Activation of Estrogen Receptor by S118 Phosphorylation ... - Cell Press

Molecular Cell, Vol. 6, 127–137, July, 2000, Copyright 2000 by Cell Press

Activation of Estrogen Receptor ␣ by S118 Phosphorylation Involves a Ligand-Dependent Interaction with TFIIH and Participation of CDK7 Dongsheng Chen,* Thilo Riedl,† Elinor Washbrook,* Paul E. Pace,* R. Charles Coombes,* Jean-Marc Egly,† and Simak Ali*‡ * CRC Laboratories Department of Cancer Medicine Division of Medicine Imperial College of Science, Technology, and Medicine Hammersmith Hospital Du Cane Road London W12 0NN United Kingdom † Institut de Genetique et de Biologie Moleculaire et Cellulaire Universite´ Louis Pasteur BP 163, F-67404, Illkirch Cedex Strasbourg France

Summary Phosphorylation of the estrogen receptor ␣ (ER␣) N-terminal transcription activation function AF1 at serine 118 (S118) modulates its activity. We show here that human ER␣ is phosphorylated by the TFIIH cyclindependent kinase in a ligand-dependent manner. Furthermore, the efficient phosphorylation of S118 requires a ligand-regulated interaction of TFIIH with AF2, the activation function located in the ligand binding domain (LBD) of ER␣. This interaction involves (1) the integrity of helix 12 of the LBD/AF2 and (2) p62 and XPD, two subunits of the core TFIIH. These findings are suggestive of a novel mechanism by which nuclear receptor activity can be regulated by ligand-dependent recruitment of modifying activities, such as kinases. Introduction Estrogen receptor ␣ (ER␣) is a member of the nuclear receptor superfamily of transcription factors, which are targets for small lipid-soluble molecules such as steroid and thyroid hormones, retinoids, and vitamin D3 (Mangelsdorf et al., 1995; Glass and Rosenfeld, 2000). These receptors are characterized by highly conserved DNA binding (DBD) and ligand binding domains (LBD). Transcription activation by ER␣ is mediated by an N-terminal domain, transcription activation function 1 (AF1), and by transcription activation function 2 (AF2) in the LBD (Gronemeyer, 1991; Tsai and O’Malley, 1994). AF2 activity is dependent on the binding of estrogen and can be blocked by estrogen antagonists. Mutational analysis has defined several important regions within the LBD of ER␣, required for AF2 activity. In particular, an amphi-

‡ To whom correspondence should be addressed (e-mail: simak.ali@

ic.ac.uk).

pathic helix near the C terminus of nuclear receptors is essential for ligand-dependent transcriptional activity. Ligand binding results in the realignment of this helix (H12), enabling interacting proteins to associate. Determination of the LBD structures for a number of other nuclear receptors indicates that ligand-induced H12 realignment is a common feature (Brzozowski et al., 1997; Moras and Gronemeyer, 1998; Shiau et al., 1998; Glass and Rosenfeld, 2000). Recent studies have identified a number of proteins termed coactivators, which associate with nuclear receptors in a ligand-dependent manner to increase transactivation, and include SRC-1/NCoA-1, TIF2/GRIP1/ NCoA-2, pCIP/ACTR/AIB1, and the TRAP/DRIP complex. Interaction between ligand-activated nuclear receptors and coactivators is mediated by a small ␣-helical motif containing the sequence LXXLL (where L denotes leucine and X is any amino acid) in coactivator proteins. The LXXLL motif is crucial, but not sufficient for the interaction, since different nuclear receptors show overlapping but distinct preferences for individual LXXLL motifs, sequences outside this motif also being important. The LXXLL coactivator motifs bind to a hydrophobic groove on the surface of the LBD that is revealed by realignment of H12 upon ligand binding (Freedman, 1999; Glass and Rosenfeld, 2000). In common with other transcription factors, nuclear receptors are phosphoproteins (Shao and Lazar, 1999). We have previously shown that ER␣ is ligand-inducibly phosphorylated (Ali et al., 1993b). Increased phosphorylation of ER␣ was also observed upon addition of the partial antagonist 4-hydroxytamoxifen (OHT), albeit to a lower extent than that seen with 17␤-oestradiol (E2). The pure anti-estrogen ICI 164, 384 increased ER␣ phosphorylation but to a smaller degree than either E2 or OHT. Serine 118 (S118) is a major phosphorylation site within AF1, mutation of which reduces transactivation by ER␣ (Ali et al., 1993b; Le Goff et al., 1994). Furthermore, S118 is phosphorylated by mitogen-activated protein kinase (MAPK) in vitro and in vivo in a ligandindependent manner (Kato et al., 1995). In addition to RNA polymerase, gene expression requires additional factors for transcription. In the case of eukaryotic RNA polymerase II (Pol II), general transcription factors (GTFs) (TFIIA, TFIIB, TFIID, TFIIE, TFIIF, and TFIIH) are required (Reinberg et al., 1998). Transactivation is believed to occur through direct or mediated interactions between transcription factors and components of the basal transcription machinery such as the TATA binding protein (TBP) and TBP-associated factors (TAFs) (which collectively make up TFIID), TFIIB and TFIIH (Roeder, 1996). TFIIH is also implicated in nucleotide excision repair (NER) and contains Cdk-activating kinase (CAK), a cyclin-dependent kinase that phosphorylates the Pol II C-terminal tail domain (CTD). CAK is composed of Cdk7, cyclin H, and MAT1. The known activities of CAK include regulation of cell cycle progression, phosphorylation of Pol II, and phosphorylation of transcription regulators (Nigg, 1996; Ko et al., 1997; Morgan 1997; Rochette-Egly et al., 1997).

Molecular Cell 128

Other cyclin-dependent kinases, Cdk8 and Cdk9 (P-Tefb), also implicated in phosphorylation of Pol II and consequently with events in transcription initiation and elongation, directly or indirectly interact with transcriptional regulators. Cdk9 phosphorylation of Pol II is enhanced by an association between HIV-1 Tat and the cyclin T1 partner of Cdk9 (for review and references, see Garber and Jones, 1999). Cdk8 (SRB10) is part of a Pol II-associated complex known as the SRB or mediator complex (SMCC), which contains several TAFs. Moreover, SMCC is similar, if not identical to the TRAP/ DRIP complex of proteins that associate with ligandactivated nuclear receptors to stimulate their activity (Ito et al., 1999; Rachez et al., 1999). The fact that liganddependent phosphorylation of S118 has been ascribed to a kinase other than MAP kinase (Joel et al. 1998), and the involvement of S118 phosphorylation in increased transcriptional activity, suggested that S118 might be a substrate for transcription regulatory kinases such as Cdk7, Cdk8, or Cdk9. Since another nuclear receptor, retinoic acid receptor ␣ (RAR␣), is phosphorylated by Cdk7 at a site within AF1 (Rochette-Egly et al., 1997), we investigated whether ligand-dependent phosphorylation of S118 is mediated by Cdk7. We report here that Cdk7 overexpression stimulates transcription activation by ER␣ by stimulating phosphorylation of S118 in a ligand-dependent manner. We show that S118 phosphorylation is mediated by an interaction between TFIIH and the ER␣ LBD in vitro and in vivo. These findings represent a novel mechanism by which nuclear receptor AF1 activity can be regulated by the LBD/AF2. Results Cdk7 and MAT1 Stimulate Transcription Activation by Human Estrogen Receptor ␣ We examined the ability of Cdk7 to modulate transcription activation by ER␣. COS-1 cells were transiently transfected with an expression plasmid encoding human ER␣ (HEG0; Figure 1A). ER␣ activated an estrogenresponsive CAT reporter gene ⬎10-fold in the presence of 17␤-estradiol (E2) (Figure 1B, lanes 1and 2). ER␣ activity in the presence of the anti-estrogens, 4-hydroxytamoxifen (OHT) and ICI 182, 780 (ICI) reflected their antagonistic properties (lanes 3 and 4). Cotransfection with Cdk7 resulted in a small increase in transactivation in the presence or absence of E2 (Figure 1B, lanes 5 and 6). MAT1 or cyclin H (Cyc H) also increased transactivation by ER␣ (lanes 10 and 14). Cotransfection of Cdk7 and MAT1, however, strongly stimulated E2-dependent transactivation (Figure 1B, compare lanes 2 and 18). Similar cooperativity was not seen using Cdk7 and Cyc H, or MAT1 and Cyc H (lanes 22 and 26), and the presence of all three subunits of CAK did not stimulate transactivation over levels obtained with Cdk7 and MAT1 (Figure 1B, compare lanes 18 and 30). No increase in transactivation was seen in the presence of ICI (Figure 1B). By contrast, Cdk7 overexpression significantly increased ER␣ activity in the presence of the partial antagonist OHT (Figure 1B, see lanes 3, 7, 19, and 31). The agonistic activity of OHT has been correlated with AF1 activation and AF2 repression (see Chen et al., 1999), suggesting that AF1 may be a target for Cdk7.

