Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Intrinsic bevacizumab resistance is associated with prolonged activation of autocrine VEGF signaling and hypoxia tolerance in colorectal cancer cells and can be overcome by nintedanib, a small molecule angiokinase inhibitor

Intrinsic bevacizumab resistance is associated with prolonged activation of autocrine VEGF... www.impactjournals.com/oncotarget/ Oncotarget, Vol. 5, No. 13 Intrinsic bevacizumab resistance is associated with prolonged activation of autocrine VEGF signaling and hypoxia tolerance in colorectal cancer cells and can be overcome by nintedanib, a small molecule angiokinase inhibitor. 1,2,3,4 1,2,3 1,2,3 Paul Mésange , Virginie Poindessous , Michèle Sabbah , Alexandre E. 1,2,3 1,2,3,5 1,2,3 Escargueil , Aimery de Gramont and Annette K. Larsen Cancer Biology and Therapeutics, Centre de Recherche Saint-Antoine Institut National de la Santé et de la Recherche Médicale U938, Paris, France Université Pierre et Marie Curie, Paris, France Universite Paris Descartes, Paris, France Department of Medical Oncology, Hôpital Saint-Antoine, Paris, France Correspondence to: Annette K. Larsen, email: [email protected] Keywords: resistance, angiogenesis inhibition, vascular endothelial growth factor (VEGF)-signaling, hypoxia, bevacizumab, ninte-danib, Received: December 9, 2013 Accepted: January 18, 2014 Published: January 20, 2014 This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. ABSTRACT: Colorectal cancer (CRC) is a common tumor type with a high mortality rate, in part due to intrinsic drug resistance. Although bevacizumab, a VEGF-directed neutralizing antibody, is particularly active in this pathology, some patients never respond for reasons not well understood. We here wish to clarify the role of autocrine VEGF signaling in the response of CRC cells to angiogenesis inhibition. Our results show that CRC cells with intrinsic bevacizumab-resistance displayed pronounced upregulation of autocrine HIF-VEGF-VEGFR signaling in response to prolonged bevacizumab exposure whereas the same signaling pathway was downregulated in bevacizumab-sensitive xenografts. Importantly, both bevacizumab-sensitive and -resistant CRC xenografts were sensitive to nintedanib, a small molecule angiokinase inhibitor, which was associated with inhibition of mTORC1. In vitro studies revealed that bevacizumab- resistant cells displayed intrinsically higher HIF-VEGF signaling intensity and hypoxia tolerance compared to their bevacizumab-sensitive counterparts. Interestingly, although nintedanib showed comparable activity toward bevacizumab-sensitive cells under normoxia and hypoxia, the drug was three-fold more toxic to the resistant cells under hypoxia, suggesting that nintedanib attenuated the survival signaling that usually protects these cells from hypoxia-mediated cell death. In conclusion, our ndings fi support a role for autocrine VEGF signaling in the survival of CRC cells to hypoxia and thus to angiogenesis inhibition. We further show that nintedanib, a small molecule angiokinase inhibitor, is active toward CRC models with intrinsic bevacizumab resistance supporting clinical trials of nintedanib in patients that do not respond to bevacizumab, alone or in combination with bevacizumab to increase angiogenesis inhibition. inhibitors. Vascular endothelial growth factor A (hereafter INTRODUCTION referred to as VEGF) is a key regulator of angiogenesis. Amplification of the VEGF locus is observed in a subset Neovascularization (sprouting angiogenesis) is of patients with metastatic colorectal cancer (mCRC) required for the growth of most solid tumors and facilitates and is associated with a remarkably aggressive disease the spread of tumor cells to secondary sites [1] providing characterized by a high incidence of vascular invasion a rational basis for the clinical use of angiogenesis www.impactjournals.com/oncotarget 4709 Oncotarget [2]. Recent data from more than 2000 mCRC patients acquired bevacizumab resistance have focused on indicate that high baseline plasma levels of VEGF might paracrine VEGF signaling and tumor-associated be a negative prognostic factor for both progression-free endothelial cells. However, the capacity of autocrine survival (PFS) and overall survival (OS) [3]. Therefore, VEGF-signaling to promote CRC cell survival under VEGF signaling is linked to invasiveness and aggressive different types of stress suggests that this pathway may disease in CRC and appears as an attractive therapeutic also play a role during angiogenesis inhibition. target. We now report that bevacizumab-resistant, but Several VEGF(R)-targeted agents are approved not bevacizumab-sensitive, CRC cells showed strong or are undergoing clinical trials for treatment of CRC. autocrine HIF-VEGF-VEGFR signaling in response to Bevacizumab (avastin), a VEGF-neutralizing monoclonal prolonged bevacizumab exposure in vivo and displayed antibody [4], was the first angiogenesis inhibitor to be intrinsically higher HIF-VEGF signaling intensity and approved and represents the current benchmark. Although hypoxia tolerance in vitro. We further show that tumors bevacizumab shows excellent activity in some patients, with intrinsic bevacizumab resistance remain sensitive to others never respond, for reasons not well understood. nintedanib, a small molecule angiokinase inhibitor. These Nintedanib (BIBF 1120) is a small molecule tyrosine data suggest that the antitumor activity of at least some kinase inhibitor that inhibits several angiokinases small molecule angiokinase inhibitors is not limited by the including VEGFR1/Flt1, VEGFR2/Flk1/KDR and mechanisms underlying natural bevacizumab resistance VEGFR3/Flt4 as well as FGFR1, FGFR2, FGFR3, and provide a rational for clinical trials of nintedanib in PDGFR-alpha, PDGFR-beta and Flt3 [5]. Nintedanib CRC patients that do not respond to bevacizumab, alone or is currently in phase III clinical trials in ovarian and in combination with bevacizumab to increase angiogenesis non-small cell lung cancer (NSCLC), where it has been inhibition. successful in combination with taxotere [6]. Although the mechanism of angiogenesis inhibitors RESULTS is not fully elucidated, it is generally believed that pruning of the tumor microvasculature will decrease blood supply thereby diminishing tumor oxygenation and promoting HT-29 xenografts are naturally bevacizumab- tumor cell death [7-11]. Hypoxia is accompanied resistant but remain sensitive to nintedanib, a by activation of the hypoxia-inducible transcription small molecule angiokinase inhibitor. factors HIF-1alpha and HIF-2alpha leading to increased expression of VEGF and other HIF targets [12, 13]. Binding of VEGF to its receptors triggers signaling Continued bevacizumab treatment of mice carrying pathways that modulate the phosphorylation, stability and/ human CRC xenografts revealed that HT-29 tumors are or activity of a variety of down-stream targets including naturally bevacizumab-resistant with only 29% growth HIF and VEGF [14-17], thereby initiating a positive feed- inhibition after 4 weeks drug exposure while the same back loop. Thus, HIF-VEGF signaling intensity emerges scheduling resulted in 68% tumor growth inhibition for as a key factor in determining the outcome of angiogenesis the bevacizumab-sensitive DLD-1 tumors (Figure 1A). inhibition. To establish if bevacizumab resistance is associated A striking feature of CRC is the capacity for both with cross-resistance to small molecules targeting the paracrine and autocrine VEGF-signaling. CRC cells are a same pathways, nintedanib was administered to animals major source of VEGF that interacts with VEGF receptors with DLD-1 or HT-29 xenografts. The results show that on tumor-associated endothelial cells thereby stimulating nintedanib displays comparable tumor growth inhibitory their growth, migration and survival [18]. In addition, activity in the two tumor models (Figure 1A). No weight CRC cells express VEGF receptors (VEGFRs) giving rise loss or other toxic side effects were noted for either to autocrine VEGF signaling [19]. Results from different bevacizumab or nintedanib (data not shown). Therefore, laboratories indicate that most, if not all, human CRC CRC tumors with intrinsic bevacizumab resistance remain cells and tumors express functional VEGFR1 [20-23] sensitive to at least some small molecule angiokinase as well as VEGFR2 [24-26]. Autocrine VEGF signaling inhibitors. promotes CRC survival under different types of stress including 5-fluorouracil exposure, low serum conditions Bevacizumab-resistant xenografts express high and anchorage-independent growth [27-29]. Interestingly, levels of VEGF. VEGFR1-mediated VEGF signaling may be particularly important for the survival of invasive CRC cells that have Since bevacizumab and nintedanib both interfere undergone the epithelial-mesenchymal transition and no with VEGF signaling, tissue extracts were prepared from longer benefit from homotypic cell-cell contact [20]. untreated DLD-1 and HT-29 tumors and the concentration Until now, most studies on both intrinsic and www.impactjournals.com/oncotarget 4710 Oncotarget of human VEGF was determined by ELISA analysis Bevacizumab has limited activity on the (Figure 1B). The amounts of VEGF were at least 4 times microvasculature in bevacizumab-resistant higher in tumor tissues from HT-29 xenografts compared xenografts. to DLD-1 xenografts (942 pg vs. 223 pg VEGF/µg protein, respectively, p < 0,001). The microvascular density of DLD-1 and HT-29 Bevacizumab is specific for human VEGF, and xenografts was compared by IHC with a CD31-directed will therefore only neutralize VEGF produced by the antibody followed by quantitative image analysis. The tumor cells, but not murine VEGF secreted by the tumor- results (Figure 1C) indicate that bevacizumab reduces associated stromal cells. To evaluate the contribution the microvascular density by 70% in DLD-1 xenografts, of stromal VEGF, tissue extracts were prepared from compared to the vehicle control. In contrast, the same untreated DLD-1 and HT-29 tumors and the concentration schedule of bevacizumab only reduced the microvascular of murine VEGF was determined by ELISA analysis. The density by 27% in HT-29 xenografts. These results show results show that murine VEGF represents a minor fraction that the microvascular density was less influenced by of the total VEGF (0,4 to 1.1%) in both tumor models bevacizumab in the bevacizumab-resistant xenografts (Figure 1B). Therefore, VEGF signaling is principally whereas nintedanib had comparable activity in the two mediated by human, tumor-derived VEGF in both DLD-1 models. and HT-29 xenografts. Figure 1: Influence of bevacizumab and nintedanib on tumor growth and angiogenesis in CRC xenografts. (A) Nude mice with DLD-1 (grey columns) or HT-29 (black columns) human CRC xenografts were dosed with vehicle (Control), bevacizumab at 5 mg/kg i.p. every 3 days (Beva), nintedanib at 12,5 mg/kg p.o. once daily (Nin 12,5) or nintedanib at 50 mg/kg p.o. once daily (Nin 50) for 4 weeks. Each treatment group corresponded to at least seven animals. Left, T/C values were determined as follows: average tumor volume of treated animals/average tumor volume of the vehicle controls x100. Right, box and whisker plot of the tumor volumes of DLD-1 (light grey boxes) or HT-29 (dark grey boxes) xenografts after 4 weeks treatment with bevacizumab or nintedanib. Lines, medians; boxes, 25th to 75th percentile interquartile ranges; whiskers, the highest and lowest values for a given treatment. (B) Total protein was extracted from DLD-1 and HT-29 xenografts (pool of 3 tumors per xenograft model) and the amounts of human (tumor-derived) and murine (stroma-derived) VEGF were determined by ELISA. (C) Top, representative images of xenografts from animals treated with vehicle or with nintedanib at 50 mg/kg p.o. once daily. CD31-positive blood vessels are outlined in red whereas the nuclei appear in blue. Bottom, quantitative image analysis of the CD31 signal for DLD-1 (grey columns) and HT-29 tumors (black columns). The data show the CD31-positive area, as % of total, and represent the average of at least 6 fields/tumor for at least 3 different tumors. The statistical analysis of experimental data was performed using a Student’s paired t-test comparing the treatment group with the vehicle control. Bars, SD; * p < 0,05; ** p < 0,01; *** p < 0,001. www.impactjournals.com/oncotarget 4711 Oncotarget effects without any striking differences between them: Bevacizumab and nintedanib show a combination 37% vs 33% decrease in DNA synthesis, 265% vs 299% of cytostatic and cytotoxic activities. increased apoptosis, and 236% vs 209% increased necrosis for bevacizumab and nintedanib, respectively (Figure 2). Tumor growth inhibition may be due to both We then compared the influence of nintedanib cytostatic (cell cycle arrest) and cytotoxic (cell death) in the two tumor models. Interestingly, both doses of effects. We therefore evaluated the influence of nintedanib induced more apoptosis in HT-29 tumors bevacizumab and nintedanib on in vivo DNA synthesis (299% and 640%, for the low and high dose of nintedanib, (EdU incorporation), apoptotic cell death (TUNEL assay) respectively) than in DLD-1 tumors (126% and 178%, and necrotic cell death (Figure 2). for the low and high dose of nintedanib, respectively), We first compared the influence of the two drugs. whereas the induction of necrosis was comparable (Figure In DLD-1 xenografts, bevacizumab and nintedanib (50 2). mg/kg) have comparable antitumor activity. Both agents Taken together, our findings indicate that treatment displayed a mixture of cytotostatic and cytotoxic effects with both bevacizumab and nintedanib resulted in without any striking differences between them: 43% a mixture of cytostatic and cytotoxic effects. The vs 46% decrease in DNA synthesis as determined by pronounced apoptosis in nintedanib-treated HT-29 tumors EdU, 197% vs 178% increase in apoptotic cell death as suggests that nintedanib may not only display antivascular determined by TUNEL, and 480% vs 353% increased effects in vivo but also be able to interfere with survival