Nuclear Receptor-Binding Sites of Coactivators Glucocorticoid ...

Nuclear Receptor-Binding Sites of Coactivators Glucocorticoid Receptor Interacting Protein 1 (GRIP1) and Steroid Receptor Coactivator 1 (SRC-1): Multiple Motifs with Different Binding Specificities

Xiu Fen Ding*, Carol M. Anderson*, Han Ma, Heng Hong, Rosalie M. Uht, Peter J. Kushner*, and Michael R. Stallcup* Departments of Pathology and of Biochemistry and Molecular Biology (X.F.D., H.M., H.H., M.R.S.) University of Southern California Los Angeles, California 90033 Metabolic Research Unit (C.M.A., R.M.U., P.J.K.) University of California San Francisco, California 94143

but specific NRs display both coactivator and NR Box preferences that may contribute to the specificity of hormonal responses. (Molecular Endocrinology 12: 302–313, 1998)

The activity of the AF-2 transcriptional activation function of nuclear receptors (NR) is mediated by the partially homologous transcriptional coactivators, glucocorticoid receptor interacting protein 1 (GRIP1)/transcriptional intermediary factor 2 (TIF2) and steroid receptor coactivator 1 (SRC-1). GRIP1 and SRC-1 bound nine different NRs and exhibited similar, but not identical, NR binding preferences. The most striking difference was seen with the androgen receptor, which bound well to GRIP1 but poorly to SRC-1. GRIP1 and SRC-1 contain three copies of the NR binding motif LXXLL (called an NR Box) in their central regions. Mutation of both NR Box II and NR Box III in GRIP1 almost completely eliminated functional and binding interactions with NRs, indicating that these two sites are crucial for most of GRIP1’s NR binding activity. Interactions of GRIP1 with the estrogen receptor were more strongly affected by mutations in NR Box II, whereas interactions with the androgen receptor and glucocorticoid receptor were more strongly affected by NR Box III mutations. One isoform of SRC-1 has an additional NR Box (NR Box IV) at its extreme C terminus with an NR-binding preference somewhat different from that of the central NRbinding domain of SRC-1. GRIP1 has no NR Box in its C-terminal region and therefore no C-terminal NR-binding function. In summary, GRIP1 and SRC-1 have overlapping NR-binding preferences,

INTRODUCTION Gene transcription is activated by transcriptional activator proteins that bind to specific regulatory DNA sequences, called enhancer elements, which are associated with the transcriptional promoters of genes (1–3). In the past few years the mechanism by which the binding of such proteins to DNA activates transcription has been further elucidated by the discovery of a class of proteins known as transcriptional coactivators. Although they generally do not bind DNA directly, these coactivators associate with DNA-bound activator proteins and are required for the transcriptional activation process to occur (4–7). Coactivators mediate the effects of transcriptional activator proteins, presumably by helping to recruit a complex of RNA polymerase II and associated basal transcription factors (a preinitiation complex) to the promoter or by activating a preinitiation complex that has already been assembled on the promoter. Some coactivators, such as the two related proteins CREB (cAMP-response element binding protein)-binding protein (CBP) and p300, mediate the effects of diverse groups of transcription factors (8–12); other coactivators, like the recently discovered class of coactivators for the nu-

0888-8809/98/$3.00/0 Molecular Endocrinology Copyright © 1998 by The Endocrine Society

302

Nuclear Receptor-Binding Sites in p160 Coactivators

clear hormone receptors (NRs) (13, 14), are functionally more specific (10, 15). The NRs are a class of hormone-regulated transcriptional activator proteins that include the receptors for the five steroid hormones, thyroid hormone, retinoids (vitamin A), and vitamin D, among others (16–19). These hormone receptors contain two transcriptional activation domains (ADs) (16, 20–22). AF-2, located within the hormone-binding domain (HBD) near the C terminus of the proteins, is highly conserved among the NRs mentioned above. AF-1 is an N-terminal AD that is not conserved in sequence. Recent studies have uncovered a large number of mammalian proteins that interact with the AF-2 transcriptional ADs of NRs (13, 14). These proteins bind to NRs that are occupied by the appropriate agonist, but generally not to ligand-free NRs or NRs occupied by antagonists (15, 23–25). Among AF-2-binding proteins are a related group of mammalian proteins of approximately 160 kDa in size, called the p160 coactivators, which have been demonstrated to serve as transcriptional coactivators for NRs in mammalian cells and in yeast (9, 15, 24–26). Steroid receptor coactivator-1a (SRC1a) (9) is another isoform of SRC-1 (15), the first NR coactivator discovered; SRC-1a contains an N-terminal PAS domain that is lacking in SRC-1 (9). Glucocorticoid receptor interacting protein 1 (GRIP1) (25, 26) and transcriptional intermediary factor 2 (TIF2) (24) are nearly identical proteins found in mouse and human cells, respectively, which share 43% sequence identity with SRC-1a (25). Several other proteins, including receptor-interacting protein 140 (RIP140) (23) and TIF1 (27, 28), which are unrelated in sequence to GRIP1 and SRC-1, also bind to NR AF-2 domains, but so far their possible roles in transcriptional activation by NRs are unclear. Transcriptional activation by NRs requires an isoform of either SRC-1 or GRIP1 as a coactivator (10, 25, 26). In addition, the participation of another coactivator, CBP or its partial homolog p300, is required; SRC-1a and GRIP1, as well as the NRs, can bind directly to CBP and p300 (9, 11). The ability of various transcription factors, coactivators, and corepressors to interact with CBP/p300 suggested that these complexes may serve as a mechanism for integrating the input from multiple signaling pathways (9). In the study reported here we characterized further the NR-binding domains and NR-binding preferences of GRIP1 and SRC-1a. Initial studies of SRC-1a and GRIP1/TIF2 suggested that these partial homologs have similar and possibly overlapping activities as NR coactivators (9, 15, 24, 25). Therefore, we looked for possible differences in the NR-binding preferences of GRIP1 and SRC-1a by conducting a comprehensive analysis of the ability of GRIP1 and SRC-1a to interact with a diverse group of NRs that includes the five steroid hormone receptors (NR class I) and four class II receptors (including representatives of the thyroid hormone, retinoid, and vitamin D receptors). Previously, it was shown that SRC-1a has two separable

303

NR-binding domains, one in the central region of the polypeptide chain and one in the C-terminal region (11). We therefore tested whether these two binding domains of SRC-1a bound the same or different sets of NRs. Like SRC-1a, GRIP1 also has an NR-binding domain in the central region of its polypeptide chain (24, 25); we asked whether GRIP1, like SRC-1a, also has an NR-binding domain in its C-terminal region. Recently a simple motif, LXXLLL (where L 5 leucine and X 5 any amino acid) and called an NR Box, was shown to be necessary and sufficient for binding of some NRs by TIF1 and RIP140 (28). Although these two proteins are unrelated in sequence to GRIP1 and SRC-1, the discovery of NR Boxes in other NR AF-2binding proteins prompted us to search for similar motifs in GRIP1 and SRC-1a and to make mutations in them to determine whether they are responsible for interactions with NR HBDs. Our study found the NRbinding domains of GRIP1 and SRC-1a to be surprisingly complex; rather than a simple, single binding site, these coactivators employ multiple NR- binding motifs with overlapping but different NR-binding preferences.