Since Cyc H overexpression did not stimulate ER␣ activity over and above that observed with Cdk7 and MAT1, it was omitted from subsequent experiments. Cdk7/MAT1 cotransfection increased transactivation even at concentrations of E2 as low as 10⫺12 M, while there was no indication of a stimulatory effect with a reporter construct lacking the ERE (17M-TATA-CAT) (Figure 1C), demonstrating that the transcriptional stimulation was ER␣ dependent. Further, these data indicate that by increasing the transcription activation ability of ER␣, Cdk7 effectively increases the sensitivity of ER␣ to E2. It is likely that transcription stimulation caused by Cdk7 is due to increased phosphorylation of ER␣. The stimulation of ER␣ activity observed with OHT suggests that Cdk7 acts by modulating AF1 function. We therefore examined the effect of Cdk7 and MAT1 on transactivation by mutants of previously identified ER␣ phosphorylation sites at amino acids 102, 104, 106, 118, and 167 in AF1 (Figure 1A; Ali et al., 1993b; Arnold et al., 1994; Le Goff et al., 1994). Mutation of S106 or of S102, S104, and S106 did not affect the transcription stimulation observed by Cdk7 and MAT1 (Figure 1D, lanes 4–7). Similarly, mutation of S167 did not alter the stimulatory effect of Cdk7/MAT1 (lanes 14 and 15). By contrast, transactivation by mutants in which S118 was replaced by alanine or by glutamic acid was not stimulated by Cdk7 and MAT1 (lanes 8–13), suggesting that S118 is a target for Cdk7 phosphorylation. These results show that Cdk7 can modulate ER␣ activity in a ligand-dependent manner. Furthermore, S118 within transcription activation function AF1 is crucial for this effect. Mutant Cdk7 (Cdk7M; Tassan et al., 1995) when expressed alone, or together with MAT1, did not stimulate ER␣ (Figure 1E, lanes 7–10). Indeed, transactivation by ER␣ was inhibited by Cdk7M (compare lane 2 with lanes 8 and 10), indicating that phosphorylation of ER␣ by Cdk7 is required for efficient transactivation. Human Estrogen Receptor ␣ Is Phosphorylated at Serine 118 In Vivo by Cdk7 HEG0-transfected COS-1 cells were 32P labeled, and extracts were immunoprecipitated using anti-ER␣ monoclonal antibody B10 (Ali et al., 1993a). E2, OHT, and ICI stimulated ER␣ phosphorylation (8-, 4-, and 2-fold, respectively) compared to the no ligand control (Figure 2A, lanes 1–5). A small increase in phosphorylation was also observed in the presence of ICI (2-fold). Cotransfection with Cdk7 and MAT1 stimulated ER␣ phosphorylation in the presence of E2 or OHT (compare lanes 3 and 4 with 15 and 16). These results parallel the transactivation data and indicate that the increased transactivation is indeed due to increased phosphorylation of ER␣. Immunoprecipitation of ER␣ followed by immunoblotting using antisera specific to ER␣ phosphorylated at S118 (␣-P-S118) showed that S118 phosphorylation was stimulated by Cdk7/MAT1 transfection in a liganddependent manner (Figure 2B). Immunoblotting using antibody B10 (HEG0) served as a control for levels of ER␣, and immunoblotting using anti-CDK7 and antiMAT1 demonstrated their overexpression. These results mirror the in vivo 32P-labeling results (Figure 2A) and

Estrogen-Regulated Phosphorylation of ER␣ by TFIIH 129

Figure 1. Transactivation by ER␣ Is Increased by Overexpression of Cdk7 in Concert with MAT1 (A) The amino acid sequence of human ER␣ (HEG0) is schematically represented and shows positions of the functional domains, activation function 1 (AF1), the ligand binding domain (LBD)/activation function-2 (AF2), and the DNA binding domain (DBD). The position of helix 12 is displayed. Also shown are the positions of mapped phosphorylation sites within AF1. Mutant ER␣ proteins lacking the LBD (HE15) or AF1 (HEG19), which were used in this study, are also represented. The amino acid sequence around identified AF1 phosphorylation sites is also shown. The sequence of the peptide used to raise ␣-P-S118 is underlined. (B) COS-1 cells were transfected with the ERE-containing CAT reporter gene, 17M-ERE-TATA-CAT (4 ␮g), and wild-type ER␣ (500 ng of HEG0), without (lanes 1–4) or with 4 ␮g of Cdk7 (lanes 5–8, 17–24, and 29–32), MAT1 (lanes 9–12, 17–20, and 25–32), and cyclin H (Cyc H; lanes 13–16, and 21–32). 17␤-estradiol (E2, 10⫺8 M), 4-hydroxytamoxifen (OHT, 10⫺7 M), or ICI 182, 780 (ICI, 10⫺7 M) was added as indicated. Since the ligands were prepared in ethanol, an equal volume of ethanol was added to the no ligand controls. The results of three independent experiments are displayed in the form of a bar chart. Transcription activation by HEG0 in the presence of E2 (lane 2) was taken as 100%. All other activities are shown relative to this. (C) COS-1 cells were transfected with 4 ␮g of 17M-ERE-TATA-CAT or 17M-TATA-CAT in the presence of HEG0 (500 ng), either without (⫺) or with 4 ␮g of Cdk7 and MAT1 (Cdk7⫹MAT1). Different concentrations of E2, ranging from 10⫺14 M to 10⫺6 M were added, as shown. Ethanol was added to the no ligand (0) controls. The results of three independent experiments are shown. Transcription activation by HEG0 with 17MERE-TATA-CAT in the presence of 10⫺8 M E2 was taken as 100%. (D) COS-1 cells were cotransfected with 4 ␮g of 17M-ERE-TATA-CAT together with 500 ng of pSG5 (lane 1), HEG0 (lanes 2 and 3), HEG0106A (lanes 4 and 5), HEG0102N, 104P, 106A (lanes 6 and 7), HEG0102N, 104P, 106A, 118A (lanes 8 and 9), HEG0118A (lanes 10 and 11), HEG0118E (lanes 12 and 13), or HEG0167A (lanes 14 and 15), and Cdk7 and MAT1 were also cotransfected, as indicated. E2 (10⫺8 M) was added in all cases. The results of three independent experiments are shown. Transcription activation by HEG0 (lane 2) was taken as 100%. (E) COS-1 cells were transfected with 17M-ERE-TATA-CAT and HEG0, as above. Cdk7, Cdk7M and MAT1 were cotransfected, as indicated. E2 (10⫺8 M) was added, as shown. The results of four independent experiments are shown. Transcription activation by HEG0 (lane 2) was taken as 100%.

