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

Learn More →

TFE3 contains two activation domains, one acidic and the other proline-rich, that synergistically activate transcription

TFE3 contains two activation domains, one acidic and the other proline-rich, that synergistically... Downloaded from https://academic.oup.com/nar/article-abstract/23/19/3865/2376375 by DeepDyve user on 06 August 2020 Nucleic Acids Research, 1995, Vol. 23, No. 19 3865-3871 TFE3 contains two activation domains, one acidic and the other proline-rich, that synergistically activate transcription Steven E. Artandi1*, Kevin Merrell3, Nicole Avitahl1 §, Kwok-Kin Wong3 and Kathryn Calame1-23* Departments of 1 Microbiology, 2Biochemistry and Molecular Biophysics and integrated Program in Cellular, Molecular and Biophysical Studies, Columbia University College of Physicians and Surgeons, New York, NY 10032, USA Received July 7, 1995; Revised and Accepted August 25, 1995 GenBank accession no. U36393 ABSTRACT TFE3 is a member of the bHLHZIP family of transcription proteins. It binds the nE3 elements in the immunoglobulin heavy chain (IgH) intronic enhancer, in Ig kappa enhancers and in some TFE3 is a basic-helix-loop-helix-zipper (bHLHZIP) IgH variable region promoters (2,3). TFE3 exists in vivo as two domain-containing protein that binds uE3 sites in regulatory elements in the immunoglobulin heavy chain alternately spliced isoforms with different activation potentials gene. The protein is a transcriptional activator that is (4). The alternately spliced exon encodes an N-terminal acidic expressed in vivo as two alternately spliced isoforms activation domain (AAD) (2). The activation potential of TFE3S, the isoform lacking this activation domain, is ~4-fold lower than with different activating properties: TFE3L contains an that of TFE3L, the isoform containing the N-terminal domain (4). N-terminal acidic activation domain; TFE3S lacks this activation domain and is a dominant negative inhibitor The TFE3S mRNA in vivo is expressed at low levels compared of TFE3L. We show that TFE3L and TFE3S contain a to TFE3L, ranging from 2 to 18% of total TFE3 mRNA in a second, C-terminal activation domain rich in proline variety of tissues and cell lines. residues. This pro-rich activation domain has activity in Cotransfection assays with TFE3L and TFE3S showed that a Gal4 fusion assay comparable to the N-terminal acidic TFE3S can act as a dominant negative. Two aspects of these data activation domain present in TFE3L. The TFE3 pro-rich were striking. First, the dominant negative effect of TFE3S on activation domain contains regions of strong homology TFE3L activity was achieved with substoichiometric amounts of with the related proteins microphthalmia and TFEB, TFE3S expression plasmid. The correlation between the plasmid suggesting that these regions are important for func- ratios at which TFE3S inhibited TFE3L activity and the ratio of tion. Using two different assays, we show that the Inl- these mRNA isoforms in vivo suggested that the dominant and C-terminal activation domains of TFE3 act syner- negative activity of TFE3S is biologically important. Second, gistically. This synergism explains in part the ability of even when TFE3S represented only 20% of the total amount of TFE3S to act as a dominant negative. Our domain input TFE3 expression plasmid, the activity of the mixture was analysis of TFE3 is incorporated into a general struc- equivalent to that of 100% TFE3S rather than an average between tural model for the TFE3 protein that predicts that the the activities of TFE3S and TFE3L. activation domains of TFE3 will be widely separated in The mechanism by which TFE3S exerts this dominant negative space. effect on TFE3L is unclear. It was originally proposed that TFE3S might poison the activity of a tetramer of TFE3 proteins (4). Tetramerization of TFE3 (5) and other bHLHZIP proteins INTRODUCTION including c-Myc (6), TFEB (7) and USF (8) has been demon- strated, however, these proteins appear to bind DNA as aimers in electromobility shift assays. Tetramerization may be important in Gene transcription is a key regulatory point in diverse develop- mediating interactions between remotely and proximally bound mental processes such as cell growth, differentiation, apoptosis and bHLHZIP proteins (5), however either dimers or tetramers of transformation. The transcription factors that control the rates of TFE3 could mediate simple activation from a proximal element transcription initiation are often tightly regulated and the activities such as the one used in the mixing experiments with TFE3S and of these proteins can be modified at a variety of levels. Eukaryotic TFE3L (4). We previously observed that TFE3 exhibits synergis- transcription factors are generally bipartite; DNA-binding domains tic activation of transcription as the number of ^E3 binding sites contact the DNA binding site and activation domains transduce are increased in a promoter (5). We wondered if synergistic regulatory information via protein-protein interactions with com- effects present in the TFE3L homodimer, but absent in the TFE3S ponents of the basal transcription apparatus (1). * To whom correspondence should be addressed Present addresses: Departments of+Medicine and ^Dermatology, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA Downloaded from https://academic.oup.com/nar/article-abstract/23/19/3865/2376375 by DeepDyve user on 06 August 2020 3866 Nucleic Acids Research, 1995, Vol. 23, No. 19 homodimer and the TFE3L-TFE3S heterodimer could contribute 0.1 mM EDTA and 400 mM NaCl. Fibroblasts from one 10 cm to the dominant negative activity of TFE3S. plate (4 x 106 cells) resuspended in 100 u.1 of lysis buffer and To understand more completely the mechanism of dominant allowed to remain on ice for 30 min then frozen in liquid nitrogen. negative inhibition by TFE3S, we characterized the activation Nuclear extract (10 |ll; 4 x 105 cell equivalents) was loaded in sequences of TFE3 and developed an assay to test how these each well of a 10% SDS-polyacrylamide gel. The nuclear proteins domains might work together. In this report, we show that a were transfered to nitrocellulose and Western blotted by standard C-terminal proline rich domain of TFE3 is an activation domain. techniques using a polyclonal anti-TFE3 antiserum. Furthermore, the proline-rich domain activates synergistically with the acidic domain of TFE3, both when these domains are in their normal context in the TFE3 molecule and when they are tethered to RESULTS different DNA binding domains bound at adjacent DNA elements. These data are incorporated into a model for TFE3 activation. TFE3L contains a second transcriptional activation domain C-terminal to the bHLHZip region MATERIALS AND METHODS The exon that is alternately spliced between isoforms TFE3L and Expression plasmids TFE3S encodes an N-terminal activation domain that is rich in acidic amino acids. TFE3S, which lacks this domain, is a weaker GAL4-TFE3PRO was constructed by cloning a Stul fragment transactivator but it does retain some transcriptional activating from the TFE3 cDNA into the £c/136II site of pSV-GAL( 1-147) ability (4). The original GAL4 fusion analyses of TFE3L showed (9), resulting in fusion of TFE3 residues 345-446 [numbering that the N-terminal acidic domain is a strong activation domain scheme as in (3)] to GAL4( 1-147). GAL4-TFE3AAD was (2). However, in that study the remainder of the protein previously described as GAL4X3-A2 and fuses 126 N-terminal containing the bHLHZIP and C-terminal domains stimulated residues from TFE3 to GAL4(1-147) (2). GAL4-VP16 has been transcription weakly when fused to the GAL4 DNA binding previously described (9). domain (GAM DBD). These data led us to speculate that TFE3L pSV2-TFE3L and pSV2-TFE3S were described previously (4). might contain a second activation domain. TFE3L-ACterm and TFE3S-ACterm represent C-terminal trun- To test this hypothesis, we constructed a fusion protein in which cations of TFE3L and TFE3S, respectively, terminating at Val-314, the GAL4 DBD was fused to the C-terminus of TFE3. We tested immediately after the leucine zipper [numbering as in (3)]. They this fusion construct and others in cotransfection experiments with were constructed by PCR using pBS-TFE3L and pBS-TFE3S as a GAL4 site-dependent reporter plasmid driving the luciferase templates. PCR fragments were digested with HindYH and BarnHL gene. and cloned into the pSV2A expression plasmid (10). Figure 1A shows that the GAL4 DBD alone, GAL4( 1-147), did not stimulate transcription. GAL4-TFE3AAD, containing an Reporter plasmids N-terminal fragment of TFE3 that includes the AAD fused to the GAL4 DBD (2), stimulated transcription 249-fold. The fusion pGAL4-TATALuciferase (gift of J. Kaplan) was generated by protein containing 102 amino acids from the C-terminus of TFE3 isolating a GAL4 site pentamer from G5E4T (gift of M. Carey) fused to the GAL4 DBD activated transcription to a similar by digestion with Xbal and HindSM. This fragment was cloned by extent, 284-fold. For comparison, activation is shown for blunt ended ligation into the BglU site of pTATA-Luciferase (5), GAL4-VP16, a fusion of