E1A and a nuclear receptor corepressor splice variant (N-CoRI) are thyroid hormone receptor coactivators that bind in the corepressor mode Xianwang Meng*, Paul Webb†, Yong-Fan Yang*, Michael Shuen‡, Ahmed F. Yousef‡, John D. Baxter†§, Joe S. Mymryk‡, and Paul G. Walfish*¶ *Department of Medicine, Endocrine Division, Mount Sinai Hospital, University of Toronto Medical School, Toronto, ON, Canada M5G 1X5; ‡Departments of Oncology, Microbiology, and Immunology, University of Western Ontario and London Regional Cancer Centre, London, ON, Canada N6A 4L6; and †Diabetes Center and Department of Medicine, University of California, San Francisco, CA 94143
Unliganded thyroid hormone (TH) receptors (TRs) and other nuclear receptors (NRs) repress transcription of hormone-activated genes by recruiting corepressors (CoRs), such as NR CoR (N-CoR) and SMRT. Unliganded TRs also activate transcription of THrepressed genes. Some evidence suggests that these effects also involve TR兾CoR contacts; however, the precise reasons that CoRs activate transcription in these contexts are obscure. Unraveling these mechanisms is complicated by the fact that it is difficult to decipher direct vs. indirect effects of TR– coregulator contacts in mammalian cells. In this study, we used yeast, Saccharomyces cerevisiae, which lack endogenous NRs and NR coregulators, to determine how unliganded TRs can activate transcription. We previously showed that adenovirus 5 early-region 1A coactivates unliganded TRs in yeast, and that these effects are blocked by TH. We show here that human adenovirus type 5 early region 1A (E1A) contains a short peptide (LDQLIEEVL amino acids 20 –28) that resembles CoR–NR interaction motifs (CoRNR boxes), and that this motif is required for TR binding and coactivation. Although fulllength N-CoR does not coactivate TR in yeast, a naturally occurring N-CoR variant (N-CoRI) and an artificial N-CoR truncation (N-CoRC) that retain CoRNR boxes but lack N-terminal repressor domains behave as potent and direct TH-repressed coactivators for unliganded TRs. We conclude that E1A and N-CoRI are naturally occurring TR coactivators that bind in the typical CoR mode and suggest that similar factors could mediate transcriptional activation by unliganded TRs in mammals. nuclear receptor coregulators 兩 gene activation
N
uclear receptors (NRs), including thyroid hormone (TH) receptors (TRs), are hormone-regulated transcription factors (1–3). TRs control growth, development, and homeostasis by binding to TH response elements (TREs) in target promoters and modulating transcription. Unliganded TRs repress THactivated genes by recruiting corepressors (CoRs) such as NR corepressor (N-CoR) and silencing mediator for retinoid and TRs (SMRT)兾TR-associated corepressor 2 (TRAC2) (4–6). Both N-CoR and SMRT are modular proteins and are comprised of N-terminal repressor domains (RDs), which bind other CoR complex components such as SIN3 and histone deacetylases, and C-terminal NR interacting domains (IDs; Fig. 1) (7). TRs directly contact short hydrophobic CoR–NR interaction (CoRNR) box motifs (CBMs; consensus I兾LXXI兾H兾 LIXXXI兾L) that are reiterated two or three times within the C-terminal ID and adopt a three-turn ␣-helical structure that docks into a hydrophobic surface on the unliganded TR ligandbinding domain (7–13). TH, predominantly triiodothyronine (T3), activates transcription by enhancing packing of TR Cterminal helix 12 over the lower part of the CoR-binding surface, simultaneously promoting CoR release and completing formation of a binding site for coactivators, such as the p160s and TRAP 220 (7). www.pnas.org兾cgi兾doi兾10.1073兾pnas.0501491102
In contrast, many TR-regulated genes are stimulated by unliganded TR and repressed by T3 (2). These TR actions are physiologically important. For example, unliganded TRs activate the thyroid-stimulating hormone (TSH) and pro-TSH releasing hormone promoters, and T3 represses transcription of both genes to mediate feedback inhibition of TH synthesis. Furthermore, actions of liganded TRs to repress gene expression are relatively common; in liver, approximately two-thirds of TR-regulated genes in mouse liver are repressed by T3 (14). Some evidence suggests that unliganded TRs activate T3repressed genes by binding CoRs (12, 15–17). Thus, TR mutations that either block or enhance CoR binding exert parallel effects on transcriptional activation by unliganded TRs. Furthermore, CoR