Molecular Cell 130

Figure 2. Phosphorylation of ER␣ by Cdk7 In Vivo (A) COS-1 cells transfected with 5 ␮g of the parental vector (pSG5) or HEG0, together with Cdk7 and/or MAT1, were labeled with [32P]phosphoric acid in the absence of ligand or in the presence of 10⫺8 M E2, 10⫺7 M OHT, or 10⫺7 M ICI. Whole-cell extracts were immunoprecipitated with monoclonal antibody B10. The immunoprecipitates were resolved by SDS-PAGE, and autoradiography (upper panel) and immunoblotting using B10 (lower panel) were performed. (B) COS-1 cells transfected with pSG5 or HEG0 without or with Cdk7 (lanes 6–13) and MAT1 were treated with ethanol, E2, OHT, ICI, and/ or phorbol myristate acetate (PMA) (100 ng/ml) for 30 min prior to harvesting. The extracts were resolved by SDS-PAGE and immunoblotted using antisera specific to ER␣ phosphorylated at S118 (P-S118) or B10 for detection of total ER␣ (HEG0). Western blotting with antibodies against Cdk7 and MAT1 was performed as controls.

indicate that Cdk7 phosphorylates ER␣ at S118 in a ligand-dependent manner, with consequent increase in transactivation. TFIIH Phosphorylates Human Estrogen Receptor ␣ at Serine 118 We next examined whether free CAK and/or TFIIH-associated CAK can phosphorylate ER␣ in vitro. Purified, recombinant ER␣ was incubated with [␥-32P] ATP in the presence of increasing amounts of purified CAK. Incubation of ER␣ with purified CAK resulted in low-level phosphorylation in the absence of ligand (Figure 3A, lane 6), which was not stimulated by E2 (lanes 3–5). No phosphorylation was observed if ER␣ or CAK was omitted from the incubation (lanes 1 and 2). Similar levels of phosphorylation were observed when purified TFIIH was used, except that addition of E2 resulted in a dramatic increase in phosphorylation (lane 8), but little increase in phosphorylation was seen in the presence of OHT or ICI (lanes 7, 9, and 10). The inability of purified CAK to phosphorylate ER␣ is not due to lower Cdk7 levels as demonstrated by immunoblotting using antiCdk7.

Figure 3. In Vitro Phosphorylation of ER␣ at S118 by TFIIH or CAK (A) Purified ER␣ (0.01 pmoles) was phosphorylated by incubation with 1, 2.5, or 5 ␮l of purified CAK or with 1 ␮l of TFIIH in the absence of ligand, or in the presence of 10⫺7 M E2, 10⫺7 M OHT, or 10⫺7 M ICI. Lane 1 shows CAK (5 ␮l) incubated in the reaction buffer in the absence of ER␣. Lane 2 shows ER␣, incubated in the absence of CAK or TFIIH. Autoradiography (top panel) and immunoblotting using B10 (␣-ER␣; middle panel) and ␣-Cdk7 (bottom panel) were performed to compare levels of each protein. (B) Whole-cell extracts of COS-1 cells transfected with pSG5, HEG0, HE15, or HEG19, immunoprecipitated using monoclonal antibodies B10 (lanes 1–12) or F3 (lanes 13–16) were phosphorylated with highly purified TFIIH. Increasing amounts of TFIIH (0.5 ␮l, lanes 2, 6, 10, and 14; 1 ␮l, lanes 3, 7, 11, and 15; and 2 ␮l, lanes 4, 8, 12, and 16) were added. Phosphorylated proteins were visualized by autoradiography (upper panel) and immunoblotting with anti-ER␣ B10 (lanes 1–12) or F3 (for HEG19; lanes 13–16), as appropriate (lower panel). Lanes 1, 5, 9, and 13 were incubated in the absence of TFIIH. E2 (10⫺7 M) was present in all samples throughout the procedure. (C) Mutants of HEG0 in which serine residues corresponding to known phosphorylation sites within AF1 were replaced by non-phosphorylatable residues were phosphorylated using 1 ␮l of TFIIH. Autoradiography (upper panel) and immunoblotting results using B10 (lower panel) are shown. E2 (10⫺7 M) was present in all samples throughout the procedure.

In vitro phosphorylation of HEG0, HE15 (lacking the LBD; Figure 1A), and HEG19 (lacking AF1) demonstrated that full-length ER␣ (HEG0) was phosphorylated by TFIIH in a dose-dependent manner in the presence of E2 (Figure 3B, lanes 5–8), whereas HE15 was phosphorylated significantly less efficiently than HEG0 (lanes 9–12). HEG19 was not detectably phosphorylated by TFIIH (lanes 13–16). Mutation of S102, S104, S106, or S167 did not affect ER␣ phosphorylation (Figure 3C, lanes 2, 3, and 5). HEG0118A, however, was not phosphorylated by TFIIH, indicating that S118 is the substrate for phosphorylation by TFIIH.


Figure 4. Association of In Vitro Synthesized ER␣ with TFIIH (A) In vitro synthesized HEG0 was incubated with purified TFIIH, followed by immunoprecipitation using antibodies against p62 and XPB subunits of TFIIH. The proteins were resolved by SDS-PAGE and visualized by autoradiography. Controls involved immunoprecipitation in the absence of TFIIH. E2 was present at a concentration of 10⫺7 M. The input lane represents 10% of the total volume of the lysate used in each immunoprecipitation. (B) [35S]-labeled HEG0, HE15, and HEG19, synthesized by in vitro translation, were incubated with TFIIH in the presence of 10⫺7 M E2 and immunoprecipitated using antibodies against p62, Cdk7, and the FLAG epitope (M2). The input lane represents 10% of the total volume of the lysate used in each immunoprecipitation. (C) [35S]-labeled HEG0 was immunoprecipitated using anti-p62 or B10 in the presence (lanes 2, 4, 6, and 8) or absence of TFIIH. E2, OHT, or ICI were present at a concentration of 10⫺7 M. The input lane represents 10% of the total volume of the lysate used in each immunoprecipitation. (D) Purified ER␣ was phosphorylated by incubation with purified TFIIH or purified CAK and immunoprecipitation using anti-Cdk7 (lanes 2–5, and 7–9) or anti-FLAG (M2, lanes 6 and 10). The immunoprecipitates were divided into four equal portions, fractionated on SDS-PAGE, and immunoblotted to detect P-S118, total ER␣, or CDK7. The fourth gel was dried down and autoradiographed. The input lane represents 10% of the total amount of the sample used in each immunoprecipitation. (E) Purified ER␣ was incubated with recombinant purified core TFIIH (rIIH5), in the presence of 10⫺7 M E2, followed by immunoprecipitation using antibodies against p62 and Cdk7. The input lane represents 10% of the amount of ER␣ used in each immunoprecipitation. (F) Sf9 cells were coinfected with baculoviruses encoding the nine subunits of TFIIH (rIIH9) or the six subunits of core TFIIH containing XPD (rIIH6) in combination with or without a baculovirus encoding FLAG-tagged human ER␣. Cell lysates were immunoprecipitated in the presence of 10⫺7 M E2 using anti-FLAG antibody. Bound proteins (B) were analyzed by SDS-PAGE and immunoblotting using antibod-