the GAL4 DBD and the potent VP16 a reporter plasmid containing a minimal TATA box upstream of activation domain (9). These data identify a second, C-terminal the firefly luciferase gene. transcriptional activation domain in TFE3L, separated in primary pGAL4/uJE3-TATALuciferase is a reporter plasmid containing a amino acid sequence from the N-terminal activation domain by pentamer of GAL4 sites and a tetramer of |XE3 sites upstream of the large bHLHZIP domain. The N-terminal activation domain the minimal promoter of pTATA-Luciferase. The GAL4 site and the C-terminal activation domain have similar transcriptional pentamer was cloned into the HindfiB. site of a pKS Bluescript activating abilities in the GAL4 fusion assay. polylinker that contained a |iE3 tetramer in the Smal site. A cassette Analysis of the C-terminal 134 amino acids of TFE3 shows the containing both multimers was excised by Xbal and fiawHI region is rich in Pro (13%), Ser (16%), Leu (13%), Gly (9%) and digestion and cloned into the BglQ site of pTATA-Luciferase by blunt ended ligation. acidic residues (15%, Asp plus Glu). This profile is similar to that of other activation domains that have been termed Pro-rich. The Pro-rich domain was originally described for CTFI; the C-ter- Transfection minal 99 amino acids of CTFI contain 19% Pro, 13% Ser, 10% Leu, and 10% Gly, but only 5% acidic residues. Oct-2, a B-cell NIH 3T3 cells were transfected by the calcium phosphate enriched member of the POU domain family, contains a technique essentially as described (5). Reporter plasmid (1 (ig) and C-terminal domain (residues 314-460) that is 13% Pro, 14% Ser, 5—10 |Xg expression plasmid were used per plate of cells. Cells were 8% Leu, 10% Gly, and 4% acidic residues. TFEB, a bHLHZIP harvested and assayed for luciferase activity as previously protein with homology to TFE3, has a 150 amino acid C-terminal described (5). domain that is 14% Pro, 13% Ser, 13% Leu, 10% Gly, and 15% acidic residues. AP2 has a 70 amino acid activation domain Western blot (residues 51-120) that is Pro and Gin rich; it is comprised of 21% Pro, 9% Ser, 9% Leu, 7% Gly, and 16% Gin. Hox4.2 contains an Whole cell extracts from transfected fibroblasts were prepared. 86 amino acid domain (residues 24-109) containing 22% Pro, 6% Cells were pelleted by centrifugation at 600 g, washed once in Ser, 6% Leu, 23% Gly, and 11% Ala. PBS, and lysed in 50 mM Tris pH 8.0,0.5% NP-40,10% glycerol, Downloaded from https://academic.oup.com/nar/article-abstract/23/19/3865/2376375 by DeepDyve user on 06 August 2020 Nucleic Acids Research, 1995, Vol. 23, No. 19 3867 amino acids (Fig. IB). Within the homologous 38 amino acids, 50% of the residues are identical among all three proteins and AAD 8HLHZIP PRO TFE3L N I V////A 45% are identical between two of the three. The conserved region is contained within the 102 amino acid C-terminal portion of TFE3 shown to be sufficient for activation (Fig. 1 A). Interesting- ly, although the proline content in the highly conserved region is GU.4 low (3/38), all three prolines are perfectly conserved. CDNA J±L The two TFE3 activation domains activate NO ACTIVATOR 1.0 + 0.2 transcription synergistically We showed previously that TFE3L is capable of synergistic GA 1.4(1-147) 1.0 ± 0.8 activation of transcription; as the number of cognate binding sites upstream of a TATA element is increased from four to eight, AAD transcription driven by TFE3L is increased by substantially more 249 ± 15 than a factor of two (5). Although proteins with a single activation domain have been shown to exhibit this phenomenon of synergy, PRO 284 ± 82 we wondered if part of the activation properties of TFE3L were GAL4-TFE3PRO L dependent upon synergistic stimulation of transcription by the separate N-terminal and C-terminal activation domains. ^ ^ VP16 1605 + 180 GAL4-VP16 We first addressed this question in the context of the nati veTFE 3 protein. Transcriptional activation was compared among TFE3L (containing both activation domains), TFE3S (containing only the 464 \ C-terminal activation domain) and TFE3L-ACterm (a truncated TFE3 GLPVPPNPGL LSLTTSSVSD SLKPEQLDIE EEGBPSTTFH VSGGPAQNAP TFEB GLPTTSPSG. ..VNMAELAQ QWKQELPSE DGPGEALMLG PEVPEPEQMP protein containing only the N-terminal activation domain). Plasmids Mi GLSLIPSTG. . .LCSPDLVN RIIKQEPVLE NCSQE encoding these proteins were cotransfected with a reporter depend- PQQPPAP PSDALLD LHFPSD.HLG DLGDPFHLG ent upon eight tandem (iE3 sites. Figure 2A shows that TFE3L TFE3 ALPPQAPLPS AAQPQSPFHH LDFSHGLSFG GGGDEGPTGY PDTLGTEHGS TFEB stimulated the promoter 105-fold. This high level of stimulation is LVQHQADLTC TT T LDLT DGTITF TNNLGTMPES Mi dependent upon synergistic activation by multiple DNA-bound DGMVGGLSG G ALSPI tAASDPLLSSVSPA TFE3 ...LEDILME E TFE3 molecules. TFE3S activated 19-fold and TFE3L-ACterm SLLPI A.S DPLFSTMSPE KKt LDLMLLD D TFEB P....FPNLS activated 9-fold. If activation by TFE3L on the multiple-site reporter RLSPV GVT DPLLSSVSPG Mi SPAYSIPRKM GS1>LEDILMD D were simply dependent upon the sum of activation by the AAD and TFE3 VSNASSRRSS FSIEEES the Pro-AD, one would predict that the activities of TFE3S and TFEB ASKASSRRSS FSMEE TFE3L-ACterm would add up to the activity of TFE3L. Since this Mi ASKTSSRRSS MSAEETEHA is not the case, the data suggest that it is the combination of acidic and Pro-rich activation domains that enables TFE3L to activate at Figure 1. (A) Analysis of activation domains of TFE3. Different protein such high levels and that the AAD and Pro activation domain regions were fused to the GAL4 DBD and tested for activation by stimulate transcription synergistically with one another. cotransfection with a GALA site-dependent reporter plasmid driving expression of the luciferase gene. GAL4-TFE3AAD was previously described as Figure 2A also shows that TFE3S-ACterm cannot activate the GAL4X3-A2 and fuses 126 N-terminal residues from TFE3 to GAL4(1-147) JIE 3 site reporter plasmid. The results with this C-terminal (2). TFE3PRO comprised the C-terminal 102 residues from TFE3 fused to the GAM DBD and GAL4-VP16 contains the activation sequences of the VP16 truncation of TFE3S confirm that the residual activity of TFE3S protein. Luciferase values from at least three independent transfections were is attributable to the Pro-rich domain. Furthermore, without the normalized to that of the reporter alone and are shown at the right for each exon encoding the 35 amino acid AAD, the remaining N-terminal activator plasmid. (B) Sequence homology among murine TFE3, TFEB and mi. and bHLHZIP domains are devoid of transactivating potential. The Pileup algorithm (GCG software) was used to compare the C-termini of these three proteins. Three regions of high homology are boxed and proline To rule out the possibility that differential protein expression residues are shown in bold. The horizontal arrow indicates the beginning of the confounds the interpretation of our results, immunoblots were 102 amino acid region of TFE3 shown to be sufficient for activation when fused performed on whole cell extracts of NIH3T3 fibroblasts frans- to the GAL4 DBD. fected with equai amounts of expression piasmid. i rc3L-ACterm was expressed at slightly lower levels than TFE3L. This small difference in expression may be accounted for by a loss of epitopes Despite similar amino acid contents, sequence homology has in the truncated protein. The polyclonal antiserum was generated not previously been demonstrated among proline-rich domains. using TFE3S as an antigen and therefore is likely to have However, we compared the sequence of the proline-rich C-termi- significant antibody activity directed against the proline-rich nus of TFE3 with the proline-rich C-termini of TFEB and C-terminus. Expression of TFE3L and TFE3L-ACterm are micophthalmia (mi) (11), two proteins in the |J.E3 family of therefore comparable. bHLHZIP domain proteins. We have cloned a murine TFEB Surprisingly, TFE3S and TFE3S-ACterm were expressed at cDNA which extents further 3' than that originally reported for significantly higher levels than TFE3L. Densitometric scanning human TFEB (12). This cDNA was sequenced to provide the of the autoradiogram showed that the TFE3S based isoforms were amino acid data shown in Figure IB. Substantial homology was expressed at -2.5-fold higher levels than their TFE3L based detected in the C-termini of TFE3, mi and TFEB. This homology counterparts. These data show that the various forms ofTFE3 extends over 52 amino acids in the extreme C-terminus of TFE3 protein used in Figure 2 A are expressed at least as well as TFE3L. and is distributed among three homology boxes containing 38 Downloaded from https://academic.oup.com/nar/article-abstract/23/19/3865/2376375 by DeepDyve user on 06 August 2020 3868 Nucleic Acids Research, 1995, Vol. 23, No. 19 |iE3 TATA p ^ LUCIFERASE FOLD ACTIVATION 0 20 40 60 80 100 120 NO ACT1VATCR TFE3L TFE3S TFE3L-ACterm TFE3S-ACterm CD u u t/) < ] CO PO ro UJ UJ UJ u_ u_ I— 1 - Endogenous TFE3 •" 71 Figure 2. (A) The activation domains of TFE3 act synergistically. Cotransfections were performed with plasmids encoding the indicated TFE3 proteins and a reporter plasmid containing eight tandem nE3 binding sites. TFE3L