContactID [101]. BioID is an engineered biotin ligase that biotinylates proteins in its near vicinity [102]. Biotinylated proteins can then be extracted and characterized by proteomics. By splitting the BioID, the biotin ligase is only activated at sites where 2 organelles of interest are in close proximity, in a similar manner to the split fluorescent system. In the context of infection, ContactID also has the capacity to identify how viral proteins may be interfacing with a given MCS. The drawback to many of the above assays is that they require a priori knowledge of which MCS proteins are of interest in a sample. Additionally, many MCS proteins are present at a low abundance and lack good/affordable antibodies for profiling them. Targeted mass spectrometry can be used to profile the abundances of all MCS proteins throughout an infection to identify which contacts and proteins are worth investigating. Given that MCS protein abundance can be used as an indicator of the extent of contact [103], this assay is a powerful tool for identifying alterations to MCSs at a systems level. Recently, an assay using the targeted mass spectrometry method of parallel reaction monitoring (PRM) was developed to simultaneously quantify the abundance of nearly all known MCS proteins in human cells [27]. This method was applied to monitor MCS proteins across infections with HCMV, HSV-1, influenza A, and the coronavirus OC43. When combined with microscopy and functional assays, this analysis [27] showed that these diverse viruses differentially regulate MCSs on a granular and cellular scale. When taken together, these tools allow researchers to identify, quantify, and assign functionalities to new MCSs, as well as to characterize MCS perturbations in diverse disease states. While several examples of virus-mediated MCS alterations have been provided in this Essay, there is still much to discover in this growing field given the abundance of both viral pathogens and MCSs. We therefore propose a workflow for the identification and characterization of MCSs of interest during any viral infection (Fig 4). First, a screening method can be used to identify MCSs of interest (Fig 4A). MCS–PRM is an efficient method for simultaneously monitoring most MCSs and the proteins that comprise them across a viral replication cycle. If access to a mass spectrometer is limited, split fluorescent probes targeted to different organelles coupled with spectral imaging can instead be used to simultaneously monitor MCSs. Second, higher resolution imaging should be employed to visualize the MCSs identified in the first step (Fig 4B). This is particularly important as a follow-up to MCS–PRM. Next, PLA can be used to identify the protein–protein interactions mediating MCSs of interest (Fig 4C). Finally, a split-TurboID system coupled with mass spectrometry can be leveraged to identify the proteins present at an MCS of interest, which can inform on the functionality of the MCS as well as potentially identify the viral proteins responsible for modulating it. This workflow can aid in expanding the understanding of infection-induced remodeling of organelle–organelle contacts and the possible coordinated functions of MCSs in virus replication and host defense. Download: PPT PowerPoint slide PNG larger image TIFF original image Fig 4. Proposed workflow for MCS identification and characterization. (A) MCSs can first be identified using MCS–PRM, which uses targeted mass spectrometry to quantify MCS protein abundances across a virus replication. Alternatively, numerous split-fluorescent probes or spectral imaging can be used to simultaneously monitor MCS formation. (B) MCSs should be visualized at high resolution, using super-resolution confocal microscopy techniques, EM/tomography or, ideally, combining the 2 to perform CLEM. (C) The protein–protein interactions that form an MCS of interest can be visualized using a PLA, while a split-TurboID system (ContactID) can identify the proteins that localize to MCSs. CLEM, correlative light and electron microscopy; EM, electron microscopy; MCS, membrane contact site; PLA, proximity ligation assay; PRM, parallel reaction monitoring. https://doi.org/10.1371/journal.pbio.3002529.g004 Conclusion and future directions Despite the diverse array of techniques available for studying MCSs and the importance of these tethers in the context of viral infection, many facets of this field remain unexplored. MCSs have traditionally been selected for characterization during viral infections based on a priori knowledge, such as regarding an organelle remodeling event (e.g., replication organelle formation) or altered flux of a small molecule (e.g., calcium signaling). The implementation of unbiased studies, such as using a mass spectrometry approach (e.g., MCS–PRM), can widen the scope of MCSs monitored throughout viral infections. It is worth considering that many of the powerful techniques previously used to study MCSs, such as ContactID [101], are yet to be applied to viral infection studies. This method has the potential to reveal mechanistic insight into viral replication strategies, particularly given that there is still limited information about how viral proteins are specifically inducing the observed changes to MCSs. One underexplored way through which viral proteins may regulate MCSs is by directly engaging host MCS proteins to form novel tethers. This could be further explored by analyzing viral proteomes for common interaction motifs found in MCS proteins, such as the FFAT domain that facilitates an interaction with VAPA/VAPB [104]. Identification of MCSs and MCS proteins targeted during an infection is critical for several reasons. First, given that MCS are lynchpins for many viral infections and that different viruses target similar MCSs for their replication, screening for affected contacts has the potential to point to novel drug targets and possibly broad-spectrum antiviral therapeutic interventions. Second, since their original discovery, the study of viruses has led to many fundamental findings about basic biology [105]. This continues to be true today and includes the study of MCSs during viral infection. Much remains to be discovered about the formation, function, and regulation of MCSs in uninfected cells. Given that viruses modulate the existing host infrastructure, viral infections have the potential to amplify normal subtle features of MCSs, revealing their importance inside and outside the context of infection. Hence, the continued study of MCS alterations during viral infection is expected to offer an important perspective for understanding both virus and host biology. Acknowledgments We would like to thank Dr. Katelyn Cook for her valuable insight. We would also like to thank the field for continually driving new and innovative frontiers in biology.