RESULTS A Comparison of the NR-Binding Specificities of Full-Length SRC-1a and GRIP1 Full-length SRC-1a and GRIP1 were tested for their relative interaction specificities with the following NRs in a yeast two-hybrid assay: class I (steroid hormone) receptors, including glucocorticoid receptor (GR), estrogen receptor (ER), androgen receptor (AR), mineralocorticoid receptor (MR), and progesterone receptor (PR); and class II receptors, including vitamin D receptor (VDR), retinoic acid receptor (RAR) a, retinoid X receptor (RXR) a, and thyroid hormone receptor (TR) b1. The coactivators were expressed as fusion proteins with the GAL4 transcriptional AD; and C-terminal NR fragments containing the complete HBD and most of the hinge region (which separates the NR DNAbinding domain from the HBD) were expressed as fusion proteins with the GAL4 DNA-binding domain (DBD). In these yeast two-hybrid assays, binding of the coactivators with the NR HBDs leads to expression of a b-galactosidase (b-gal) gene controlled by a GAL4 enhancer element. Both full-length coactivators exhibited binding to all of the NR HBDs tested in the presence of a suitable agonist for each NR (Fig. 1A). RARa and RXRa, but none of the other NRs tested, also exhibited some coactivator binding in the absence of ligand (data not shown) (25). As controls, each GAL4 DBD-NR HBD fusion protein was coexpressed with the GAL4 AD that lacked an attached coactivator. In these tests there was little or no activity in the presence (Fig. 1A) or absence (not shown) of the appropriate NR ligand. The GAL4 AD-GRIP1 fusion protein was also inactive by itself (25, 26).

MOL ENDO · 1998 304

Vol 12 No. 2

ious GAL4 DBD-NR HBD fusion proteins were very similar (Fig. 1B). Note that in a previous study (25) the fusion proteins for the RAR, RXR, and TR HBDs were expressed at higher levels than the fusion proteins for the steroid receptors and VDR because a different expression vector was used for the RAR, RXR, and TR fusion proteins. Our current studies have indicated that the overexpression of the RARa fusion protein in the yeast two-hybrid assays of the previous study (25) caused an artifactual reduction (presumably due to a squelching effect) in the level of b-gal activity observed in the presence of retinoic acid; as a result, retinoic acid appeared to reduce the level of interaction between GRIP1 and RARa in the yeast two-hybrid system. In the current study, with all NR fusion proteins expressed at the same low levels, the interaction between GRIP1 and all NR HBDs, including RARa, was enhanced by agonist binding (data not shown). NR Binding Activity and Preferences of the Central and C-Terminal Domains of SRC-1a and GRIP1

Fig. 1. Interaction of SRC-1a and GRIP1 with NR HBDs in the Yeast Two-Hybrid System Yeast containing an integrated b-galactosidase reporter gene controlled by a GAL4-binding site were transformed with expression plasmids coding for a GAL4 DBD-NR HBD fusion protein and a second fusion protein of GAL4 AD with full-length SRC-1a or GRIP1 (see Fig. 2, A and C). After transformation, yeast were grown with the appropriate hormone for 15 h, and the resulting cell extracts were assayed for b-gal activity. Hormones and concentrations used are specified in Materials and Methods. A, b-gal activity from yeast two-hybrid assays represents the level of interaction between the indicated NR HBD and SRC-1a (left panel) or GRIP1 (right panel) (striped bars). Activity of the NR fusion protein in the absence of a coactivator is also shown as a control (black bars). B, Immunoblots, performed with an antibody against GAL4 DBD, were used to determine relative expression levels of the various GAL4 DBD-NR HBD fusion proteins. Top panel, Expression in yeast containing SRC-1a fused with GAL4 AD; bottom panel, expression in yeast containing GRIP1 fused with GAL4 AD; NO, extract from yeast lacking a GAL4 DBD-NR HBD fusion protein.

While all of the NRs interacted with SRC-1a and GRIP1, there were some differences in the relative NR preferences of the two coactivators. Most notable was that AR, relative to the other NRs, bound weakly to SRC-1a but bound strongly to GRIP1. Immunoblot experiments using an antibody against the GAL4 DBD indicated that the relative expression levels of the var-

SRC-1a has two functionally separable NR-binding domains, one in the central region of the polypeptide and the other at the C terminus (11). We examined the NR-binding preference of each domain in the yeast two-hybrid system; coactivator fragments were fused with the GAL4 AD, and NR HBDs were fused with the GAL4 DBD (Fig. 2A). The central and C-terminal domains of SRC-1a had overlapping but nonidentical NR-binding specificities (Fig. 2B). ER, PR, VDR, RARa, and TRb1 HBDs bound more strongly to the central SRC-1a domain than to the C-terminal domain. In contrast, GR and AR HBDs bound preferentially to the C-terminal domain; in fact, there was almost no AR binding to the central domain. MR and RXRa HBDs bound with approximately equal strength to both SRC-1a domains. The presence of an NR-binding domain in the central region of the GRIP1 polypeptide chain has been established previously (24, 25). The NR-binding strength and preference of the central GRIP1 fragment (GRIP132021121) were essentially the same as those of full-length GRIP1 (Fig. 2D). This region shares partial homology with the central SRC-1a NR-binding domain (25). Since GRIP1 and SRC-1a also have extensive partial sequence homology in the C-terminal region (25), we tested whether GRIP1 also has a C-terminal NR-binding domain comparable to that of the SRC-1a C-terminal region described above. Neither GRIP1112121462 (data not shown) nor a longer Cterminal GRIP1 fragment GRIP177521462 (Fig. 2D) had any NR-binding activity. The longer C-terminal GRIP1 fragment contains the CBP/p300-binding site (Fig. 2C), which was previously mapped in SRC-1a (11). This allowed us to perform a control experiment to demonstrate that GRIP177521462 bound strongly to a C-terminal p300 fragment (amino acids 1856–2414) in the yeast two-hybrid assay (Fig. 2D) and was thus