TFIIH Associates with Human Estrogen Receptor ␣ The differential ability of free and TFIIH-associated CAK to phosphorylate S118 could be due to a requirement for an interaction between CAK and ER␣, mediated by TFIIH. In vitro synthesized HEG0 was coimmunoprecipitated with TFIIH by antibodies to the p62 and XPB subunits of TFIIH, indicating that ER␣ can interact with TFIIH (Figure 4A, lanes 2–5). Similarly, HEG0 and HEG19 were brought down by antibodies to p62 and Cdk7 but not by an irrelevant antibody (anti-FLAG antibody, M2) (Figure 4B, see lanes 2–4). No association with HE15 was evident, indicating that the interaction between TFIIH and ER␣ is mediated by the LBD and not by AF1 or the DBD. The association between HEG0 and TFIIH was stimulated by E2 (Figure 4C) and was also enhanced by the presence of OHT, albeit less so than in the presence of E2. No interaction was evident in the presence of ICI. These results indicate that the LBD of ER␣ can interact with TFIIH in a ligand-dependent manner. Immunoprecipitation using anti-Cdk7 demonstrated that there is only a weak E2-independent interaction between ER␣ and CAK (Figure 4D), suggesting that phosphorylation of ER␣ by Cdk7 occurs efficiently only in the context of TFIIH. Purified ER␣ interacted with core TFIIH (rIIH5; lacking CAK and XPD; Tirode et al., 1999) in the presence of E2 and weakly in the absence of ligand (Figure 4E). Coinfection of Sf9 insect cells with baculoviruses encoding FLAG-tagged ER␣, and the subunits of the core TFIIH containing XPD (rIIH6) or all nine subunits of TFIIH (rIIH9), followed by immunoprecipitation using antiFLAG antibody demonstrated that ER␣ interacts with rIIH9 (Figure 4F, lanes 1 and 2) and rIIH6 (lanes 4 and 5). TFIIH (rIIH9 or rIIH6) was not immunoprecipitated by anti-FLAG antibody in coinfections lacking baculoviruses encoding FLAG-tagged ER␣ (Figure 4F, lanes 3 and 6). Taken together, these results show that ER␣ directly interacts with one or more subunits of core TFIIH in an E2-regulated manner. Interaction between ER␣ and CAK was detectable but was weaker at least by one order of magnitude, and the interaction was not ligand regulated. These interactions are reflected in the differential abilities of TFIIH and CAK to phosphorylate S118, indicative of a requirement of ER␣ association with TFIIH for S118 phosphorylation. Estrogen Receptor ␣ Associates with the p62 and XPD Subunits of TFIIH TFIIH is comprised of nine subunits and includes the three subunits of CAK (Cdk7, cyclin H, and MAT1), XPB, XPD, and p62 (de Laat et al., 1999). GST pulldowns using the LBD of murine ER␣ (amino acids 313–599) fused to GST (GST-AF2) demonstrated that the CAK subunits and XPB did not interact with AF2 (Figure 5A), whereas p62 and XPD did associate (Figure 5A). p62 associated with GST-AF2 in the presence of E2 or OHT but not in the absence of ligand or in the presence of ICI (compare

ies against various subunits of core TFIIH, the CAK subcomplex, or ER␣ (anti-FLAG) as indicated on the left. The load lane (L) represents 8% of the total volume of lysate used in each immunoprecipitation.

Molecular Cell 132

Figure 5. ER␣ Association with Subunits of TFIIH (A) Pulldowns of in vitro synthesized Cdk7, MAT1, cyclin H, p62, XPD, and XPB using GST or GST-AF2 were performed. 10⫺7 M E2 was present throughout the procedure. The input lanes represent 10% of the total volume of the lysate used for the pulldowns for each protein. (B) GST pulldowns of [35S]-labeled p62 and XPD were performed using GST and GST-AF2 in the absence of ligand or in the presence of 10⫺7 M E2, OHT, or ICI. The input lanes represent 10% of the total volume of the lysate used for the pulldowns for each protein. (C) GST pulldowns of [35S]-labeled p62 (upper panel) and XPD (lower panel) were performed using GST and GST-AF2 in the presence of 10⫺7 M E2. Pulldowns were also performed using AF2 containing mutations within the AF2 core/helix 12 region of the LBD. The input lane represents 10% of the total volume of lysate used for the pulldowns for each protein. (D) Pulldowns were performed using GST, GST-AF2, or GST-AF2M in the presence of 10⫺7 M E2. GST-AF2M contains alanines at positions 543 and 544 of murine ER␣ in place of leucines. A peptide corresponding to amino acid residues 91–106 of human p62 (P1) was used to compete with p62 for binding to GST-AF2. The leucine residues corresponding to amino acids 100 and 101 were replaced with alanines in


lanes 4 and 5 with lanes 3 and 6). The association between GST-AF2 and XPD was ligand independent, although E2 and OHT stimulated the interaction, suggestive of mechanistic differences between the association of p62 and XPD with AF2. We next tested whether association of p62 and XPD with AF2 is mechanistically similar to coactivator association with AF2. Substitution of the hydrophobic residues at positions 543 and 544 of murine ER␣ within H12 by alanines abolishes transcriptional activity, as well as binding to the coactivator SRC-1 (Danielian et al., 1992; Kalkhoven et al., 1998). This mutant did not bind p62 (Figure 5C). However, p62 association with mutants in which amino acid residues at positions 547 and 548 or at positions 542, 546, and 549 have been mutated (which also abolish transcriptional activity and association with SRC-1 [Danielian et al., 1992; Kalkhoven et al., 1998]) was reduced, but not lost. Mutation of glutamic acid at position 546 to alanine also resulted in reduced association with p62, while mutation of tyrosine 541 to phenylalanine had no effect on the association, although reduced association with another putative coactivator RIP-140, but not SRC-1, has been described (White et al., 1997). None of these mutations significantly reduced association with XPD, suggesting that AF2 core sequences do not play a significant role in the interaction between the ER␣ LBD/AF2 and XPD. Thus, AF2 association with p62 in vitro is mechanistically similar, although not identical, to AF2 interactions with coactivator proteins. An LXXLL Motif in p62 Mediates Its Interaction with Estrogen Receptor ␣ A region with the sequence AVKDLLQQLL (amino acids 91–102), conforming to the recently described LXXLL motif, is present in human p62. Association of p62 with GST-AF2 was reduced as increasing concentrations of a peptide corresponding to amino acids 89–107 of human p62 (P1) were added (Figure 5D, lanes 5–8). Competitions performed using increasing concentrations of a peptide (P2), in which the leucine residues corresponding to amino acids 101 and 102 were replaced by alanines, did not compete (lanes 9–12). Similarly, the interaction of ER␣ with TFIIH was competed by P1 but not by P2 (Figure 5E). In a yeast two-hybrid assay, fusion of amino acids 89–105 of p62 were sufficient for an interaction with