overexpression potentiates unliganded TR activity in these settings. Nevertheless, the reasons that a CoR might be able to activate transcription in some contexts are not clear. One group proposed that TR–CoR complexes activate gene transcription indirectly by sequestration of limiting levels of histone deacetylases (HDACs) associated with other transcription factors at the TSH and pro-TSH releasing hormone promoters (16). Others observed that TRs bind weakly to unusual TREs within negatively regulated promoters (17–19), and it has been proposed that these TREs force the TR–CoR complex to adopt a conformation that recruits auxiliary coactivators, rather than CoR complex components such as HDACs (17). Indeed, N-CoR binds to the p160 coactivator ACTR, and unliganded TRs may be able to activate transcription by nucleating formation of a complex that contains both coregulators (20). Thus, overall, the mechanisms proposed to date assume that the stimulatory effects of CoRs on unliganded TRs either are mediated through indirect mechanisms or depend upon promoter-context-specific contributions from additional NR coregulators. CoR genes are differentially spliced and can encode both full-length CoRs and truncated variants that retain C-terminal IDs but lack N-terminal RDs (4, 6, 21–23). Transfection of a gene encoding the truncated N-CoR variant (N-CoRI; amino acids 1539–2441; Fig. 1) reverses transcriptional repression by unliganded TRs at T3-inducible reporters and targeted expression of N-CoRI in livers of transgenic mice activates T3-inducible genes in the absence of TH (21, 22). Based on these observations, it has been proposed that N-CoRI acts as dominant negative Abbreviations: TH, thyroid hormone; TR, TH receptor; NR, nuclear receptor; T3, triiodothyronine; TRE, TH response element; N-CoR, NR corepressor; SMRT, silencing mediator for retinoid receptors and TRs; ID, interacting domain; RD, repressor domain; CoRNR, CoR–NR interaction; CBM, CoRNR box motif; E1A, human adenovirus type 5 early region 1A; Triac, L-triiodothyroacetic acid; TSH, thyroid-stimulating hormone. §J.D.B. has proprietary interests in, and serves as a consultant and Deputy Director to, Karo
Bio AB, which has commercial interests in this area of research. ¶To
whom correspondence should be addressed. E-mail:
[email protected].
© 2005 by The National Academy of Sciences of the USA
PNAS 兩 May 3, 2005 兩 vol. 102 兩 no. 18 兩 6267– 6272
BIOCHEMISTRY
Contributed by John D. Baxter, February 28, 2005
RDs behave like E1A and coactivate unliganded TRs. We propose that truncated CoRs can be coactivators that mediate transcriptional activation by unliganded TRs in mammalian cells. Materials and Methods Yeast Strains and Media. S. cerevisiae wild-type strains YPH499
(MAT␣, ura3, lys2, ade2, trp1, his3, and leu2) were used for transcriptional activation assays. Yeast transformants were grown in minimal medium (0.67% yeast nitrogen base兾2% glucose) routinely supplemented with adenine (40 mg per liter) and lysine (40 mg per liter). Assay of TR-Dependent Transcriptional Activation in Yeast. hTR1-
YEp56, hTR1-YEp46, E1A1– 82-pAS1L, and the -gal reporter plasmids TRE-F2 ⫻ 1 were described previously (24–26). Wildtype and mutant E1A1– 82 and N-CoR derivatives (4, 21, 26) were expressed as Gal4-DNA-binding domain fusions from the vector pAS1L with LEU2 selectable marker or pRS424-ADH with TRP1 marker E1A1– 82, and N-CoR point mutants were generated by site-directed mutagenesis (QuikChange, Stratagene), as described (24–26). Selected hTR1 ligand-binding domain mutants (12, 33) were subcloned into the Yep46 yeast expression vector as described (24–26). GST Pull-Down Assays and Synthetic Peptide Competition Analyses. Fig. 1. Structure of N-CoR and adenovirus 5 E1A. (Upper) A schematic depiction of full-length N-CoR (amino acids 1–2453); N-CoR truncations; and the longer E1A splice variant, 289R. RDs in the N-CoR N terminus are shown with darker shading, and NR IDs (1–3) in the C terminus, each containing CBMs (black bars), are shown in light shading. Also shown are schematic representations of N-CoRI (amino acids 1539 –2453) and N-CoRc (amino acids 1944 – 2453). The position of the E1A CBM and conserved regions (CR)1–3, which are important for E1A activity but not TR binding, are marked. (Lower) A comparison of the CoRNR box consensus [(I兾L)XX(I兾H兾L)IXXX(I兾L)] with similar motifs from the corepressors N-CoR, SMRT, and E1A.