305

stably produced and structurally intact. Thus, GRIP1 has no C-terminal NR-binding function comparable to that of SRC-1a. Identification of Two NR Box Motifs in GRIP1 and Their Relative Contributions to the NR-Binding Strength and Preference of GRIP1 in Yeast Two- Hybrid Assays When the NR Box motif LXXLLL was initially reported as an NR-binding sequence in TIF1 and RIP140 (28), which are not related to GRIP1 and SRC-1, we looked for similar motifs in GRIP1. While there were no perfect matches for this sequence in GRIP1, there were many sequences that partially matched the TIF1/RIP140 consensus sequence. We decided to focus on two motifs, NR Box II1 with the first leucine at position 690 and NR Box III with the first leucine at position 745, because they alone met the following three criteria: First, they conformed to the sequence FLXXLL (where F 5 hydrophobic, L 5 leucine, and X 5 any amino acid), found in the TIF1 and RIP140 NR Boxes (Fig. 3A). Second, they were highly conserved in GRIP1 and SRC-1a; NR Box II and Box III are located in regions where at least 10 consecutive amino acids are conserved between GRIP1 and SRC-1 (Fig. 3A). Third, they were located within the minimum NR-binding domains defined for SRC-1a (11) and GRIP1 (25) (our unpublished data). Another possible NR Box motif in this region (NR Box I at position 641) did not fit these criteria well and was therefore not analyzed in this study. To investigate whether NR Box II and NR Box III were important for NR binding by GRIP1, we changed the last two leucines in each of these FLXXLL motifs to alanines, resulting in FLXXAA (Fig. 3B). Analogous substitutions were previously shown to eliminate NR binding by the TIF1 NR Box motif (28). Binding of full-length wild-type and mutant GRIP1 species to various NR HBDs was tested in yeast two-hybrid assays in the presence of the appropriate agonist (Fig. 3C). Altering leucines L693 and L694 to alanines in NR Box II of GRIP1 caused a drastic loss (.90%) of binding to ER, a severe but less dramatic reduction in binding (70–80%) to AR, and #50% loss of binding to the other NRs. In contrast, substituting alanines for

Fig. 2. Interaction of Central and C-Terminal Fragments of SRC-1a and GRIP1 with Various NR HBDs in Yeast TwoHybrid Assays A, The diagram shows the fusion proteins used in the yeast two-hybrid system. NR HBDs were fused with the GAL4 DBD (white box). hSRC-1a fragments were fused with the GAL4 AD (ovals). Numbers associated with coactivator fragments indicate amino acid positions. Vertical black bars with Roman numerals indicate NR Box motifs (FLXXLL consensus); the associated amino acid number corresponds to the first L in the consensus motif. Black regions indicate the CBP/p300 binding site. B, b-gal activity from yeast two-hybrid assays represents the level of interaction between hSRC-1a frag-

1 To promote conformity and clarity in the NR coactivator field, we have adopted the NR Box numbering system suggested by Drs. P. Chambon and H. Gronemeyer (28a).

ments and the specific NR HBD indicated on the x-axis; activity is expressed as percent of the activity obtained for the same NR HBD interacting with full-length hSRC-1a (Fig. 1A). C, The diagram shows the fusion proteins used in the yeast two- hybrid assays for the interaction of NR HBDs with GRIP1 fragments. Symbols and amino acid numbering are as in panel A. D, b-gal activity from yeast two-hybrid assays represents the level of interaction between GRIP1 fragments and various NR HBDs, as in panel C.

MOL ENDO · 1998 306

Vol 12 No. 2

MR, PR, and RXRa (Fig. 3C). None of these mutations caused any loss of binding of GRIP1 to p300 (Fig. 3C), indicating that the mutations did not compromise the stability or overall structural integrity of GRIP1. These mutational studies suggest that NR Box II and NR Box III can account for most of the NR-binding activity of GRIP1, with each motif contributing distinct but overlapping specificities for the various NRs. The GRIP132021121 fragment containing NR Boxes I, II, and III bound all the NRs as efficiently as full-length GRIP1 (Fig. 2D). GRIP173021121, which contained only NR Box III, bound most NRs as efficiently as full-length GRIP1 but bound ER very weakly (data not shown). Thus, both this deletion analysis and the NR Box point mutations discussed above (Fig. 3) indicated that NR Box II is primarily responsible for ER interactions with GRIP1. A Fourth NR Box at the C Terminus of SRC-1a, but Not GRIP1

Fig. 3. Yeast Two-Hybrid Assays of NR HBDs Binding to GRIP1 Containing Mutations in NR Box Motifs A, NR Box motifs from TIF1, RIP140, GRIP1, and hSRC-1a, with the consensus sequence below; * indicates C terminus. B, Mutant NR Box II and NR Box III motifs of GRIP1 with the amino acid substitutions shown in larger type. C, Yeast two-hybrid assays were used to compare binding of NR HBDs to full-length GRIP1 with a wild-type sequence or mutations in NR Box II, NR Box III, or both. The GAL4 DBD-NR HBD fusion proteins and the GAL4 AD-GRIP1.FL fusion proteins are as in Fig. 2C. Yeast were incubated with the appropriate hormone before preparing cell extracts for b-gal assays. Amino acid substitutions: NRBIIm, L693A1L694A; NRBIIIm, L748A1L749A. b-gal activity is expressed as percent of the activity obtained for the same NR HBD interacting with wild-type GRIP1 (Fig. 1A).

leucines L748 and L749 in NR Box III reduced binding by more than 95% to AR, reduced binding by more than 70% to GR and ER, but had little if any effect on binding to the remaining NRs. Altering the two Nterminal leucines L744 and L745 to alanines in NR Box III produced virtually the same phenotype as altering L748 and L749 to alanines (data not shown). When both NR Box II and NR Box III were altered in the same GRIP1 molecule, binding to most of the NRs was almost completely eliminated; however, this mutant GRIP1 retained about 10–30% of wild type binding to

We investigated why SRC-1a, but not GRIP1, has a C-terminal NR-binding function. NR Box motifs were responsible for NR binding in the central binding domain of GRIP1 and, by inference from sequence homologies, SRC-1a. This finding prompted us to focus on an additional motif that matched the NR Box consensus sequence and was located within the defined C-terminal NR-binding region of SRC-1a (Figs. 3A and 4A). This fourth motif, which we call NR Box IV, was located at the extreme C terminus of SRC-1a in a 50-amino acid region that has no homologous counterpart in GRIP1. To test the importance of NR Box IV for the NR-binding function of the C-terminal SRC-1a domain, we first demonstrated that the C-terminal 206 amino acids of SRC-1a (SRC-1a123621441), as well as the longer C-terminal fragment SRC-1a78921441, was sufficient for strong NR binding (Fig. 4). Deletion of the extreme eight C-terminal amino acids, including NR Box IV, from the SRC-1a78921441 fragment essentially eliminated NR binding but left the ability to bind p300 intact (Fig. 4B). Thus, NR Box IV is essential for NR binding by the C-terminal NR-binding domain of SRC1a. As a corollary to this finding, we conclude that the inability of the GRIP1 C-terminal region to bind NR is due to the absence of the NR Box IV motif in this region of GRIP1. Effects of NR Box II and III Mutations and of Competing NR Box II and III Peptides on Binding of ER, GR, and TR to GRIP1 in Vitro The yeast two-hybrid system data indicated that NR Box II is more important than NR Box III for ER binding but that the opposite is true for GR binding to GRIP1 (Fig. 3). As a further test of this different NR specificity of the two NR Box motifs, binding of GRIP1 with ER was measured in vitro, using glutathione S-transferase (GST) fusion proteins. Full-length GRIP1, wild-type or containing NR Box mutations, was synthesized in vitro


307

Fig. 5. Effect of NR Box Mutations on Interaction of GRIP1 with ER HBD in Vitro Full-length, radiolabeled GRIP1 (wild-type or containing mutations in NR Box sites) was synthesized in vitro and then incubated with Sepharose beads containing immobilized GST, GST-ERHBD (with or without bound estradiol, E2), or GST-CBP204122240; bound GRIP1 was eluted and observed by SDS-PAGE and autoradiography. The GST fusion proteins and the affinity binding assays are described in Materials and Methods; the amino acid substitutions in NR Box II (NRBIIm) and NR Box III (NRBIIIm) are described in Fig. 3.