necrosis for bevacizumab and nintedanib, respectively. signaling in the CRC cells. (Figure 2). In HT-29 xenografts, bevacizumab and low dose nintedanib have comparable antitumor activity. Both agents displayed a mixture of cytostatic and cytotoxic Figure 2: Cytostatic and cytotoxic effects of bevacizumab and nintedanib in CRC xenografts. Influence of bevacizumab and nintedanib on (A) DNA synthesis, (B) apoptosis and (C) necrosis in DLD-1 (grey columns) and HT-29 (black columns) tumors. Animals with CRC xenografts were dosed with vehicle (Control), bevacizumab at 5 mg/kg i.p. every 3 days (Beva), nintedanib at 12,5 mg/ kg p.o. once daily (Nin 12,5) or nintedanib at 50 mg/kg p.o. once daily (Nin 50) for 4 weeks. The photos illustrate typical staining patterns. V, viable tissue; N, necrotic tissue. DNA synthesis is determined as the ratio between EdU-positive cells and the total number of viable cells and corresponds to the average of 6 fields/tumor (each field representing approximately 1700 cells) from at least 3 different tumors. Apoptosis is expressed as the area of TUNEL-positive cells, as % of the total area of viable cells, and is the average of 6 fields/tumor for at least 3 different tumors. For necrosis, the data indicates the surface of necrotic cells as % of the total surface and is the average of 6 fields/ tumor for at least 3 different tumors. The statistical analysis of experimental data was performed using a Student’s paired t-test comparing the treatment group with the vehicle control. Bars, SD; * p < 0,05; ** p < 0,01; *** p < 0,001. www.impactjournals.com/oncotarget 4712 Oncotarget treatments with comparable antitumor activity, for most Angiogenesis inhibition has markedly different markers (Figure 3), although the downregulation of HIF- influence on HIF-VEGF signaling in bevacizumab 2alpha was more marked for nintedanib-treated tumors sensitive and -resistant tumors. than for bevacizumab-treated tumors. In clear contrast, for HT-29 tumors, the signal for all markers was up to Microvascular pruning limits oxygen supply to 3-fold higher following bevacizumab exposure than after the tumor thereby activating the hypoxia-inducible exposure to nintedanib (12,5 mg/kg), in spite of their transcription factors, HIF-1alpha and HIF-2alpha. This comparable antitumor and antivascular activities. In is accompanied by transcriptional upregulation of HIF particular, for the nintedanib- and bevacizumab-treated targets like VEGF, thereby promoting autocrine VEGF tumors, the signal for pVEGFR1 was 110% vs 192% while signaling. The presence of HIF-1alpha, HIF-2alpha, VEGF the signal for pVEGFR2 was 75% vs 205%, respectively. and the active autophosphorylated forms of VEGFR1 These findings likely reflect a direct inhibitory effect of and VEGFR2 in the two xenograft models was revealed nintedanib on the tumor cell-associated VEGF receptors, by IHC followed by quantitative image analysis (Figure consistent with the potent angiokinase inhibitory activity 3). The results revealed striking differences between of this compound. DLD-1 and HT-29 xenografts, since all treatments were Taken together, these results reveal both tumor- and accompanied by attenuation of the HIF-VEGF-VEGFR drug-specific differences with respect to HIF-VEGF- axis in DLD-1 tumors, but activation of the same signaling VEGFR signaling. For the bevacizumab-sensitive components in HT-29 xenografts. DLD-1 tumors, all treatments were accompanied by For DLD-1 tumors, there were no marked differences downregulation of the HIF-VEGF-VEGFR axis whereas between nintedanib (50 mg/kg) and bevacizumab, two Figure 3: Influence of bevacizumab and nintedanib on HIF-VEGF-VEGFR signaling in CRC xenografts. Nude mice with DLD-1 (grey columns) or HT-29 (black columns) human CRC xenografts were dosed with vehicle (Control), bevacizumab at 5 mg/ kg i.p. every 3 days (Beva), nintedanib at 12,5 mg/kg p.o. once daily (Nin 12,5) or nintedanib at 50 mg/kg p.o. once daily (Nin 50) for 4 weeks. The expression of HIF-1alpha, HIF-2alpha, VEGF, pVEGFR1, pVEGFR2 and pS6 was determined by immunohistochemistry followed by quantitative image analysis. The photos illustrate the typical staining patterns for tumors derived from animals treated with bevacizumab (Beva) or nintedanib (Nin, 12,5 mg/kg) that provided comparable tumor growth inhibition. For the quantitative analysis of the signal intensity, the data represent the average fluorescence intensity of treated tumors compared to the treatment intensity of control tumors and are the average of 6 fields/tumor for at least 3 different tumors. The statistical analysis of experimental data was performed using a Student’s paired t-test comparing the treatment group with the vehicle control. Bars, SD; * p < 0,05; ** p < 0,01; *** p < 0,001. www.impactjournals.com/oncotarget 4713 Oncotarget the same treatments resulted in activation of autocrine Nintedanib, but not bevacizumab, attenuates VEGF signaling in the bevacizumab-resistant HT-29 mTORC1 activity in bevacizumab-resistant tumors. Furthermore, for HT-29 tumors, bevacizumab xenografts. treatment was associated with much stronger activation of autocrine VEGF signaling than was the case for VEGF receptors and other RTKs mediate multiple nintedanib. downstream signaling pathways that may be integrated at the level of mTORC1, the mammalian target of rapamycin (mTOR) complex 1 [30]. Recent findings suggest that inhibition of mTORC1 activity, as measured by tumor levels of Ser240/Ser244-phosphorylated S6 (pS6) is a robust biomarker for the antitumor activity of at least some protein kinase inhibitors such as inhibitors of mutant BRAF and PI3K [31, 32]. In the present study, all treatments reduced the expression of pS6 in DLD-1 tumors, which was most pronounced for bevacizumab and nintedanib at the 50 mg/kg dose (Figure 3). In HT- 29 tumors, nintedanib induced a marked, dose-dependent decrease of pS6 levels reaching 12% for the high dose of nintedanib, compared to the vehicle control. In clear contrast, bevacizumab increased the levels of pS6 to ~130% in the same xenograft model. Bevacizumab and nintedanib show synergistic activity toward bevacizumab-resistant xenografts. The important differences in the activities of bevacizumab and nintedanib toward HT-29 xenografts prompted us to explore the combination of the two agents (Figure 4). Bevacizumab and low dose nintedanib showed comparable activity toward HT-29 xenografts as single Figure 4: Influence of bevacizumab, nintedanib and agents with approximately 27% tumor growth inhibition. their combination on tumor growth, microvessel In comparison, the combination of bevacizumab and density and mTORC1 activity in bevacizumab- nintedanib was accompanied by 65% tumor growth resistant xenografts. Nude mice with HT-29 human CRC inhibition (p < 0,001, bevacizumab alone vs bevacizumab xenografts were dosed with vehicle (Control), bevacizumab at and nintedanib). The combination was active toward both 5 mg/kg i.p. every 3 days (Beva), nintedanib at 12,5 mg/kg p.o. endothelial and tumor cells. The microvascular density, once daily (Nin 12,5) or with a combination of bevacizumab and nintedanib for 4 weeks. Each treatment group corresponded as determined by CD31 staining, decreased by ~28% to at least seven animals. (A) Average tumor growth of HT- when bevacizumab or nintedanib were given alone, to 29 tumorgrafts in mice treated with vehicle (black squares), 63% when the two agents were given together (p < 0,001, bevacizumab (black circles), nintedanib (white circles) or their bevacizumab alone vs bevacizumab and nintedanib). In combination (light grey circles). (B) Box and whisker plot of addition, nintedanib was able to counteract the activation the tumor volumes of HT-29 xenografts after 4 weeks treatment of mTORC1 in the tumor cells, as measured by pS6, with bevacizumab, nintedanib or their combination. Lines, from 131% (compared to the vehicle control) when medians; boxes, 25th to 75th percentile interquartile ranges; bevacizumab was given alone to 72% when the two agents whiskers, the highest and lowest values for a given treatment. were given together (p < 0,001, bevacizumab alone vs The expression of CD31 (C) and pS6 (D) was determined by immunohistochemistry followed by quantitative image analysis. bevacizumab and nintedanib). The photos illustrate typical staining patterns for tumors derived from animals treated with bevacizumab or with a combination Brief hypoxia is accompanied by HIF-1alpha of bevacizumab plus nintedanib. For the quantitative analysis of expression in both DLD-1 and HT-29 cells. the signal intensity, the data represent the average fluorescence intensity of treated tumors compared to the treatment intensity of control tumors and are the averages of 6 fields/tumor for at The marked differences in HIF expression between least 3 different tumors. The statistical analysis of experimental DLD-1 and HT-29 xenografts raise the question if DLD-1 data was performed using a Student’s paired t-test comparing the cells are unable to activate HIF, unable to maintain HIF treatment group with the vehicle control. Bars, SD; * p < 0,05; signaling, or rather if HIF signaling intensity in DLD-1 ** p < 0,01; *** p < 0,001. www.impactjournals.com/oncotarget 4714 Oncotarget tumors is weak compared to HT-29 tumors. To answer 24 hrs hypoxia is accompanied by HIF-1alpha induction this question, DLD-1 and HT-29 cells were exposed to in both cell lines whereas the expression of HIF-2alpha acute hypoxia (1% O ) for 24 hrs and the expression of remains marginal (Figure 5A). HIF-1alpha and HIF-2alpha was determined by Western To establish if the HIF-1alpha protein was blot analysis. The results show that the expression of the transcriptionally active, the expression of its downstream two HIF proteins is marginal under normoxia. In contrast, target, VEGF, was determined by qRT-PCR. Acute hypoxia was accompanied by a ~10-fold increase in VEGF expression for HT-29 cells, compared to the corresponding normoxic cells, whereas the corresponding increase in VEGF expression was ~3.5-fold for DLD-1 cells. Therefore, DLD-1 cells show a functional hypoxia response as indicated by upregulation of VEGF mRNA, although to a lesser degree than HT-29 (Figure 5B). Bevacizumab and nintedanib upregulate HIF- 1alpha in both bevasizumab-sensitive and -resistant cells. We next explored if angiogenesis inhibitors might have a direct effect on HIF-VEGF signaling in CRC cells. Our results indicate that the presence of nintedanib or bevacizumab had no detectable effect on the expression of HIF-1alpha or HIF-2alpha under normoxia (Figure 5A). However, under hypoxia, nintedanib activated the expression of HIF-1alpha in a dose-dependent manner in both cell lines (2 to 3-fold, compared to the normoxia control) (Figure 5A). In addition, bevacizumab was able Figure 5: Influence of bevacizumab and nintedanib on CRC cells under normoxia and hypoxia. (A) Western blot analysis of HIF-1alpha, HIF-2alpha and beta-actin in DLD- 1 and HT-29 cells after 24 hrs exposure to nintedanib (0,01, 0,1 or 1 µM) or bevacizumab (250 µg/ml) under normoxia or hypoxia (1% O ). (B) Induction of VEGF mRNA in DLD-1 (grey columns) and HT-29 (black columns) cells following 24 hrs exposure to nintedanib (0,01, 0,1 or 1µM) or bevacizumab (250 µg/ml) under normoxia or hypoxia. VEGF expression was normalized with 36B4 mRNA and is the average of three independent experiments. The statistical analysis of experimental data was performed using a Student’s paired t-test comparing the treatment group with the vehicle control. Bars, SD; * p < 0,05; ** p < 0,01; *** p < 0,001. (C) Hypoxia-induced apoptosis in DLD-1 (grey columns) and HT-29 (black columns) cells after hypoxia (1% O ) exposure for the indicated times as measured by the TUNEL assay (left) or by the formation of catalytically active cleaved caspase 3 (right). Values are the averages of at least two independent experiments done in duplicate. The statistical analysis of experimental data was performed using a Student’s paired t-test comparing the treatment group with the Figure 6: HIF-VEGF-VEGFR signaling in CRC vehicle control. Bars, SD; * p < 0,05; ** p < 0,01; *** p < 0,001. cells. Hypoxia exposure activates HIF, thereby increasing the (D) Viability of DLD-1 cells (left) or HT-29 cells (right) after expression of VEGF, a HIF-transcriptional target. Subsequent 24 hrs exposure to the indicated concentrations of nintedanib binding of VEGF to its receptor on either CRC cells (autocrine under normoxia (grey circles) or hypoxia (black circles) for pathway) or tumor-associated endothelial cells (paracrine 24 hrs followed by 96 hrs post-incubation under normoxia and pathway) leads to activation of VEGFR and its downstream MTT determination. Bars indicate SD and are shown when signaling pathways. Receptor tyrosine kinases like VEGFR they exceed symbol size. The data represents three independent are able to activate and/or stabilize HIF and VEGF, thereby experiments each done in duplicate. promoting a positive feed-back loop. www.impactjournals.com/oncotarget 4715 Oncotarget to induce HIF-1alpha almost 4-fold in the HT-29 cells days) exposure to the angiokinase inhibitor nintedanib whereas its influence was marginal in DLD-1 cells (Figure showed that DLD-1 and HT-29 xenografts are both 5A). sensitive to nintedanib at doses with no detectable toxic The drug-induced increase of HIF-1alpha protein side effects. These findings suggest that the antitumor had modest influence on the expression of VEGF mRNA activity of at least some small molecule angiokinase in hypoxic DLD-1 cells (Figure 5B). In clear contrast, inhibitors is not limited by the mechanisms underlying for hypoxic HT-29 cells, nintedanib-exposure was natural bevacizumab resistance. accompanied by a dose-dependent increase in mRNA Next, we characterized the differences between VEGF expression that reached ~2,4 times the levels of bevacizumab-sensitive and -resistant cells, with special the untreated controls, whereas bevacizumab increased attention to VEGF signaling. Comparison of the vascular mRNA VEGF expression ~1.8-fold. Therefore, although density in DLD-1 and HT-29 tumors by IHC analysis drug exposure was associated with increased HIF-1alpha reveals that bevacizumab treatment is associated with accumulation under short-term hypoxia in both cell lines, important vascular pruning in the sensitive DLD- VEGF mRNA expression was only increased for HT-29 1, but not in the resistant HT-29 xenografts. In clear cells. contrast, nintedanib showed comparable effects on the microvascular density in both xenograft models. To establish if the intrinsic bevacizumab resistance of HT-29 Bevacizumab-resistant cells show increased xenografts might be linked to elevated levels of stroma- hypoxia tolerance. derived VEGF, the concentrations of human and murine VEGF were determined. The results show that for both Next, DLD-1 and HT-29 cells were exposed to tumor types, murine VEGF represents, at the most, 1% hypoxia for up to 48 hrs and the fraction of apoptotic of the total amount of VEGF and is therefore unlikely cells was determined by the TUNEL assay (Figure to influence the response to angiogenesis inhibition. 