Crystal Structure of a Mutant hERα Ligand-binding Domain Reveals Key Structural Features for the Mechanism of Partial Agonism

The crystal structure of a triple cysteine to serine mutant ERα ligand-binding domain (LBD), complexed with estradiol, shows that despite the presence of a tightly bound agonist ligand, the protein exhibits an antagonist-like conformation, similar to that observed in raloxifen and 4-hydroxytamoxifen-bound structures. This mutated receptor binds estradiol with wild type affinity and displays transcriptional activity upon estradiol stimulation, but with limited potency (about 50%). This partial activity is efficiently repressed in antagonist competition assays. The comparison with available LBD structures reveals key features governing the positioning of helix H12 and highlights the importance of cysteine residues in promoting an active conformation. Furthermore the present study reveals a hydrogen bond network connecting ligand binding to protein trans conformation. These observations support a dynamic view of H12 positioning, where the control of the equilibrium between two stable locations determines the partial agonist character of a given ligand.1QKT The crystal structure of a triple cysteine to serine mutant ERα ligand-binding domain (LBD), complexed with estradiol, shows that despite the presence of a tightly bound agonist ligand, the protein exhibits an antagonist-like conformation, similar to that observed in raloxifen and 4-hydroxytamoxifen-bound structures. This mutated receptor binds estradiol with wild type affinity and displays transcriptional activity upon estradiol stimulation, but with limited potency (about 50%). This partial activity is efficiently repressed in antagonist competition assays. The comparison with available LBD structures reveals key features governing the positioning of helix H12 and highlights the importance of cysteine residues in promoting an active conformation. Furthermore the present study reveals a hydrogen bond network connecting ligand binding to protein trans conformation. These observations support a dynamic view of H12 positioning, where the control of the equilibrium between two stable locations determines the partial agonist character of a given ligand.1QKT ligand-binding domain estrogen receptor selective ER modulator activation factor 1 chloramphenicol acetyltransferase polymerase chain reaction 4-morpholinepropanesulfonic acid root mean square Steroid hormones regulate the transcription of target genes in the cell by binding to transcription regulators that belong to the superfamily of nuclear receptors. All members of this family display a modular structure composed of six domains (A–F). The E region constitutes the ligand-binding domain (LBD)1 containing a ligand-dependant transactivation function (AF-2) (1Mangelsdorf D.J. Evans R.M. Cell. 1995; 83: 841-850Abstract Full Text PDF PubMed Scopus (2805) Google Scholar, 2Katzenellenbogen J.A. Katzenellenbogen B.S. Chem. Biol. 1996; 3: 529-536Abstract Full Text PDF PubMed Scopus (151) Google Scholar). The transcriptional activity of nuclear receptors is mediated by interactions with the transcriptional machinery through various corepressors and coactivators (3McKenna N.J. Lanz R.B. O'Malley B.W. Endocr. Rev. 1999; 20: 321-344Crossref PubMed Scopus (1629) Google Scholar). Their ability to modulate gene expression in a ligand-regulated manner is based on the position of helix H12 carrying the AF2-AD transactivation function (4Ruff M. Gangloff M. Wurtz J.M. Moras D. Breast Cancer Res. 2000; 2: 353-359Crossref PubMed Scopus (102) Google Scholar). Several positions of H12 have been observed (5Moras D. Gronemeyer H. Curr. Opin. Cell Biol. 1998; 10: 384-391Crossref PubMed Scopus (688) Google Scholar). In the absence of ligand, H12 has been shown to be exposed to solvent (6Bourguet W. Ruff M. Chambon P. Gronemeyer H. Moras D. Nature. 