Fig. 4. Role of NR Box IV in NR Binding by the C-Terminal SRC-1a Domain A, Diagram of GAL4 fusion proteins used to assess interaction of NR HBDs with hSRC-1a fragments in yeast twohybrid assays. Symbols and amino acid numbering are as in Fig. 2A. B, Yeast two-hybrid assays for interaction of various NR HBDs or the C-terminal p300185622414 fragment with C-terminal fragments of SRC-1a containing or lacking NR Box IV.

and tested for specific, hormone-dependent binding to an immobilized fusion protein consisting of GST and the ER HBD (GST-ERHBD). Wild-type GRIP1 bound specifically to GST-ERHBD in the presence, but not in the absence, of estradiol (Fig. 5). A mutation in NR Box II caused a severe but not complete loss of ER binding, while the analogous mutation in NR Box III caused a more modest reduction in binding. Mutation of both NR Box II and NR Box III eliminated ER binding. In contrast, the wild-type GRIP1 and all three mutants bound at similar levels to a GST-CBP fusion protein, demonstrating that the GRIP1 mutations had specific effects on ER binding rather than general effects on the structural integrity of the entire GRIP1 protein. Peptides of 13 amino acids representing NR Box II or NR Box III were next used as competitors in similar NR-GRIP1-binding assays in vitro. ER HBD translated in vitro bound to a GST-GRIP173021121 fusion protein in an estradiol-dependent manner (Fig. 6, top panel). When NR Box peptides were added as competitors to this reaction, the NR Box II peptide was a much more effective competitor than the NR Box III peptide. Similarly, for specific binding of hormone-activated TRb1

to GST-GRIP1, the NR Box II peptide was a more effective competitor than the NR Box III peptide (Fig. 6, bottom panel), although the difference in competitor activity of the two peptides was less dramatic than with ER. In contrast, when a labeled GR DBD-HBD fragment was bound to GST-GRIP1 in the presence of dexamethasone, both peptides were relatively weak competitors, but the NR Box III peptide was a better competitor than the NR Box II peptide (Fig. 6, middle panel). Densitometric analysis of the autoradiogram indicated that 5 mg of NR Box III peptide gave 50% competition, while there was no competition by 5 mg of NR Box II peptide. Several controls demonstrated that the inhibitory effect of these peptides were specific for the NR-GRIP1 interaction. Substitution of alanines for the two C-terminal leucines or substitution of glutamic acid for the penultimate leucine in either NR Box peptide eliminated its ability to compete with GRIP1 for binding to ER, indicating the importance of the NR Box motif within these peptides (data not shown). Furthermore, the wild-type NR Box peptides failed to inhibit ER binding to the TATA-binding protein and also failed to inhibit GRIP1 binding to CBP in similar GST pull-down assays (data not shown). The relatively weak effect of the competitor peptides on GR binding to GRIP1 is presumably because GR binds more strongly than ER or TR to the specific GST-GRIP1 fusion protein used. The GST-GRIP1 protein contained only amino acids 730-1121 of GRIP1 and thus contained NR Box III but not NR Box I or II. As shown above (Fig. 3) NR Box II is more important than NR Box III for strong ER binding activity; GR has the opposite NR Box preference, and TR binds well to both NR Boxes.

MOL ENDO · 1998 308

Vol 12 No. 2

Fig. 6. NR Box II and NR Box III Peptides as Competitors for GRIP1 Binding to ER, GR, and TR HBDs in Vitro Radiolabeled ER HBD (top panel), GR DBD-HBD (middle panel), or TRb1 (bottom panel) were synthesized in vitro and incubated with Sepharose-immobilized GST or GST-GRIP173021121 in the presence of the indicated amounts of NR Box II peptide (KHKILHRLLQDSS) or NR Box III peptide (ENALLRYLLDKDD). Where indicated, ER contained bound estradiol, GR contained bound dexamethasone, and TR contained bound T3. Bound NR proteins or fragments were eluted and observed as in Fig. 5.

Fig. 7. Effect of NR Box II and NR Box III Mutations on GRIP1 Coactivator Function in Mammalian Cells HeLa cells were transiently transfected as described in Materials and Methods with the indicated expression vector for an NR or the GAL4-CBP fusion protein, a suitable reporter gene (5 mg) for each NR or for GAL4, and a GRIP1 expression vector (1 mg) coding for full-length GRIP1 with a wild-type (wt) sequence or with mutations in NR Box II (NRBIIm), NR Box III (NRBIIIm), or both. Transfected cells were grown without (white bars) or with (black bars) the appropriate hormone, and cell extracts were tested for reporter gene activity, i.e. luciferase or b-gal enzyme activity. NR Box mutations were as described in Fig. 3B. NR expression vectors and reporter genes were used as follows: A, Full-length hER with the G400V mutation (0.5 mg) and MMTV-ERE-luc; B, MMTV-CAT was activated by the endogenous GR in HeLa cells; C, full-length hTRb1 (0.2 mg) and MMTV-TREpal-luc. D, GAL4 DBD-CBP206022174 fusion protein (1 mg) and (GALRE)5-e1b-luc reporter gene.

Effects of NR Box Mutations on the Coactivator Function of GRIP1 in Mammalian Cells Full-length GRIP1 containing mutations in NR Box II, NR Box III, or both were tested for their ability to serve as coactivators for ER, TRb1, and GR in transiently transfected HeLa cells (Fig. 7). Reporter genes controlled by an appropriate hormone response element were cotransfected with an expression vector for the corresponding nuclear receptor (except that the endogenous GR in HeLa cells was used in Fig. 7B) and an expression vector for mutant or wild-type GRIP1. Cells were then incubated in the presence or absence of hormone, and reporter gene product was measured in the resulting cell extracts by using the appropriate

enzyme assay. Wild-type GRIP1 enhanced the hormone- dependent activities of ER, GR, and TR by several fold (Fig. 7). Mutation of either NR Box II or III caused moderate to severe loss of GRIP1 coactivator function for all three NRs, and simultaneous mutation of NR Boxes II and III almost completely eliminated the ability of GRIP1 to enhance transcriptional activation by the NRs. The NR Box II mutation caused a more severe loss of function than the NR Box III mutation when GRIP1 was tested with ER (Fig. 7A) or TRb1 (Fig. 7C); in contrast, the NR Box III mutation was more severe with GR (Fig. 7B). As a control for the integrity of the mutant GRIP1 species we took advantage of the ability of CBP and GRIP1 to enhance or trigger each other’s transcrip-


tional activation function (P. Webb, P. J. Kushner, et al., submitted). Although the mechanism of coactivation by GRIP1 and SRC-1 is still not understood, these unpublished results suggest that the association of a p160 coactivator (i.e. GRIP1 or SRC-1) and CBP or p300 with a DNA-bound NR at the promoter serves two important functions: first, it serves to recruit the coactivators to the promoter; and second, the interactions between NRs, p160 coactivators, and CBP/ p300 are necessary to trigger the full activity of the transcriptional activation functions of each of these proteins. In our control experiment (Fig. 7D) the GRIP1 mutants were tested for their ability to enhance transcriptional activation by a GAL4 DBD-CBP fusion protein, using a reporter gene with GAL4-binding sites. The NR Box mutations had no effect on the ability of GRIP1 to activate CBP (Fig. 7D), demonstrating that the NR Box II and III mutations selectively affected the NR-binding activity of GRIP1 rather than the overall structural integrity or level of expression of GRIP1 protein.