inhibitor of unliganded TR action by displacing endogenous N-CoR and SMRT from the TR CoR-binding surface. However, two groups found that N-CoRI potentiates unliganded TR action at the negatively regulated pro-TSH-releasing hormone promoter (19, 21). Although the effects of full-length N-CoR and N-CoRI were not directly compared in these studies, this finding suggests that N-CoRI may not behave exclusively as a dominant negative inhibitor of unliganded TRs, and that it is able to modulate TR activity by different mechanisms. It is difficult to determine the precise functional consequences of particular TR–coregulator interactions in mammalian cells, because of complicating effects of endogenous NR coregulators. We have therefore used the budding yeast Saccharomyces cerevisiae to reconstruct individual TR transcription complexes and analyze their activities (24–26); yeast contains conserved eukaryotic general transcription factors and chromatin remodeling enzymes but lacks endogenous NRs and NR coregulators (27– 32). Although coexpression of TRs and p160s was sufficient to reconstitute T3-dependent transcriptional activation (24, 25), we also found, surprisingly, an adenovirus-encoded coactivator, human adenovirus type 5 early region 1A (E1A), behaves as a specific coactivator for unliganded TRs in yeast, thereby establishing a model to dissect how unliganded TRs activate transcription (26). Here, we investigate how E1A coactivates unliganded TRs and examine the effects of CoRs on TR activity in yeast. We find that E1A contains a consensus CBM and activates unliganded TRs by binding in a similar mode to CoRs. This represents a previously unrecognized observation of a coactivator that binds TR, or other NRs, in the CoR mode. Although full-length N-CoR suppresses E1A-induced unliganded TR activity in yeast, truncated versions of N-CoR that lack N-terminal 6268 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0501491102
E1A1– 82 and C-terminal residues 1944–2453 of mouse N-CoR were cloned into the EcoRI兾SalI sites of pGEX4T1 (GSTE1A1– 82 and GST-N-CoRC). GST-E1A1– 82 and mutants ⌬4–25 or ⌬30–49, or GST-N-CoRC were expressed in Escherichia coli BL21 cells and prepared by standard methods. Full-length 35S-labeled hTR1 was synthesized by in vitro transcription and translation (TNT, Promega). Equivalent amounts of GST or GST-fusion protein were used for in vitro binding assays as described (11). The sequence of the N-CoR CBM1 competition peptide corresponds to residues 2272–2291: N-ASNLGLEDIIRKALMGSFDD-C. Yeast Two-Hybrid Assay. Full-length human TR1 and the
I280K mutant were expressed as LexA-DNA-binding domain fusions by using the bait vector pBAIT. Sequences encoding amino acids 1–82 of E1A or N-CoRC were expressed as prey in vector pJG4-5. L40 yeast was transformed with bait and prey plasmids and grown overnight in 2% galactose and minimal media containing all essential amino acids except leucine, and tryptophan. Protein–protein interactions were monitored by using the LacZ reporter gene and assayed as Miller units per mg of protein (26). Results E1A Utilizes a CoRNR Box to Bind and Coactivate Unliganded TR. To
delineate the mechanism whereby adenovirus E1A serves as a coactivator for unliganded TRs in yeast, we first defined the E1A residues required for the effect more precisely. We previously determined that E1A amino acids 1–82, expressed as a GAL4 fusion protein to facilitate nuclear localization (E1A1– 82), were necessary and sufficient to coactivate unliganded TRs in yeast, irrespective of the nature of the TRE and the presence or absence of the retinoid X receptor TR heterodimer partner (26). We also found that this effect depends on E1A amino acid residues 4–29. We therefore examined the effects of internal deletions and mutations within E1A1– 82 on TR coactivation and TR binding (Fig. 2). In accordance with previous results, E1A1– 82 coactivated unliganded TR, and this effect was down-regulated by the TR agonist L-triiodothyroacetic acid (Triac), used here because, unlike T3, it is taken up efficiently by yeast (Fig. 2 A). Confirming previous results, there is little effect of Triac in the absence of a cotransfected coregulator. Deletion of residues 1–14, 4–25, and Meng et al.