murine ER␣ LBD/AF2 in the presence of E2 (about 25% of the activity observed with DBD-SV40 T antigen and p53-AD), but not in the absence of ligand (Figure 5F). AF2M, in which leucines 543 and 544 of murine ER␣ have been replaced by alanines, did not interact with DBD-p6289–105. Amino acids 93 and 96 of p62 were mutated without affecting the interaction with AF2, whereas mutation of any of the leucines at positions 98 and 99 or 101 and 102 completely abrogated the interaction (Figure 5G; p62E, p62F). Mutation of the leucines at positions 101 and 102 (GST-p62AA) dramatically reduced interaction of full-length p62 with ER␣ (Figure 5H). Thus, the LXXLL motif present in p62 is central to E2-dependent interaction of p62 with ER␣. Ligand-Dependent Phosphorylation of Serine 118 In Vivo Requires an Interaction between TFIIH and Estrogen Receptor ␣ We next examined whether S118 phosphorylation requires recruitment of TFIIH by AF2. Incubation of ER␣ with TFIIH reduced E2-induced phosphorylation by TFIIH to background levels, whereas P2 did not (Figure 6A, lanes 2–4). The low-level phosphorylation by CAK was unaffected by either peptide (lanes 6–9). Phosphorylation of HEG0539A, 540A (leucine residues 539 and 540 of human ER␣ correspond to leucine residues 543 and 544 of murine ER␣, respectively) by TFIIH was also significantly reduced when compared to HEG0 (Figure 6B). These results are evidence that ligand-dependent phosphorylation of S118 requires an association of ER␣ with TFIIH. The in vivo requirement for an association between TFIIH and ER␣, for S118 phosphorylation, was tested by cotransfecting p62 and ER␣. In order to distinguish between TFIIH- and MAPK-mediated phosphorylation of S118, the effect of p62 overexpression on PMA-induced MAPK activation (and S118 phosphorylation) was also examined. Overexpression of the CL100 phosphatase, which inactivates MAPK, acted to further distinguish the two signaling pathways. S118 phosphorylation was stimulated 4-fold by E2 and 7-fold by PMA and was greatest in the presence of E2 and PMA (9-fold) (Figure 6C, lanes 4–6). MAPK (P-ERK1/P-ERK2) activation was induced by treatment with PMA, but not by addition of E2. CL100 overexpression inhibited MAPK activation by PMA (lanes 7–10) and reduced PMA-stimulated S118

peptide P2. Peptides P1 and P2 were added at concentrations of 2 (lanes 5 and 9), 6 (lanes 6 and 10), 12 (lanes 7 and 11), and 24 ␮M (lanes 8 and 12). The input lane represents 10% of the total volume of the lysate used for the pulldowns. (E) HEG0 synthesized in vitro in the presence of [35S]-labeled methionine was incubated with purified TFIIH in the presence of 10⫺7 M E2, followed by immunoprecipitation using antibodies against the FLAG epitope (M2) or p62. Increasing concentrations of P1 and P2 (as in D) were used to compete with TFIIH for binding to HEG0. The input lane represents 10% of the total volume of the HEG0 lysate used in each immunoprecipitation. (F) Yeast two-hybrid assay was used to assess the ability of the small region of p62, which, containing the p62 LXXLL motif, fused to the GAL4 DNA binding domain (DBD-p62 89–105), to interact with murine ER␣ AF2, fused to the VP16 activation domain (AF2-AD), in the presence and absence of 10⫺6 M E2. The interaction was also tested with vector (VP16 only) or VP16-AF2 containing a substitution of leucine residues at positions 543 and 544 to alanines (AF2M-AD). p53-AD and the SV40 T antigen fused to the GAL4 DBD (DBD-T) were used as positive controls for interaction. Interactions were assayed by determining the activity of an integrated GAL4-regulated ␤-galactosidase gene. The results of four independent experiments are displayed in the form of a bar chart. (G) Yeast two-hybrid assays were performed as in (F) using DBD-p62 89–105 containing amino acid susbtitutions as shown, together with AF2-AD in the presence of 10⫺6 M E2. The results of four independent experiments are displayed in the form of a bar chart. (H) Pulldowns were performed using GST, GST-p62, or GST-p62AA in the presence or absence of 10⫺7 M E2. Leucine residues at positions 101 and 102 of p62 were replaced with alanines to give GST-p62AA. Immunoblotting was performed to reveal ER␣. The input lane represents 10% of the amount of purified ER␣ used for the pulldowns.

Molecular Cell 134

requires the integrity of helix 12 of the LBD/AF2 on the one hand, and on the other, the p62 and XPD subunits of TFIIH. Furthermore, a LXXLL motif in p62 is crucial for TFIIH association with AF2. These findings are suggestive of a novel mechanism by which nuclear receptor activity can be regulated by ligand-dependent recruitment of modifying activities such as kinases.

Figure 6. Phosphorylation of Serine 118 Is Regulated by Association of TFIIH with AF2 (A) Purified ER␣ (0.01 pmoles) was phosphorylated using TFIIH (1 ␮l) or CAK (5 ␮l), as described for Figure 3A, except that peptides P1 (lanes 4 and 8) or P2 (lanes 5 and 9) were added to the phosphorylation reactions at a concentration of 24 ␮M. E2 (10⫺7 M) was added as shown. (B) Extracts of COS-1 cells transfected with HEG0 or HEG0539A, 540A were immunoprecipitated using monoclonal antibody B10 and phosphorylated with purified CAK (5 ␮l) or TFIIH (1 ␮l). Phosphorylated proteins were visualized by autoradiography (upper panel) and immunoblotting with B10 (lower panel). (C) COS-1 cells were transfected with 5 ␮g of pSG5, HEG0118A, HEG0, or HEG0539A, 540A, with CL100, p62, or p62101A, 102A (p62-AA), as appropriate. E2 (10⫺7 M) and PMA (100 ng/ml) was added 30 min prior to harvesting, where indicated. Cell lysates were resolved by SDSPAGE, and immunoblotting was performed for total ER␣ (B10), ER␣ phosphorylated at S118 (P-S118), total ERK2, or phosphorylated ERK1/ERK2 (P-ERK1/P-ERK2).

phosphorylation, whereas E2-stimulated phosphorylation of S118 was unaffected (compare lanes 4 and 8 and lanes 6 and 10). Cotransfection with p62 resulted in a significant increase in S118 phosphorylation in the presence of E2 (12-fold) but did not affect PMA-stimulated MAPK or S118 phosphorylation (Figure 6C, lanes 11–14). P62AA did not stimulate S118 phosphorylation (lanes 15–18). These results suggest that S118 phosphorylation in vivo requires association of ER␣ with p62 (and TFIIH). In agreement with the in vitro data (Figure 6B), phosphorylation of HEG0539A, 540A was only weakly increased by E2, whereas PMA-stimulated phosphorylation was similar to that of wild-type ER␣ (lanes 19–22). Taken together, these results indicate that E2-dependent phosphorylation of S118 requires a ligand-dependent interaction between TFIIH and the LBD/AF2 in vivo as well as in vitro. Discussion In this report, we show that human ER␣ is phosphorylated by the TFIIH kinase Cdk7, in a ligand-dependent manner, with a consequent stimulation of transactivation. Furthermore, efficient phosphorylation of S118 in AF1 of ER␣ requires a ligand-regulated interaction of TFIIH with the ligand binding domain. This interaction