1995; 375: 377-382Crossref PubMed Scopus (1042) Google Scholar). Ligand binding triggers a conformational change that results in the repositioning of H12 on the core of the LBD, closing the ligand binding pocket like a lid (7Egea P.F. Mitschler A. Rochel N. Ruff M. Chambon P. Moras D. EMBO J. 2000; 19: 2592-2601Crossref PubMed Google Scholar). This is referred to as the mouse trap mechanism (8Renaud J.P. Rochel N. Ruff M. Vivat V. Chambon P. Gronemeyer H. Moras D. Nature. 1995; 378: 681-689Crossref PubMed Scopus (1012) Google Scholar). In agonist-bound LBDs a surface suitable for coactivator binding is then created (9Darimont B.D. Wagner R.L. Apriletti J.W. Stallcup M.R. Kushner P.J. Baxter J.D. Fletterick R.J. Yamamoto K.R. Genes Dev. 1998; 12: 3343-3356Crossref PubMed Scopus (816) Google Scholar, 10Gampe Jr., R.T. Montana V.G. Lambert M.H. Miller A.B. Bledsoe R.K. Milburn M.V. Kliewer S.A. Willson T.M. Xu H.E. Mol. Cell. 2000; 5: 545-555Abstract Full Text Full Text PDF PubMed Scopus (505) Google Scholar, 11Nolte R.T. Wisely G.B. Westin S. Cobb J.E. Lambert M.H. Kurokawa R. Rosenfeld M.G. Willson T.M. Glass C.K. Milburn M.V. Nature. 1998; 395: 137-143Crossref PubMed Scopus (1645) Google Scholar, 12Shiau A.K. Barstad D. Loria P.M. Cheng L. Kushner P.J. Agard D.A. Greene G.L. Cell. 1998; 95: 927-937Abstract Full Text Full Text PDF PubMed Scopus (2210) Google Scholar). In most antagonist-bound complexes (11Nolte R.T. Wisely G.B. Westin S. Cobb J.E. Lambert M.H. Kurokawa R. Rosenfeld M.G. Willson T.M. Glass C.K. Milburn M.V. Nature. 1998; 395: 137-143Crossref PubMed Scopus (1645) Google Scholar, 12Shiau A.K. Barstad D. Loria P.M. Cheng L. Kushner P.J. Agard D.A. Greene G.L. Cell. 1998; 95: 927-937Abstract Full Text Full Text PDF PubMed Scopus (2210) Google Scholar), H12 has been observed positioned in a structurally conserved cleft where the LXXLL motif of the coactivator molecule binds. These observations suggest a mechanism for antagonism where H12 and the coactivator compete for a common binding site. Note that the agonist position of H12 is unique, whereas its position in antagonist-bound complexes is not. Therefore knowledge of the features responsible for inducing and stabilizing a given conformation is a key step in understanding the initial events of nuclear receptor transactivation. Several crystal structures of both ER isotypes (ERα and ERβ) have been solved in complex with natural and synthetic ligands (12Shiau A.K. Barstad D. Loria P.M. Cheng L. Kushner P.J. Agard D.A. Greene G.L. Cell. 1998; 95: 927-937Abstract Full Text Full Text PDF PubMed Scopus (2210) Google Scholar, 13Brzozowski A.M. Pike A.C. Dauter Z. Hubbard R.E. Bonn T. Engstrom O. Ohman L. Greene G.L. Gustafsson J.A. Carlquist M. Nature. 1997; 389: 753-758Crossref PubMed Scopus (2900) Google Scholar, 14Tanenbaum D.M. Wang Y. Williams S.P. Sigler P.B. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 5998-6003Crossref PubMed Scopus (585) Google Scholar, 15Pike A.C. Brzozowski A.M. Hubbard R.E. Bonn T. Thorsell A.G. Engstrom O. Ljunggren J. Gustafsson J.A. Carlquist M. EMBO J. 1999; 18: 4608-4618Crossref PubMed Scopus (903) Google Scholar, 16Eiler, S., Gangloff, M., Duclaud, S., Moras, D., and Ruff, M. (2001)Protein Expression Purif., in press.Google Scholar). The natural ligand 17β-estradiol acts as a pure agonist on both isotypes. Others typified by EM-800 and ICI164,384 are described as pure antagonists (17MacGregor J.I. Jordan V.C. Pharmacol. Rev. 1998; 50: 151-196PubMed Google Scholar). A third category of ligands displaying cell-type and promoter dependence in ER regulation are referred to as selective ER modulators (SERMs) (18Katzenellenbogen B.S. Katzenellenbogen J.A. Breast Cancer Res. 2000; 2: 335-344Crossref PubMed Scopus (244) Google Scholar). SERMs such as raloxifen and 4-hydroxytamoxifen efficiently antagonize the AF2, but not the AF1 function, and act as a pure antagonist (19Barkhem T. Carlsson B. Nilsson Y. Enmark E. Gustafsson J. Nilsson S. Mol. Pharmacol. 