26–35 (⌬1–14, ⌬4–25, and ⌬26–35, respectively) abolished E1A coactivator function, whereas deletion of residues 2–9 (⌬2–9), 30–49 (⌬30–49), and 48–60 (⌬48–60) did not. Thus, E1A residues 10–29 are essential for coactivation of unliganded TRs. Inspection of E1A residues 10–29 reveals a sequence that resembles a CoRNR box at amino acids 20–28 (Fig. 1 Lower). Ala substitutions within this putative CoRNR box at Leu-28 (L28A); Leu-23 and Ile-24 (A2); or Leu-20, Leu-23, Ile-24, and Leu-28 (A4) abrogated TR coactivator function (Fig. 2B). Other point mutations outside the CoRNR box did not (R2G, H3N, I4A, and V10A), although an Ala substitution at Ile-11 (I11A) did result in some loss of function. Thus, E1A contains a CoRNR box-like sequence required for coactivation of unliganded TR in yeast. The presence of the E1A CoRNR box suggested an obvious explanation for the ability of E1A to coactivate TRs in the absence, but not in the presence, of TH; E1A binds TRs in a similar mode to CoRs. To confirm this idea, we examined interactions of TR with bacterially expressed E1A1– 82 in GST pull-down assays in vitro. As expected, TR bound to E1A1– 82 in Meng et al.
the absence of TH, and TR–E1A1– 82 interactions were abrogated by deletion of E1A amino acid residues 4–25, which removes part of the CoRNR box, but not by deletion of residues 30–49, adjacent to the CoRNR box (Fig. 2C) and by the 2A (L23A and I24A) and 4A (L20A, L23A, I24A, and L28A) point mutations within the CoRNR box (not shown). Furthermore, and contrary to previous reports of TR–E1A interactions in vitro (34), TR binding to E1A1– 82 was decreased in the presence of T3 just as TR–CoR interactions are decreased by T3 (Fig. 2 C and D). Finally, TR1 binding to E1A was inhibited by short peptides (20–30 aa) that overlap N-CoR CoRNR box 1 (Fig. 2D) and also N-CoR CoRNR box 3 and SMRT CoRNR box 1 (not shown). Thus, E1A is a TR coactivator that binds in the typical CoR mode. N-CoRI Is a Coactivator for Unliganded TR. Because E1A utilizes a CBM to coactivate unliganded TRs, we examined the possibility that naturally occurring CoRs or truncated CoRs might also coactivate unliganded TRs in yeast. Fig. 3A confirms our previous finding that full-length N-CoR (N-CoRFL) does not coacPNAS 兩 May 3, 2005 兩 vol. 102 兩 no. 18 兩 6269
BIOCHEMISTRY
Fig. 2. The E1A CBM is required for TR coactivation. (A) E1A amino acids 4 –29 are required for TR coactivation. Shown are the results of -gal assays performed with a yeast strain containing a single copy of the chicken lysozyme (F2) TRE and expressing hTR1, wild-type E1A1– 82, or E1A1– 82 internal deletion mutants or empty vector. Cultures were treated with either vehicle (hatched bars) or 10⫺7 M Triac (solid bars). -Gal activities were expressed as Miller units per mg of protein. Data shown were pooled from three independent experiments and calculated as mean ⫾ SE. (B) Point mutations in the E1A CBM abrogate TR coactivation as for A, except that E1A1– 82 point mutants were used. A2, Ala substitution for L23 and I24; A4, Ala substitution at L20, L23, L28, and I24. (C) TR interactions with E1A require the CBM. Shown is an exposure of a 10% SDS–polyacrylamide gel loaded with input labeled TR or TR retained on GST beads in the presence of GST-E1A proteins ⫾ T3 (10⫺6 M). (D) CBM peptide blocks TR binding to E1A and N-CoR. The sequence of the N-CoR CBM1 peptide is shown (Upper) with key hydrophobic residues that are NR contact points boxed in black. (Lower) An exposure of a 10% SDS–polyacrylamide gel loaded with TRs retained on GST-E1A or GST-N-CoR columns plus increasing amounts of CBM peptide (0.3–10 g). Experiments were performed in the absence of TH except where indicated.