DISCUSSION NR-Binding Preferences of Full-Length GRIP1 and SRC-1a and Their Central and C-Terminal Fragments GRIP1 and SRC-1 are related proteins that both serve as coactivators for nuclear hormone receptors (9, 15, 24–26). Although the specific protein isoforms expressed by the GRIP1 and SRC-1 genes in various cell types have not been determined, it appears that most tissues express both of these genes (15, 24, 25). The reason for coexpression of both of these coactivator genes in most tissues is not known. The two proteins could have redundant functions or they could serve as coactivators for different or overlapping subsets of nuclear receptors. We found that GRIP1 and SRC-1a both interact with all class I and class II NRs tested and exhibited only minor differences in their relative preferences to bind NRs. Perhaps the only major difference was that AR bound poorly to SRC-1a relative to other NRs, whereas GRIP1 bound AR just as avidly as most other NRs. In terms of NR binding preferences, these results suggest that GRIP1 and SRC-1a may have largely redundant functions. Whether other aspects of the coactivator functions of GRIP1 and SRC-1a (e.g. how they interact with the transcription machinery) are the same or different remains to be studied. Our results indicate that the ability of SRC-1a and GRIP1 to bind a wide range of NRs is accomplished by similar but not identical structural solutions. SRC-1a has two separable NR-binding domains, one in the central region of the polypeptide chain and the other at the C-terminal end (11). Both NR-binding domains of SRC-1a bound a wide range of NRs, but with some differences in NR specificity. The central domain of

309

SRC-1a failed to bind AR and bound GR poorly, while the C-terminal domain bound ER, VDR, RAR, and TR poorly, relative to the central domain. In contrast to SRC-1a, GRIP1 has only the central NR-binding domain; this domain efficiently bound all of the NRs tested and thus has a somewhat broader NR-binding repertoire than the central domain of SRC-1a. Roles of NR Boxes in NR Binding by GRIP1 and SRC-1a Our discovery that NR Box motifs in GRIP1 and SRC-1a are essential for NR binding by these NR coactivators, combined with the previous demonstration that these motifs are responsible for NR binding by TIF1 and RIP140 (28), which are unrelated to GRIP1 and SRC-1, indicates that the NR Box motif is widely used for interaction with the AF-2 transactivation domains of NRs. While this manuscript was in preparation, two other groups reported the involvement of these motifs in NR binding by SRC-1a (10, 29). Torchia et al. (10) designated six sequences in SRC-1a that resembled the NR Box motif and called them leucinecharged domains (LCD). LCD1, LCD2, and LCD3 correspond to the NR Boxes I, II, and III described here. These three NR Boxes are conserved among SRC-1a, GRIP1, and the recently discovered p/CIP, which represents a third genetically distinct member of the p160 coactivator family (10). LCD6 is the same as NR Box IV of SRC-1a. Heery et al. (29) demonstrated that small SRC-1a fragments representing each of the NR Boxes I-IV of SRC-1a bound ER HBD. Torchia et al. (10), using small fragments containing various combinations of NR Boxes I, II, and III, found that NR Box II was the most important motif in the central NR-binding domain of SRC-1a for binding ER and RAR; NR Box IV at the C terminus also bound ER and RAR. The results of these studies on the role of NR Boxes in SRC-1a binding of ER and RAR (10, 29) will be compared, below, with our studies on the role of NR Boxes in GRIP1 binding to a broader spectrum of NRs. It is important to note that, in spite of the extensive partial homology between SRC-1a and GRIP1, their overall homology in the region of the central NR-binding domain is relatively low (25). Therefore, the relative contributions of the individual NR Box motifs to the binding of specific NRs may not be exactly the same in these two coactivators, as exemplified by the nonidentical NR binding preferences exhibited by the central NR-binding domains of GRIP1 and SRC-1a. Mutations in NR Boxes II and III of GRIP1 were used to assess the contributions of these motifs to NRbinding function and specificity in the context of the intact coactivator. Mutations in either NR Box II or NR Box III of GRIP1 caused partial loss of NR-binding activity in vitro and in vivo, as well as the ability of GRIP1 to serve as a NR coactivator in mammalian cells. Simultaneous amino acid substitutions in NR Boxes II and III almost completely eliminated these activities, demonstrating that these two NRs are nec-

MOL ENDO · 1998 310

essary and probably responsible for the majority of the NR-binding activity of GRIP1. The combined data from NR-binding studies in vitro and in vivo and studies to measure coactivator activity in mammalian cells demonstrated that NR Boxes II and III had overlapping but somewhat different NR-binding preferences. GRIP1’s interactions with ER and TR were highly dependent on NR Box II, while interactions with GR and AR depended more on NR Box III. Torchia et al. (10) made similar conclusions about the relative roles of these two NR Boxes for ER binding by SRC-1a. In our experiments, the individual NR Box mutations caused a severe loss of GRIP1’s coactivator activity for TR in mammalian cells but caused only minor losses of TR binding in yeast two-hybrid assays. The relatively minor effects of the individual NR Box mutations on TR binding in the yeast two-hybrid assays may be due to the extreme sensitivity of the yeast two-hybrid system for detecting even weak interactions or to differences between the yeast and mammalian systems. Presence of NR Box IV in C-Terminal Region of SRC-1a but Not GRIP1 Explains GRIP1’s Lack of a C-Terminal NR-Binding Function The recent studies on NR Box motifs in SRC-1a demonstrated that small fragments containing NR Box IV are sufficient for binding ER and RAR (10, 29). Here we demonstrated in the context of the intact C-terminal NR-binding domain of SRC-1a that this motif is essential for binding a wide range of NRs. Thus, NR Box IV is necessary and sufficient for the C-terminal NRbinding function of SRC-1a. These findings provide the basis for understanding why the GRIP1 C-terminal region cannot bind NR: GRIP1 lacks a NR Box motif in its C-terminal region. While GRIP1 and SRC-1a share approximately 43% amino acid sequence identity that extends through most of the length of the polypeptide chain, there are a few regions in which each coactivator contains unique sequences not found in the other. The C-terminal 54 amino acids of SRC-1a, which includes NR Box IV, have no homologous region in GRIP1 (25). Some isoforms of SRC-1 contain the C-terminal NR Box IV motif and some do not (Refs. 9, 11, 15, and the GenBank files cited therein), presumably due to alternative splicing patterns. Our results suggest that SRC-1 isoforms that lack the C-terminal NR Box will not have a C-terminal NR-binding function. Since AR and GR bound poorly to the central NR-binding domain of SRC-1a (Fig. 2B), SRC-1 isoforms lacking NR Box IV would be predicted to bind AR and GR poorly. Thus, if there is differential expression of SRC-1 isoforms containing or lacking NR Box IV in different cell types, it could affect the ability of the cells to support glucocorticoid and androgen responses.