1998; 54: 105-112Crossref PubMed Scopus (708) Google Scholar) in ERβ, which seems to lack a functional AF1 domain (20McInerney E.M. Weis K.E. Sun J. Mosselman S. Katzenellenbogen B.S. Endocrinology. 1998; 139: 4513-4522Crossref PubMed Scopus (234) Google Scholar). The features responsible for inducing a given conformation and stabilizing it are crucial to the definition of the optimal stereochemical and biophysical specificity of a ligand. Here we present the comparison of the wild type hERα LBD crystal structure (16Eiler, S., Gangloff, M., Duclaud, S., Moras, D., and Ruff, M. (2001)Protein Expression Purif., in press.Google Scholar) with that of a mutant protein complexed with estradiol, where three cysteine residues were mutated in serine. The mutant protein binds estradiol with wild type affinity but has limited transcriptional capacity. In the structure of the Cys → Ser triple mutant hERα LBD, we observed an antagonist conformation despite the presence of a tightly bound estradiol in the ligand-binding cavity. This antagonist conformation, together with the transcriptional activity of the single, double, and triple cysteine to serine mutant receptors, supports the view of the agonist-antagonist equilibrium of H12 and gives some insight into the molecular mechanism for the conformational switch that drives the receptor in an agonist or antagonist conformation. The Cys → Ser triple mutant hERα LBD (Lys302 → Pro552), in fusion with six histidine residues is produced using the pET15b/Escherichia coli BL21(DE3) expression system and purified by a zinc affinity column, ion exchange, and gel filtration. The purification procedure is similar to that of the wild type ER LBD (16Eiler, S., Gangloff, M., Duclaud, S., Moras, D., and Ruff, M. (2001)Protein Expression Purif., in press.Google Scholar). Crystals were obtained by vapor diffusion at 4 and 17 °C using hanging drops made by mixing 1 μl of protein solution (2.5 mg/ml) with 1 μl of reservoir solution (12% polyethylene glycol 8000, 0.4 m NaCl, 100 mm imidazole, pH = 6.9). Prior to data collection, crystals were flash-cooled in liquid ethane after a fast soaking in a cryoprotectant buffer (20% glycerol, 15% polyethylene glycol 8000, 0.4 m NaCl, 100 mm imidazole, pH = 6.9). Crystals belong to the space group P6522 with cell parameters a =b = 58.6 Å, c = 276.02 Å, α = β = 90°, γ = 120° (one monomer in the asymmetric unit, 45% solvent). X-ray data were collected at 120 K in a nitrogen gas stream using synchrotron radiation (European Synchrotron Radiation Facility Laboratoire pour l'Utilisation du Rayonnement Éléctromagnétique). The diffracted intensities were processed using the programs DENZO and SCALEPACK (21Otwinoswski Z. Minor W. Methods Enzymol. 1997; 276: 307-326Crossref Scopus (38253) Google Scholar). Experimental phases were obtained using gold and platinum derivatives. The multiple isomorphous replacement analysis was performed using CCP4 (22Collaborative Computational Project No. 4 Acta Crystallogr. Sect. D Biol. Crystallogr. 1994; 50: 760-763Crossref PubMed Scopus (19668) Google Scholar) and SHARP (23de la Fortelle E. Bricogne G. Methods Enzymol. 