5C left). The results show that hypoxia exposure was Interestingly, HT-29 tumors express at least four times accompanied by rapid apoptosis in DLD-1 cells to reach more human VEGF, compared to the DLD-1 tumors, almost 17% apoptotic cells by 48 hrs, whereas the level which likely contributes to the bevacizumab-resistant of apoptosis stayed below 5% for HT-29. These findings phenotype. were confirmed by a caspase 3 assay that measures the To establish the mechanistic basis for tumor formation of catalytically active cleaved caspase 3 (Figure growth inhibition, the levels of DNA synthesis, apoptosis 5C right). and necrosis were determined. The results show that prolonged exposure to the two drugs resulted in a Nintedanib shows increased toxicity toward mixture of cytostatic and cytotoxic effects, independent bevacizumab-resistant cells under hypoxia. of the tumor model. However, nintedanib exposure was associated with significantly more apoptosis in HT-29 than in DLD-1 xenografts, suggesting that in this tumor model, To determine the influence of hypoxia on nintedanib nintedanib may not only act on the microvasculature, but toxicity, cells were exposed to different doses of nintedanib also attenuate the survival response of the tumor cells to for 24 hrs under hypoxia or normoxia followed by 96 hrs environmental stress. post-exposure under normoxia and MTT determination. To characterize the signaling intensity of the HIF- The results show that nintedanib is slightly (~30%) VEGF signaling pathway (summarized in Figure 6), more toxic for DLD-1 cells under hypoxia compared to the expression of HIF-1alpha, HIF-2alpha, VEGF and normoxia (Figure 5D left). In comparison, nintedanib was the active, autophosphorylated forms of VEGFR1 and almost 3 times more toxic to HT-29 cells under hypoxia, VEGFR2 was determined by IHC followed by quantitative suggesting that nintedanib attenuates the survival signaling image analysis. The results revealed striking differences that usually protects HT-29 from hypoxia-mediated cell between the two tumor models. Drug treatment was death (Figure 5D right). accompanied by a general downregulation of all signaling components of the HIF-VEGF-VEGFR autocrine loop in DISCUSSION DLD-1 tumors, in clear contrast to HT-29 tumors, where the same pathway was upregulated. Interestingly, for HT- Bevacizumab shows excellent activity in a subset 29 tumors, treatment with bevacizumab was accompanied of mCRC patients while others never respond. A key by much higher activation of autocrine VEGF signaling question is if bevacizumab-resistance is limited to this than nintedanib, thereby identifying a key difference particular agent or rather is indicative of a general between the two agents. resistance to angiogenesis inhibition. To clarify this mTORC1 integrates multiple signaling pathways question, we identified the HT-29 xenograft model as downstream of activated receptor tyrosine kinases. naturally bevacizumab-resistant in contrast to DLD-1 Accordingly, recent findings suggest that the degree of xenografts that are bevacizumab-sensitive. Prolonged (28 www.impactjournals.com/oncotarget 4716 Oncotarget mTORC1 activity, as determined by the levels of the drives the initial response to hypoxia whereas HIF- Ser240/Ser244 phosphorylated form of the ribosomal 2alpha is needed for the chronic response [34]. Next, we S6 protein (pS6), serve as a robust biomarker for the determined the capacity of HIF to increase the expression antitumor activity of at least some signaling pathways of VEGF mRNA. The results reveal that although VEGF inhibitors. Quantitative IHC analysis showed that was induced by hypoxia in both cell lines, the increase in treatment with both bevacizumab and nintedanib was VEGF levels was more prominent for HT-29 cells (~10- accompanied by downregulation of mTORC1 activity in fold increase) than for DLD-1 cells (~3-fold increase). DLD-1 tumors. In clear contrast, bevacizumab exposure Interestingly, the differences in VEGF expression in vitro was associated with upregulation of mTORC1 activity were comparable to the differences in VEGF expression in HT-29 tumors whereas nintedanib treatment was observed for the corresponding xenografts. Therefore, associated with a marked decrease of mTORC1 activity, the in vivo differences in basal VEGF levels appear to be compared to the vehicle control. For the bevacizumab- a result of an intrinsically higher HIF-VEGF signaling treated HT-29 tumors, the increase in mTORC1 intensity in bevacizumab-resistant cells. activity is fully coherent with the elevated levels of Unexpectedly, the hypoxia-mediated up-regulation active, phosphorylated VEGFR1 and VEGFR2. In of HIF-1alpha protein was further increased in the contrast, the strong downregulation of mTORC1 in the presence of nintedanib and bevacizumab, suggesting nintedanib-treated HT-29 tumors was unexpected. The that the two agents can influence tumor cells directly downregulation is unlikely to rely on VEGF-signaling under cellular stress. The activity of nintedanib is alone, but likely involves other nintedanib targets like consistent with previous findings, where nintedanib was FGFR (fibroblast growth factor receptor) and PDGF shown to modulate cell cycle progression and viability (platelet-derived growth factor receptor receptor) family of a wide range of CRC cell lines [35]. The influence members [5]. of bevacizumab is more unexpected considering that Another unexpected finding was that HIF-1alpha bevacizumab is a ligand-targeted agent that is unable was upregulated in the nintedanib-treated HT-29 tumors to enter the tumor cells. However, two recent articles in spite of the marked downregulation of mTORC1, report that chronic in vitro exposure of CRC cells to a well-established negative regulator of HIF-1alpha bevacizumab was accompanied by the emergence of a translation. The upregulation of HIF-1alpha is coherent more migratory and invasive phenotype [36, 37], which is with the loss of tumor-associated microvessels that coherent with a direct effect of bevacizumab on the tumor results in tumor hypoxia and activation of autocrine cells, as observed here. VEGF signaling. Furthermore, it should be noted that the HT-29 cells were more resistant to hypoxia-induced regulation of HIF-1alpha is complex and is mediated at apoptosis compared to DLD-1 cells consistent with their several levels including transcription, translation as well bevacizumab-resistant phenotype in vivo. If the increased as protein stability. Even on the translational level, our survival of HT-29 cells under hypoxia depends on their previous results suggest that HIF-1alpha expression can potent kinase activity including autocrine VEGF-signaling, be modulated by (at least) two different agents, rapamycin we would predict that nintedanib should be more cytotoxic via mTORC1, and irinotecan, that acts in a mTORC1- under hypoxic conditions than under normoxia. In independent manner [33]. contrast, we would expect a smaller difference for DLD- Interestingly, the beva-resistance of the HT-29 1 cells that are inherently hypoxia-sensitive and appear xenografts could be overcome by combining bevacizumab to have weaker angiokinase signaling. In agreement, with low dose nintedanib. The additional antitumor effect the results showed almost three-fold increased activity was characterized by a diminution, rather than an increase, of nintedanib under hypoxia in HT-29 cells, whereas of mTORC1 activity as well as by increased inhibition of nintedanib was only marginally more active under hypoxia the tumor microvasculature. toward DLD-1 cells. The increased activity of nintedanib A crucial question is if bevacizumab-sensitive under hypoxia is a noticeable feature, since other tyrosine cells are unable to activate HIF-signaling in response kinase inhibitors like gefitinib have been reported to show to hypoxia, unable to maintain HIF activation, or if the less, rather than more, cytotoxic activity under hypoxia HIF signaling pathway is functional, but less prominent [38]. compared to bevacizumab-resistant cells. To answer this Taken together, our data indicate that intrinsic question, DLD-1 and HT-29 cells were incubated under bevacizumab-resistance is multifactorial and associated hypoxia for 24 hrs and the induction of HIF-1alpha and with i) the high levels of VEGF in the tumor environment HIF-2alpha was determined. The results show that both that renders the endothelial cells more resistant to cell lines express little, if any, HIF-1alpha or HIF-2alpha bevacizumab, ii) strong VEGFR1 and VEGFR2 signaling under normoxia. In contrast, hypoxia is accompanied by in the tumor cells that leads to mTORC1 activation as robust HIF-1alpha induction in both cell lines whereas the well as iii) resistance to hypoxia-induced apoptosis. We induction of HIF-2alpha is marginal. These findings are further show that all these resistance mechanisms can coherent with previous work reporting that HIF-1alpha be overcome by nintedanib. These findings suggest that www.impactjournals.com/oncotarget 4717 Oncotarget the antitumor activity of at least some small molecule MetaMorph software (Universal Imaging Corporation) for angiokinase inhibitors is not limited by the mechanisms quantitative image analysis. underlying natural bevacizumab resistance and provide Blood vessel density is expressed as the CD31- a rational for clinical trials of nintedanib in CRC positive area, in % of total, and represents the averages patients that do not respond to bevacizumab, alone or in of at least 6 fields/tumor for at least 3 different tumors. combination with bevacizumab. DNA synthesis is determined as the ratio between EdU- positive cells and the total number of viable cells and is the average of 6 fields/tumor (each field representing MATERIALS AND METHODS approximately 1700 cells) from at least 3 different tumors. Apoptosis is expressed as the % area of TUNEL-positive cells compared to the total area of viable cells, and is the Xenograft models. average of 6 fields/tumor for at least 3 different tumors. For necrosis, the data indicates the surface of necrotic cells The antitumor activity of bevacizumab and as % of the total surface and is the average of 6 fields/ nintedanib was evaluated in athymic mice (female tumor for at least 3 different tumors. NMRI-Fox 1nu, 6 weeks old) from Taconic (Skensved, For the quantitative analysis of the signal intensity Denmark) bearing DLD-1 or HT-29 CRC xenografts. Two for HIF-1alpha, HIF-2alpha, VEGF, pVEGFR1, (HT-29) or five (DLD-1) million cells were injected into pVEGFR2 and pS6, the data represents the average the right flank, and the treatments were started when the fluorescence intensity of treated tumors compared to the tumors were palpable (median tumor volume ~100 mm ). treatment intensity of control tumors and is the average of The animals were weighted daily and the tumor size 6 fields/tumor for at least 3 different tumors. was determined three times per week. Tumor volumes (mm ) were calculated according to formula: [(length² x ELISA assay for VEGF. width)/2]. Treated/Control (T/C) values were calculated as follows: average tumor volume of treated animals/average Healthy tumor tissues were collected from untreated, tumor volume of control animals x100. Each treatment frozen tumors (three tumors per xenograft model), and group was composed of at least 7 animals. Animals were protein extracts were prepared in RIPA buffer according treated according to institutional guidelines. to the manufacturer’s instructions. VEGF levels were determined by Quantikine ELISA (R&D System # DVE00 Immunohistochemistry. (human, tumor-derived VEGF) and MMV00 (murine, stroma-derived VEGF). The values represent the average Biomarker analysis was carried out with tumors of 3 independent experiments, each done in duplicate. collected after 28 days of treatment. To measure in vivo DNA synthesis, the thymidine analog 5-ethynyl-2’- Tumor cells, apoptosis and viability assays. deoxyuridine (EdU, Life Technologies) was administered 48 hrs before sacrifice (500 µg i.p.). The incorporated EdU HT-29 cells were kindly provided by Richard was revealed by a fluorescent-azide coupling reaction Camalier, Division of Cancer Treatment and Diagnosis (Click-iT # C10337, Life Technologies) of paraffin- tumor repository (NCI, US) whereas DLD-1 cells were embedded tumor samples and counterstained by DAPI a generous gift from Richard Hamelin (Saint-Antoine to reveal the nuclei of individual cells. The proportion of Research Center, Paris, France). apoptotic tumor cells was scored by the TUNEL assay (In To determine the induction of hypoxia-induced Situ Cell Death Detection kit, Roche Applied Science # apoptosis, HT-29 and DLD-1 cells were seeded on 11684795910). coverslips in DMEM medium with 5% fetal calf serum The following antibodies were used for under hypoxia (1% O ). Cover slips were removed after immunohistochemistry (IHC) analysis: anti-CD31 (BD 0, 12, 24 and 48 hrs hypoxia and subjected to TUNEL Bioscience # 550274), anti-HIF-1alpha (BD Bioscience # analysis. Alternatively, hypoxia-induced apoptosis was 610958), anti-HIF-2alpha (Novus Biological NB-100122), determined by the capacity of catalytically active cleaved anti-VEGF (Santa Cruz # sc-152), anti-pVEGFR1 caspase-3 to generate fluorescent reaction products (Millipore 07-758) that recognizes phospho-Tyr1213 (Caspase 3 fluorescence assay kit, Cayman chemical, VEGFR1, anti-pVEGFR2 (Santa Cruz # sc-101819) #10009135). that recognizes phospho-Tyr1175 VEGFR2, and anti- Cellular viability was determined by the MTT pS6 (Cell Signaling # 2211) that recognizes phospho- (methylthiazolyldiphenyl-tetrazolium bromide) Ser240/244 S6. The relevant Cy3-conjugated secondary viability assay as described previously [39] with minor antibodies were obtained from Jackson ImmunoResearch. modifications. Cells were incubated with different All images were captured by a fluorescence microscope concentrations of nintedanib under normoxia or hypoxia and the fluorescence intensities were determined by the www.impactjournals.com/oncotarget 4718 Oncotarget (1% O ) for 24 hrs followed by 96 hrs post-incubation 2. Vlajnic T, Andreozzi MC, Schneider S, Tornillo L, under normoxia and MTT determination. Karamitopoulou E, Lugli A, Ruiz C, Zlobec I, Terracciano L. VEGFA gene locus (6p12) amplification identifies a small but highly aggressive subgroup of colorectal cancer Real-time RT (reverse transcription)-PCR and patients. Mod Pathol. 