contains both activation domains and TFE3S is the isoform that lacks the N-terminal activation domain. TFE3L-ACterm and TFE3S-ACterm were engineered to include a stop codon after the leucine zipper of TFE3L and TFE3S, respectively. TFE3L-ACterm contains only the N-terminal activation domain, and TFE3S-ACterm contains neither activation domain. (B) Western blot of extracts from transfected NIH3T3 cells. Cells were transfected with 20 p.g of the expression plasmid indicated. Western blots using a polyclonal anti-TFE3S anu'serum were performed on whole cell extracts. Lane 1, untransfected cells; lane 2, TFE3S-ACterm; lane 3, TFE3L-ACterm; lane 4, TFE3L; lane 5, TFE3S. The positions of endogenous TFE3 related proteins and the migration of molecular weight standards are shown. Therefore, the lower transactivation potential of the shorter reproduced when the activation domains were joined to separate proteins is an innate property of these proteins and is not due to DNA-binding domains and bound to the same DNA template. decreased expression. This strengthens our conclusion that the N- Because this approach changes the spacing and, possibly, orienta- and C-terminal activation domains of TFE3L act synergistically. tion of the activation domains relative to their positions in the native protein, it might reveal different functional consequences of domain interactions. A reporter plasmid containing five The activation domains of TFE3 activate synergistically multimerized GAL4 binding sites and four multimerized (iE3 when tethered to different DNA-binding domains binding sites upstream of a TATA box and the luciferase gene was To examine further the capacity of TFE3's activation domains to constructed. Cotransfections were performed with TFE3 proteins act synergistically, we asked if synergistic activation could be containing either the AAD or the Pro-rich domain and with hybrid ran cted term 3L- :erm Downloaded from https://academic.oup.com/nar/article-abstract/23/19/3865/2376375 by DeepDyve user on 06 August 2020 Nucleic Acids Research, 1995, Vol. 23, No. 19 3869 example, proline residues might provide unstructured 'punctu- ation' between small structured regions which contact the basal GAL4 UE3 TATA ^ . LUCIFERASE 35 - transcription machinery. Alternatively, conserved regions and not 1 I I I I I—I I I I I—l~l—M the prolines may be functionally important. Recent dissection of both glutamine and acidic domains has shown that hydrophobic residues, not the prevalent glutamine and acidic residues, are required for interaction of these domains with their protein targets (13,14). The prevalent amino acids upon which classification of activation domains has been based may serve a more general structural role or an as yet unidentified function. Importance of synergistic activation by TFE3 activation domains We have shown that multiple DNA-bound molecules of TFE3 synergistically activate transcription and that this synergism is < E < dependent upon the presence of both activation domains. This < Q- a> < CO was demonstrated within the context of the native TFE3 protein UJ S 3 LL and when the AAD and the Pro-rich activation domain were fused I- h- £ CT _J to different DNA binding domains. In what circumstances is Zl U. 3 3 S 3 synergy between TFE3's activation domains likely to be import- <5 S ant biologically? TFE3 has been shown to mediate communica- tion between promoter and enhancer elements. The HLHZIP Figure 3 . The activation domains of TFE3 act synergistically when tethered to domain controls interaction between widely separated elements different DNA binding domains on the same DNA template. TFE3 proteins and the activation domains determine synergistic activation (5). containing a single activation domain were transfected either alone or in The immunoglobulin heavy chain gene contains a (iE3 site in its combination with GAL-4-fusion constructs expressing a single TFE3 activa- tion domain. Activity was measured from a reporter plasmid with five GAL4 V H promoters and in the intronic enhancer. TFE3 may facilitate sites and four p£ 3 sites. association of these elements. Once these elements are in close proximity, synergistic activation by TFE3 molecules bound in the promoter and in the enhancer is likely to contribute to activation of the IgH gene. proteins comprised of the GAL4 DNA-binding domain fused to either the TFE3 AAD or the Pro-rich domain. GAL4-TFE3AAD Demonstration of synergistic activation by TFE3L's activation domains also provides insight into the mechanism of inhibition of and GAL4-TFE3PRO each activated this reporter plasmid ~9-fold (Fig. 3). Note that the fold activation is lower than in TFE3L activity by the alternately spliced isoform TFE3S. TFE3S Figure 1 because the basal activity of the GAL4/|xE3 reporter is mRNA is expressed at lower concentrations than TFE3L mRNA, higher than the GAL4 reporter due to activation by endogenous ranging from 2 to 18 % of TFE3L in all tissues and cell lines flE3 binding proteins. TFE3S and TFE3L-ACterm yield similar tested. The activity of TFE3L is approximately three to four times greater than TFE3S. When cotransfected at substoichiometric levels of activation of this reporter plasmid. Cotransfection of GAL4-TFE3Pro and TFE3L-ACterm results in 28-fold activa- amounts similar to those that exist in vivo, TFE3S inhibits the tion, -2-fold higher than the sums of activation by either protein activity of TFE3L (4). In fact, in experiments in which TFE3S alone. Reversing the positions of the activation domains by comprised only 20% of the input mixture of TFE3 isoforms, transfecting GAL4-TFE3AAD and TFE3S did not result in activity was indistinguishable from that of TFE3S alone. synergistic activation. Thus, synergistic activation by AAD and How is inhibition at substoichiometric ratios possible? We Pro-rich activation domain can be recapitulated by removing suggest that two mechanisms play a role. First, the presence of these domains from their normal protein context and tethering one short monomer in a mixed dimer of TFE3L and TFE3S (L-S) them to different DNA-binding domains; however, this syner- appears to destroy synergy between activation domains. If there gism is not reproduced in every geometric arrangement. were no synergy between the activation domains, a mixed dimer (L-S) would have an activation potential exactly intermediate to DISCUSSION that of TFE3L and TFE3S dimers, assuming equal dissociation constants for dimer formation and for DNA binding. This is not A new activation domain for TFE3 the case because activity is low with subtoichiometric amounts of TFE3S—a situation where primarily L-S and L-L dimers are We have shown that the C-terminal 102 residues of TFE3 are expected to be present (4). In view of the synergy observed in this sufficient to confer strong activation on a hybrid protein when work, we suggest that lack of inter-monomer synergy accounts fused to the DBD of GAL4. Although this region is proline-rich for the low activity of L-S dimers. Thus, the model is that activity and has a similar amino acid content to other proline-rich domains, of L-L dimers is dependent upon the synergistic combination of sequence comparisons among TFE3 family members revealed AAD and Pro-rich domains and that removing one AAD from the highly conserved sub-domains, containing only three prolines, dimer reduces the activity of the mixed dimer, rendering it much located C-terminal to a highly proline-rich portion of the region. It closer to the activity of the S-S dimer. Second, our data suggest will be interesting to determine if these conserved subdomains are that TFE3S protein is more stable that TFE3L protein which either necessary or sufficient for transcriptional activation. It may would lead to a higher molar ratio of TFE3S/TFE3L protein be that both prolines and the conserved regions are necessary. For Downloaded from https://academic.oup.com/nar/article-abstract/23/19/3865/2376375 by DeepDyve user on 06 August 2020 3870 Nucleic Acids Research, 1995, Vol. 23, No. 19 relative to the RNA ratio. Nevertheless, the differences in protein expression are not large enough to produce a molar excess of TFE3S when TFE3L and TFE3S and transfected in a 4:1 plasmid ratio. Therefore, disruption of synergism among the activation domains of TFE3 appears to be an important component of the dominant negative action of TFE3S. The generation of alternate forms of transcription factors from the same gene which is seen in other proteins as well, including FosB, LAP and CREM, is an economical means of regulating the activity of key proteins (15,16). In the absence of synergistic effects such as the ones demonstrated here, dominant negative forms of these proteins would have to be generated at significantly higher levels to inhibit the activities of the more strongly activating forms. It is likely that nature exploits the effects of activation domain synergism in such a way that dramatic changes in gene expression can be achieved by more subtle alterations in the concentrations of isoforms generated from the same transcription factor gene. A model for TFE3 activation Stimulation of transcription by TFE3 is mediated by two separate activation