Fig. 3. N-CoRI is a coactivator for unliganded TRs. (A) Triac blocks actions of unliganded activators. Shown are the results of -gal assays performed as in Fig. 2 A, with yeast strains expressing E1A1– 82, N-CoRFL, N-CoRI, and N-CoRC as galactosidase fusions in the presence of increasing concentrations of Triac (10⫺9⫺10⫺5 M). (B) N-CoR CBM mutations block TR coactivation as in A, with wild-type or mutated N-CoRC. Point mutations were all Ala substitutions. Leu-2277, Ile-2281 (M1), and Leu-2285 are in CBM1; Ile-2073 is in CBM2 (M2); and Ile-1953 is in CBM3 (M3). A201 combines mutations at Ile-2281 and Ile-2073 to disrupt CBM1 and -2 (M1⫹2), A202 combines mutations at Ile-2281 and Ile-1953 to disrupt CBM1 and -3 (M1⫹3), A203 combines mutations at Ile-2073 and Ile-1953 to disrupt CBM2 and the remaining CBM3 core (M2⫹3), and A204 combines mutations at all three residues to disrupt all CBMs (M1⫹2⫹3).
tivate unliganded TRs in yeast (26). N-CoR did inhibit E1A coactivation of TRs, proving that it is expressed in a functional form in this system (not shown) and confirming that it does not behave as a coactivator when bound to the TR CoR-binding site. By contrast, the naturally occurring truncated corepressor NCoRI and an artificial N-CoR truncation that retains two intact CoRNR boxes (N-CoRC amino acids 1944–2453) both behaved as potent coactivators for unliganded TRs. Moreover, activities of both N-CoR truncations were suppressed by Triac, although greater Triac concentrations (10⫺6 M) were required to suppress N-CoRI and N-CoRC activity than E1A activity (10⫺8 M), possibly reflecting the fact that the N-CoR truncations contain multiple CoRNR boxes that could result in tighter binding of the N-CoR truncations to the TR than E1A. Similar results were also observed at a reporter driven by the negatively regulated proTSH releasing hormone promoter (not shown), as reported also for E1A (26). To determine whether N-CoR CoRNR boxes are required for TR coactivation, we examined the effects of Ala substitutions in all three motifs (Fig. 3B). N-CoRc contains intact functional CoRNR boxes 1 and 2, and CoRNR box 3 is not functional because of deletion of sequences N-terminal to the core motif 6270 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0501491102
Fig. 4. E1A and N-CoR coactivation requires the TR CoR-binding surface. (A) Mutations in the TR CoR-binding surface disrupt coactivation by E1A and N-CoR truncations. Shown are results of -gal assays performed as in Fig. 2 A, with yeast strains expressing E1A1– 82 or N-CoRC as galactosidase fusions along with wild-type TRs or TR mutants. (B) Interaction of E1A1– 82 with hTR ligand-binding domain is impaired by the TR I280K mutant. Shown are results of a yeast two-hybrid experiment performed with a LexA-DNA-binding domain to either wild-type hTR or the I280K mutant as bait and E1A1– 82-AD as prey.
that are essential for TR binding (11, 35, 36). Accordingly, mutations in CoRNR boxes 1 and 2 partially reduced TR coactivation and combinations of mutations in CoRNR boxes 1 and 2 (A201) or in all three CoRNR boxes (A204) abolished TR coactivation. Thus, N-CoRI and another N-CoR truncation that lacks N-terminal RDs behave as coactivators for unliganded TR at positive and negative TREs, and these effects require intact CoRNR boxes. Because yeast lacks endogenous CoRs, we conclude, contrary to existing models, that N-CoRI activates unliganded TRs directly and not simply by displacing endogenous CoRs. E1A and N-CoRC Coactivate TR by Binding the Corepressor-Binding Surface. Finally, we examined the requirement for the TR
corepressor-binding surface in coactivation by E1A1– 82 and N-CoRC. Fig. 4A shows that TR mutants, previously characterized for effects upon TR兾N-CoR interactions in vitro and unliganded TR activities in transfection assays in vivo, inhibited TR coactivation by E1A and N-CoRC in yeast (12). Most notably, mutations within residues that lie under the usual position of H12 in liganded TR and form an essential part of the corepressor-binding surface (I280K and C309W) impaired TR coactivation by E1A and N-CoRC. These mutations did not enhance the basal activity of unliganded TRs in yeast in the absence of E1A and N-CoRC (not shown). This suggests that Meng et al.