2011; 24: 1404-1412. Western blot analysis. 3. Jürgensmeier JM, Schmoll HJ, Robertson JD, Brooks L, Taboada M, Morgan SR, Wilson D, Hoff PM. Prognostic Total RNA was extracted from CRC cells and predictive value of VEGF, sVEGFR-2 and CEA in using the TRIzol® RNA purification reagent. RNA mCRC studies comparing cediranib, bevacizumab and quantity and purity were determined by using a chemotherapy. Br J Cancer. 2013; 108: 1316-1323. NanoDrop ND-1000. Total RNA (1 μg) from each 4. Ferrara N, Hillan KJ, Gerber HP, Novotny W. Discovery sample was reverse transcribed and real-time RT-PCR and development of bevacizumab, an anti-VEGF antibody measurements were performed as described previously for treating cancer. Nat Rev Drug Discov. 2004; 3: 391-400. [40, 41] using a Mx3000P apparatus (Agilent) with the 5. Hilberg F, Roth GJ, Krssak M, Kautschitsch S, corresponding SYBR Green kit. PCR primers were Sommergruber W, Tontsch-Grunt U, Garin-Chesa P, designed with Primer 3 (Agilent) as follows: VEGF, Bader G, Zoephel A, Quant J, Heckel A, Rettig WJ. BIBF upper, 5’-CGAAGTGGTGAAGTTCATGGATG-3’, 1120: triple angiokinase inhibitor with sustained receptor lower, 5’-TTCTGTATCAGTC TTTCCTGGTGAG. blockade and good antitumor efficacy. Cancer Res. 2008; 36B4 (also known as RPLP0), upper, 68: 4774-4782. 5′-GATTGGCTACCCAACTGTTG-3′; lower, 6. Reck M, Kaiser R, Mellemgaard A, Douillard JY, Orlov S, 5′-CAGGGGCAGCAGCCAC AAA-3′. Gene expression Krzakowski MJ, Von Pawel J, Gottfried M, Bondarenko was normalized to 36B4. I, Liao M, Barrueco J, Gaschler-Markefski B, Novello S. Western blot analysis was carried out as described Nintedanib (BIBF 1120) plus docetaxel in NSCLC patients previously [42]. The primary antibodies were directed progressing after first-line chemotherapy: LUME Lung 1, a against HIF-1alpha (BD Bioscience # 610958), HIF- randomized, double-blind phase III trial. J Clin Oncol. 31, 2alpha (Novus Biological # NB-100122) or beta-actin 2013 (suppl; abstr LBA8011). (Santa Cruz # sc-47778), while the corresponding secondary antibodies were purchased from Jackson 7. Willett CG, Duda DG, di Tomaso E, Boucher Y, ImmunoResearch. Ancukiewicz M, Sahani DV, Lahdenranta J, Chung DC, Fischman AJ, Lauwers GY, Shellito P, Czito BG, Wong TZ, Statistical analysis. Paulson E, Poleski M, Vujaskovic Z, et al. Efficacy, safety, and biomarkers of neoadjuvant bevacizumab, radiation therapy,and fluorouracil in rectal cancer: a multidisciplinary The statistical analysis of experimental data was phase II study. J Clin Oncol. 2009; 27: 3020-3026. performed using a Student’s paired t-test, and results are 8. Van der Veldt AA, Lubberink M, Bahce I, Walraven M, presented as mean ± standard deviation (SD). de Boer MP, Greuter HN, Hendrikse NH, Eriksson J, Windhorst AD, Postmus PE, Verheul HM, Serné EH, ACKNOWLEDGMENTS Lammertsma AA, Smit EF. Rapid decrease in delivery of chemotherapy to tumors after anti-VEGF therapy: This work was financed in part by research funding implications for scheduling of anti-angiogenic drugs. to Annette K. Larsen from Boehringer Ingelheim. The Cancer Cell. 2012; 21: 82-91. sponsors had no role in the study design, data collection 9. Pastuskovas CV, Mundo EE, Williams SP, Nayak TK, Ho and analysis, interpretation of the results, the preparation J, Ulufatu S, Clark S, Ross S, Cheng E, Parsons-Reponte of the manuscript, or the decision to submit the manuscript K, Cain G, Van Hoy M, Majidy N, Bheddah S, dela for publication. We like to acknowledge the contribution Cruz Chuh J, Kozak KR, et al. Effects of anti-VEGF on of Olivier Bernadini, Tatiana Ledent and Delphine Muller pharmacokinetics, biodistribution, and tumor penetration from the animal facilities at Saint-Antoine Research of trastuzumab in a preclinical breast cancer model. Mol Center and Sylvie Dumont and Fatiha Merabtene from Cancer Ther. 2012; 11: 752-762. the pathology platform at Hôpital Saint-Antoine. Paul 10. Eichten A, Adler AP, Cooper B, Griffith J, Wei Y, Mésange is supported by a Mouna Nasrallah fellowship. Yancopoulos GD, Lin HC, Thurston G. Rapid decrease in tumor perfusion following VEGF blockade predicts long- REFERENCES term tumor growth inhibition in preclinical tumor models. Angiogenesis. 2013; 16: 429-441. 1. Chung AS, Lee J, Ferrara N. Targeting the tumour 11. Rapisarda A, Melillo G. Overcoming disappointing results vasculature: insights from physiological angiogenesis. Nat with antiangiogenic therapy by targeting hypoxia. Nat Rev Rev Cancer. 2010; 10: 505-514. Clin Oncol. 2012; 9: 378-390. www.impactjournals.com/oncotarget 4719 Oncotarget 12. Shweiki D, Neeman M, Itin A, Keshet E. Induction 23. Naik S, Dothager RS, Marasa J, Lewis CL, Piwnica-Worms of vascular endothelial growth factor expression by D. Vascular Endothelial Growth Factor Receptor-1 Is hypoxia and by glucose deficiency in multicell spheroids: Synthetic Lethal to Aberrant {beta}-Catenin Activation in implications for tumor angiogenesis. Proc Natl Acad Sci Colon Cancer. Clin Cancer Res. 2009; 15: 7529-7537. USA. 1995; 92: 768-772. 24. Mulkeen AL, Silva T, Yoo PS, Schmitz JC, Uchio E, Chu 13. Maxwell PH, Dachs GU, Gleadle JM, Nicholls LG, Harris E, Cha C. Short interfering RNA-mediated gene silencing AL, Stratford IJ, Hankinson O, Pugh CW, Ratcliffe PJ. of vascular endothelial growth factor: effects on cellular Hypoxia-inducible factor-1 modulates gene expression in proliferation in colon cancer cells. Arch Surg. 2006; 141: solid tumors and influences both angiogenesis and tumor 367-374. growth. Proc Natl Acad Sci USA. 1997; 94: 8104-8109. 25. Giatromanolaki A, Koukourakis MI, Sivridis E, 14. von Marschall Z, Cramer T, Höcker M, Finkenzeller G, Chlouverakis G, Vourvouhaki E, Turley H, Harris AL, Wiedenmann B, Rosewicz S. Dual mechanism of vascular Gatter KC. Activated VEGFR2/KDR pathway in tumour endothelial growth factor upregulation by hypoxia in human cells and tumour associated vessels of colorectal cancer. Eur hepatocellular carcinoma. Gut. 2001; 48: 87-96. J Clin Invest. 2007; 37: 878-886. 15. Das B, Yeger H, Tsuchida R, Torkin R, Gee MF, Thorner 26. Calvani M, Trisciuoglio D, Bergamaschi C, Shoemaker RH, PS, Shibuya M, Malkin D, Baruchel S. A hypoxia-driven Melillo G. Differential involvement of vascular endothelial vascular endothelial growth factor/Flt1 autocrine loop growth factor in the survival of hypoxic colon cancer cells. interacts with hypoxia-inducible factor-1alpha through Cancer Res. 2008; 68: 285-291. mitogen-activated protein kinase/extracellular signal- 27. Masuda K, Teshima-Kondo S, Mukaijo M, Yamagishi N, regulated kinase 1/2 pathway in neuroblastoma. Cancer Nishikawa Y, Nishida K, Kawai T, Rokutan K. A novel Res. 2005; 65: 7267-7275. tumor-promoting function residing in the 5’ non-coding 16. Harada H, Itasaka S, Kizaka-Kondoh S, Shibuya K, region of vascular endothelial growth factor mRNA. PLoS Morinibu A, Shinomiya K, Hiraoka M. The Akt/mTOR Med. 2008; 5: e94. pathway assures the synthesis of HIF-1alpha protein in a 28. Yin Y, Cao LY, Wu WQ, Li H, Jiang Y, Zhang HF. glucose- and reoxygenation-dependent manner in irradiated Blocking effects of siRNA on VEGF expression in human tumors. J Biol Chem. 2009; 284: 5332-5342. colorectal cancer cells. World J Gastroenterol. 2010; 16: 17. Nilsson MB, Zage PE, Zeng L, Xu L, Cascone T, Wu HK, 1086-1092. Saigal B, Zweidler-McKay PA, Heymach JV. Multiple 29. Samuel S, Fan F, Dang LH, Xia L, Gaur P, Ellis LM. receptor tyrosine kinases regulate HIF-1alpha and HIF- Intracrine vascular endothelial growth factor signaling in 2alpha in normoxia and hypoxia in neuroblastoma: survival and chemoresistance of human colorectal cancer implications for antiangiogenic mechanisms of multikinase cells. Oncogene. 2011; 30: 1205-1212. inhibitors. Oncogene 2010; 29: 2938-2949. 30. Laplante M, Sabatini DM. mTOR signaling at a glance. J 18. Larsen AK, Ouaret D, El Ouadrani K, Petitprez A. Cell Sci. 2009; 122(Pt 20): 3589-3594. Targeting EGFR and VEGF(R) pathway cross-talk in tumor 31. Corcoran RB, Rothenberg SM, Hata AN, Faber AC, Piris survival and angiogenesis. Pharmacol Ther. 2011; 131: 80- A, Nazarian RM, Brown RD, Godfrey JT, Winokur D, Walsh J, Mino-Kenudson M, Maheswaran S, Settleman J, 19. Larsen AK, de Gramont A, Poindessous V, Bouygues A, Wargo JA, Flaherty KT, Haber DA, Engelman JA. TORC1 Ayadi M, Mésange P. Functions and clinical implications Suppression Predicts Responsiveness to RAF and MEK of autocrine VEGF signaling in colorectal cancer. Current Inhibition in BRAF-Mutant Melanoma. Sci Transl Med. Colorectal Cancer Reports. 2013; 9: 270-277. 2013; 5: 196ra98. 20. Bates RC, Goldsmith JD, Bachelder RE, Brown C, Shibuya 32. Elkabets M, Vora S, Juric D, Morse N, Mino-Kenudson M, M, Oettgen P, Mercurio AM. Flt-1-dependent survival Muranen T, et al. Tao J, Campos AB, Rodon J, Ibrahim characterizes the epithelial-mesenchymal transition of YH, Serra V, Rodrik-Outmezguine V, Hazra S, Singh S, colonic organoids. Curr Biol. 2003; 13: 1721-1727. Kim P, Quadt C, et al. mTORC1 Inhibition Is Required 21. Fan F, Wey JS, McCarty MF Belcheva A, Liu W, Bauer for Sensitivity to PI3K p110α Inhibitors in PIK3CA-Mutant TW Somcio RJ, Wu Y, Hooper A, Hicklin DJ, Ellis LM. Breast Cancer. Sci Transl Med. 2013; 5: 196ra99. Expression and function of vascular endothelial growth 33. Pencreach E, Guérin E, Nicolet C, Lelong-Rebel I, factor receptor-1 on human colorectal cancer cells. Voegeli AC, Oudet P, Larsen AK, Gaub MP, Guenot D. Oncogene. 2005; 24: 2647-2653. Marked activity of irinotecan and rapamycin combination 22. Okita NT, Yamada Y, Takahari D, Hirashima Y, Matsubara toward colon cancer cells in vivo and in vitro is mediated J, Kato K, Hamaguchi T, Shirao K, Shimada Y, Taniguchi through cooperative modulation of the mammalian target H, Shimoda T. Vascular endothelial growth factor receptor of rapamycin/hypoxia-inducible factor-1alpha axis. Clin expression as a prognostic marker for survival in colorectal Cancer Res. 2009; 15:1297-1307. cancer. Jpn J Clin Oncol. 2009; 39: 595-600. 34. Koh MY, Powis G. Passing the baton: the HIF switch. www.impactjournals.com/oncotarget 4720 Oncotarget Trends Biochem Sci. 2012; 37: 364-372. 35. Poindessous V, Ouaret D, El Ouadrani K, Battistella A, Mégalophonos VF, Kamsu-Kom N, Petitprez A, Escargueil AE, Boudou P, Dumont S, Cervera P, Fléjou JF, André T, Tournigand C, Chibaudel B, de Gramont A, Larsen AK. EGFR- and VEGF(R)-targeted small molecules show synergistic activity in colorectal cancer models refractory to combinations of monoclonal antibodies. Clin Cancer Res. 2011; 17: 6522-6530. 36. Fan F, Samuel S, Gaur P, Lu J, Dallas NA, Xia L, Bose D, Ramachandran V, Ellis LM. Chronic exposure of colorectal cancer cells to bevacizumab promotes compensatory pathways that mediate tumour cell migration. Br J Cancer. 2011; 104: 1270-1277. 37. Yamagishi N, Teshima-Kondo S, Masuda K, Nishida K, Kuwano Y, Dang DT, Dang LH, Nikawa T, Rokutan K. Chronic inhibition of tumor cell-derived VEGF enhances the malignant phenotype of colorectal cancer cells. BMC Cancer. 2013; 13: 229. 38. Minakata K, Takahashi F, Nara T, Hashimoto M, Tajima K, Murakami A, Nurwidya F, Yae S, Koizumi F, Moriyama H, Seyama K, Nishio K, Takahashi K. Hypoxia induces gefitinib resistance in non-small-cell lung cancer with both mutant and wild-type epidermal growth factor receptors. Cancer Sci. 2012; 103: 1946-1954. 39. Soares DG, Escargueil AE, Poindessous V, Sarasin A, de Gramont A, Bonatto D, Henriques JA, Larsen AK. Replication and homologous recombination repair regulate DNA double-strand break formation by the antitumor alkylator ecteinascidin 743. Proc Natl Acad Sci USA. 2007; 104: 13062-13067. 40. Fritah A, Saucier C, De Wever O, Bracke M, Bieche I, Lidereau R, Fritah Gespach C, Drouot S, Redeuilh G, Sabbah M. Role of WISP-2/CCN5 in the maintenance of a differentiated and noninvasive phenotype in human breast cancer cells. Mol Cell Biol. 2008; 28: 1114-1123. 41. Petitprez A, Poindessous V, Ouaret D, Regairaz M, Bastian G, Guérin E, Escargueil AE, Larsen AK. Acquired irinotecan resistance is accompanied by stable modifications of cell cycle dynamics independent of MSI status. Int J Oncol. 2013; 42: 1644-1653. 42. Escargueil AE, Poindessous V, Soares DG, Sarasin A, Cook PR, Larsen AK. Influence of irofulven, a transcription-coupled repair-specific antitumor agent, on RNA polymerase activity, stability and dynamics in living mammalian cells. J Cell Sci. 2008; 121: 1275-1283. www.impactjournals.com/oncotarget 4721 Oncotarget http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Oncotarget Unpaywall

Intrinsic bevacizumab resistance is associated with prolonged activation of autocrine VEGF signaling and hypoxia tolerance in colorectal cancer cells and can be overcome by nintedanib, a small molecule angiokinase inhibitor

Download PDF
13 pages

Loading next page...