domains, one rich in acidic residues and located in the amino terminus, and one rich in Pro residues and located in the carboxy terminus. Analysis of many other transcription factors has shown they are often comprised of multiple activation domains. Typically, these domains are in contiguous segments of the protein (17—19) although sometimes activation domains are separated in the primary amino acid sequence (20). Frequently, activation domains are comprised of multiple subdomains, which, when deleted, reduce the activity of the protein in a gradual fashion, without evidence of synergistic activation. This has been shown for GAL4 (20), C/EBP (19) and USF (18), among others. The two activation domains of TFE3L, however, are not subdomains of a larger activation domain and are likely to be separated in three Figure 4. Model for general structure of TFE3. The TFE3L protein is proposed dimensional space by the relatively large alpha helices which to form a dumbbell shaped protein with N-terminal (bottom) and C-ternunal confer DNA binding and subunit dimerization. We propose a (top) activation domains widely separated in space. Activation domains are model for the general structure of the TFE3 protein that takes into depicted as spheres. The bHLHZIP domain and DNA were generated as a account the data presented above, the location of the domains in nbbon structure from the max-DNA coordinates (21) using Molscript. the primary amino acid sequence, and specific structural details of the bHLHZIP-DNA complex of Max determined by X-ray crystallography (21). domain turns back toward the other, ending up in close proximity. We propose that the TFE3 protein dimer, when bound to DNA, assumes a 'dumbbell shape'; the shaft of the dumbbell represents The extended nature of the bHLHZIP structure makes this less the bHLHZIP domain and the ends of the dumbbell represent the likely. AADs and the Pro-rich activation domains (Fig. 4). The bHLHZIP Targets for the activation domains of TFE3 have not yet been domain that separates the activation domains is an unusual motif identified, but candidates exist. TBP (22), TFIIB (23), TAFTI40 in a non-structural protein. It is an extended domain comprised of (24) and TAFII60 (14) have been shown to interact with other two parallel ot-helices connected by a short loop. One helix AADs. The Pro-rich domain of CTFI has been shown to require contains the basic region that binds DNA and the helix 1 portion TFIIB (25) as well as TAFs (26) for transactivation. However, it of the HLH domain. The other helix contains the helix2 portion is also possible that other targets will actually be demonstrated to of the HLH domain as well as the leucine zipper. The axis of the mediate TFE3 activation. The minimal sequence homology among HLHZIP domain is perpendicular to the DNA axis. The basic members of a given family of activation domains make a single region a-helix passes through the major groove of the DNA protein target unlikely, hi fact, TAFQ110 interacts with the Gin-rich binding site and terminates on the other side. Thus, sequences sequences of Spl and CREB, but not those of Antennapedia or N-terminal to the basic region must at least begin on the other side bicoid (13,27). of a plane that is perpendicular to the HLHZIP domain and that The proposed dumbbell shape of TFE3 and our data that show contains the DNA molecule. The dumbbell model predicts that synergistic interaction of both activation domains suggest a three the AAD and the Pro-rich domains will be widely separated in dimensional view of how TFE3 might direct formation of a space and will lie on opposite sides of the DNA molecule. The stereospecific preinitiation complex. Transactivation of a pro- alternative, which we cannot rule out, is that each activation moter by TFE3 would therefore involve molecular connections Downloaded from https://academic.oup.com/nar/article-abstract/23/19/3865/2376375 by DeepDyve user on 06 August 2020 Nucleic Acids Research, 1995, Vol. 23, No. 19 3871 between TBP and TFE3 both above and below the plane of the Aggarwaal for critically reading the manuscript. S.A., K.W. and DNA helix. Thus, TAFII40 or TFIIB might bridge between the K.M. were supported by an NIGMS M.S.T.P. training grant and TFIID complex and the AAD of TFE3 on one side of this plane, this work was supported by USPH grant GM29361 to K.C. while TFIIB or another TAF might bridge between the Pro-rich domain and the TFIID complex on the other side of this plane. REFERENCES Of the many TFE3, USF and Myc/max related proteins that comprise the bHLHZIP family of proteins, only TFE3, TFEB and 1 Tjian, R. and Maniatis, T. (1994) Cell, 77, 5-8. mi contain a proline-rich region C-terminal to the ZIP domain. 2 Beckmann, H., Su, L.-K. and Kadesch, T. (1990) Genes Dev., 4, 167-179. 3 Roman, C , Matera, G., Cooper, C , Artandi, S., Blain, S., Ward, D. and USF, c-Myc and max terminate shortly after the ZIP domain. In Calame, K. (1992) Mol. Cell. Biol., 12, 817-827. addition to having different activation sequences, the overall 4 Roman, C , Cohn, L. and Calame, K. (1991) Science, 254, 94-97. geometric shapes and structures of these proteins are likely to be 5 Artandi, S. E., Cooper, C , Shrivastava, A. and Calame, K. (1994) Mol. different. The criteria that dictate whether a gene is regulated by Cell. Biology, 14, 704-7716. TFE3, USF or c-Myc/max are not known. It is tempting to 6 Dang, C. V., McGuire, M , Buckmire, M. and Lee, W. M. F. (1989) Nature, 337, 664-666. speculate that the stereospecific complex formed by promoter- 7 Fisher, D. E., Carr, C. S., Parent, L. A. and Sharp, P. A. (1991) Genes Dev., bound factors and components of the basal transcription apparatus 5, 2342-2352. will encourage or disfavor incorporation of a sequence specific 8 Ferre-D'Amare, A., Pognonec, P., Roeder, R. G. and Burley, S. K. (1994) factor based upon the sequence of its activation domains and their EMBOJ., 13, 180-189. geometric organization. 9 Sadowski, I. and Ptashne, M. (1989) Nucleic Acids Res., 17, 7539. 10 deWet, J. R., Wood, K. V., DeLuca, M., Helsinki, D. R. and Subramani, S. (1987) Mol. Cell. Biol., 7, 725. Similarity between TFE3 and Oct-2 11 Hodgkinson, C. A., Moore, K. J., Nakayama, A., Steingrimsson, E., Copeland, N. G., Jenkins, N. A. and Arnheiter, H. (1993) Cell, 74, To our knowledge, Oct-2 is the only protein previously shown to 395-404. contain multiple activation domains that activate synergistically 12 Carr, C. and Sharp, P. (1990) Mol. Cell. Biol., 10, 4384-88. 13 Gill, G., Pascal, E., Tseng, Z. H. and Tjian, R. (1994) Proc. Natl. Acad. (28). The overall organization of the Oct-2 protein is similar to Sci. USA, 91, 192-196. that of TFE3. Each protein is comprised of a DNA binding 14 Thut, C. J., Chen, J.-L., Klemm, R. and Tjian, R. (1995) Science, 267, domain sandwiched between an N-terminal activation domain (Gin-rich in Oct-2) and a C-terminal Pro-rich activation domain. 15 Foulkes, N. and Sassone-Corsi, P. (1992) Cell, 68, 411-414. The recent X-ray crystal structure of the Oct-1 POU domain (29) 16 Descombes, P. and Schibler, U. (1991) Cell, 67, 569-579. 17 Courey, A. J. and Tjian, R. (1988) Cell, 55, 887-898. shows that the POU specific domain contacts the 5' half of the 18 Kirschbaum, B. J., Pognonec, P. and Roeder, R. G. (1992) Mol. Celt. Biol., binding site. This domain is connected by a flexible linker to the 12(11), 5094-5101. POU homeodomain which contacts the 3 ' half of the OCT site on 19 Friedman, A. D. and McKnight, S. L. (1990) Genes Dev., 4, 1416-1426. the opposite face of the DNA double helix. Therefore, bHLHZIP 20 Ma, J. and Ptashne, M. (1987) Cell, 48, 847-853. domains and POU domains achieve placement of their N- and 21 Ferre-D'Amare, A. R., Prendergast, G. C , Ziff, E. B. and Burley, S. K. (.1993) Nature, 363, 38^5 . C-termini on opposite sides of the DNA molecule by using 22 Stringer, K. F , Infles, C. J. and Greenblatt, J. (1990) Nature, 345, completely different binding modules. If the N- and C-terminal 783-786. activation domains of Oct-2 and TFE3 are indeed located on 23 Lin, Y.-S., Ha, I., Maldonado, E., Reinberg, D. and Green, M. R. (1991) opposite sides of the DNA molecule, it suggests formation of a Nature, 353, 569-571. preinitiation complex circumferentially around the double helix 24 Goodrich, J. A., Hoey, T., Thut, C. J., Admon, A. and Tjian, R. (1993) Cell, 75,519-530. in elements such as the VH promoters that contain adjacent 25 Kim, T. K. and Roeder, R. G. (1994) Proc. Natl. Acad. Sci. USA, 91, binding sites for both factors. 4170-4174. 26 Tanese, N., Pugh, B. F. and Tjian, R. (1991) Genes Dev., 5, 2212-2224. 27 Hoey, T , Wemzierl, R. O. J., Gill, G., Chen, J.-L., Dynlacht, B. D. and ACKNOWLEDGEMENTS Tjian, R. (1993) Cell, 72, 247-260. 28 Tanaka, M., Clouston, W. M. and Herr, W. (1994) Mol. Cell. Biol., 14, We thank members of the Calame lab for helpful discussions. We 6046-6055. are grateful to Dr V. Racaniello for assistance with the Molscript 29 Klemm, J. D., Rould, M. A., Aurora, R., Herr, W. and Pabo, C. O. (1994) software, to Dr M. Carey for providing constructs and to Dr A. Cell, 77, 21-32. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Nucleic Acids Research Oxford University Press