Discussion The yeast system permits us to reconstruct TR-dependent gene regulation pathways and determine effects of particular TR– coregulator interactions in a eukaryotic cell background that is devoid of complicating influences of endogenous NRs or NR coactivators and CoRs (1, 7). Here, we used yeast to investigate how E1A potentiates the activity of unliganded TRs and how various forms of N-CoR affect TR activity. Our data indicate that E1A contains a CoRNR box-like sequence (LDQLIEEVL) at amino acids 20–28 that is required for TR coactivation and binding, and that E1A interactions with TR are blocked by T3 and short CoRNR box peptides in vitro and by mutations in the TR CoR-binding surface in yeast and in vitro. Thus, E1A is a naturally occurring TR coactivator that binds in the CoR mode. In accordance with previous results, full-length N-CoR does not coactivate unliganded TR in yeast (26), although it does block E1A action on TR, proving that it is expressed in yeast in a functional form that can compete for the TR CoR-binding surface. These data confirm that the intrinsic activity of fulllength N-CoR is repression. By contrast, a naturally occurring N-CoR splice variant (NCoRI) and an artificial N-CoR truncation (N-CoRC) behave as potent coactivators for unliganded TRs. These effects require CoRNR boxes and the integrity of the TR CoR-binding surface. Thus, the N-CoR C terminus harbors a TR coactivation function whose activity is unmasked by truncation of the N-terminal RDs, and N-CoRI and N-CoRC are, like E1A, coactivators for unliganded TRs. Our results comprise previously unreported evidence for the existence of specific coactivators that potentiate unliganded TRs by adopting the typical CoR-binding mode. There are several examples of CoRs that inhibit actions of liganded NRs by binding to the typical coactivator-binding surface. These include RIP140, SHP, and DAX1 (37–39). Thus, we now have examples of both coactivators and corepressors that bind to the corepressor-binding surface and coactivators and corepressors that bind to the coactivator-binding surface. Although we are not advocating a change in the terminology, it is evident that the current usage of the coactivator- and corepressor-binding sites does not fully convey the properties of these surfaces. 1. Cheng, S. Y. (2000) Rev. Endocr. Metab. Disord. 1, 9–18. 2. Yen, P. M. (2001) Physiol. Rev. 81, 1097–1142. 3. Mangelsdorf, D. J., Thummel, C., Beato, M., Herrlich, P., Schutz, G., Umesono, K., Blumberg, B., Kastner, P., Mark, M., Chambon, P., et al. (1995) Cell 83, 835–839. 4. Horlein, A. J., Naar, A. M., Heinzel, T., Torchia, J., Gloss, B., Kurokawa, R., Ryan, A., Kamel, Y., Soderstrom, M., Glass, C. K., et al. (1995) Nature 377, 397–404. 5. Chen, J. D. & Evans, R. M. (1995) Nature 377, 454–457. 6. Sande, S. & Privalsky, M. L. (1996) Mol. Endocrinol. 10, 813–825. 7. Glass, C. K. & Rosenfeld, M. G. (2000) Genes Dev. 14, 121–141. 8. Hu, X. & Lazar, M. A. (1999) Nature 402, 93–96.
Meng et al.