Modulation of Estrogen Receptor ␣ Activity through Phosphorylation of S118 in Activation Function AF1 by Cdk7 S118 is phosphorylated in vivo, is required for efficient transactivation by ER␣ (Ali et al., 1993b), and is phosphorylated in a ligand-independent manner by MAPK (Kato et al., 1995). However, S118 phosphorylation is also induced by estrogen binding to the receptor in a MAPK-independent manner (Joel et al., 1998). We noted that phosphorylation at a similar serine-proline motif within AF1 of RAR␣ is mediated by CAK and TFIIH (Rochette-Egly et al., 1997). In transactivation assays, Cdk7, cyclin H, and MAT1 significantly stimulated transactivation by ER␣ in the presence of E2. Furthermore, this stimulation required the presence of S118, and indeed, Cdk7 overexpression resulted in a ligand-dependent increase in S118 phosphorylation in vivo. In vitro, ER␣ was phosphorylated by TFIIH in a ligand-dependent manner at S118. Taken together, these data indicate that ligand-dependent phosphorylation of S118 is mediated by the TFIIH kinase Cdk7 and results in stimulation of AF1 activity. Phosphorylation of Serine 118 of Human Estrogen Receptor ␣ by TFIIH-Associated CAK but Not by Free CAK TFIIH-associated CAK phosphorylated ER␣ in vitro in an E2-dependent manner, whereas only low-level phosphorylation was observed with free CAK. Also, note that TRAP/DRIP coactivator complex that associates with several nuclear receptors, including ER␣, contains the Pol II CTD kinase Cdk8. If present as a contaminant in the purified TFIIH preparations, Cdk8 could be responsible for the differential ability of free CAK and TFIIH to phosphorylate ER␣. However, immunoblotting demonstrated that Cdk8 is not present in the purified TFIIH preparation (data not shown). Furthermore, recombinant TFIIH (rIIH9) reconstituted from baculovirus phosphorylated ER␣, whereas recombinant TFIIH containing mutant Cdk7 did not (data not shown). Differential substrate specificity of free and TFIIH-associated CAK has also been described for phosphorylation of RAR␣ (RochetteEgly et al., 1997). TFIIE␣ and the Pol II largest subunit IIA are similarly more efficiently performed by TFIIHassociated CAK than by free CAK, whereas Cdk2 is a better substrate for CAK than for TFIIH-associated CAK (Rossignol et al., 1997; Yankulov and Bentley, 1997). In transactivation assays, we also noted that overexpression of any one CAK subunit stimulated ER␣ to a small extent. Cdk7 and MAT1 coexpression, on the other hand, resulted in a 3-fold increase in ER␣ activity, indicative of a requirement for MAT1. Cyclin H, together with Cdk7 and MAT1, stimulated ER␣ to the same extent as Cdk7 and MAT1 alone, whereas overexpression of Cdk7 with cyclin H did not stimulate ER␣ any better than either


subunit alone. The requirement for both Cdk7 and MAT1, but not of cyclin H, for efficient stimulation of transactivation may be a reflection of the presence of limiting amounts of Cdk7 and/or MAT1. In regard to the requirement for MAT1, in addition to Cdk7 for ER␣ activation and phosphorylation in vivo, RAR␣ is also more efficiently phosphorylated by the purified, recombinant CAK complex than by the Cdk7/ cyclin H complex (Rochette-Egly et al., 1997). Other work has demonstrated that MAT1 regulates Cdk7 substrate specificity, switching its substrate preference from Cdk2 to the Pol II CTD (Yankulov and Bentley, 1997). It is possible that the cooperativity between Cdk7 and MAT1 seen here reflects a similar requirement for MAT1 for Cdk7 phosphorylation of S118. Ligand-Dependent Association of TFIIH with Estrogen Receptor ␣ Results in S118 Phosphorylation TFIIH-associated CAK phosphorylated S118 considerably more efficiently than free CAK. Similarly, ER␣ interacted with TFIIH in vitro in a ligand-regulated manner. Moreover, phosphorylation by and interaction with TFIIH was ligand dependent, whereas the weak phosphorylation by and interaction with CAK were ligand independent. The interaction between ER␣ and TFIIH required the presence of the LBD/AF2. Furthermore, pulldown experiments revealed that the p62 and XPD subunits of TFIIH associate with AF2 in a ligand-dependent manner. The interaction between ER␣ LBD and p62 was mediated by an LXXLL-type motif with the sequence LLQQLL, located between amino acids 96–102 of human p62. By contrast, XPD association with AF2 was unaffected by helix 12 mutations, suggesting that XPD and p62 association with AF2 involves different regions of the LBD. Furthermore, the functional significance of XPD association with AF2 is unclear at present, given that a p62 peptide was sufficient to prevent TFIIH association with ER␣. Coactivator interaction with ER␣ AF2 is largely E2 dependent, although association in the absence of ligand has been demonstrated for SRC-1a (Kalkhoven et al., 1998). We also observed interaction between ER␣ and TFIIH in the absence of ligand. However, ER␣ also associated with TFIIH in the presence of OHT and ICI. While this may be attributable to the low ligand selectivity of XPD interaction with AF2, p62 associated with AF2 in the presence of the partial antagonist OHT. Structural studies show that OHT binding to the LBD/AF2 should prevent an association between AF2 and LXXLL-type motifs by occlusion of the LXXLL region binding goove by helix 12 (Shiau et al., 1998). Our results suggest that LXXLL-mediated interactions with the LBD are possible in the case of the OHT-bound receptor. Inhibition of TFIIH association with ER␣ by peptide competition abolished phosphorylation of ER␣. In vivo, overexpression of p62 increased phosphorylation of S118 in the presence of E2. Mutating the LXXLL motif prevented stimulation of phosphorylation. Finally, mutation of key leucine residues in helix 12 of human ER␣ (HEG0539A, 540A) prevented association with p62 and phosphorylation by TFIIH in vitro, and this mutant was not significantly phosphorylated in response to E2 in vivo. Taken together, these results strongly indicate that S118

phosphorylation by CAK requires a ligand-regulated interaction between TFIIH and the ER␣ LBD. A Novel Mechanism for Ligand Regulation of Transactivation by the Estrogen Receptor Transcription is a highly regulated process involving alteration of chromatin structure, recruitment of Pol II and GTFs to the promoter in preinitiation complex (PIC) formation, and promoter clearance. Transcription factors can stimulate any or all of these events, either by direct interactions with GTFs or indirectly by recruitment of global regulators, such as coactivators. These global regulators mediate the action of transcription factors, at least in part, by covalent modification of chromatin, Pol II, and/or GTFs. Covalent modifications include chromatin remodeling through histone acetylation/deacetylation. The TRAP/DRIP complex contains a subset (or all) of the proteins in the Pol II-interacting SMCC, which phosphorylates the Pol II CTD (Parvin and Young, 1998). The nuclear receptor superfamily of transcription factors has evolved an exquisite mechanism, whereby ligand binding to the LBD elicits their transactivation potential, by regulating these steps in transcription. Our results with ER␣ and TFIIH indicate that ligand binding can, in addition, regulate nuclear receptor activity by inducing their modification through direct recruitment of modifying enzymes. At least one other nuclear receptor, RAR␣, is also phosphorylated by and associates with TFIIH. As is the case for ER␣, RAR␣ is phosphorylated at a site in AF1 and the association appears to involve AF2. TFIIH recruitment and phosphorylation might therefore be a feature of many nuclear receptors. Indeed, many other nuclear receptors are phosphorylated at Ser-Pro motifs and may be targets for phosphorylation by CAK (see Shao and Lazar, 1999). Other transcription regulators, for example, p53, Oct factors, and E2F, are phosphorylated by and associate with components of CAK and/or TFIIH, and their phosphorylation may be dependent on interaction with CAK and/or TFIIH (Xiao et al., 1994; Leveillard et al., 1996; Inamoto et al., 1997; Schneider et al., 1998). Similarly, yeast GAL4 is phosphorylated by SRB10/Cdk8, a component of the SMCC, at a site known to be important for efficient transactivation (Hirst et al., 1999). The recent demonstration that HATs can acetylate p53 and GATA-1, resulting in an augmentation in their DNA binding activity (Gu and Roeder, 1997; Boyes et al., 1998), raises the possibility that transcription factor activity can be regulated by interaction with HATs. Our findings with ER␣ describe an important mechanism by which transcription factor recruitment of coactivators and GTFs that have modifying activities regulates transcription factor activities, in addition to targeting other substrates such as histones and Pol II. Experimental Procedures Expression Plasmids, Antibodies, Peptides, and Purified Proteins The ER␣ constructs and reporter genes have previously been described (see Chen et al., 1999). Cdk7 and MAT1 were kindly provided by Dr. E. Nigg. Drs. T. Makela and D. O. Morgan provided cyclin H. XPB and XPD were provided by Drs. S. Rademakers and J. H. J. Hoeijmakers. GST-AF2 and mutants were kindly provided by Dr. M. Parker. Yeast expression plasmids were obtained from Dr. D. Heery.