1997; 276: 472-494Crossref PubMed Scopus (1796) Google Scholar) packages. It enabled the construction of the complete model. An initial map was calculated to 3-Å resolution using multiple isomorphous replacement phases and solvent flattening using SOLOMON (22Collaborative Computational Project No. 4 Acta Crystallogr. Sect. D Biol. Crystallogr. 1994; 50: 760-763Crossref PubMed Scopus (19668) Google Scholar,23de la Fortelle E. Bricogne G. Methods Enzymol. 1997; 276: 472-494Crossref PubMed Scopus (1796) Google Scholar). Refinement was performed with CNS (24Brüger A.T. Adams P.D. Clore G.M. Delano W.L. Gros P. Grosse-Kuntstleve R.W. Jiang J.S. Kuszewski J. Nilges M. Pannu N.S. Read R.J. Rice M.L. Simonson T. Warren G.L. Acta Crystallogr. Sect. D Biol. Crystallogr. 1998; 54: 905-921Crossref PubMed Scopus (16919) Google Scholar) using bulk solvent corrections. All data between 15- and 2.2-Å resolution were included with no sigma cutoffs (TableI).Table IData processing, phase determination and refinement statistics of Cys → Ser triple mutant structure (P6522, a = b = 58.6 Å, c = 276.0 Å)Space group P6522NativePt1Pt2AuResolution (Å)2.02.93.42.9Unique reflections20231691744156793Completeness (%)99.998.397.096.8Multiplicity10.05.03.04.0Rsym (%) (last shell (%))5.1 (33)6.96.26.5Number of sites884Rcullis(centric/acentric/anomal)0.64/0.64/0.910.65/0.59/0.860.79/0.83/0.94Phasing power (centric/acentric/anomal)1.9/2.5/1.42.1/2.9/2/21.4/1.6/1.2fom (40–2.9 Å) (centric/acentric)0.7/0.67fom after solvent flattening and phase extension (40 → 2 Å)0.87Structure refinementResolution (Å)15–2.2Working set14756Test set715Number of water molecules466Working R factor (%)22.3Free R factor (%)27.3Rmsd bond lengths (Å)0.0081Rmsd bond angles (degree)1.172AverageB-factor34.3Rsym = ∑h∑i‖I(h) − 〈I(h)i〉‖/∑h∑i I(h)i, where 〈I(h)i〉 is the average intensity of reflection h, ∑h is the sum of the measurements of reflection h. Rcullis = lack of closure/isomorphous difference. Phasing power =Fh/lack of closure. Rmsd, root mean square deviation. fom, figure of merit. Open table in a new tab Rsym = ∑h∑i‖I(h) − 〈I(h)i〉‖/∑h∑i I(h)i, where 〈I(h)i〉 is the average intensity of reflection h, ∑h is the sum of the measurements of reflection h. Rcullis = lack of closure/isomorphous difference. Phasing power =Fh/lack of closure. Rmsd, root mean square deviation. fom, figure of merit. The dimer interface was calculated with the Grasp package (25Nichols A. Honig B. J. Comp. Chem. 1991; 12: 435-445Crossref Scopus (1158) Google Scholar). The buried interface, calculated with non-hydrogen atoms only, was obtained with the excluded area method, which calculates the accessible surface regions of protomer A buried by protomer B. pSG5-HEGO (26Tora L. Mullick A. Metzger D. Ponglikitmongkol M. Park I. Chambon P. EMBO J. 1989; 8: 1981-1986Crossref PubMed Scopus (370) Google Scholar), pG4m polyII-ER(DEF) (27Brou C. Chaudhary S. Davidson I. Lutz Y. Wu J. Egly J.M. Tora L. Chambon P. EMBO J. 1993; 12: 489-499Crossref PubMed Scopus (151) Google Scholar) for full-length ER and GAL-ER eucaryotic expression, respectively, are used in transactivation assays. Vit-tk-CAT (28Klein-Hitpass L. Schorpp M. Wagner U. Ryffel G.U. Cell. 1986; 46: 1053-1061Abstract Full Text PDF PubMed Scopus (558) Google Scholar) for full-length ER transcriptional activity and 17m-tk-CAT (29Chen J.D. Evans R.M. Nature. 