 
/lp/unpaywall/intrinsic-bevacizumab-resistance-is-associated-with-prolonged-qqLMaN9FNu

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

Publisher
Unpaywall
ISSN
1949-2553
DOI
10.18632/oncotarget.1671
Publisher site
See Article on Publisher Site

Abstract

www.impactjournals.com/oncotarget/ Oncotarget, Vol. 5, No. 13 Intrinsic bevacizumab resistance is associated with prolonged activation of autocrine VEGF signaling and hypoxia tolerance in colorectal cancer cells and can be overcome by nintedanib, a small molecule angiokinase inhibitor. 1,2,3,4 1,2,3 1,2,3 Paul Mésange , Virginie Poindessous , Michèle Sabbah , Alexandre E. 1,2,3 1,2,3,5 1,2,3 Escargueil , Aimery de Gramont and Annette K. Larsen Cancer Biology and Therapeutics, Centre de Recherche Saint-Antoine Institut National de la Santé et de la Recherche Médicale U938, Paris, France Université Pierre et Marie Curie, Paris, France Universite Paris Descartes, Paris, France Department of Medical Oncology, Hôpital Saint-Antoine, Paris, France Correspondence to: Annette K. Larsen, email: [email protected] Keywords: resistance, angiogenesis inhibition, vascular endothelial growth factor (VEGF)-signaling, hypoxia, bevacizumab, ninte-danib, Received: December 9, 2013 Accepted: January 18, 2014 Published: January 20, 2014 This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. ABSTRACT: Colorectal cancer (CRC) is a common tumor type with a high mortality rate, in part due to intrinsic drug resistance. Although bevacizumab, a VEGF-directed neutralizing antibody, is particularly active in this pathology, some patients never respond for reasons not well understood. We here wish to clarify the role of autocrine VEGF signaling in the response of CRC cells to angiogenesis inhibition. Our results show that CRC cells with intrinsic bevacizumab-resistance displayed pronounced upregulation of autocrine HIF-VEGF-VEGFR signaling in response to prolonged bevacizumab exposure whereas the same signaling pathway was downregulated in bevacizumab-sensitive xenografts. Importantly, both bevacizumab-sensitive and -resistant CRC xenografts were sensitive to nintedanib, a small molecule angiokinase inhibitor, which was associated with inhibition of mTORC1. In vitro studies revealed that bevacizumab- resistant cells displayed intrinsically higher HIF-VEGF signaling intensity and hypoxia tolerance compared to their bevacizumab-sensitive counterparts. Interestingly, although nintedanib showed comparable activity toward bevacizumab-sensitive cells under normoxia and hypoxia, the drug was three-fold more toxic to the resistant cells under hypoxia, suggesting that nintedanib attenuated the survival signaling that usually protects these cells from hypoxia-mediated cell death. In conclusion, our ndings fi support a role for autocrine VEGF signaling in the survival of CRC cells to hypoxia and thus to angiogenesis inhibition. We further show that nintedanib, a small molecule angiokinase inhibitor, is active toward CRC models with intrinsic bevacizumab resistance supporting clinical trials of nintedanib in patients that do not respond to bevacizumab, alone or in combination with bevacizumab to increase angiogenesis inhibition. inhibitors. Vascular endothelial growth factor A (hereafter INTRODUCTION referred to as VEGF) is a key regulator of angiogenesis. Amplification of the VEGF locus is observed in a subset Neovascularization (sprouting angiogenesis) is of patients with metastatic colorectal cancer (mCRC) required for the growth of most solid tumors and facilitates and is associated with a remarkably aggressive disease the spread of tumor cells to secondary sites [1] providing characterized by a high incidence of vascular invasion a rational basis for the clinical use of angiogenesis www.impactjournals.com/oncotarget 4709 Oncotarget [2]. Recent data from more than 2000 mCRC patients acquired bevacizumab resistance have focused on indicate that high baseline plasma levels of VEGF might paracrine VEGF signaling and tumor-associated be a negative prognostic factor for both progression-free endothelial cells. However, the capacity of autocrine survival (PFS) and overall survival (OS) [3]. Therefore, VEGF-signaling to promote CRC cell survival under VEGF signaling is linked to invasiveness and aggressive different types of stress suggests that this pathway may disease in CRC and appears as an attractive therapeutic also play a role during angiogenesis inhibition. target. We now report that bevacizumab-resistant, but Several VEGF(R)-targeted agents are approved not bevacizumab-sensitive, CRC cells showed strong or are undergoing clinical trials for treatment of CRC. autocrine HIF-VEGF-VEGFR signaling in response to Bevacizumab (avastin), a VEGF-neutralizing monoclonal prolonged bevacizumab exposure in vivo and displayed antibody [4], was the first angiogenesis inhibitor to be intrinsically higher HIF-VEGF signaling intensity and approved and represents the current benchmark. Although hypoxia tolerance in vitro. We further show that tumors bevacizumab shows excellent activity in some patients, with intrinsic bevacizumab resistance remain sensitive to others never respond, for reasons not well understood. nintedanib, a small molecule angiokinase inhibitor. These Nintedanib (BIBF 1120) is a small molecule tyrosine data suggest that the antitumor activity of at least some kinase inhibitor that inhibits several angiokinases small molecule angiokinase inhibitors is not limited by the including VEGFR1/Flt1, VEGFR2/Flk1/KDR and mechanisms underlying natural bevacizumab resistance VEGFR3/Flt4 as well as FGFR1, FGFR2, FGFR3, and provide a rational for clinical trials of nintedanib in PDGFR-alpha, PDGFR-beta and Flt3 [5]. Nintedanib CRC patients that do not respond to bevacizumab, alone or is currently in phase III clinical trials in ovarian and in combination with bevacizumab to increase angiogenesis non-small cell lung cancer (NSCLC), where it has been inhibition. successful in combination with taxotere [6]. Although the mechanism of angiogenesis inhibitors RESULTS is not fully elucidated, it is generally believed that pruning of the tumor microvasculature will decrease blood supply thereby diminishing tumor oxygenation and promoting HT-29 xenografts are naturally bevacizumab- tumor cell death [7-11]. Hypoxia is accompanied resistant but remain sensitive to nintedanib, a by activation of the hypoxia-inducible transcription small molecule angiokinase inhibitor. factors HIF-1alpha and HIF-2alpha leading to increased expression of VEGF and other HIF targets [12, 13]. Binding of VEGF to its receptors triggers signaling Continued bevacizumab treatment of mice carrying pathways that modulate the phosphorylation, stability and/ human CRC xenografts revealed that HT-29 tumors are or activity of a variety of down-stream targets including naturally bevacizumab-resistant with only 29% growth HIF and VEGF [14-17], thereby initiating a positive feed- inhibition after 4 weeks drug exposure while the same back loop. Thus, HIF-VEGF signaling intensity emerges scheduling resulted in 68% tumor growth inhibition for as a key factor in determining the outcome of angiogenesis the bevacizumab-sensitive DLD-1 tumors (Figure 1A). inhibition. To establish if bevacizumab resistance is associated A striking feature of CRC is the capacity for both with cross-resistance to small molecules targeting the paracrine and autocrine VEGF-signaling. CRC cells are a same pathways, nintedanib was administered to animals major source of VEGF that interacts with VEGF receptors with DLD-1 or HT-29 xenografts. The results show that on tumor-associated endothelial cells thereby stimulating nintedanib displays comparable tumor growth inhibitory their growth, migration and survival [18]. In addition, activity in the two tumor models (Figure 1A). No weight CRC cells express VEGF receptors (VEGFRs) giving rise loss or other toxic side effects were noted for either to autocrine VEGF signaling [19]. Results from different bevacizumab or nintedanib (data not shown). Therefore, laboratories indicate that most, if not all, human CRC CRC tumors with intrinsic bevacizumab resistance remain cells and tumors express functional VEGFR1 [20-23] sensitive to at least some small molecule angiokinase as well as VEGFR2 [24-26]. Autocrine VEGF signaling inhibitors. promotes CRC survival under different types of stress including 5-fluorouracil exposure, low serum conditions Bevacizumab-resistant xenografts express high and anchorage-independent growth [27-29]. Interestingly, levels of VEGF. VEGFR1-mediated VEGF signaling may be particularly important for the survival of invasive CRC cells that have Since bevacizumab and nintedanib both interfere undergone the epithelial-mesenchymal transition and no with VEGF signaling, tissue extracts were prepared from longer benefit from homotypic cell-cell contact [20]. untreated DLD-1 and HT-29 tumors and the concentration Until now, most studies on both intrinsic and www.impactjournals.com/oncotarget 4710 Oncotarget of human VEGF was determined by ELISA analysis Bevacizumab has limited activity on the (Figure 1B). The amounts of VEGF were at least 4 times microvasculature in bevacizumab-resistant higher in tumor tissues from HT-29 xenografts compared xenografts. to DLD-1 xenografts (942 pg vs. 223 pg VEGF/µg protein, respectively, p < 0,001). The microvascular density of DLD-1 and HT-29 Bevacizumab is specific for human VEGF, and xenografts was compared by IHC with a CD31-directed will therefore only neutralize VEGF produced by the antibody followed by quantitative image analysis. The tumor cells, but not murine VEGF secreted by the tumor- results (Figure 1C) indicate that bevacizumab reduces associated stromal cells. To evaluate the contribution the microvascular density by 70% in DLD-1 xenografts, of stromal VEGF, tissue extracts were prepared from compared to the vehicle control. In contrast, the same untreated DLD-1 and HT-29 tumors and the concentration schedule of bevacizumab only reduced the microvascular of murine VEGF was determined by ELISA analysis. The density by 27% in HT-29 xenografts. These results show results show that murine VEGF represents a minor fraction that the microvascular density was less influenced by of the total VEGF (0,4 to 1.1%) in both tumor models bevacizumab in the bevacizumab-resistant xenografts (Figure 1B). Therefore, VEGF signaling is principally whereas nintedanib had comparable activity in the two mediated by human, tumor-derived VEGF in both DLD-1 models. and HT-29 xenografts. Figure 1: Influence of bevacizumab and nintedanib on tumor growth and angiogenesis in CRC xenografts. (A) Nude mice with DLD-1 (grey columns) or HT-29 (black columns) human CRC xenografts were dosed with vehicle (Control), bevacizumab at 5 mg/kg i.p. every 3 days (Beva), nintedanib at 12,5 mg/kg p.o. once daily (Nin 12,5) or nintedanib at 50 mg/kg p.o. once daily (Nin 50) for 4 weeks. Each treatment group corresponded to at least seven animals. Left, T/C values were determined as follows: average tumor volume of treated animals/average tumor volume of the vehicle controls x100. Right, box and whisker plot of the tumor volumes of DLD-1 (light grey boxes) or HT-29 (dark grey boxes) xenografts after 4 weeks treatment with bevacizumab or nintedanib. Lines, medians; boxes, 25th to 75th percentile interquartile ranges; whiskers, the highest and lowest values for a given treatment. (B) Total protein was extracted from DLD-1 and HT-29 xenografts (pool of 3 tumors per xenograft model) and the amounts of human (tumor-derived) and murine (stroma-derived) VEGF were determined by ELISA. (C) Top, representative images of xenografts from animals treated with vehicle or with nintedanib at 50 mg/kg p.o. once daily. CD31-positive blood vessels are outlined in red whereas the nuclei appear in blue. Bottom, quantitative image analysis of the CD31 signal for DLD-1 (grey columns) and HT-29 tumors (black columns). The data show the CD31-positive area, as % of total, and represent the average of at least 6 fields/tumor for at least 3 different tumors. The statistical analysis of experimental data was performed using a Student’s paired t-test comparing the treatment group with the vehicle control. Bars, SD; * p < 0,05; ** p < 0,01; *** p < 0,001. www.impactjournals.com/oncotarget 4711 Oncotarget effects without any striking differences between them: Bevacizumab and nintedanib show a combination 37% vs 33% decrease in DNA synthesis, 265% vs 299% of cytostatic and cytotoxic activities. increased apoptosis, and 236% vs 209% increased necrosis for bevacizumab and nintedanib, respectively (Figure 2). Tumor growth inhibition may be due to both We then compared the influence of nintedanib cytostatic (cell cycle arrest) and cytotoxic (cell death) in the two tumor models. Interestingly, both doses of effects. We therefore evaluated the influence of nintedanib induced more apoptosis in HT-29 tumors bevacizumab and nintedanib on in vivo DNA synthesis (299% and 640%, for the low and high dose of nintedanib, (EdU incorporation), apoptotic cell death (TUNEL assay) respectively) than in DLD-1 tumors (126% and 178%, and necrotic cell death (Figure 2). for the low and high dose of nintedanib, respectively), We first compared the influence of the two drugs. whereas the induction of necrosis was comparable (Figure In DLD-1 xenografts, bevacizumab and nintedanib (50 2). mg/kg) have comparable antitumor activity. Both agents Taken together, our findings indicate that treatment displayed a mixture of cytotostatic and cytotoxic effects with both bevacizumab and nintedanib resulted in without any striking differences between them: 43% a mixture of cytostatic and cytotoxic effects. The vs 46% decrease in DNA synthesis as determined by pronounced apoptosis in nintedanib-treated HT-29 tumors EdU, 197% vs 178% increase in apoptotic cell death as suggests that nintedanib may not only display antivascular determined by TUNEL, and 480% vs 353% increased effects in vivo but also be able to interfere with survival necrosis for