Molecular Cell 136

Dr. S. Keyse kindly provided CL100. Other yeast two-hybrid constructs were made in pAS2–1 (Clontech). SV40 T antigen in pAS2–1 (DBD-T) and p53 cloned in pACT2 (p53-AD) were obtained from Clontech. Anti-Cdk7 has previously been described (Adamczewski et al., 1996), as have the ER␣ antibodies B10 and F3 (Ali et al., 1993a). ERK2, XPB, and p62 antibodies were from Santa Cruz Biotechnology. Antibodies to phosphorylated ERK1/ERK2 were from NEB. Purification of TFIIH, rIIH5, rIIH6, rIIH9, and CAK has previously been described (Gerard et al., 1991; Rossignol et al., 1997; Tirode et al., 1999). Purified recombinant human ER␣ was purchased from Calbiochem. ER␣ expressing baculovirus was kindly provided by Dr. W. L. Kraus (Kraus and Kadonaga, 1998). Cells, Transfections, CAT Assays, In Vivo Labeling, Immunoprecipitations, and Immunoblotting COS-1 cells were maintained and transfected as described previously. Whole-cell extracts, CAT assays, in vivo [32P] labeling, immunoprecipitations, and immunoblotting procedures have also been described (Pace et al., 1997; Chen et al., 1999). Baculovirus Coinfections and Immunoprecipitations Baculoviruses encoding TFIIH subunits, coinfections of Sf9 cells, and lysate preparation have been described previously (Tirode et al., 1999). Immunoprecipitations were performed on lysates followed by extensive washing with essentially the same buffer containing 250 mM NaCl. In Vitro Phosphorylation In vitro phosphorylation of purified recombinant ER␣ by purified CAK or TFIIH was performed essentially as described (RochetteEgly et al., 1997). ER␣ mutants were phosphorylated as described (Chen et al. 1999). Immunoprecipitation and GST Pulldown Experiments In vitro transcription and translation were performed using TNT (Promega) in the presence of [35S]methionine. GST fusion proteins were induced and lysates prepared as described (Cavailles et al., 1995). Coimmunoprecipitation of in vitro synthesized, [35S]methioninelabeled HEG0, HE15, and HEG19 with TFIIH was performed using 1 ␮l of purified TFIIH with antibodies to p62, XPB, or Cdk7, unless otherwise stated in the text. Yeast Two-Hybrid Assay Interactions between AF2 and p62 were determined by coexpression of the p62-GAL4 (DBD) fusion and AF2 fused to the acidic activation domain of VP16 in the yeast strain Y187, containing a ␤-galactosidase gene under the regulation of the GAL4 DNA binding element (17 M). ␤-galactosidase activities were assayed according to manufacturer’s protocols (Clontech). Acknowledgments We are grateful to D. M. Heery, J. H. J. Hoeijmakers, S. M. Keyse, T. Makela, D. O. Morgan, E. A. Nigg, D. Metzger, P. Chambon, M. G. Parker, and S. Rademakers for their kind gifts of plasmids. We also thank Dr. Alan Wakeling (Zeneca) for providing ICI 182, 780 and Dr. W. L. Kraus for ER␣ baculoviruses. We thank M. Chipoulet for expert technical assistance. We are especially indebted to L. Buluwela and J. H. White for critical reading of the manuscript. This work was funded by the Cancer Research Campaign, the Mandeville Trust, the Human Frontier Science Program, as well as the Association de la Recherche Contre le Cancer. T. R. was funded by the DAAD Hoschulsonderprogramm III. Received November 29, 1999; revised May 3, 2000. References Adamczewski, J.P., Rossignol, M., Tassan, J.P., Nigg, E.A., Moncollin, V., and Egly, J.M. (1996). MAT1, cdk7 and cyclin H form a kinase complex which is UV light-sensitive upon association with TFIIH. EMBO J. 15, 1877–1884.

Ali, S., Lutz, Y., Bellocq, J.-P., Chenard-Neu, M.-P., Rouyer, N., and Metzger, D. (1993a). Production and characterisation of monoclonal antibodies recognising defined regions of the human oestrogen receptor. Hybridoma 12, 391–405. Ali, S., Metzger, D., Bornert, J.M., and Chambon, P. (1993b). Modulation of transcriptional activation by ligand-dependent phosphorylation of the human oestrogen receptor A/B region. EMBO J. 12, 1153–1160. Arnold, S.F., Obourn, J.D., Jaffe, H., and Notides, A.C. (1994). Serine 167 is the major estradiol-induced phosphorylation site on the human estrogen receptor. Mol. Endocrinol. 8, 1208–1214. Boyes, J., Byfield, P., Nakatani, Y., and Ogryzko, V. (1998). Regulation of activity of the transcription factor GATA-1 by acetylation. Nature 396, 594–598. Brzozowski, A.M., Pike, A.C., Dauter, Z., Hubbard, R.E., Bonn, T., Engstrom, O., Ohman, L., Greene, G.L., Gustafsson, J.A., and Carlquist, M. (1997). Molecular basis of agonism and antagonism in the oestrogen receptor. Nature 389, 753–758. Cavailles, V., Dauvois, S., L’Horset, F., Lopez, G., Hoare, S., Kushner, P.J., and Parker, M.G. (1995). Nuclear factor RIP140 modulates transcriptional activation by the estrogen receptor. EMBO J. 14, 3741– 3751. Chen, D., Pace, P.E., Coombes, R.C., and Ali, S. (1999). Phosphorylation of human estrogen receptor alpha by protein kinase A regulates dimerization. Mol. Cell. Biol. 19, 1002–1015. Danielian, P.S., White, R., Lees, J.A., and Parker, M.G. (1992). Identification of a conserved region required for hormone dependent transcriptional activation by steroid hormone receptors. EMBO J. 11, 1025–1033. de Laat, W.L., Jaspers, N.G., and Hoeijmakers, J.H. (1999). Molecular mechanism of nucleotide excision repair. Genes Dev. 13, 768–785. Freedman, L.P. (1999). Increasing the complexity of coactivation in nuclear receptor signaling. Cell 97, 5–8. Garber, M.E., and Jones, K.A. (1999). HIV-1 Tat: coping with negative elongation factors. Curr. Opin. Immunol. 11, 460–465. Gerard, M., Fischer, L., Moncollin, V., Chipoulet, J.M., Chambon, P., and Egly, J.M. (1991). Purification and interaction properties of the human RNA polymerase B(II) general transcription factor BTF2. J. Biol. Chem. 266, 20940–20945. Glass, C.K., and Rosenfeld, M.G. (2000). The coregulator exchange in transcriptional functions of nuclear receptors. Genes Dev. 14, 121–141. Gronemeyer, H. (1991). Transcription activation by estrogen and progesterone receptors. Annu. Rev. Genet. 25, 89–123. Gu, W., and Roeder, R.G. (1997). Activation of p53 sequence-specific DNA binding by acetylation of the p53 C-terminal domain. Cell 90, 595–606. Hirst, M., Kobor, M.S., Kuriakose, N., Greenblatt, J., and Sadowski, I. (1999). GAL4 is regulated by the RNA polymerase II holoenzymeassociated cyclin-dependent protein kinase SRB10/CDK8. Mol. Cell 3, 673–678. Inamoto, S., Segil, N., Pan, Z.Q., Kimura, M., and Roeder, R.G. (1997). The cyclin-dependent kinase-activating kinase (CAK) assembly factor, MAT1, targets and enhances CAK activity on the POU domains of octamer transcription factors. J. Biol. Chem. 272, 29852–29858. Ito, M., Yuan, C.X., Malik, S., Gu, W., Fondell, J.D., Yamamura, S., Fu, Z.Y., Zhang, X., Qin, J., and Roeder, R.G. (1999). Identity between TRAP and SMCC complexes indicates novel pathways for the function of nuclear receptors and diverse mammalian activators. Mol. Cell 3, 361–370. Joel, P.B., Traish, A.M., and Lannigan, D.A. (1998). Estradiol-induced phosphorylation of serine 118 in the estrogen receptor is independent of p42/p44 mitogen-activated protein kinase. J. Biol. Chem. 273, 13317–13323. Kalkhoven, E., Valentine, J.E., Heery, D.M., and Parker, M.G. (1998). Isoforms of steroid receptor co-activator 1 differ in their ability to potentiate transcription by the oestrogen receptor. EMBO J. 17, 232–243. Kato, S., Endoh, H., Masuhiro, Y., Kitamoto, T., Uchiyama, S., Sasaki,