1995; 377: 454-457Crossref PubMed Scopus (1694) Google Scholar) for GAL-ER activity are used as reporter plasmids. CMV-βGal served as an internal control to normalize for transfection efficiency. Bluescript KS+ plasmid was used as carrier DNA. The procaryotic expression system pET15b-LBD/BL21(DE3) (Novagen) was used for estradiol binding ability and structural studies of the LBD. The mutations were generated in different constructs to allow structural and functional studies. Cysteine to serine mutations at positions 381, 417, and 530 were introduced in the LBD cloned in NdeI-BamHI sites of pET15b by PCR-assisted mutagenesis, using Deep Vent DNA polymerase (Biolabs) and the appropriate oligonucleotides. Triple mutant Cys → Ser His-tagged LBD was used in the structural studies and ligand binding assays. For transactivation assays the single, double, or triple mutants, in all possible combinations, were brought into the full-length receptor (pSG5-ER26) by digesting the LBD with the restriction enzymes HindIII-BglII. This fragment of 252 base pairs sharing the mutation C381S and/or C417S was inserted in the HindIII-BglII sites of pSG5-ER. The presence of another BglII cleavage site in pSG5 causes the loss of a fragment of 547 base pairs. ThisBglII-BglII fragment with or without the C530S mutation was reinserted in the vector. TheBglII-BglII C530S fragment was obtained by PCR-assisted site-directed mutagenesis. To remove the AF1 contribution, the triple mutant was also brought into theXhoI-BamHI sites of the pG4m vector by PCR cloning using the pSG5-ER triple mutant as template, leading to the eucaryotic expression of GAL-ER(DEF). Based on structural observations, another triple mutant was designed (E339A,E419A,K531A). These mutations were successively generated by PCR mutagenesis in the DEF region (282) subcloned in theXhoI-BamHI sites of pG4m. These mutations were also brought in the LBD subcloned in theNdeI-BamHI sites of pET 15b. All constructs were verified by automated DNA sequencing. COS1 cells, an estrogen receptor-deficient cell line, were transferred in phenol red-free Dulbecco's modified Eagle's medium supplemented with 5% charcoal-dextran-treated fetal calf serum and antibiotics (40 μg/ml gentamicin, 0.1 mg/ml streptomycin, 500 units/ml specillin). Cells were plated in six-well dishes (Costar) at a density of ∼5 × 105 cells/well in a humidified 5% CO2atmosphere at 37 °C. The cells were transfected 5–6 h later by the calcium phosphate coprecipitation technique with 0.2 μg of wild type or mutant receptor plasmid, 2 μg of reporter plasmid, 0.5 μg of CMV-βGal (internal control plasmid), and 7.3 μg of Bluescript KS+ carrier DNA. These plasmids were mixed in 420 μl of 10 mm Tris-HCl, pH 8, 0.05 mm EDTA. 60 μl of 2 m CaCl2 was added by dripping. The mixture was dripped in 2× 480 μl of HBS (280 mmNaCl, 50 mm HEPES, 1.5 mmNa2HPO4·12H2O, pH 7.12) and left 30 min at room temperature before being dispersed on the cells. 300 μl were taken per well. The precipitate remained in contact with the cells for 15 h. After this exposure, the cells were washed with phenol red-free Dulbecco's modified Eagle's medium and antibiotics. Cells were then incubated in culture medium containing the indicated concentrations of estradiol for 24 h. Lysis was achieved in 300 μl/well in 10 mm MOPS, 10 mmNaCl, 1 mm EGTA, 1% Triton X-100, pH 6.5, for 30 min at room temperature. The cellular lysates were centrifuged for 10 min at 16,000 × g. The CAT was quantified by enzyme-linked immunosorbent assay (Roche Molecular Biochemicals) according to the manufacturer's recommendations. The amount of CAT was standardized for transfection efficiency with the β-galactosidase activity in each lysate. The basal level was defined as the CAT activity in cells transfected with the reporter plasmid in the absence of receptor plasmid. Each experiment was performed at least three times in duplicate. Results are expressed in relative CAT activity in percent of maximal wild type receptor activity (Fig. 3). We have used the pET15b-LBD/BL21(DE3) system to produce hERα LBD for ligand binding assays. E. coli BL21(DE3) expressing native or mutant LBD were lysed by sonification in the binding buffer (1 mSB201, 50 mm NaCl, 50 mm Tris-HCl, pH 8, 1 mm EDTA, 1 mm dithiothreitol). After