bevacizumab and nintedanib, respectively. signaling in the CRC cells. (Figure 2). In HT-29 xenografts, bevacizumab and low dose nintedanib have comparable antitumor activity. Both agents displayed a mixture of cytostatic and cytotoxic Figure 2: Cytostatic and cytotoxic effects of bevacizumab and nintedanib in CRC xenografts. Influence of bevacizumab and nintedanib on (A) DNA synthesis, (B) apoptosis and (C) necrosis in DLD-1 (grey columns) and HT-29 (black columns) tumors. Animals with CRC xenografts were dosed with vehicle (Control), bevacizumab at 5 mg/kg i.p. every 3 days (Beva), nintedanib at 12,5 mg/ kg p.o. once daily (Nin 12,5) or nintedanib at 50 mg/kg p.o. once daily (Nin 50) for 4 weeks. The photos illustrate typical staining patterns. V, viable tissue; N, necrotic tissue. DNA synthesis is determined as the ratio between EdU-positive cells and the total number of viable cells and corresponds to the average of 6 fields/tumor (each field representing approximately 1700 cells) from at least 3 different tumors. Apoptosis is expressed as the area of TUNEL-positive cells, as % of the total area of viable cells, and is the average of 6 fields/tumor for at least 3 different tumors. For necrosis, the data indicates the surface of necrotic cells as % of the total surface and is the average of 6 fields/ tumor for at least 3 different tumors. The statistical analysis of experimental data was performed using a Student’s paired t-test comparing the treatment group with the vehicle control. Bars, SD; * p < 0,05; ** p < 0,01; *** p < 0,001. www.impactjournals.com/oncotarget 4712 Oncotarget treatments with comparable antitumor activity, for most Angiogenesis inhibition has markedly different markers (Figure 3), although the downregulation of HIF- influence on HIF-VEGF signaling in bevacizumab 2alpha was more marked for nintedanib-treated tumors sensitive and -resistant tumors. than for bevacizumab-treated tumors. In clear contrast, for HT-29 tumors, the signal for all markers was up to Microvascular pruning limits oxygen supply to 3-fold higher following bevacizumab exposure than after the tumor thereby activating the hypoxia-inducible exposure to nintedanib (12,5 mg/kg), in spite of their transcription factors, HIF-1alpha and HIF-2alpha. This comparable antitumor and antivascular activities. In is accompanied by transcriptional upregulation of HIF particular, for the nintedanib- and bevacizumab-treated targets like VEGF, thereby promoting autocrine VEGF tumors, the signal for pVEGFR1 was 110% vs 192% while signaling. The presence of HIF-1alpha, HIF-2alpha, VEGF the signal for pVEGFR2 was 75% vs 205%, respectively. and the active autophosphorylated forms of VEGFR1 These findings likely reflect a direct inhibitory effect of and VEGFR2 in the two xenograft models was revealed nintedanib on the tumor cell-associated VEGF receptors, by IHC followed by quantitative image analysis (Figure consistent with the potent angiokinase inhibitory activity 3). The results revealed striking differences between of this compound. DLD-1 and HT-29 xenografts, since all treatments were Taken together, these results reveal both tumor- and accompanied by attenuation of the HIF-VEGF-VEGFR drug-specific differences with respect to HIF-VEGF- axis in DLD-1 tumors, but activation of the same signaling VEGFR signaling. For the bevacizumab-sensitive components in HT-29 xenografts. DLD-1 tumors, all treatments were accompanied by For DLD-1 tumors, there were no marked differences downregulation of the HIF-VEGF-VEGFR axis whereas between nintedanib (50 mg/kg) and bevacizumab, two Figure 3: Influence of bevacizumab and nintedanib on HIF-VEGF-VEGFR signaling in CRC xenografts. Nude mice with DLD-1 (grey columns) or HT-29 (black columns) human CRC xenografts were dosed with vehicle (Control), bevacizumab at 5 mg/ kg i.p. every 3 days (Beva), nintedanib at 12,5 mg/kg p.o. once daily (Nin 12,5) or nintedanib at 50 mg/kg p.o. once daily (Nin 50) for 4 weeks. The expression of HIF-1alpha, HIF-2alpha, VEGF, pVEGFR1, pVEGFR2 and pS6 was determined by immunohistochemistry followed by quantitative image analysis. The photos illustrate the typical staining patterns for tumors derived from animals treated with bevacizumab (Beva) or nintedanib (Nin, 12,5 mg/kg) that provided comparable tumor growth inhibition. For the quantitative analysis of the signal intensity, the data represent the average fluorescence intensity of treated tumors compared to the treatment intensity of control tumors and are the average of 6 fields/tumor for at least 3 different tumors. The statistical analysis of experimental data was performed using a Student’s paired t-test comparing the treatment group with the vehicle control. Bars, SD; * p < 0,05; ** p < 0,01; *** p < 0,001. www.impactjournals.com/oncotarget 4713 Oncotarget the same treatments resulted in activation of autocrine Nintedanib, but not bevacizumab, attenuates VEGF signaling in the bevacizumab-resistant HT-29 mTORC1 activity in bevacizumab-resistant tumors. Furthermore, for HT-29 tumors, bevacizumab xenografts. treatment was associated with much stronger activation of autocrine VEGF signaling than was the case for VEGF receptors and other RTKs mediate multiple nintedanib. downstream signaling pathways that may be integrated at the level of mTORC1, the mammalian target of rapamycin (mTOR) complex 1 [30]. Recent findings suggest that inhibition of mTORC1 activity, as measured by tumor levels of Ser240/Ser244-phosphorylated S6 (pS6) is a robust biomarker for the antitumor activity of at least some protein kinase inhibitors such as inhibitors of mutant BRAF and PI3K [31, 32]. In the present study, all treatments reduced the expression of pS6 in DLD-1 tumors, which was most pronounced for bevacizumab and nintedanib at the 50 mg/kg dose (Figure 3). In HT- 29 tumors, nintedanib induced a marked, dose-dependent decrease of pS6 levels reaching 12% for the high dose of nintedanib, compared to the vehicle control. In clear contrast, bevacizumab increased the levels of pS6 to ~130% in the same xenograft model. Bevacizumab and nintedanib show synergistic activity toward bevacizumab-resistant xenografts. The important differences in the activities of bevacizumab and nintedanib toward HT-29 xenografts prompted us to explore the combination of the two agents (Figure 4). Bevacizumab and low dose nintedanib showed comparable activity toward HT-29 xenografts as single Figure 4: Influence of bevacizumab, nintedanib and agents with approximately 27% tumor growth inhibition. their combination on tumor growth, microvessel In comparison, the combination of bevacizumab and density and mTORC1 activity in bevacizumab- nintedanib was accompanied by 65% tumor growth resistant xenografts. Nude mice with HT-29 human CRC inhibition (p < 0,001, bevacizumab alone vs bevacizumab xenografts were dosed with vehicle (Control), bevacizumab at and nintedanib). The combination was active toward both 5 mg/kg i.p. every 3 days (Beva), nintedanib at 12,5 mg/kg p.o. endothelial and tumor cells. The microvascular density, once daily (Nin 12,5) or with a combination of bevacizumab and nintedanib for 4 weeks. Each treatment group corresponded as determined by CD31 staining, decreased by ~28% to at least seven animals. (A) Average tumor growth of HT- when bevacizumab or nintedanib were given alone, to 29 tumorgrafts in mice treated with vehicle (black squares), 63% when the two agents were given together (p < 0,001, bevacizumab (black circles), nintedanib (white circles) or their bevacizumab alone vs bevacizumab and nintedanib). In combination (light grey circles). (B) Box and whisker plot of addition, nintedanib was able to counteract the activation the tumor volumes of HT-29 xenografts after 4 weeks treatment of mTORC1 in the tumor cells, as measured by pS6, with bevacizumab, nintedanib or their combination. Lines, from 131% (compared to the vehicle control) when medians; boxes, 25th to 75th percentile interquartile ranges; bevacizumab was given alone to 72% when the two agents whiskers, the highest and lowest values for a given treatment. were given together (p < 0,001, bevacizumab alone vs The expression of CD31 (C) and pS6 (D) was determined by immunohistochemistry followed by quantitative image analysis. bevacizumab and nintedanib). The photos illustrate typical staining patterns for tumors derived from animals treated with bevacizumab or with a combination Brief hypoxia is accompanied by HIF-1alpha of bevacizumab plus nintedanib. For the quantitative analysis of expression in both DLD-1 and HT-29 cells. the signal intensity, the data represent the average fluorescence intensity of treated tumors compared to the treatment intensity of control tumors and are the averages of 6 fields/tumor for at The marked differences in HIF expression between least 3 different tumors. The statistical analysis of experimental DLD-1 and HT-29 xenografts raise the question if DLD-1 data was performed using a Student’s paired t-test comparing the cells are unable to activate HIF, unable to maintain HIF treatment group with the vehicle control. Bars, SD; * p < 0,05; signaling, or rather if HIF signaling intensity in DLD-1 ** p < 0,01; *** p < 0,001. www.impactjournals.com/oncotarget 4714 Oncotarget tumors is weak compared to HT-29 tumors. To answer 24 hrs hypoxia is accompanied by HIF-1alpha induction this question, DLD-1 and HT-29 cells were exposed to in both cell lines whereas the expression of HIF-2alpha acute hypoxia (1% O ) for 24 hrs and the expression of remains marginal (Figure 5A). HIF-1alpha and HIF-2alpha was determined by Western To establish if the HIF-1alpha protein was blot analysis. The results show that the expression of the transcriptionally active, the expression of its downstream two HIF proteins is marginal under normoxia. In contrast, target, VEGF, was determined by qRT-PCR. Acute hypoxia was accompanied by a ~10-fold increase in VEGF expression for HT-29 cells, compared to the corresponding normoxic cells, whereas the corresponding increase in VEGF expression was ~3.5-fold for DLD-1 cells. Therefore, DLD-1 cells show a functional hypoxia response as indicated by upregulation of VEGF mRNA, although to a lesser degree than HT-29 (Figure 5B). Bevacizumab and nintedanib upregulate HIF- 1alpha in both bevasizumab-sensitive and -resistant cells. We next explored if angiogenesis inhibitors might have a direct effect on HIF-VEGF signaling in CRC cells. Our results indicate that the presence of nintedanib or bevacizumab had no detectable effect on the expression of HIF-1alpha or HIF-2alpha under normoxia (Figure 5A). However, under hypoxia, nintedanib activated the expression of HIF-1alpha in a dose-dependent manner in both cell lines (2 to 3-fold, compared to the normoxia control) (Figure 5A). In addition, bevacizumab was able Figure 5: Influence of bevacizumab and nintedanib on CRC cells under normoxia and hypoxia. (A) Western blot analysis of HIF-1alpha, HIF-2alpha and beta-actin in DLD- 1 and HT-29 cells after 24 hrs exposure to nintedanib (0,01, 0,1 or 1 µM) or bevacizumab (250 µg/ml) under normoxia or hypoxia (1% O ). (B) Induction of VEGF mRNA in DLD-1 (grey columns) and HT-29 (black columns) cells following 24 hrs exposure to nintedanib (0,01, 0,1 or 1µM) or bevacizumab (250 µg/ml) under normoxia or hypoxia. VEGF expression was normalized with 36B4 mRNA and is the average of three independent experiments. The statistical analysis of experimental data was performed using a Student’s paired t-test comparing the treatment group with the vehicle control. Bars, SD; * p < 0,05; ** p < 0,01; *** p < 0,001. (C) Hypoxia-induced apoptosis in DLD-1 (grey columns) and HT-29 (black columns) cells after hypoxia (1% O ) exposure for the indicated times as measured by the TUNEL assay (left) or by the formation of catalytically active cleaved caspase 3 (right). Values are the averages of at least two independent experiments done in duplicate. The statistical analysis of experimental data was performed using a Student’s paired t-test comparing the treatment group with the Figure 6: HIF-VEGF-VEGFR signaling in CRC vehicle control. Bars, SD; * p < 0,05; ** p < 0,01; *** p < 0,001. cells. Hypoxia exposure activates HIF, thereby increasing the (D) Viability of DLD-1 cells (left) or HT-29 cells (right) after expression of VEGF, a HIF-transcriptional target. Subsequent 24 hrs exposure to the indicated concentrations of nintedanib binding of VEGF to its receptor on either CRC cells (autocrine under normoxia (grey circles) or hypoxia (black circles) for pathway) or tumor-associated endothelial cells (paracrine 24 hrs followed by 96 hrs post-incubation under normoxia and pathway) leads to activation of VEGFR and its downstream MTT determination. Bars indicate SD and are shown when signaling pathways. Receptor tyrosine kinases like VEGFR they exceed symbol size. The data represents three independent are able to activate and/or stabilize HIF and VEGF, thereby experiments each done in duplicate. promoting a positive feed-back loop. www.impactjournals.com/oncotarget 4715 Oncotarget to induce HIF-1alpha almost 4-fold in the HT-29 cells days) exposure to the angiokinase inhibitor nintedanib whereas its influence was marginal in DLD-1 cells (Figure showed that DLD-1 and HT-29 xenografts are both 5A). sensitive to nintedanib at doses with no detectable toxic The drug-induced increase of HIF-1alpha protein side effects. These findings suggest that the antitumor had modest influence on the expression of VEGF mRNA activity of at least some small molecule angiokinase in hypoxic DLD-1 cells (Figure 5B). In clear contrast, inhibitors is not limited by the mechanisms underlying for hypoxic HT-29 cells, nintedanib-exposure was natural bevacizumab resistance. accompanied by a dose-dependent increase in mRNA Next, we characterized the differences between VEGF expression that reached ~2,4 times the levels of bevacizumab-sensitive and -resistant cells, with special the untreated controls, whereas bevacizumab increased attention to VEGF signaling. Comparison of the vascular mRNA VEGF expression ~1.8-fold. Therefore, although density in DLD-1 and HT-29 tumors by IHC analysis drug exposure was associated with increased HIF-1alpha reveals that bevacizumab treatment is associated with accumulation under short-term hypoxia in both cell lines, important vascular pruning in the sensitive DLD- VEGF mRNA expression was only increased for HT-29 1, but not in the resistant HT-29 xenografts. In clear cells. contrast, nintedanib showed comparable effects on the microvascular density in both xenograft models. To establish if the intrinsic bevacizumab resistance of HT-29 Bevacizumab-resistant cells show increased xenografts might be linked to elevated levels of stroma- hypoxia tolerance. derived VEGF, the concentrations of human and murine VEGF were determined. The results show that for both Next, DLD-1 and HT-29 cells were exposed to tumor types, murine VEGF represents, at the most, 1% hypoxia for up to 48 hrs and the fraction of apoptotic of the total amount of VEGF and is therefore unlikely cells was determined by the TUNEL assay (Figure to influence the response to angiogenesis inhibition. 