Vol 12 No. 2

Roles of Other Sequences in GRIP1 and SRC-1a That Resemble the NR Box Motif This study has focused on NR Boxes II and III, which are conserved in GRIP1 and SRC-1a, and NR Box IV, found only in SRC-1a. In addition, there are three other sequences that partially or substantially resemble the NR Box consensus FLXXLL and are partially or substantially conserved in GRIP1 and SRC-1a (10, 25). The NR Box I sequence KLLQLLTT begins at amino acid 640 of GRIP1. Heery et al. (29) found that the homologous SRC-1a NR Box I sequence KLVQLLTTT bound ER. However, Torchia et al. (10) found that deletion of NR Box I from a small SRC-1a fragment that also contained NR Box II and/or NR Box III had little if any effect on ER and RAR binding. In addition, our mutational analyses of NR Boxes II and III (Fig. 3) and our results with GRIP1 fragments that lack NR Box I (data not shown) indicated that NR Boxes II and III can account for most of the NR binding activity of GRIP1 and therefore suggest that NR Box I may play, at most, a redundant role in NR binding. Torchia et al. (10) designated two additional sequences that resemble the NR Box consensus sequence as LCD4 and LCD5 but did not test their activity; these motifs are within the CBP/p300-binding region of SRC-1a and GRIP1. While LCD5 is mostly conserved between GRIP1 and SRC-1a, LCD4 in GRIP1 only partially resembles the consensus NR Box sequence. Furthermore, our deletion analysis of the C-terminal domain of SRC-1a demonstrated that the region containing LCD4 and LCD5 is neither necessary nor sufficient for the NR binding activity of the Cterminal domain of SRC-1a (Fig. 4). Rather, as discussed above, NR Box IV (LCD6) is necessary and sufficient for the C-terminal NR-binding activity of SRC-1a. Heery et al. (29) also found LCD4 to be inactive in ER binding, but they did not test LCD5. Together, these results suggest that LCD4 and LCD5 are not involved in NR binding. Why Do NR Coactivators Have Multiple NR-Binding Motifs? Thus, NR binding by the p160 coactivators GRIP1 and SRC-1a is accomplished by a surprisingly complex strategy. Rather than relying on a single structure to bind NRs, GRIP1 and SRC-1a have multiple motifs that contribute differentially to the binding of different NRs. The reason for multiple NR-binding motifs rather than a single one remains to be investigated. Perhaps nature found it difficult to design, through evolution, a single NR-binding sequence that could efficiently bind all of the NRs; instead, perhaps the multiple NR Boxes with overlapping but nonidentical NR preferences solved the problem of how these coactivators could interact with a broad range of NRs. Another possible reason for multiple NR Boxes might be to allow each coactivator molecule to interact with more than one NR monomer. For example, using multiple NR Boxes,


one coactivator molecule could interact with both members of an NR dimer bound to a hormone response element; or when tandem hormone response elements occur in a promoter, one coactivator molecule could conceivably interact with one NR monomer in each of two different NR dimers bound to the tandem enhancer elements. The last scenario could conceivably explain some types of synergism that result when multiple NR dimers bind to tandem hormone response elements.

MATERIALS AND METHODS Plasmids Yeast expression plasmids coding for fusion proteins of GAL4 DBD and the HBDs of GR, ER, PR, AR, MR, and VDR were described previously (25). Similar yeast expression plasmids for GAL4 DBD fused to the HBDs of TRb1, RARa, and RXRa were constructed by inserting EcoRI-BamHI (for TR and RAR) or EcoRI-PstI (for RXR) cDNA fragments into pGBT9 (CLONTECH, Palo Alto, CA) as follows: hTRb1, amino acids 202–461; hRARa, amino acids 155–462; hRXRa, amino acids 200–462. The yeast expression vector coding for fusion proteins of GAL4 AD and full-length GRIP1 (amino acids 5–1462), called pGAD424.GRIP1/FL, was also described previously (25). Expression vectors coding for fusion proteins of GAL4 AD with fragments of GRIP1 or with full-length hSRC-1a or fragments of hSRC-1a were similarly constructed in pGAD424 (CLONTECH). The hSRC-1a sequences and amino acid numbering are according to Spencer et al. (GenBank accession number U90661), with one exception. According to the GenBank file, the C-terminal amino acid sequence of hSRC-1a, including NR Box IV, is LRQQLLTE. However, we sequenced the cDNA clone, which was kindly provided by Dr. Ming-Jer Tsai, and found that the true encoded amino acid sequence is LLQQLLTE, so that the NR Box IV motif is conserved between hSRC-1a and mSRC-1a (9). For expression of full-length GRIP1 in vitro and in mammalian cells, a 4.7-kb EcoRI fragment containing the entire open reading frame of GRIP1 (25) was subcloned into pSG5 (30), which has both T7 and SV40 promoters. The mammalian expression vectors HE0 coding for full-length hER with a G400V mutation (30) and hTRb1 wt (W. Feng and P. J. Kushner, in preparation) have been described. The GAL4 DBDCBP expression vector, encoding a fusion of the GAL4 DBD with amino acids 2060–2174 of CBP, has been described (12). The reporter gene plasmid mouse mammary tumor virus (MMTV)-chloramphenicol acetyltransferase (CAT), containing the MMTV long terminal repeat, has been described (21). In MMTV-thyroid response element (TRE)-luciferase and MMTV-ERE-luciferase (31) the major GREs located between 2190 and 288 of the MMTV long terminal repeat have been deleted and replaced with a single palindromic TRE or estrogen response element (ERE). The reporter (GALRE)5, containing five GAL4 response elements upstream of the e1b promoter, will be described elsewhere (P. Webb and P. J. Kushner, in preparation). Bacterial expression vectors for GST-ERHBD have been described (32). GST-GRIP1 encodes a fusion of GST to amino acids 730-1121 of GRIP1 (26). GST-CBP encodes a fusion of GST to amino acids 2041–2240 of CBP (33). The expression vector for the ER LBD has been described (32). The in vitro transcription/translation vector pSP64/rGR407C (34) contains the SP6 promoter upstream of the rat GRcoding region.