H., Masushige, S., Gotoh, Y., Nishida, E., Kawashima, H., et al. (1995). Activation of the estrogen receptor through phosphorylation by mitogen-activated protein kinase. Science 270, 1491–1494.

Tsai, M.J., and O’Malley, B.W. (1994). Molecular mechanisms of action of steroid/thyroid receptor superfamily members. Annu. Rev. Biochem. 63, 451–486.

Ko, L.J., Shieh, S.Y., Chen, X., Jayaraman, L., Tamai, K., Taya, Y., Prives, C., and Pan, Z.Q. (1997). p53 is phosphorylated by CDK7cyclin H in a p36MAT1-dependent manner. Mol. Cell. Biol. 17, 7220– 7229.

White, R., Sjoberg, M., Kalkhoven, E., and Parker, M.G. (1997). Ligand-independent activation of the oestrogen receptor by mutation of a conserved tyrosine. EMBO J. 16, 1427–1435.

Kraus, W.L., and Kadonaga, J.T. (1998). p300 and estrogen receptor cooperatively activate transcription via differential enhancement of initiation and reinitiation. Genes Dev. 12, 331–342. Le Goff, P., Montano, M.M., Schodin, D.J., and Katzenellenbogen, B.S. (1994). Phosphorylation of the human estrogen receptor. Identification of hormone-regulated sites and examination of their influence on transcriptional activity. J. Biol. Chem. 269, 4458–4466. Leveillard, T., Andera, L., Bissonnette, N., Schaeffer, L., Bracco, L., Egly, J.M., and Wasylyk, B. (1996). Functional interactions between p53 and the TFIIH complex are affected by tumour-associated mutations. EMBO J. 15, 1615–1624. Mangelsdorf, D.J., Thummel, C., Beato, M., Herrlich, P., Schutz, G., Umesono, K., Blumberg, B., Kastner, P., Mark, M., Chambon, P., and Evans, R.M. (1995). The nuclear receptor superfamily: the second decade. Cell 83, 835–839. Moras, D., and Gronemeyer, H. (1998). The nuclear receptor ligandbinding domain: structure and function. Curr. Opin. Cell Biol. 10, 384–391. Morgan, D.O. (1997). Cyclin-dependent kinases: engines, clocks, and microprocessors. Annu. Rev. Cell Dev. Biol. 13, 261–291. Nigg, E.A. (1996). Cyclin-dependent kinase 7: at the cross-roads of transcription, DNA repair and cell cycle control? Curr. Opin. Cell Biol. 8, 312–317. Pace, P., Taylor, J., Suntharalingam, S., Coombes, R.C., and Ali, S. (1997). Human estrogen receptor beta binds DNA in a manner similar to and dimerizes with estrogen receptor alpha. J. Biol. Chem. 2, 25832–25838. Parvin, J.D., and Young, R.A. (1998). Regulatory targets in the RNA polymerase II holoenzyme. Curr. Opin. Genet. Dev. 8, 565–570. Rachez, C., Lemon, B.D., Suldan, Z., Bromleigh, V., Gamble, M., Naar, A.M., Erdjument-Bromage, H., Tempst, P., and Freedman, L.P. (1999). Ligand-dependent transcription activation by nuclear receptors requires the DRIP complex. Nature 398, 824–828. Reinberg, D., Orphanides, G., Ebright, R., Akoulitchev, S., Carcamo, J., Cho, H., Cortes, P., Drapkin, R., Flores, O., Ha, I., et al. (1998). The RNA polymerase II general transcription factors: past, present, and future. Cold Spring Harb. Symp. Quant. Biol. 63, 83–103. Rochette-Egly, C., Adam, S., Rossignol, M., Egly, J.M., and Chambon, P. (1997). Stimulation of RAR alpha activation function AF-1 through binding to the general transcription factor TFIIH and phosphorylation by CDK7. Cell 90, 97–107. Roeder, R.G. (1996). Nuclear RNA polymerases: role of general initiation factors and cofactors in eukaryotic transcription. Methods Enzymol. 273, 165–171. Rossignol, M., Kolb-Cheynel, I., and Egly, J.M. (1997). Substrate specificity of the cdk-activating kinase (CAK) is altered upon association with TFIIH. EMBO J. 16, 1628–1637. Schneider, E., Montenarh, M., and Wagner, P. (1998). Regulation of CAK kinase activity by p53. Oncogene 17, 2733–2741. Shao, D., and Lazar, M.A. (1999). Modulating nuclear receptor function: may the phos be with you. J. Clin. Invest. 103, 1617–1618. Shiau, A.K., Barstad, D., Loria, P.M., Cheng, L., Kushner, P.J., Agard, D.A., and Greene, G.L. (1998). The structural basis of estrogen receptor/coactivator recognition and the antagonism of this interaction by tamoxifen. Cell 95, 927–937. Tassan, J.P., Jacquenot, M., Fry, A.M., Frutiger, S., Hughes, G.J., and Nigg, E.A. (1995). In vitro assembly of a functional human CDK7cyclin H complex requires MAT1, a novel 36 kDa RING finger protein. EMBO J. 14, 5608–5617. Tirode, F., Busso, D., Coin, F., and Egly, J.-M. (1999). Reconstitution of the transcription factor TFIIH: assignment of functions for the enzymatic subunits, XPB, XPD, and cdk7. Mol. Cell 3, 87–95.

Xiao, H., Pearson, A., Coulombe, B., Truant, R., Zhang, S., Regier, J.L., Triezenberg, S.J., Reinberg, D., Flores, O., Ingles, C.J., et al. (1994). Binding of basal transcription factor TFIIH to the acidic activation domains of VP16 and p53. Mol. Cell. Biol. 14, 7013–7024. Yankulov, K.Y., and Bentley, D.L. (1997). Regulation of CDK7 substrate specificity by MAT1 and TFIIH. EMBO J. 16, 1638–1646.