5C left). The results show that hypoxia exposure was Interestingly, HT-29 tumors express at least four times accompanied by rapid apoptosis in DLD-1 cells to reach more human VEGF, compared to the DLD-1 tumors, almost 17% apoptotic cells by 48 hrs, whereas the level which likely contributes to the bevacizumab-resistant of apoptosis stayed below 5% for HT-29. These findings phenotype. were confirmed by a caspase 3 assay that measures the To establish the mechanistic basis for tumor formation of catalytically active cleaved caspase 3 (Figure growth inhibition, the levels of DNA synthesis, apoptosis 5C right). and necrosis were determined. The results show that prolonged exposure to the two drugs resulted in a Nintedanib shows increased toxicity toward mixture of cytostatic and cytotoxic effects, independent bevacizumab-resistant cells under hypoxia. of the tumor model. However, nintedanib exposure was associated with significantly more apoptosis in HT-29 than in DLD-1 xenografts, suggesting that in this tumor model, To determine the influence of hypoxia on nintedanib nintedanib may not only act on the microvasculature, but toxicity, cells were exposed to different doses of nintedanib also attenuate the survival response of the tumor cells to for 24 hrs under hypoxia or normoxia followed by 96 hrs environmental stress. post-exposure under normoxia and MTT determination. To characterize the signaling intensity of the HIF- The results show that nintedanib is slightly (~30%) VEGF signaling pathway (summarized in Figure 6), more toxic for DLD-1 cells under hypoxia compared to the expression of HIF-1alpha, HIF-2alpha, VEGF and normoxia (Figure 5D left). In comparison, nintedanib was the active, autophosphorylated forms of VEGFR1 and almost 3 times more toxic to HT-29 cells under hypoxia, VEGFR2 was determined by IHC followed by quantitative suggesting that nintedanib attenuates the survival signaling image analysis. The results revealed striking differences that usually protects HT-29 from hypoxia-mediated cell between the two tumor models. Drug treatment was death (Figure 5D right). accompanied by a general downregulation of all signaling components of the HIF-VEGF-VEGFR autocrine loop in DISCUSSION DLD-1 tumors, in clear contrast to HT-29 tumors, where the same pathway was upregulated. Interestingly, for HT- Bevacizumab shows excellent activity in a subset 29 tumors, treatment with bevacizumab was accompanied of mCRC patients while others never respond. A key by much higher activation of autocrine VEGF signaling question is if bevacizumab-resistance is limited to this than nintedanib, thereby identifying a key difference particular agent or rather is indicative of a general between the two agents. resistance to angiogenesis inhibition. To clarify this mTORC1 integrates multiple signaling pathways question, we identified the HT-29 xenograft model as downstream of activated receptor tyrosine kinases. naturally bevacizumab-resistant in contrast to DLD-1 Accordingly, recent findings suggest that the degree of xenografts that are bevacizumab-sensitive. Prolonged (28 www.impactjournals.com/oncotarget 4716 Oncotarget mTORC1 activity, as determined by the levels of the drives the initial response to hypoxia whereas HIF- Ser240/Ser244 phosphorylated form of the ribosomal 2alpha is needed for the chronic response [34]. Next, we S6 protein (pS6), serve as a robust biomarker for the determined the capacity of HIF to increase the expression antitumor activity of at least some signaling pathways of VEGF mRNA. The results reveal that although VEGF inhibitors. Quantitative IHC analysis showed that was induced by hypoxia in both cell lines, the increase in treatment with both bevacizumab and nintedanib was VEGF levels was more prominent for HT-29 cells (~10- accompanied by downregulation of mTORC1 activity in fold increase) than for DLD-1 cells (~3-fold increase). DLD-1 tumors. In clear contrast, bevacizumab exposure Interestingly, the differences in VEGF expression in vitro was associated with upregulation of mTORC1 activity were comparable to the differences in VEGF expression in HT-29 tumors whereas nintedanib treatment was observed for the corresponding xenografts. Therefore, associated with a marked decrease of mTORC1 activity, the in vivo differences in basal VEGF levels appear to be compared to the vehicle control. For the bevacizumab- a result of an intrinsically higher HIF-VEGF signaling treated HT-29 tumors, the increase in mTORC1 intensity in bevacizumab-resistant cells. activity is fully coherent with the elevated levels of Unexpectedly, the hypoxia-mediated up-regulation active, phosphorylated VEGFR1 and VEGFR2. In of HIF-1alpha protein was further increased in the contrast, the strong downregulation of mTORC1 in the presence of nintedanib and bevacizumab, suggesting nintedanib-treated HT-29 tumors was unexpected. The that the two agents can influence tumor cells directly downregulation is unlikely to rely on VEGF-signaling under cellular stress. The activity of nintedanib is alone, but likely involves other nintedanib targets like consistent with previous findings, where nintedanib was FGFR (fibroblast growth factor receptor) and PDGF shown to modulate cell cycle progression and viability (platelet-derived growth factor receptor receptor) family of a wide range of CRC cell lines [35]. The influence members [5]. of bevacizumab is more unexpected considering that Another unexpected finding was that HIF-1alpha bevacizumab is a ligand-targeted agent that is unable was upregulated in the nintedanib-treated HT-29 tumors to enter the tumor cells. However, two recent articles in spite of the marked downregulation of mTORC1, report that chronic in vitro exposure of CRC cells to a well-established negative regulator of HIF-1alpha bevacizumab was accompanied by the emergence of a translation. The upregulation of HIF-1alpha is coherent more migratory and invasive phenotype [36, 37], which is with the loss of tumor-associated microvessels that coherent with a direct effect of bevacizumab on the tumor results in tumor hypoxia and activation of autocrine cells, as observed here. VEGF signaling. Furthermore, it should be noted that the HT-29 cells were more resistant to hypoxia-induced regulation of HIF-1alpha is complex and is mediated at apoptosis compared to DLD-1 cells consistent with their several levels including transcription, translation as well bevacizumab-resistant phenotype in vivo. If the increased as protein stability. Even on the translational level, our survival of HT-29 cells under hypoxia depends on their previous results suggest that HIF-1alpha expression can potent kinase activity including autocrine VEGF-signaling, be modulated by (at least) two different agents, rapamycin we would predict that nintedanib should be more cytotoxic via mTORC1, and irinotecan, that acts in a mTORC1- under hypoxic conditions than under normoxia. In independent manner [33]. contrast, we would expect a smaller difference for DLD- Interestingly, the beva-resistance of the HT-29 1 cells that are inherently hypoxia-sensitive and appear xenografts could be overcome by combining bevacizumab to have weaker angiokinase signaling. In agreement, with low dose nintedanib. The additional antitumor effect the results showed almost three-fold increased activity was characterized by a diminution, rather than an increase, of nintedanib under hypoxia in HT-29 cells, whereas of mTORC1 activity as well as by increased inhibition of nintedanib was only marginally more active under hypoxia the tumor microvasculature. toward DLD-1 cells. The increased activity of nintedanib A crucial question is if bevacizumab-sensitive under hypoxia is a noticeable feature, since other tyrosine cells are unable to activate HIF-signaling in response kinase inhibitors like gefitinib have been reported to show to hypoxia, unable to maintain HIF activation, or if the less, rather than more, cytotoxic activity under hypoxia HIF signaling pathway is functional, but less prominent [38]. compared to bevacizumab-resistant cells. To answer this Taken together, our data indicate that intrinsic question, DLD-1 and HT-29 cells were incubated under bevacizumab-resistance is multifactorial and associated hypoxia for 24 hrs and the induction of HIF-1alpha and with i) the high levels of VEGF in the tumor environment HIF-2alpha was determined. The results show that both that renders the endothelial cells more resistant to cell lines express little, if any, HIF-1alpha or HIF-2alpha bevacizumab, ii) strong VEGFR1 and VEGFR2 signaling under normoxia. In contrast, hypoxia is accompanied by in the tumor cells that leads to mTORC1 activation as robust HIF-1alpha induction in both cell lines whereas the well as iii) resistance to hypoxia-induced apoptosis. We induction of HIF-2alpha is marginal. These findings are further show that all these resistance mechanisms can coherent with previous work reporting that HIF-1alpha be overcome by nintedanib. These findings suggest that www.impactjournals.com/oncotarget 4717 Oncotarget the antitumor activity of at least some small molecule MetaMorph software (Universal Imaging Corporation) for angiokinase inhibitors is not limited by the mechanisms quantitative image analysis. underlying natural bevacizumab resistance and provide Blood vessel density is expressed as the CD31- a rational for clinical trials of nintedanib in CRC positive area, in % of total, and represents the averages patients that do not respond to bevacizumab, alone or in of at least 6 fields/tumor for at least 3 different tumors. combination with bevacizumab. DNA synthesis is determined as the ratio between EdU- positive cells and the total number of viable cells and is the average of 6 fields/tumor (each field representing MATERIALS AND METHODS approximately 1700 cells) from at least 3 different tumors. Apoptosis is expressed as the % area of TUNEL-positive cells compared to the total area of viable cells, and is the Xenograft models. average of 6 fields/tumor for at least 3 different tumors. For necrosis, the data indicates the surface of necrotic cells The antitumor activity of bevacizumab and as % of the total surface and is the average of 6 fields/ nintedanib was evaluated in athymic mice (female tumor for at least 3 different tumors. NMRI-Fox 1nu, 6 weeks old) from Taconic (Skensved, For the quantitative analysis of the signal intensity Denmark) bearing DLD-1 or HT-29 CRC xenografts. Two for HIF-1alpha, HIF-2alpha, VEGF, pVEGFR1, (HT-29) or five (DLD-1) million cells were injected into pVEGFR2 and pS6, the data represents the average the right flank, and the treatments were started when the fluorescence intensity of treated tumors compared to the tumors were palpable (median tumor volume ~100 mm ). treatment intensity of control tumors and is the average of The animals were weighted daily and the tumor size 6 fields/tumor for at least 3 different tumors. was determined three times per week. Tumor volumes (mm ) were calculated according to formula: [(length² x ELISA assay for VEGF. width)/2]. Treated/Control (T/C) values were calculated as follows: average tumor volume of treated animals/average Healthy tumor tissues were collected from untreated, tumor volume of control animals x100. Each treatment frozen tumors (three tumors per xenograft model), and group was composed of at least 7 animals. Animals were protein extracts were prepared in RIPA buffer according treated according to institutional guidelines. to the manufacturer’s instructions. VEGF levels were determined by Quantikine ELISA (R&D System # DVE00 Immunohistochemistry. (human, tumor-derived VEGF) and MMV00 (murine, stroma-derived VEGF). The values represent the average Biomarker analysis was carried out with tumors of 3 independent experiments, each done in duplicate. collected after 28 days of treatment. To measure in vivo DNA synthesis, the thymidine analog 5-ethynyl-2’- Tumor cells, apoptosis and viability assays. deoxyuridine (EdU, Life Technologies) was administered 48 hrs before sacrifice (500 µg i.p.). The incorporated EdU HT-29 cells were kindly provided by Richard was revealed by a fluorescent-azide coupling reaction Camalier, Division of Cancer Treatment and Diagnosis (Click-iT # C10337, Life Technologies) of paraffin- tumor repository (NCI, US) whereas DLD-1 cells were embedded tumor samples and counterstained by DAPI a generous gift from Richard Hamelin (Saint-Antoine to reveal the nuclei of individual cells. The proportion of Research Center, Paris, France). apoptotic tumor cells was scored by the TUNEL assay (In To determine the induction of hypoxia-induced Situ Cell Death Detection kit, Roche Applied Science # apoptosis, HT-29 and DLD-1 cells were seeded on 11684795910). coverslips in DMEM medium with 5% fetal calf serum The following antibodies were used for under hypoxia (1% O ). Cover slips were removed after immunohistochemistry (IHC) analysis: anti-CD31 (BD 0, 12, 24 and 48 hrs hypoxia and subjected to TUNEL Bioscience # 550274), anti-HIF-1alpha (BD Bioscience # analysis. Alternatively, hypoxia-induced apoptosis was 610958), anti-HIF-2alpha (Novus Biological NB-100122), determined by the capacity of catalytically active cleaved anti-VEGF (Santa Cruz # sc-152), anti-pVEGFR1 caspase-3 to generate fluorescent reaction products (Millipore 07-758) that recognizes phospho-Tyr1213 (Caspase 3 fluorescence assay kit, Cayman chemical, VEGFR1, anti-pVEGFR2 (Santa Cruz # sc-101819) #10009135). that recognizes phospho-Tyr1175 VEGFR2, and anti- Cellular viability was determined by the MTT pS6 (Cell Signaling # 2211) that recognizes phospho- (methylthiazolyldiphenyl-tetrazolium bromide) Ser240/244 S6. The relevant Cy3-conjugated secondary viability assay as described previously [39] with minor antibodies were obtained from Jackson ImmunoResearch. modifications. Cells were incubated with different All images were captured by a fluorescence microscope concentrations of nintedanib under normoxia or hypoxia and the fluorescence intensities were determined by the www.impactjournals.com/oncotarget 4718 Oncotarget (1% O ) for 24 hrs followed by 96 hrs post-incubation 2. Vlajnic T, Andreozzi MC, Schneider S, Tornillo L, under normoxia and MTT determination. Karamitopoulou E, Lugli A, Ruiz C, Zlobec I, Terracciano L. VEGFA gene locus (6p12) amplification identifies a small but highly aggressive subgroup of colorectal cancer Real-time RT (reverse transcription)-PCR and patients. Mod Pathol. 