311

Mutations in the NR Box sites of GRIP1 were introduced into the pGAD424.GRIP1/FL and pSG5.GRIP1/FL vectors using the QuikChange Site-Directed Mutagenesis Kit (Stratagene) and verified by sequencing. Protein-Protein Interaction Assays Yeast two-hybrid assays for interaction of coactivators with NR HBDs were performed as described previously (26) except as follows. Yeast culture and b-gal assays were performed and quantified in 96-well microtiter dishes with a Dynatech MR4000 plate reader as described (35) except that o-nitrophenyl b-D-galactopyranoside was used as substrate. Where indicated, yeast cultures were incubated for 15 h before harvest with various hormones at the following concentrations: 10 mM deoxycorticosterone for GR; 100 nM estradiol for ER; 100 nM dihydrotestosterone for AR; 500 nM progesterone for PR; 10 mM corticosterone for MR; 10 mM T3 for TR; 1 mM 1,25-dihydroxy-vitamin D3 for VDR; 10 mM alltrans-retinoic acid for RAR; 10 mM 9-cis-retinoic acid for RXR. Data shown are the mean and SD for the results from three independent yeast transformants and are representative of two or more independent experiments. For in vitro binding assays, proteins were translated in vitro in the presence of [35S]methionine, using the TNT Coupled Reticulocyte Lysate System (Promega, Madison, WI). When GR fragments were translated, separate translations were performed in the presence and absence of 10 mM dexamethasone. GST fusion proteins were prepared as described previously (32). For all of the binding assays except those involving GST-GRIP1, a volume of the bead suspension containing 10 mg total protein was incubated with 1–2 ml 35 S-labeled in vitro translated protein in buffer IPAB-150 (20 mM HEPES, 150 mM KCl, 10 mM MgCl2, 10% glycerol, 1 mM dithiothreitol, 0.1% NP-40, 0.1% Triton X-100, and a protease inhibitor cocktail, pH 7.9) supplemented with 20 mg/ml BSA, in the presence of either 100 nM estradiol or vehicle, for a total volume of 150 ml. After incubation for 90 min at 4 C, beads were washed four times in IPAB-150. Beads and inputlabeled proteins were then subjected to SDS-PAGE and visualized by fluorography. The experiments involving GST-GRIP1 binding to ER, GR, or TR in the presence of various doses of peptides were performed essentially as described (26). Briefly, GST-GRIP1 beads containing 10 mg total protein were incubated with 5 ml in vitro translated protein in buffer NETN (100 mM NaCl, 1 mM EDTA, 20 mM Tris-HCl, pH 8.0, and 0.01% NP-40), in the presence of 100 nM estradiol (for ER), 10 mM dexamethasone (for GR), 100 nM T3 (for TR), or vehicle, and 1 ml peptide (amounts specified in each experiment) or vehicle, for a total volume of 50 ml. After incubation for 1 h at 4 C, beads were washed four times in NETN and subjected to SDS-PAGE and fluorography. NR Box peptides were synthesized by the University of California at San Francisco Biomolecular Resource Center. Immunoblotting Yeast extracts were prepared by a urea-SDS method (36). Electrophoresis and blotting methods were described previously (37). All incubations for blocking and immunostaining were performed at room temperature. Blots were incubated 3 h in blocking solution consisting of TBST (10 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.1% Tween-20) containing 1% BSA and 0.02% sodium azide. The primary antibody, mouse monoclonal RK5C1 (Santa Cruz Biotechnology, Santa Cruz, CA) against GAL4 DBD, was diluted in blocking solution to 0.1 mg/ml and incubated with the blot for 1 h. The blot was washed three times for 15 min each in TBST, blocked again for 30 min in TBST plus BSA plus sodium azide, and incubated for 45 min in TBST containing the secondary antibody, goat anti-mouse IgG coupled to horseradish peroxidase (Pro-

MOL ENDO · 1998 312

mega), at a dilution of 1:2500. After three more 15-min washes in TBST, the Amersham (Arlington Heights, IL) enhanced chemiluminescence (ECL) system was used to visualize the immunostaining pattern. Cell Culture and Transfections HeLa cells were maintained in DME H-16/F-12 Coon’s modified medium without phenol red (Sigma), supplemented with 10% iron-supplemented calf serum (Sigma). Cells were transfected by electroporation, using plasmids indicated in each experiment. In each cuvette 1–2 million cells were suspended in 0.5 ml PBS containing 0.1% glucose and 10 mg/ml Biobrene (Applied Biosystems, Foster City, CA). Cells were electroporated at 0.24 kV, 960 mFarads in a Bio-Rad Gene Pulser apparatus (Bio-Rad Laboratories, Richmond, CA). After electroporation, cells were resuspended in medium, plated into six-well dishes, and treated immediately with vehicle or hormones as follows: 100 nM estradiol for ER, 100 nM dexamethasone for GR, and 100 nM T3 for TR. After 40 h, cells were washed with PBS and lysed with 200 ml lysis buffer (100 mM Tris-HCl, pH 7.8, 0.1% Triton X-100, 1 mM dithiothreitol). Luciferase activity was measured using the Luciferase Assay System (Promega). CAT assays were performed as previously described (38). CAT activities were defined as the increase in counts per unit time, corrected for background CAT activity. Both CAT and luciferase activities were corrected for efficiency of transfection. Transfection efficiency was monitored by cotransfection with a plasmid containing b-gal reporter gene driven by the actin promoter, which was a gift of Michael Garabedian (New York University). b-gal activity was measured using the Galacto-Light Plus chemiluminescent assay (Tropix, Bedford, MA). Luciferase and CAT activities shown are the means and SDs of triplicate wells from a single experiment and are representative of at least three independent experiments.

Acknowledgments We thank Drs. Hinrich Gronemeyer and Pierre Chambon (Universite´ Louis Pasteur, Paris, France) for generously communicating their unpublished results on the NR Boxes of TIF2. In this manuscript we have adopted their NR Box naming and numbering system. We thank Dr. Ming-Jer Tsai (Baylor College of Medicine, Houston, TX) for the SRC-1a cDNA; Dr. David Livingston (Dana-Farber Cancer Institute, Boston, MA) for the p300 cDNA; Dr. John Chrivia (Saint Louis University, St. Louis, MO) for the GAL4-CBP expression vector; Dr. Yoshihiro Nakatani (NIH, Bethesda, MD) for the GST-CBP expression vector; Dr. Keith Yamamoto (University of California, San Francisco, CA) for GR expression vectors; Dr. Brian West (University of California, San Francisco, CA) for hTRb1 expression vector; Ms. Jeanette Shinsako (University of California, San Francisco, CA) for expert technical assistance; and Dr. Paul Webb and Dr. Weijun Feng (University of California, San Francisco, CA) for helpful discussions.

Received August 1, 1997. Revision received October 24, 1997. Accepted November 10, 1997. Address requests for reprints to: Michael R. Stallcup, Department of Pathology, HMR 301, University of Southern California, 2011 Zonal Avenue, Los Angeles, California 90033. E-mail: [email protected]. This work was supported by USPHS Grants DK-43093 (to M.R.S.), DK-51083 (to P.J.K.), and K08 DK-02335 (to R.M.U). from the National Institute of Diabetes and Digestive and Kidney Disease, and by AIBS Grant 562 (to P.J.K.) from the U. S. Army Breast Cancer Research Program. *These authors contributed equally to the work described.