TFE3 contains two activation domains, one acidic and the other proline-rich, that synergistically activate transcription

Download PDF
7 pages

Loading next page...
 
/lp/oxford-university-press/tfe3-contains-two-activation-domains-one-acidic-and-the-other-proline-N0avOWQpZD

References (0)

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

Publisher
Oxford University Press
Copyright
© 1995 Oxford University Press
ISSN
0305-1048
eISSN
1362-4962
DOI
10.1093/nar/23.19.3865
Publisher site
See Article on Publisher Site

Abstract

Downloaded from https://academic.oup.com/nar/article-abstract/23/19/3865/2376375 by DeepDyve user on 06 August 2020 Nucleic Acids Research, 1995, Vol. 23, No. 19 3865-3871 TFE3 contains two activation domains, one acidic and the other proline-rich, that synergistically activate transcription Steven E. Artandi1*, Kevin Merrell3, Nicole Avitahl1 §, Kwok-Kin Wong3 and Kathryn Calame1-23* Departments of 1 Microbiology, 2Biochemistry and Molecular Biophysics and integrated Program in Cellular, Molecular and Biophysical Studies, Columbia University College of Physicians and Surgeons, New York, NY 10032, USA Received July 7, 1995; Revised and Accepted August 25, 1995 GenBank accession no. U36393 ABSTRACT TFE3 is a member of the bHLHZIP family of transcription proteins. It binds the nE3 elements in the immunoglobulin heavy chain (IgH) intronic enhancer, in Ig kappa enhancers and in some TFE3 is a basic-helix-loop-helix-zipper (bHLHZIP) IgH variable region promoters (2,3). TFE3 exists in vivo as two domain-containing protein that binds uE3 sites in regulatory elements in the immunoglobulin heavy chain alternately spliced isoforms with different activation potentials gene. The protein is a transcriptional activator that is (4). The alternately spliced exon encodes an N-terminal acidic expressed in vivo as two alternately spliced isoforms activation domain (AAD) (2). The activation potential of TFE3S, the isoform lacking this activation domain, is ~4-fold lower than with different activating properties: TFE3L contains an that of TFE3L, the isoform containing the N-terminal domain (4). N-terminal acidic activation domain; TFE3S lacks this activation domain and is a dominant negative inhibitor The TFE3S mRNA in vivo is expressed at low levels compared of TFE3L. We show that TFE3L and TFE3S contain a to TFE3L, ranging from 2 to 18% of total TFE3 mRNA in a second, C-terminal activation domain rich in proline variety of tissues and cell lines. residues. This pro-rich activation domain has activity in Cotransfection assays with TFE3L and TFE3S showed that a Gal4 fusion assay comparable to the N-terminal acidic TFE3S can act as a dominant negative. Two aspects of these data activation domain present in TFE3L. The TFE3 pro-rich were striking. First, the dominant negative effect of TFE3S on activation domain contains regions of strong homology TFE3L activity was achieved with substoichiometric amounts of with the related proteins microphthalmia and TFEB, TFE3S expression plasmid. The correlation between the plasmid suggesting that these regions are important for func- ratios at which TFE3S inhibited TFE3L activity and the ratio of tion. Using two different assays, we show that the Inl- these mRNA isoforms in vivo suggested that the dominant and C-terminal activation domains of TFE3 act syner- negative activity of TFE3S is biologically important. Second, gistically. This synergism explains in part the ability of even when TFE3S represented only 20% of the total amount of TFE3S to act as a dominant negative. Our domain input TFE3 expression plasmid, the activity of the mixture was analysis of TFE3 is incorporated into a general struc- equivalent to that of 100% TFE3S rather than an average between tural model for the TFE3 protein that predicts that the the activities of TFE3S and TFE3L. activation domains of TFE3 will be widely separated in The mechanism by which TFE3S exerts this dominant negative space. effect on TFE3L is unclear. It was originally proposed that TFE3S might poison the activity of a tetramer of TFE3 proteins (4). Tetramerization of TFE3 (5) and other bHLHZIP proteins INTRODUCTION including c-Myc (6), TFEB (7) and USF (8) has been demon- strated, however, these proteins appear to bind DNA as aimers in electromobility shift assays. Tetramerization may be important in Gene transcription is a key regulatory point in diverse develop- mediating interactions between remotely and proximally bound mental processes such as cell growth, differentiation, apoptosis and bHLHZIP proteins (5), however either dimers or tetramers of transformation. The transcription factors that control the rates of TFE3 could mediate simple activation from a proximal element transcription initiation are often tightly regulated and the activities such as the one used in the mixing experiments with TFE3S and of these proteins can be modified at a variety of levels. Eukaryotic TFE3L (4). We previously observed that TFE3 exhibits synergis- transcription factors are generally bipartite; DNA-binding domains tic activation of transcription as the number of ^E3 binding sites contact the DNA binding site and activation domains transduce are increased in a promoter (5). We wondered if synergistic regulatory information via protein-protein interactions with com- effects present in the TFE3L homodimer, but absent in the TFE3S ponents of the basal transcription apparatus (1). * To whom correspondence should be addressed Present addresses: Departments of+Medicine and ^Dermatology, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA Downloaded from https://academic.oup.com/nar/article-abstract/23/19/3865/2376375 by DeepDyve user on 06 August 2020 3866 Nucleic Acids Research, 1995, Vol. 23, No. 19 homodimer and the TFE3L-TFE3S heterodimer could contribute 0.1 mM EDTA and 400 mM NaCl. Fibroblasts from one 10 cm to the dominant negative activity of TFE3S. plate (4 x 106 cells) resuspended in 100 u.1 of lysis buffer and To understand more completely the mechanism of dominant allowed to remain on ice for 30 min then frozen in liquid nitrogen. negative inhibition by TFE3S, we characterized the activation Nuclear extract (10 |ll; 4 x 105 cell equivalents) was loaded in sequences of TFE3 and developed an assay to test how these each well of a 10% SDS-polyacrylamide gel. The nuclear proteins domains might work together. In this report, we show that a were transfered to nitrocellulose and Western blotted by standard C-terminal proline rich domain of TFE3 is an activation domain. techniques using a polyclonal anti-TFE3 antiserum. Furthermore, the proline-rich domain activates synergistically with the acidic domain of TFE3, both when these domains are in their normal context in the TFE3 molecule and when they are tethered to RESULTS different DNA binding domains bound at adjacent DNA elements. These data are incorporated into a model for TFE3 activation. TFE3L contains a second transcriptional activation domain C-terminal to the bHLHZip region MATERIALS AND METHODS The exon that is alternately spliced between isoforms TFE3L and Expression plasmids TFE3S encodes an N-terminal activation domain that is rich in acidic amino acids. TFE3S, which lacks this domain, is a weaker GAL4-TFE3PRO was constructed by cloning a Stul fragment transactivator but it does retain some transcriptional activating from the TFE3 cDNA into the £c/136II site of pSV-GAL( 1-147) ability (4). The original GAL4 fusion analyses of TFE3L showed (9), resulting in fusion of TFE3 residues 345-446 [numbering that the N-terminal acidic domain is a strong activation domain scheme as in (3)] to GAL4( 1-147). GAL4-TFE3AAD was (2). However, in that study the remainder of the protein previously described as GAL4X3-A2 and fuses 126 N-terminal containing the bHLHZIP and C-terminal domains stimulated residues from TFE3 to GAL4(1-147) (2). GAL4-VP16 has been transcription weakly when fused to the GAL4 DNA binding previously described (9). domain (GAM DBD). These data led us to speculate that TFE3L pSV2-TFE3L and pSV2-TFE3S were described previously (4). might contain a second activation domain. TFE3L-ACterm and TFE3S-ACterm represent C-terminal trun- To test this hypothesis, we constructed a fusion protein in which cations of TFE3L and TFE3S, respectively, terminating at Val-314, the GAL4 DBD was fused to the C-terminus of TFE3. We tested immediately after the leucine zipper [numbering as in (3)]. They this fusion construct and others in cotransfection experiments with were constructed by PCR using pBS-TFE3L and pBS-TFE3S as a GAL4 site-dependent reporter plasmid driving the luciferase templates. PCR fragments were digested with HindYH and BarnHL gene. and cloned into the pSV2A expression plasmid (10). Figure 1A shows that the GAL4 DBD alone, GAL4( 1-147), did not stimulate transcription. GAL4-TFE3AAD, containing an Reporter plasmids N-terminal fragment of TFE3 that includes the AAD fused to the GAL4 DBD (2), stimulated transcription 249-fold. The fusion pGAL4-TATALuciferase (gift of J. Kaplan) was generated by protein containing 102 amino acids from the C-terminus of TFE3 isolating a GAL4 site pentamer from G5E4T (gift of M. Carey) fused to the GAL4 DBD activated transcription to a similar by digestion with Xbal and