centrifugation at 14,000 × g during 1 h at 4 °C, the soluble fraction (crude extract) was used for receptor quantification and dissociation constant (Kd) determination. The total amount of protein in the crude extract was quantified using the Bradford technique. The binding assays were all performed in the presence of a total protein concentration of 5 mg/ml to avoid retention of the receptor in complex with the ligand, thus the crude extract was diluted with soluble proteins from untransformed bacteria. Receptor quantification was achieved in the presence of 10−8m[6,7-3H]estradiol (E. I. du Pont de Nemours & Co.E. I.) with (for nonspecific binding) or without (for total binding) an excess of cold estradiol (2.10−6m) and increasing amounts of crude extract. After 5 h at 4 °C, bound (B) and free (F) ligands were separated by dextran-coated charcoal (4% Norit A charcoal, 0.4% dextran T-70 in the binding buffer). This mixture was left on ice for 5 min and centrifuged at 12,000 × gfor 5 min. The supernatant was removed for scintillation counting (30Rafestin-Oblin M.E. Couette B. Radanyi C. Lombes M. Baulieu E.E. J. Biol. Chem. 1989; 264: 9304-9309Abstract Full Text PDF PubMed Google Scholar). Specific binding was plotted against the volume of crude extract for receptor quantification. For the Kd determination the crude extract was incubated with increasing concentrations (from 10−10 to 10−7m) of radiolabeled estradiol at 4 °C overnight. Each measure was done in triplicate for Scatchard analysis. The variation of B/F as a function of B was analyzed as described previously (31Claire M. Rafestin-Oblin M.E. Michaud A. Corvol P. Venot A. Roth-Meyer C. Boisvieux J.F. Mallet A. FEBS Lett. 1978; 88: 295-299Crossref PubMed Scopus (45) Google Scholar). The structure of the triple mutant ERα LBD (Fig. 1 b) exhibits the predominantly α-helical fold observed for all nuclear receptors. The superposition over the wild type structure in complex with estradiol (16Eiler, S., Gangloff, M., Duclaud, S., Moras, D., and Ruff, M. (2001)Protein Expression Purif., in press.Google Scholar) (Fig. 1 a) leads to an r.m.s. deviation of 0.54 Å over 211 Cα atoms (H1, H3-H8, H9-H11). The most striking conformational difference between these two structures is the different positioning of helix H12 and the concomitant shortening of helices H3 and H11 (Fig. 1, a and b). In the mutant, the activation helix is in the antagonist position as observed in the raloxifen and tamoxifen complexes (12Shiau A.K. Barstad D. Loria P.M. Cheng L. Kushner P.J. Agard D.A. Greene G.L. Cell. 1998; 95: 927-937Abstract Full Text Full Text PDF PubMed Scopus (2210) Google Scholar, 13Brzozowski A.M. Pike A.C. Dauter Z. Hubbard R.E. Bonn T. Engstrom O. Ohman L. Greene G.L. Gustafsson J.A. Carlquist M. Nature. 1997; 389: 753-758Crossref PubMed Scopus (2900) Google Scholar). Both antagonist structures superimpose very well to that of the mutant LBD. The r.m.s. deviation is 0.5 Å over 213 Cα atoms (H1, H3-H8, H9-H11, H12) and 0.5 Å over 238 Cα atoms (H1-H6, H7-H11, H12) for the raloxifen and tamoxifen complexes, respectively (Fig.1 c). The loop L8–9 is not seen in raloxifen, and the structure of the loop L1–3 is closer to the triple mutant in the complex with tamoxifen than that with raloxifen. Wild type and mutant homodimers can be superimposed with a r.m.s. deviation value of 0.64 Å over 418 Cα atoms. The helices H9 and H10 form the core of the interface and contribute to more than 70% of it. Despite this good match, the contributions of the secondary structure elements to the interface, spanning the helices H7 to H11, differ among the two forms. Due to the antagonist conformation of the mutant structure, helices H7, H9, and H10 exhibit a