Vol 12 No. 2

REFERENCES 1. Roeder RG 1996 The role of general initiation factors in transcription by RNA polymerase II. Trends Biochem Sci 21:327–335 2. Pugh BF 1996 Mechanisms of transcription complex assembly. Curr Opin Cell Biol 8:303–311 3. Ptashne M, Gann A 1997 Transcriptional activation by recruitment. Nature 386:569–577 4. Drapkin R, Merino A, Reinberg D 1993 Regulation of RNA polymerase II transcription. Curr Opin Cell Biol 5:469–476 5. Tjian R, Maniatis T 1994 Transcriptional activation: a complex puzzle with few easy pieces. Cell 77:5–8 6. Guarente L 1995 Transcriptional coactivators in yeast and beyond. Trends Biochem Sci 20:517–521 7. Verrijzer CP, Tjian R 1996 TAFs mediate transcriptional activation and promoter selectivity. Trends Biochem Sci 21:338–342 8. Kwok RPS, Lundblad JR, Chrivia JC, Richards JP, Ba¨chinger HP, Brennan RG, Roberts SGE, Green MR, Goodman RH 1994 Nuclear protein CBP is a coactivator for the transcription factor CREB. Nature 370:223–229 9. Kamei Y, Xu L, Heinzel T, Torchia J, Kurokawa R, Gloss B, Lin S-C, Heyman RA, Rose DW, Glass CK, Rosenfeld MG 1996 A CBP integrator complex mediates transcriptional activation and AP-1 inhibition by nuclear receptors. Cell 85:403–414 10. Torchia J, Rose DW, Inostroza J, Kamei Y, Westin S, Glass CK, Rosenfeld MG 1997 The transcriptional coactivator p/CIP binds CBP and mediates nuclear-receptor function. Nature 387:677–684 11. Yao T-P, Ku G, Zhou N, Scully R, Livingston DM 1996 The nuclear hormone receptor coactivator SRC-1 is a specific target of p300. Proc Natl Acad Sci USA 93:10626–10631 12. Swope DL, Mueller CL, Chrivia JC 1996 CREB-binding protein activates transcription through multiple domains. J Biol Chem 271:28138–28145 13. Horwitz KB, Jackson TA, Bain DL, Richer JK, Takimoto GS, Tung L 1996 Nuclear receptor coactivators and corepressors. Mol Endocrinol 10:1167–1177 14. Glass CK, Rose DW, Rosenfeld MG 1997 Nuclear receptor coactivators. Curr Opin Cell Biol 9:222–232 15. On˜ate SA, Tsai SY, Tsai M-J, O’Malley BW 1995 Sequence and characterization of a coactivator for the steroid hormone receptor superfamily. Science 270: 1354–1357 16. Evans RM 1988 The steroid and thyroid hormone receptor superfamily. Science 240:889–895 17. Tsai M-J, O’Malley BW 1994 Molecular mechanisms of action of steroid/thyroid receptor superfamily members. Annu Rev Biochem 63:451–486 18. Mangelsdorf DJ, Evans RM 1995 The RXR heterodimers and orphan receptors. Cell 83:841–850 19. Truss M, Beato M 1993 Steroid hormone receptors: interaction with deoxyribonucleic acid and transcription factors. Endocr Rev 14:459–476 20. Kumar V, Green S, Stack G, Berry M, Jin J-R, Chambon P 1987 Functional domains of the human estrogen receptor. Cell 51:941–951 21. Hollenberg SM, Evans RM 1988 Multiple and cooperative transactivation domains of the human glucocorticoid receptor. Cell 55:899–906 22. Durand B, Saunders M, Gaudon C, Roy B, Losson R, Chambon P 1994 Activation function 2 (AF-2) of retinoic acid receptor and 9-cis retinoic acid receptor: presence of a conserved autonomous constitutive activating domain and influence of the nature of the response element on AF-2 activity. EMBO J 13:5370–5382 23. Cavaille`s V, Dauvois S, L’Horset F, Lopez G, Hoare S, Kushner PJ, Parker MG 1995 Nuclear factor RIP140


modulates transcriptional activation by the estrogen receptor. EMBO J 14:3741–3751 24. Voegel JJ, Heine MJS, Zechel C, Chambon P, Gronemeyer H 1996 TIF2, a 160 kDa transcriptional mediator for the ligand-dependent activation function AF-2 of nuclear receptors. EMBO J 15:3667–3675 25. Hong H, Kohli K, Garabedian MJ, Stallcup MR 1997 GRIP1, a transcriptional coactivator for the AF-2 transactivation domain of steroid, thyroid, retinoid, and vitamin D receptors. Mol Cell Biol 17:2735–2744 26. Hong H, Kohli K, Trivedi A, Johnson DL, Stallcup MR 1996 GRIP1, a novel mouse protein that serves as a transcriptional co-activator in yeast for the hormone binding domains of steroid receptors. Proc Natl Acad Sci USA 93:4948–4952 27. Le Douarin B, Zechel C, Garnier J-M, Lutz Y, Tora L, Pierrat B, Heery D, Gronemeyer H, Chambon P, Losson R 1995 The N-terminal part of TIF1, a putative mediator of the ligand-dependent activation function (AF-2) of nuclear receptors, is fused to B-raf in the oncogenic protein T18. EMBO J 14:2020–2033 28. Le Douarin B, Nielsen AL, Garnier J-M, Ichinose H, Jeanmougin F, Losson R, Chambon P 1996 A possible involvement of TIF1a and TIF1b in the epigenetic control of transcription by nuclear receptors. EMBO J 15: 6701–6715 28a.Voegel JJ, Heine MJS, Tini M, Vivat V, Chambon P, Gronemeyer H, The coactivator TIF2 contains three nuclear receptor binding motifs and mediates transactivation through CBP binding-dependent and -independent pathways. EMBO J, in press 29. Heery DM, Kalkhoven E, Hoare S, Parker MG 1997 A signature motif in transcriptional co-activators mediates binding to nuclear receptors. Nature 387:733–736

313

30. Green S, Issemann I, Sheer E 1988 A versatile in vivo and in vitro eukaryotic expression vector for protein engineering. Nucleic Acids Res 16:369 31. Umesono K, Evans RM 1989 Determinants of target gene specificity for steroid/thyroid hormone receptors. Cell 57:1139–1146 32. Webb P, Lopez GN, Uht RM, Kushner PJ 1995 Tamoxifen activation of the estrogen receptor/AP-1 pathway: potential origin for the cell-specific estrogen-like effects of antiestrogens. Mol Endocrinol 9:443–456 33. Yang X-J, Ogryzko VV, Nishikawa J-i, Howard BH, Nakatani Y 1996 A p300/CBP-associated factor that competes with the adenoviral oncoprotein E1A. Nature 382:319–324 34. Godowski PJ, Rusconi S, Miesfeld R, Yamamoto KR 1987 Glucocorticoid receptor mutants that are constitutive activators of transcriptional enhancement. Nature 325:365–368 35. In˜iguez-Lluhı´ JA, Lou DY, Yamamoto KR 1997 Three amino acid substitutions selectively disrupt the activation but not the repression function of the glucocorticoid receptor N terminus. J Biol Chem 272:4149–4156 36. Printen JA, Sprague GF 1994 Protein-protein interactions in the yeast pheromone response pathway: Ste5p interacts with all members of the MAP kinase cascade. Genetics 138:609–619 37. Northrop JP, Gametchu B, Harrison RW, Ringold GM 1985 Characterization of wild type and mutant glucocorticoid receptors from rat hepatoma and mouse lymphoma cells. J Biol Chem 260:6398–6403 38. Webb P, Lopez GN, Greene GL, Baxter JD, Kushner PJ 1992 The limits of the cellular capacity to mediate an estrogen response. Mol Endocrinol 6:157–167