Our data force us to reconsider the interpretation of previous studies yet are also consistent with published results. E1A was originally reported to act as a coactivator for liganded TRs (34). Our reinspection of the data reveals, however, that E1A also enhanced TR activity in the absence of T3. Thus, we propose that E1A enhances unliganded TR activity directly in mammalian cells by contacts with the TR CoR-binding surface and liganded TR activity indirectly, possibly by contacts with TR-associated coactivators. These observations would not exclude the possibility that additional mechanisms could allow E1A to be a coactivator in other contexts. Our data are also consistent with observations that N-CoRI enhances unliganded TR activity in transfections and reactivates the THinducible gene program in livers of thyroidectomized transgenic mice (21, 22) but suggest that these effects could stem, at least in part, from direct TR coactivation rather than reversal of TR dominant-negative activities. Finally, our data suggest a single coherent explanation for the apparently diverse behaviors of N-CoRI at TH-induced and repressed genes; N-CoRI reverses transcriptional inhibition by unliganded TRs at T3-induced genes and potentiates transcriptional activation by unliganded TRs at negatively regulated genes, because it is a potent and generalized coactivator for unliganded TRs, irrespective of the nature of the promoter and TRE (19, 21, 22, 26). As developed in the Introduction, unliganded TRs can paradoxically activate transcription of negatively regulated genes by binding CoRs, but the reasons that the CoRs can activate transcription in these contexts are not clear. Our demonstration that the N-CoR C terminus harbors a potent TR transcriptional activation function suggests an explanation for these effects. Although the activity of this N-CoR C-terminal transactivation function will often be masked by competing activities of Nterminal RDs and associated proteins in mammalian cells, we propose that unliganded TRs could either activate transcription by preferentially recruiting truncated CoRs such as N-CoRI or, possibly, full-length CoRs that lack associated CoR complex components such as histone deacetylases. We predict that TR兾 CoR contacts would lead to transcriptional activation in either of these contexts. We thank C. Glass (University of California at San Diego, La Jolla) for the GST-TR1 fusion protein; A. N. Hollenberg (Harvard Medical School, Boston) for N-CoRI as well as technical and consultative advice; R. Y. Wang for technical assistance; and C. Walfish, T. Fridman, and B. Stork for assistance in manuscript preparation. This work was supported by a Diana Meltzer Abramsky Research Fellowship from the Thyroid Foundation of Canada (to X.M.); Canadian Institutes of Health Operational Grants (MOP-49448 to P.G.W. and MOP-14631 to J.S.M.); the Mount Sinai Hospital Foundation of Toronto and Department of Medicine Research Funds, the Julius Kuhl and Temmy Latner兾Dynacare Family Foundations, and Sensium Technologies, Inc. (to P.G.W.); a joint Canadian Institutes of Health Research兾London Regional Cancer Centre Studentship Award (to M.S.); and National Institutes of Health Grants DK51281, DK41482, and DK51281 (to J.D.B.). A.F.Y. is a Scholar of the Canadian Institutes of Health Research兾University of Western Ontario Strategic Training Initiative in Cancer Research and Technology Transfer. 9. Perissi, V., Staszewski, L. M., McInerney, E. M., Kurokawa, R., Krones, A., Rose, D. W., Lambert, M. H., Milburn, M. V., Glass, C. K. & Rosenfeld, M. G. (1999) Genes Dev. 13, 3198–3208. 10. Nagy, L., Kao, H. Y., Love, J. D., Li, C., Banayo, E., Gooch, J. T., Krishna, V., Chatterjee, K., Evans, R. M. & Schwabe, J. W. (1999) Genes Dev. 13, 3209–3216. 11. Webb, P., Anderson, C. M., Valentine, C., Nguyen, P., Marimuthu, A., West, B. L., Baxter, J. D. & Kushner, P. J. (2000) Mol. Endocrinol. 14, 1976–1985. 12. Marimuthu, A., Feng, W., Tagami, T., Nguyen, H., Jameson, J. L., Fletterick, R. J., Baxter, J. D. & West, B. L. (2002) Mol. Endocrinol. 16, 271–286. 13. Xu, H. E., Stanley, T. B., Montana, V. G., Lambert, M. H., Shearer, B. G., Cobb, J. E., McKee, D. D., Galardi, C. M., Plunket, K. D., Nolte, R. T., et al. (2002) Nature 415, 813–817. PNAS 兩 May 3, 2005 兩 vol. 102 兩 no. 18 兩 6271
BIOCHEMISTRY
neither cofactor works by displacing an unknown endogenous yeast corepressor that interacts with the TR兾CoR-binding surface on the ligand-binding domain. In addition, two TR mutations (451X and E457K) that impair p160 binding in vitro and p160 coactivation in mammalian cells (33), and in our hands in yeast (not shown), did not affect coactivation by E1A or N-CoR in this system, confirming that this effect does not require the classical coactivator-binding surface. TR I280K also exhibited impairment in E1A1– 82 and N-CoR binding in yeast two-hybrid assays (Fig. 4B) and in GST pulldowns (not shown), confirming that it inhibits E1A and N-CoR binding directly. Thus, E1A and N-CoRC enhance TR activity by binding the TR CoR-binding surface and not the classical coactivator-binding surface.