HindSM. This fragment was cloned by extent, 284-fold. For comparison, activation is shown for blunt ended ligation into the BglU site of pTATA-Luciferase (5), GAL4-VP16, a fusion of the GAL4 DBD and the potent VP16 a reporter plasmid containing a minimal TATA box upstream of activation domain (9). These data identify a second, C-terminal the firefly luciferase gene. transcriptional activation domain in TFE3L, separated in primary pGAL4/uJE3-TATALuciferase is a reporter plasmid containing a amino acid sequence from the N-terminal activation domain by pentamer of GAL4 sites and a tetramer of |XE3 sites upstream of the large bHLHZIP domain. The N-terminal activation domain the minimal promoter of pTATA-Luciferase. The GAL4 site and the C-terminal activation domain have similar transcriptional pentamer was cloned into the HindfiB. site of a pKS Bluescript activating abilities in the GAL4 fusion assay. polylinker that contained a |iE3 tetramer in the Smal site. A cassette Analysis of the C-terminal 134 amino acids of TFE3 shows the containing both multimers was excised by Xbal and fiawHI region is rich in Pro (13%), Ser (16%), Leu (13%), Gly (9%) and digestion and cloned into the BglQ site of pTATA-Luciferase by blunt ended ligation. acidic residues (15%, Asp plus Glu). This profile is similar to that of other activation domains that have been termed Pro-rich. The Pro-rich domain was originally described for CTFI; the C-ter- Transfection minal 99 amino acids of CTFI contain 19% Pro, 13% Ser, 10% Leu, and 10% Gly, but only 5% acidic residues. Oct-2, a B-cell NIH 3T3 cells were transfected by the calcium phosphate enriched member of the POU domain family, contains a technique essentially as described (5). Reporter plasmid (1 (ig) and C-terminal domain (residues 314-460) that is 13% Pro, 14% Ser, 5—10 |Xg expression plasmid were used per plate of cells. Cells were 8% Leu, 10% Gly, and 4% acidic residues. TFEB, a bHLHZIP harvested and assayed for luciferase activity as previously protein with homology to TFE3, has a 150 amino acid C-terminal described (5). domain that is 14% Pro, 13% Ser, 13% Leu, 10% Gly, and 15% acidic residues. AP2 has a 70 amino acid activation domain Western blot (residues 51-120) that is Pro and Gin rich; it is comprised of 21% Pro, 9% Ser, 9% Leu, 7% Gly, and 16% Gin. Hox4.2 contains an Whole cell extracts from transfected fibroblasts were prepared. 86 amino acid domain (residues 24-109) containing 22% Pro, 6% Cells were pelleted by centrifugation at 600 g, washed once in Ser, 6% Leu, 23% Gly, and 11% Ala. PBS, and lysed in 50 mM Tris pH 8.0,0.5% NP-40,10% glycerol, Downloaded from https://academic.oup.com/nar/article-abstract/23/19/3865/2376375 by DeepDyve user on 06 August 2020 Nucleic Acids Research, 1995, Vol. 23, No. 19 3867 amino acids (Fig. IB). Within the homologous 38 amino acids, 50% of the residues are identical among all three proteins and AAD 8HLHZIP PRO TFE3L N I V////A 45% are identical between two of the three. The conserved region is contained within the 102 amino acid C-terminal portion of TFE3 shown to be sufficient for activation (Fig. 1 A). Interesting- ly, although the proline content in the highly conserved region is GU.4 low (3/38), all three prolines are perfectly conserved. CDNA J±L The two TFE3 activation domains activate NO ACTIVATOR 1.0 + 0.2 transcription synergistically We showed previously that TFE3L is capable of synergistic GA 1.4(1-147) 1.0 ± 0.8 activation of transcription; as the number of cognate binding sites upstream of a TATA element is increased from four to eight, AAD transcription driven by TFE3L is increased by substantially more 249 ± 15 than a factor of two (5). Although proteins with a single activation domain have been shown to exhibit this phenomenon of synergy, PRO 284 ± 82 we wondered if part of the activation properties of TFE3L were GAL4-TFE3PRO L dependent upon synergistic stimulation of transcription by the separate N-terminal and C-terminal activation domains. ^ ^ VP16 1605 + 180 GAL4-VP16 We first addressed this question in the context of the nati veTFE 3 protein. Transcriptional activation was compared among TFE3L (containing both activation domains), TFE3S (containing only the 464 \ C-terminal activation domain) and TFE3L-ACterm (a truncated TFE3 GLPVPPNPGL LSLTTSSVSD SLKPEQLDIE EEGBPSTTFH VSGGPAQNAP TFEB GLPTTSPSG. ..VNMAELAQ QWKQELPSE DGPGEALMLG PEVPEPEQMP protein containing only the N-terminal activation domain). Plasmids Mi GLSLIPSTG. . .LCSPDLVN RIIKQEPVLE NCSQE encoding these proteins were cotransfected with a reporter depend- PQQPPAP PSDALLD LHFPSD.HLG DLGDPFHLG ent upon eight tandem (iE3 sites. Figure 2A shows that TFE3L TFE3 ALPPQAPLPS AAQPQSPFHH LDFSHGLSFG GGGDEGPTGY PDTLGTEHGS TFEB stimulated the promoter 105-fold. This high level of stimulation is LVQHQADLTC TT T LDLT DGTITF TNNLGTMPES Mi dependent upon synergistic activation by multiple DNA-bound DGMVGGLSG G ALSPI tAASDPLLSSVSPA TFE3 ...LEDILME E TFE3 molecules. TFE3S activated 19-fold and TFE3L-ACterm SLLPI A.S DPLFSTMSPE KKt LDLMLLD D TFEB P....FPNLS activated 9-fold. If activation by TFE3L on the multiple-site reporter RLSPV GVT DPLLSSVSPG Mi SPAYSIPRKM GS1>LEDILMD D were simply dependent upon the sum of activation by the AAD and TFE3 VSNASSRRSS FSIEEES the Pro-AD, one would predict that the activities of TFE3S and TFEB ASKASSRRSS FSMEE TFE3L-ACterm would add up to the activity of TFE3L. Since this Mi ASKTSSRRSS MSAEETEHA is not the case, the data suggest that it is the combination of acidic and Pro-rich activation domains that enables TFE3L to activate at Figure 1. (A) Analysis of activation domains of TFE3. Different protein such high levels and that the AAD and Pro activation domain regions were fused to the GAL4 DBD and tested for activation by stimulate transcription synergistically with one another. cotransfection with a GALA site-dependent reporter plasmid driving expression of the luciferase gene. GAL4-TFE3AAD was previously described as Figure 2A also shows that TFE3S-ACterm cannot activate the GAL4X3-A2 and fuses 126 N-terminal residues from TFE3 to GAL4(1-147) JIE 3 site reporter plasmid. The results with this C-terminal (2). TFE3PRO comprised the C-terminal 102 residues from TFE3 fused to the GAM DBD and GAL4-VP16 contains the activation sequences of the VP16 truncation of TFE3S confirm that the residual activity of TFE3S protein. Luciferase values from at least three independent transfections were is attributable to the Pro-rich domain. Furthermore, without the normalized to that of the reporter alone and are shown at the right for each exon encoding the 35 amino acid AAD, the remaining N-terminal activator plasmid. (B) Sequence homology among murine TFE3, TFEB and mi. and bHLHZIP domains are devoid of transactivating potential. The Pileup algorithm (GCG software) was used to compare the C-termini of these three proteins. Three regions of high homology are boxed and proline To rule out the possibility that differential protein expression residues are shown in bold. The horizontal arrow indicates the beginning of the confounds the interpretation of our results, immunoblots were 102 amino acid region of TFE3 shown to be sufficient for activation when fused performed on whole cell extracts of NIH3T3 fibroblasts frans- to the GAL4 DBD. fected with equai amounts of expression piasmid. i rc3L-ACterm was expressed at slightly lower levels than TFE3L. This small difference in expression may be accounted for by a loss of epitopes Despite similar amino acid contents, sequence homology has in the truncated protein. The polyclonal antiserum was generated not previously been demonstrated among proline-rich domains. using TFE3S as an antigen and therefore is likely to have However, we compared the sequence of the proline-rich C-termi- significant antibody activity directed against the proline-rich nus of TFE3 with the proline-rich C-termini of TFEB and C-terminus. Expression of TFE3L and TFE3L-ACterm are micophthalmia (mi) (11), two proteins in the |J.E3 family of therefore comparable. bHLHZIP domain proteins. We have cloned a murine TFEB Surprisingly, TFE3S and TFE3S-ACterm were expressed at cDNA which extents further 3' than that originally reported for significantly higher levels than TFE3L. Densitometric scanning human TFEB (12). This cDNA was sequenced to provide the of the autoradiogram showed that the TFE3S based isoforms were amino acid data shown in Figure IB. Substantial homology was expressed at -2.5-fold higher levels than their TFE3L based detected in the C-termini of TFE3, mi and TFEB. This homology counterparts. These data show that the various forms ofTFE3 extends over 52 amino acids in the extreme C-terminus of TFE3 protein used in Figure 2 A are expressed at least as well as TFE3L. and is distributed among three homology boxes containing 38 Downloaded from https://academic.oup.com/nar/article-abstract/23/19/3865/2376375 by DeepDyve user on 06 August 2020 3868 Nucleic Acids Research, 1995, Vol. 23, No. 19 |iE3 TATA p ^ LUCIFERASE FOLD ACTIVATION 0 20 40 60 80 100 120 NO ACT1VATCR TFE3L TFE3S TFE3L-ACterm TFE3S-ACterm CD u u t/) < ] CO PO ro UJ UJ UJ u_ u_ I— 1 - Endogenous TFE3 •" 71 Figure 2. (A) The activation domains of TFE3 act synergistically. Cotransfections were performed with plasmids