2011; 24: 1404-1412. Western blot analysis. 3. Jürgensmeier JM, Schmoll HJ, Robertson JD, Brooks L, Taboada M, Morgan SR, Wilson D, Hoff PM. Prognostic Total RNA was extracted from CRC cells and predictive value of VEGF, sVEGFR-2 and CEA in using the TRIzol® RNA purification reagent. RNA mCRC studies comparing cediranib, bevacizumab and quantity and purity were determined by using a chemotherapy. Br J Cancer. 2013; 108: 1316-1323. NanoDrop ND-1000. Total RNA (1 μg) from each 4. Ferrara N, Hillan KJ, Gerber HP, Novotny W. Discovery sample was reverse transcribed and real-time RT-PCR and development of bevacizumab, an anti-VEGF antibody measurements were performed as described previously for treating cancer. Nat Rev Drug Discov. 2004; 3: 391-400. [40, 41] using a Mx3000P apparatus (Agilent) with the 5. Hilberg F, Roth GJ, Krssak M, Kautschitsch S, corresponding SYBR Green kit. PCR primers were Sommergruber W, Tontsch-Grunt U, Garin-Chesa P, designed with Primer 3 (Agilent) as follows: VEGF, Bader G, Zoephel A, Quant J, Heckel A, Rettig WJ. BIBF upper, 5’-CGAAGTGGTGAAGTTCATGGATG-3’, 1120: triple angiokinase inhibitor with sustained receptor lower, 5’-TTCTGTATCAGTC TTTCCTGGTGAG. blockade and good antitumor efficacy. Cancer Res. 2008; 36B4 (also known as RPLP0), upper, 68: 4774-4782. 5′-GATTGGCTACCCAACTGTTG-3′; lower, 6. Reck M, Kaiser R, Mellemgaard A, Douillard JY, Orlov S, 5′-CAGGGGCAGCAGCCAC AAA-3′. Gene expression Krzakowski MJ, Von Pawel J, Gottfried M, Bondarenko was normalized to 36B4. I, Liao M, Barrueco J, Gaschler-Markefski B, Novello S. Western blot analysis was carried out as described Nintedanib (BIBF 1120) plus docetaxel in NSCLC patients previously [42]. The primary antibodies were directed progressing after first-line chemotherapy: LUME Lung 1, a against HIF-1alpha (BD Bioscience # 610958), HIF- randomized, double-blind phase III trial. J Clin Oncol. 31, 2alpha (Novus Biological # NB-100122) or beta-actin 2013 (suppl; abstr LBA8011). (Santa Cruz # sc-47778), while the corresponding secondary antibodies were purchased from Jackson 7. Willett CG, Duda DG, di Tomaso E, Boucher Y, ImmunoResearch. Ancukiewicz M, Sahani DV, Lahdenranta J, Chung DC, Fischman AJ, Lauwers GY, Shellito P, Czito BG, Wong TZ, Statistical analysis. Paulson E, Poleski M, Vujaskovic Z, et al. Efficacy, safety, and biomarkers of neoadjuvant bevacizumab, radiation therapy,and fluorouracil in rectal cancer: a multidisciplinary The statistical analysis of experimental data was phase II study. J Clin Oncol. 2009; 27: 3020-3026. performed using a Student’s paired t-test, and results are 8. Van der Veldt AA, Lubberink M, Bahce I, Walraven M, presented as mean ± standard deviation (SD). de Boer MP, Greuter HN, Hendrikse NH, Eriksson J, Windhorst AD, Postmus PE, Verheul HM, Serné EH, ACKNOWLEDGMENTS Lammertsma AA, Smit EF. Rapid decrease in delivery of chemotherapy to tumors after anti-VEGF therapy: This work was financed in part by research funding implications for scheduling of anti-angiogenic drugs. to Annette K. Larsen from Boehringer Ingelheim. The Cancer Cell. 2012; 21: 82-91. sponsors had no role in the study design, data collection 9. Pastuskovas CV, Mundo EE, Williams SP, Nayak TK, Ho and analysis, interpretation of the results, the preparation J, Ulufatu S, Clark S, Ross S, Cheng E, Parsons-Reponte of the manuscript, or the decision to submit the manuscript K, Cain G, Van Hoy M, Majidy N, Bheddah S, dela for publication. We like to acknowledge the contribution Cruz Chuh J, Kozak KR, et al. Effects of anti-VEGF on of Olivier Bernadini, Tatiana Ledent and Delphine Muller pharmacokinetics, biodistribution, and tumor penetration from the animal facilities at Saint-Antoine Research of trastuzumab in a preclinical breast cancer model. Mol Center and Sylvie Dumont and Fatiha Merabtene from Cancer Ther. 2012; 11: 752-762. the pathology platform at Hôpital Saint-Antoine. Paul 10. Eichten A, Adler AP, Cooper B, Griffith J, Wei Y, Mésange is supported by a Mouna Nasrallah fellowship. Yancopoulos GD, Lin HC, Thurston G. Rapid decrease in tumor perfusion following VEGF blockade predicts long- REFERENCES term tumor growth inhibition in preclinical tumor models. Angiogenesis. 2013; 16: 429-441. 1. Chung AS, Lee J, Ferrara N. Targeting the tumour 11. Rapisarda A, Melillo G. Overcoming disappointing results vasculature: insights from physiological angiogenesis. Nat with antiangiogenic therapy by targeting hypoxia. Nat Rev Rev Cancer. 2010; 10: 505-514. Clin Oncol. 2012; 9: 378-390. www.impactjournals.com/oncotarget 4719 Oncotarget 12. Shweiki D, Neeman M, Itin A, Keshet E. Induction 23. Naik S, Dothager RS, Marasa J, Lewis CL, Piwnica-Worms of vascular endothelial growth factor expression by D. Vascular Endothelial Growth Factor Receptor-1 Is hypoxia and by glucose deficiency in multicell spheroids: Synthetic Lethal to Aberrant {beta}-Catenin Activation in implications for tumor angiogenesis. Proc Natl Acad Sci Colon Cancer. Clin Cancer Res. 2009; 15: 7529-7537. USA. 1995; 92: 768-772. 24. Mulkeen AL, Silva T, Yoo PS, Schmitz JC, Uchio E, Chu 13. Maxwell PH, Dachs GU, Gleadle JM, Nicholls LG, Harris E, Cha C. Short interfering RNA-mediated gene silencing AL, Stratford IJ, Hankinson O, Pugh CW, Ratcliffe PJ. of vascular endothelial growth factor: effects on cellular Hypoxia-inducible factor-1 modulates gene expression in proliferation in colon cancer cells. Arch Surg. 2006; 141: solid tumors and influences both angiogenesis and tumor 367-374. growth. Proc Natl Acad Sci USA. 1997; 94: 8104-8109. 25. Giatromanolaki A, Koukourakis MI, Sivridis E, 14. von Marschall Z, Cramer T, Höcker M, Finkenzeller G, Chlouverakis G, Vourvouhaki E, Turley H, Harris AL, Wiedenmann B, Rosewicz S. Dual mechanism of vascular Gatter KC. Activated VEGFR2/KDR pathway in tumour endothelial growth factor upregulation by hypoxia in human cells and tumour associated vessels of colorectal cancer. Eur hepatocellular carcinoma. Gut. 2001; 48: 87-96. J Clin Invest. 2007; 37: 878-886. 15. Das B, Yeger H, Tsuchida R, Torkin R, Gee MF, Thorner 26. Calvani M, Trisciuoglio D, Bergamaschi C, Shoemaker RH, PS, Shibuya M, Malkin D, Baruchel S. A hypoxia-driven Melillo G. Differential involvement of vascular endothelial vascular endothelial growth factor/Flt1 autocrine loop growth factor in the survival of hypoxic colon cancer cells. interacts with hypoxia-inducible factor-1alpha through Cancer Res. 2008; 68: 285-291. mitogen-activated protein kinase/extracellular signal- 27. Masuda K, Teshima-Kondo S, Mukaijo M, Yamagishi N, regulated kinase 1/2 pathway in neuroblastoma. Cancer Nishikawa Y, Nishida K, Kawai T, Rokutan K. A novel Res. 2005; 65: 7267-7275. tumor-promoting function residing in the 5’ non-coding 16. Harada H, Itasaka S, Kizaka-Kondoh S, Shibuya K, region of vascular endothelial growth factor mRNA. PLoS Morinibu A, Shinomiya K, Hiraoka M. The Akt/mTOR Med. 2008; 5: e94. pathway assures the synthesis of HIF-1alpha protein in a 28. Yin Y, Cao LY, Wu WQ, Li H, Jiang Y, Zhang HF. glucose- and reoxygenation-dependent manner in irradiated Blocking effects of siRNA on VEGF expression in human tumors. J Biol Chem. 2009; 284: 5332-5342. colorectal cancer cells. World J Gastroenterol. 2010; 16: 17. Nilsson MB, Zage PE, Zeng L, Xu L, Cascone T, Wu HK, 1086-1092. Saigal B, Zweidler-McKay PA, Heymach JV. Multiple 29. Samuel S, Fan F, Dang LH, Xia L, Gaur P, Ellis LM. receptor tyrosine kinases regulate HIF-1alpha and HIF- Intracrine vascular endothelial growth factor signaling in 2alpha in normoxia and hypoxia in neuroblastoma: survival and chemoresistance of human colorectal cancer implications for antiangiogenic mechanisms of multikinase cells. Oncogene. 2011; 30: 1205-1212. inhibitors. Oncogene 2010; 29: 2938-2949. 30. Laplante M, Sabatini DM. mTOR signaling at a glance. J 18. Larsen AK, Ouaret D, El Ouadrani K, Petitprez A. Cell Sci. 2009; 122(Pt 20): 3589-3594. Targeting EGFR and VEGF(R) pathway cross-talk in tumor 31. Corcoran RB, Rothenberg SM, Hata AN, Faber AC, Piris survival and angiogenesis. Pharmacol Ther. 2011; 131: 80- A, Nazarian RM, Brown RD, Godfrey JT, Winokur D, Walsh J, Mino-Kenudson M, Maheswaran S, Settleman J, 19. Larsen AK, de Gramont A, Poindessous V, Bouygues A, Wargo JA, Flaherty KT, Haber DA, Engelman JA. TORC1 Ayadi M, Mésange P. Functions and clinical implications Suppression Predicts Responsiveness to RAF and MEK of autocrine VEGF signaling in colorectal cancer. Current Inhibition in BRAF-Mutant Melanoma. Sci Transl Med. Colorectal Cancer Reports. 2013; 9: 270-277. 2013; 5: 196ra98. 20. Bates RC, Goldsmith JD, Bachelder RE, Brown C, Shibuya 32. Elkabets M, Vora S, Juric D, Morse N, Mino-Kenudson M, M, Oettgen P, Mercurio AM. Flt-1-dependent survival Muranen T, et al. Tao J, Campos AB, Rodon J, Ibrahim characterizes the epithelial-mesenchymal transition of YH, Serra V, Rodrik-Outmezguine V, Hazra S, Singh S, colonic organoids. Curr Biol. 2003; 13: 1721-1727. Kim P, Quadt C, et al. mTORC1 Inhibition Is Required 21. Fan F, Wey JS, McCarty MF Belcheva A, Liu W, Bauer for Sensitivity to PI3K p110α Inhibitors in PIK3CA-Mutant TW Somcio RJ, Wu Y, Hooper A, Hicklin DJ, Ellis LM. Breast Cancer. Sci Transl Med. 2013; 5: 196ra99. Expression and function of vascular endothelial growth 33. Pencreach E, Guérin E, Nicolet C, Lelong-Rebel I, factor receptor-1 on human colorectal cancer cells. Voegeli AC, Oudet P, Larsen AK, Gaub MP, Guenot D. Oncogene. 2005; 24: 2647-2653. Marked activity of irinotecan and rapamycin combination 22. Okita NT, Yamada Y, Takahari D, Hirashima Y, Matsubara toward colon cancer cells in vivo and in vitro is mediated J, Kato K, Hamaguchi T, Shirao K, Shimada Y, Taniguchi through cooperative modulation of the mammalian target H, Shimoda T. Vascular endothelial growth factor receptor of rapamycin/hypoxia-inducible factor-1alpha axis. Clin expression as a prognostic marker for survival in colorectal Cancer Res. 2009; 15:1297-1307. cancer. Jpn J Clin Oncol. 2009; 39: 595-600. 34. Koh MY, Powis G. Passing the baton: the HIF switch. www.impactjournals.com/oncotarget 4720 Oncotarget Trends Biochem Sci. 2012; 37: 364-372. 35. Poindessous V, Ouaret D, El Ouadrani K, Battistella A, Mégalophonos VF, Kamsu-Kom N, Petitprez A, Escargueil AE, Boudou P, Dumont S, Cervera P, Fléjou JF, André T, Tournigand C, Chibaudel B, de Gramont A, Larsen AK. EGFR- and VEGF(R)-targeted small molecules show synergistic activity in colorectal cancer models refractory to combinations of monoclonal antibodies. Clin Cancer Res. 2011; 17: 6522-6530. 36. Fan F, Samuel S, Gaur P, Lu J, Dallas NA, Xia L, Bose D, Ramachandran V, Ellis LM. Chronic exposure of colorectal cancer cells to bevacizumab promotes compensatory pathways that mediate tumour cell migration. Br J Cancer. 2011; 104: 1270-1277. 37. Yamagishi N, Teshima-Kondo S, Masuda K, Nishida K, Kuwano Y, Dang DT, Dang LH, Nikawa T, Rokutan K. Chronic inhibition of tumor cell-derived VEGF enhances the malignant phenotype of colorectal cancer cells. BMC Cancer. 2013; 13: 229. 38. Minakata K, Takahashi F, Nara T, Hashimoto M, Tajima K, Murakami A, Nurwidya F, Yae S, Koizumi F, Moriyama H, Seyama K, Nishio K, Takahashi K. Hypoxia induces gefitinib resistance in non-small-cell lung cancer with both mutant and wild-type epidermal growth factor receptors. Cancer Sci. 2012; 103: 1946-1954. 39. Soares DG, Escargueil AE, Poindessous V, Sarasin A, de Gramont A, Bonatto D, Henriques JA, Larsen AK. Replication and homologous recombination repair regulate DNA double-strand break formation by the antitumor alkylator ecteinascidin 743. Proc Natl Acad Sci USA. 2007; 104: 13062-13067. 40. Fritah A, Saucier C, De Wever O, Bracke M, Bieche I, Lidereau R, Fritah Gespach C, Drouot S, Redeuilh G, Sabbah M. Role of WISP-2/CCN5 in the maintenance of a differentiated and noninvasive phenotype in human breast cancer cells. Mol Cell Biol. 2008; 28: 1114-1123. 41. Petitprez A, Poindessous V, Ouaret D, Regairaz M, Bastian G, Guérin E, Escargueil AE, Larsen AK. Acquired irinotecan resistance is accompanied by stable modifications of cell cycle dynamics independent of MSI status. Int J Oncol. 2013; 42: 1644-1653. 42. Escargueil AE, Poindessous V, Soares DG, Sarasin A, Cook PR, Larsen AK. Influence of irofulven, a transcription-coupled repair-specific antitumor agent, on RNA polymerase activity, stability and dynamics in living mammalian cells. J Cell Sci. 2008; 121: 1275-1283. www.impactjournals.com/oncotarget 4721 Oncotarget

Journal

Oncotarget – Unpaywall

Published: Jan 20, 2014

There are no references for this article.