smaller contact surface area (1475 Å2) compared with the wild type (1686 Å2). The overall structure of the pocket is similar in the wild type and the Cys → Ser triple mutant. All side chains of the hydrophobic residues lining the pocket are at the same position. This explains the fact that the dissociation constants at equilibrium (Kd) between the wild type and the Cys → Ser triple mutant for estradiol are very close (Table II and Fig.2). The main differences are found on the 17-OH and 3-OH side of estradiol (Fig. 3). Due to the antagonist position of H12, the cavity in the triple mutant is not sealed as in the wild type structure. On the O17 side (d-ring side) of estradiol, the cavity reaches the surface of the protein and results in a much larger volume than the wild type ligand binding pocket. This channel is partially filled with water molecules, forming numerous hydrogen bonds with the protein. On the 3-OH side (A-ring side) an open narrow tunnel filled with water molecules is present in the mutant. In the wild type, this tunnel is almost closed and only the water molecule interacting with the 3-OH of estradiol is present. In the Cys → Ser triple mutant the estradiol A-ring superimposes perfectly with its equivalent group in the wild type complex, whereas thed-ring is slightly shifted, as shown by the displacement of the C17, which moves 0.5 Å closer to helices H3 and H12 (Fig. 3).Table IIAffinity constant (KD in nm), estradiol efficient concentration (EC50 in nm), and raloxifene-inhibiting concentration (IC50 in nm) are compared for wild type and mutant GAL-ERsGAL-ERTransactivation yieldEC50IC50KD%nmWild type1002.7 ± 0.91.8 ± 0.40.26 ± 0.05C381S + C417S + C530S48 ± 49.8 ± 4.50.7 ± 0.10.92 ± 0.09E339A + E419A + K531A62 ± 65.4 ± 3.70.7 ± 0.21.39 ± 0.09 Open table in a new tab The superposition of the Cys → Ser triple mutant with the hERα-raloxifen complex reveals that Asp351, which anchors the ammonium moiety of raloxifen, adopts the same conformation in both structures. All the helices, including H12, match perfectly (r.m.s. deviation: 0.5 Å), the protruding chain of raloxifen fitting perfectly in the water channel observed in the mutant structure. Furthermore, in this structure, electron density could be observed for the loop 11–12, a region that was not seen in the ER LBD/raloxifen structure. This loop includes the C terminus of the shortened helix H11, in particular Lys529, which points toward the ligand binding pocket channel. ER pure antagonists exhibit acidic moieties in their protruding chain and are thus unlikely to interact with Asp351 as do raloxifen and tamoxifen. The present structure suggests Lys529 as a potential hydrogen bond partner for pure antagonist ligands bearing sulfinyl-like (ICI182780) (32Dauvois S. White R. Parker M.G. J. Cell Sci. 1993; 106: 1377-1388Crossref PubMed Google Scholar) or sulfonyl-like (RU58668) (33Devin-Leclerc J. Meng X. Delahaye F. Leclerc P. Baulieu E.E. Catelli M.G. Mol. Endocrinol. 1998; 12: 842-854Crossref PubMed Scopus (63) Google Scholar) groups in their protruding chain. Such a contact would cross the AF2 AD groove and hamper the agonist positioning of H12. Interestingly each single Cys → Ser mutation (C381S,C417S; C530S) contributes equally to the observed transactivation reduction for the triple mutant receptor (Fig.4 and Table II). Each single Cys → Ser mutation decreases the full-length receptor's activity by about 20%, whereas double mutations reduce CAT activity by ∼40%, and the triple mutant exhibits a 56% decrease. Maximal wild type activity could not be restored even in the presence of saturating estradiol concentrations. Moreover each time a cysteine is mutated to serine the ligand dose-response curve of the mutant receptor is