encoding the indicated TFE3 proteins and a reporter plasmid containing eight tandem nE3 binding sites. TFE3L contains both activation domains and TFE3S is the isoform that lacks the N-terminal activation domain. TFE3L-ACterm and TFE3S-ACterm were engineered to include a stop codon after the leucine zipper of TFE3L and TFE3S, respectively. TFE3L-ACterm contains only the N-terminal activation domain, and TFE3S-ACterm contains neither activation domain. (B) Western blot of extracts from transfected NIH3T3 cells. Cells were transfected with 20 p.g of the expression plasmid indicated. Western blots using a polyclonal anti-TFE3S anu'serum were performed on whole cell extracts. Lane 1, untransfected cells; lane 2, TFE3S-ACterm; lane 3, TFE3L-ACterm; lane 4, TFE3L; lane 5, TFE3S. The positions of endogenous TFE3 related proteins and the migration of molecular weight standards are shown. Therefore, the lower transactivation potential of the shorter reproduced when the activation domains were joined to separate proteins is an innate property of these proteins and is not due to DNA-binding domains and bound to the same DNA template. decreased expression. This strengthens our conclusion that the N- Because this approach changes the spacing and, possibly, orienta- and C-terminal activation domains of TFE3L act synergistically. tion of the activation domains relative to their positions in the native protein, it might reveal different functional consequences of domain interactions. A reporter plasmid containing five The activation domains of TFE3 activate synergistically multimerized GAL4 binding sites and four multimerized (iE3 when tethered to different DNA-binding domains binding sites upstream of a TATA box and the luciferase gene was To examine further the capacity of TFE3's activation domains to constructed. Cotransfections were performed with TFE3 proteins act synergistically, we asked if synergistic activation could be containing either the AAD or the Pro-rich domain and with hybrid ran cted term 3L- :erm Downloaded from https://academic.oup.com/nar/article-abstract/23/19/3865/2376375 by DeepDyve user on 06 August 2020 Nucleic Acids Research, 1995, Vol. 23, No. 19 3869 example, proline residues might provide unstructured 'punctu- ation' between small structured regions which contact the basal GAL4 UE3 TATA ^ . LUCIFERASE 35 - transcription machinery. Alternatively, conserved regions and not 1 I I I I I—I I I I I—l~l—M the prolines may be functionally important. Recent dissection of both glutamine and acidic domains has shown that hydrophobic residues, not the prevalent glutamine and acidic residues, are required for interaction of these domains with their protein targets (13,14). The prevalent amino acids upon which classification of activation domains has been based may serve a more general structural role or an as yet unidentified function. Importance of synergistic activation by TFE3 activation domains We have shown that multiple DNA-bound molecules of TFE3 synergistically activate transcription and that this synergism is < E < dependent upon the presence of both activation domains. This < Q- a> < CO was demonstrated within the context of the native TFE3 protein UJ S 3 LL and when the AAD and the Pro-rich activation domain were fused I- h- £ CT _J to different DNA binding domains. In what circumstances is Zl U. 3 3 S 3 synergy between TFE3's activation domains likely to be import- <5 S ant biologically? TFE3 has been shown to mediate communica- tion between promoter and enhancer elements. The HLHZIP Figure 3 . The activation domains of TFE3 act synergistically when tethered to domain controls interaction between widely separated elements different DNA binding domains on the same DNA template. TFE3 proteins and the activation domains determine synergistic activation (5). containing a single activation domain were transfected either alone or in The immunoglobulin heavy chain gene contains a (iE3 site in its combination with GAL-4-fusion constructs expressing a single TFE3 activa- tion domain. Activity was measured from a reporter plasmid with five GAL4 V H promoters and in the intronic enhancer. TFE3 may facilitate sites and four p£ 3 sites. association of these elements. Once these elements are in close proximity, synergistic activation by TFE3 molecules bound in the promoter and in the enhancer is likely to contribute to activation of the IgH gene. proteins comprised of the GAL4 DNA-binding domain fused to either the TFE3 AAD or the Pro-rich domain. GAL4-TFE3AAD Demonstration of synergistic activation by TFE3L's activation domains also provides insight into the mechanism of inhibition of and GAL4-TFE3PRO each activated this reporter plasmid ~9-fold (Fig. 3). Note that the fold activation is lower than in TFE3L activity by the alternately spliced isoform TFE3S. TFE3S Figure 1 because the basal activity of the GAL4/|xE3 reporter is mRNA is expressed at lower concentrations than TFE3L mRNA, higher than the GAL4 reporter due to activation by endogenous ranging from 2 to 18 % of TFE3L in all tissues and cell lines flE3 binding proteins. TFE3S and TFE3L-ACterm yield similar tested. The activity of TFE3L is approximately three to four times greater than TFE3S. When cotransfected at substoichiometric levels of activation of this reporter plasmid. Cotransfection of GAL4-TFE3Pro and TFE3L-ACterm results in 28-fold activa- amounts similar to those that exist in vivo, TFE3S inhibits the tion, -2-fold higher than the sums of activation by either protein activity of TFE3L (4). In fact, in experiments in which TFE3S alone. Reversing the positions of the activation domains by comprised only 20% of the input mixture of TFE3 isoforms, transfecting GAL4-TFE3AAD and TFE3S did not result in activity was indistinguishable from that of TFE3S alone. synergistic activation. Thus, synergistic activation by AAD and How is inhibition at substoichiometric ratios possible? We Pro-rich activation domain can be recapitulated by removing suggest that two mechanisms play a role. First, the presence of these domains from their normal protein context and tethering one short monomer in a mixed dimer of TFE3L and TFE3S (L-S) them to different DNA-binding domains; however, this syner- appears to destroy synergy between activation domains. If there gism is not reproduced in every geometric arrangement. were no synergy between the activation domains, a mixed dimer (L-S) would have an activation potential exactly intermediate to DISCUSSION that of TFE3L and TFE3S dimers, assuming equal dissociation constants for dimer formation and for DNA binding. This is not A new activation domain for TFE3 the case because activity is low with subtoichiometric amounts of TFE3S—a situation where primarily L-S and L-L dimers are We have shown that the C-terminal 102 residues of TFE3 are expected to be present (4). In view of the synergy observed in this sufficient to confer strong activation on a hybrid protein when work, we suggest that lack of inter-monomer synergy accounts fused to the DBD of GAL4. Although this region is proline-rich for the low activity of L-S dimers. Thus, the model is that activity and has a similar amino acid content to other proline-rich domains, of L-L dimers is dependent upon the synergistic combination of sequence comparisons among TFE3 family members revealed AAD and Pro-rich domains and that removing one AAD from the highly conserved sub-domains, containing only three prolines, dimer reduces the activity of the mixed dimer, rendering it much located C-terminal to a highly proline-rich portion of the region. It closer to the activity of the S-S dimer. Second, our data suggest will be interesting to determine if these conserved subdomains are that TFE3S protein is more stable that TFE3L protein which either necessary or sufficient for transcriptional activation. It may would lead to a higher molar ratio of TFE3S/TFE3L protein be that both prolines and the conserved regions are necessary. For Downloaded from https://academic.oup.com/nar/article-abstract/23/19/3865/2376375 by DeepDyve user on 06 August 2020 3870 Nucleic Acids Research, 1995, Vol. 23, No. 19 relative to the RNA ratio. Nevertheless, the differences in protein expression are not large enough to produce a molar excess of TFE3S when TFE3L and TFE3S and transfected in a 4:1 plasmid ratio. Therefore, disruption of synergism among the activation domains of TFE3 appears to be an important component of the dominant negative action of TFE3S. The generation of alternate forms of transcription factors from the same gene which is seen in other proteins as well, including FosB, LAP and CREM, is an economical means of regulating the activity of key proteins (15,16). In the absence of synergistic effects such as the ones demonstrated here, dominant negative forms of these proteins would have to be generated at significantly higher levels to inhibit the activities of the more strongly activating forms. It is likely that nature exploits the effects of activation domain synergism in such a way that dramatic changes in gene expression can be achieved by more subtle alterations in the concentrations of isoforms generated from the same transcription factor gene. A model for TFE3 activation Stimulation of transcription by TFE3 is