14. Feng, X., Jiang, Y., Meltzer, P. & Yen, P. M. (2000) Mol. Endocrinol. 14, 947–955. 15. Tagami, T., Madison, L. D., Nagaya, T. & Jameson, J. L. (1997) Mol. Cell. Biol. 17, 2642–2648. 16. Tagami, T., Park, Y. & Jameson, J. L. (1999) J. Biol. Chem. 274, 22345–53. 17. Berghagen, H., Ragnhildstveit, E., Krogsrud, K., Thuestad, G., Apriletti, J. & Saatcioglu, F. (2002) J. Biol. Chem. 277, 49517–49522. 18. Shibusawa, N., Hollenberg, A. N. & Wondisford, F. E. (2003) J. Biol. Chem. 278, 732–738. 19. Satoh, T., Monden, T., Ishizuka, T., Mitsuhashi, T., Yamada, M. & Mori, M. (1999) Mol. Cell Endocrinol. 154, 137–149. 20. Li, X., Kimbrel, E. A., Kenan, D. J. & McDonnell, D. P. (2002) Mol. Endocrinol. 16, 1482–1491. 21. Hollenberg, A. N., Monden, T., Madura, J. P., Lee, K. & Wondisford, F. E. (1996) J. Biol. Chem. 271, 28516–28520. 22. Feng, X., Jiang, Y., Meltzer, P. & Yen, P. M. (2001) J. Biol. Chem. 276, 15066–15072. 23. Ordentlich, P., Downes, M., Xie, W., Genin, A., Spinner, N. B. & Evans, R. M. (1999) Proc. Natl. Acad. Sci. USA 96, 2639–2644. 24. Walfish, P. G., Yoganathan, T., Yang, Y. F., Hong, H., Butt, T. R. & Stallcup, M. R. (1997) Proc. Natl. Acad. Sci. USA 94, 3697–3702. 25. Anafi, M., Yang, Y. F., Barlev, N. A., Govindan, M. V., Berger, S. L., Butt, T. R. & Walfish, P. G. (2000) Mol. Endocrinol. 14, 718–732. 26. Meng, X., Yang, Y. F., Cao, X., Govindan, M. V., Shuen, M., Hollenberg, A. N., Mymryk, J. S. & Walfish, P. G. (2003) Mol. Endocrinol. 17, 1095–1105.
6272 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0501491102
27. 28. 29. 30.
31.
32. 33.
34. 35. 36. 37. 38. 39.
Mymryk, J. S. & Smith, M. M. (1997) Biochem. Cell Biol. 75, 95–102. Frisch, S. M. & Mymryk, J. S. (2002) Nat. Rev. Mol. Cell. Biol. 3, 441–452. Peterson, C. L. & Tamkun, J. W. (1995) Trends Biochem. Sci. 20, 143–146. Sterner, D. E., Grant, P. A., Roberts, S. M., Duggan, L. J., Belotserkovskaya, R., Pacella, L. A., Winston, F., Workman, J. L. & Berger, S. L. (1999) Mol. Cell. Biol. 19, 86–98. Grant, P. A., Duggan, L., Cote, J., Roberts, S. M., Brownell, J. E., Candau, R., Ohba, R., Owen-Hughes, T., Allis, C. D., Winston, F., et al. (1997) Genes Dev. 11, 1640–1650. Roberts, S. M. & Winston, F. (1997) Genetics 147, 451–465. Feng, W., Ribeiro, R. C. J., Wagner, R. L., Nguyen, H., Apriletti, J. W., Fletterick, R. J., Baxter, J. D., Kushner, P. J. & West, B. L. (1998) Science 280, 1747–1749. Wahlstrom, G. M., Vennstrom, B. & Bolin, M. B. (1999) Mol. Endocrinol. 13, 1119–1129. Cohen, R. N., Putney, A., Wondisford, F. E. & Hollenberg, A. N. (2000) Mol. Endocrinol. 14, 900–914. Cohen, R. N., Brzostek, S., Kim, B., Chorev, M., Wondisford, F. E. & Hollenberg, A. N. (2001) Mol. Endocrinol. 15, 1049–1061. Cavailles, V., Dauvois, S., L’Horset, F., Lopez, G., Hoare, S., Kushner, P. J. & Parker, M. G. (1995) EMBO J. 14, 3741–3751. Lee, Y. K. & Moore, D. D. (2002) J. Biol. Chem. 277, 2463–2467. Zhang, H., Thomsen, J. S., Johansson, L., Gustafsson, J. A. & Treuter, E. (2000) J. Biol. Chem. 275, 39855–39859.
Meng et al.