slightly shifted to the right, leading to a 6-fold shift in the ligand-efficient concentration to trigger half-maximal activity (EC50 = 4.0 ± 1.5 nm) for the triple mutant, compared with the wild type receptor (EC50 = 0.7 ± 0.1 nm, Table II). To investigate the contribution of the AF1 on transcriptional activity, we used the chimeric receptors (GAL-ER) wild type and triple mutant on which the contribution of the ligand-independent transactivation function AF1 is removed. The Cys → Ser triple mutant displays 48% activity compared with the wild type GAL-ER, which is very close to the value observed for the full-length triple mutant receptor (44%, Fig. 2 b). These data showed that the residual transactivation activity in the triple mutant is not due to AF1. Antagonist competition assays with raloxifen revealed that this SERM represses more efficiently the estradiol-stimulated CAT activity of the Cys → Ser triple mutant GAL-ER than that of the wild type (Fig. 2 c). These data suggest that the activation helix can be more easily displaced from its optimal position in the triple mutant context than in the wild type. The cysteine mutation at position 381 induces a destabilization of the agonist position of H12, which is most likely due to a solvating effect. This residue is located in helix H4, and its side chain is directed toward the solvent and is located in the agonist binding groove of H12. This residue is accessible in the mutant structure where H12 is in the antagonist position. In the wild type structure the cysteine residue is precluded from the solvent by helix H12. In the present structure this residue, which is now serine, is still solvent-accessible and is involved in a water-mediated hydrogen bond network lining the helix H12 agonist binding groove. In the mutant receptor, a positioning of H12 in the agonist groove is possible but would require the desolvation of the serine residue, a process more energetically costly for a serine than for a cysteine. The C530S mutation disrupts the hydrophobic contact between Cys530 and Tyr526 (Fig. 5 b) and contributes to the shortening by one turn of helix H11 at its C terminus end, compared with the wild type structure. A serine residue exhibits different solvating properties and favors a coil structure with a surface-exposed side chain, as observed in the Cys → Ser triple mutant. The shortening of H11 and the subsequent lengthening of loop 11–12 allow H12 to reach the coactivator binding groove, as observed in the tamoxifen-ER complex. In the wild type structure, Cys417 is located in the rather flexible loop 6–7. Its side chain is involved in numerous hydrophobic contacts with neighboring residues (it forms van der Waals contacts with Phe337 inside an hydrophobic core composed of the N-terminal parts of H3 (Phe337, Leu345) and of the β-sheet (Leu410, Leu408; Fig.5 a). The substitution by a serine residue, by disrupting these hydrophobic contacts, is likely to be the triggering factor that induces the shortening of helix H3 by one turn at its N terminus. The conformational reorganization includes the last 10 residues of loop 1–3. Interestingly, the tamoxifen complex is nearly identical to the triple mutant in this region, whereas the raloxifen-bound structure exhibits a wild type conformation without shortening. Note that the overall Cys → Ser triple mutant structure is closer to that of the tamoxifen-bound LBD than to the raloxifen one. Nevertheless some differences remain between the mutant and tamoxifen structures, especially in the loop 6–7 region, which is shifted by more than 3.0 Å (Glu419 and Gly420) toward the core of the protein. This large movement in the antagonist structure is most likely induced by tamoxifen, whose aromatic rin