mediated by two separate activation domains, one rich in acidic residues and located in the amino terminus, and one rich in Pro residues and located in the carboxy terminus. Analysis of many other transcription factors has shown they are often comprised of multiple activation domains. Typically, these domains are in contiguous segments of the protein (17—19) although sometimes activation domains are separated in the primary amino acid sequence (20). Frequently, activation domains are comprised of multiple subdomains, which, when deleted, reduce the activity of the protein in a gradual fashion, without evidence of synergistic activation. This has been shown for GAL4 (20), C/EBP (19) and USF (18), among others. The two activation domains of TFE3L, however, are not subdomains of a larger activation domain and are likely to be separated in three Figure 4. Model for general structure of TFE3. The TFE3L protein is proposed dimensional space by the relatively large alpha helices which to form a dumbbell shaped protein with N-terminal (bottom) and C-ternunal confer DNA binding and subunit dimerization. We propose a (top) activation domains widely separated in space. Activation domains are model for the general structure of the TFE3 protein that takes into depicted as spheres. The bHLHZIP domain and DNA were generated as a account the data presented above, the location of the domains in nbbon structure from the max-DNA coordinates (21) using Molscript. the primary amino acid sequence, and specific structural details of the bHLHZIP-DNA complex of Max determined by X-ray crystallography (21). domain turns back toward the other, ending up in close proximity. We propose that the TFE3 protein dimer, when bound to DNA, assumes a 'dumbbell shape'; the shaft of the dumbbell represents The extended nature of the bHLHZIP structure makes this less the bHLHZIP domain and the ends of the dumbbell represent the likely. AADs and the Pro-rich activation domains (Fig. 4). The bHLHZIP Targets for the activation domains of TFE3 have not yet been domain that separates the activation domains is an unusual motif identified, but candidates exist. TBP (22), TFIIB (23), TAFTI40 in a non-structural protein. It is an extended domain comprised of (24) and TAFII60 (14) have been shown to interact with other two parallel ot-helices connected by a short loop. One helix AADs. The Pro-rich domain of CTFI has been shown to require contains the basic region that binds DNA and the helix 1 portion TFIIB (25) as well as TAFs (26) for transactivation. However, it of the HLH domain. The other helix contains the helix2 portion is also possible that other targets will actually be demonstrated to of the HLH domain as well as the leucine zipper. The axis of the mediate TFE3 activation. The minimal sequence homology among HLHZIP domain is perpendicular to the DNA axis. The basic members of a given family of activation domains make a single region a-helix passes through the major groove of the DNA protein target unlikely, hi fact, TAFQ110 interacts with the Gin-rich binding site and terminates on the other side. Thus, sequences sequences of Spl and CREB, but not those of Antennapedia or N-terminal to the basic region must at least begin on the other side bicoid (13,27). of a plane that is perpendicular to the HLHZIP domain and that The proposed dumbbell shape of TFE3 and our data that show contains the DNA molecule. The dumbbell model predicts that synergistic interaction of both activation domains suggest a three the AAD and the Pro-rich domains will be widely separated in dimensional view of how TFE3 might direct formation of a space and will lie on opposite sides of the DNA molecule. The stereospecific preinitiation complex. Transactivation of a pro- alternative, which we cannot rule out, is that each activation moter by TFE3 would therefore involve molecular connections Downloaded from https://academic.oup.com/nar/article-abstract/23/19/3865/2376375 by DeepDyve user on 06 August 2020 Nucleic Acids Research, 1995, Vol. 23, No. 19 3871 between TBP and TFE3 both above and below the plane of the Aggarwaal for critically reading the manuscript. S.A., K.W. and DNA helix. Thus, TAFII40 or TFIIB might bridge between the K.M. were supported by an NIGMS M.S.T.P. training grant and TFIID complex and the AAD of TFE3 on one side of this plane, this work was supported by USPH grant GM29361 to K.C. while TFIIB or another TAF might bridge between the Pro-rich domain and the TFIID complex on the other side of this plane. REFERENCES Of the many TFE3, USF and Myc/max related proteins that comprise the bHLHZIP family of proteins, only TFE3, TFEB and 1 Tjian, R. and Maniatis, T. (1994) Cell, 77, 5-8. mi contain a proline-rich region C-terminal to the ZIP domain. 2 Beckmann, H., Su, L.-K. and Kadesch, T. (1990) Genes Dev., 4, 167-179. 3 Roman, C , Matera, G., Cooper, C , Artandi, S., Blain, S., Ward, D. and USF, c-Myc and max terminate shortly after the ZIP domain. In Calame, K. (1992) Mol. Cell. Biol., 12, 817-827. addition to having different activation sequences, the overall 4 Roman, C , Cohn, L. and Calame, K. (1991) Science, 254, 94-97. geometric shapes and structures of these proteins are likely to be 5 Artandi, S. E., Cooper, C , Shrivastava, A. and Calame, K. (1994) Mol. different. The criteria that dictate whether a gene is regulated by Cell. Biology, 14, 704-7716. TFE3, USF or c-Myc/max are not known. It is tempting to 6 Dang, C. V., McGuire, M , Buckmire, M. and Lee, W. M. F. (1989) Nature, 337, 664-666. speculate that the stereospecific complex formed by promoter- 7 Fisher, D. E., Carr, C. S., Parent, L. A. and Sharp, P. A. (1991) Genes Dev., bound factors and components of the basal transcription apparatus 5, 2342-2352. will encourage or disfavor incorporation of a sequence specific 8 Ferre-D'Amare, A., Pognonec, P., Roeder, R. G. and Burley, S. K. (1994) factor based upon the sequence of its activation domains and their EMBOJ., 13, 180-189. geometric organization. 9 Sadowski, I. and Ptashne, M. (1989) Nucleic Acids Res., 17, 7539. 10 deWet, J. R., Wood, K. V., DeLuca, M., Helsinki, D. R. and Subramani, S. (1987) Mol. Cell. Biol., 7, 725. Similarity between TFE3 and Oct-2 11 Hodgkinson, C. A., Moore, K. J., Nakayama, A., Steingrimsson, E., Copeland, N. G., Jenkins, N. A. and Arnheiter, H. (1993) Cell, 74, To our knowledge, Oct-2 is the only protein previously shown to 395-404. contain multiple activation domains that activate synergistically 12 Carr, C. and Sharp, P. (1990) Mol. Cell. Biol., 10, 4384-88. 13 Gill, G., Pascal, E., Tseng, Z. H. and Tjian, R. (1994) Proc. Natl. Acad. (28). The overall organization of the Oct-2 protein is similar to Sci. USA, 91, 192-196. that of TFE3. Each protein is comprised of a DNA binding 14 Thut, C. J., Chen, J.-L., Klemm, R. and Tjian, R. (1995) Science, 267, domain sandwiched between an N-terminal activation domain (Gin-rich in Oct-2) and a C-terminal Pro-rich activation domain. 15 Foulkes, N. and Sassone-Corsi, P. (1992) Cell, 68, 411-414. The recent X-ray crystal structure of the Oct-1 POU domain (29) 16 Descombes, P. and Schibler, U. (1991) Cell, 67, 569-579. 17 Courey, A. J. and Tjian, R. (1988) Cell, 55, 887-898. shows that the POU specific domain contacts the 5' half of the 18 Kirschbaum, B. J., Pognonec, P. and Roeder, R. G. (1992) Mol. Celt. Biol., binding site. This domain is connected by a flexible linker to the 12(11), 5094-5101. POU homeodomain which contacts the 3 ' half of the OCT site on 19 Friedman, A. D. and McKnight, S. L. (1990) Genes Dev., 4, 1416-1426. the opposite face of the DNA double helix. Therefore, bHLHZIP 20 Ma, J. and Ptashne, M. (1987) Cell, 48, 847-853. domains and POU domains achieve placement of their N- and 21 Ferre-D'Amare, A. R., Prendergast, G. C , Ziff, E. B. and Burley, S. K. (.1993) Nature, 363, 38^5 . C-termini on opposite sides of the DNA molecule by using 22 Stringer, K. F , Infles, C. J. and Greenblatt, J. (1990) Nature, 345, completely different binding modules. If the N- and C-terminal 783-786. activation domains of Oct-2 and TFE3 are indeed located on 23 Lin, Y.-S., Ha, I., Maldonado, E., Reinberg, D. and Green, M. R. (1991) opposite sides of the DNA molecule, it suggests formation of a Nature, 353, 569-571. preinitiation complex circumferentially around the double helix 24 Goodrich, J. A., Hoey, T., Thut, C. J., Admon, A. and Tjian, R. (1993) Cell, 75,519-530. in elements such as the VH promoters that contain adjacent 25 Kim, T. K. and Roeder, R. G. (1994) Proc. Natl. Acad. Sci. USA, 91, binding sites for both factors. 4170-4174. 26 Tanese, N., Pugh, B. F. and Tjian, R. (1991) Genes Dev., 5, 2212-2224. 27 Hoey, T , Wemzierl, R. O. J., Gill, G., Chen, J.-L., Dynlacht, B. D. and ACKNOWLEDGEMENTS Tjian, R. (1993) Cell, 72, 247-260. 28 Tanaka, M., Clouston, W. M. and Herr, W. (1994) Mol. Cell. Biol., 14, We thank members of the Calame lab for helpful discussions. We 6046-6055. are grateful to Dr V. Racaniello for assistance with the Molscript 29 Klemm, J. D., Rould, M. A., Aurora, R., Herr, W. and Pabo, C. O. (1994) software, to Dr M. Carey for providing constructs and to Dr A. Cell, 77, 21-32.

Journal

Nucleic Acids Research – Oxford University Press

Published: Oct 11, 1995

There are no references for this article.