Binding of hsp90 to the Glucocorticoid Receptor Requires a Specific 7-Amino Acid Sequence at the Amino Terminus of the Hormone-binding Domain*

The glucocorticoid receptor (GR) HBD must be bound to the protein chaperone hsp90 in order to acquire the high affinity steroid binding conformation. Despite this crucial role of hsp90, its binding site in GR remains poorly defined. Large portions of the GR HBD have been implicated and no similarity has been established between steroid receptor HBDs and the catalytic domains of the protein kinases (e.g. pp60 src , Raf) that also form stable heterocomplexes with hsp90. Thus, it has been thought that some general property of the proteins, such as exposure of hydrophobic residues in partially denatured regions, determines the assembly of stable hsp90 heterocomplexes. In this work, we have studied fusion proteins containing glutathione S-transferase (GST) and very short amino-terminal truncations just before and at the beginning of the rat GR HBD that are otherwise intact to the carboxyl terminus. Overexpression in COS cells of the chimeras GST537C and GST547C was found to yield receptors that were bound to hsp90 and had wild-type steroid binding affinity. However, removal of 7 more amino acids to form GST554C resulted in a fusion protein that did not bind either hsp90 or steroid. Additional mutations revealed that the role of these 7 amino acids was neither to provide a spacer between protein domains nor to expose a protein surface by introducing a bend in the conserved α-helix. Instead, these observations support a model in which the sequence of the 7 amino acids directly or indirectly affects hsp90 binding to the GR HBD. Thus, a region of GR that has not been thought to be relevant for hsp90 binding is now seen to be of critical importance, and these data argue strongly against the commonly accepted model of receptor-hsp90 heterocomplex assembly in which the chaperone initially interacts nonspecifically with hydrophobic regions of the partially denatured HBD and subsequently assists its folding to the steroid binding confirmation.

The hormone-binding domain (HBD) 1 of the glucocorticoid receptor (GR) is comprised of the carboxyl-terminal one-third of the protein (for review, see Refs. 1 and 2). However, the HBD cannot bind steroid unless the receptor is bound to the chaperone hsp90 (3,4). hsp90 is a component of a multiprotein chaperone system (including hsp70, p60, and p23) that directs the ATP-dependent assembly of hsp90 into complexes with a variety of transcription factors and protein kinases (for review, see Refs. 5 and 6). It is not known what properties permit these proteins to form relatively stable complexes with hsp90, whereas most proteins do not. It is clear, though, that hsp90 binds directly to the GR HBD (7,8). When the GR HBD is fused to another transcription factor, the fusion protein is bound to hsp90 and binds steroid (9). Furthermore, the activity of the fused transcription factor is usually controlled by steroid binding (for review, see Ref. 10).
Several approaches have been taken to determine the region of the GR HBD involved in steroid binding and hsp90 binding. When GR was translated in reticulocyte lysates, only the fulllength translation product was recovered in association with hsp90 (11), suggesting that the HBD is completely formed before the original hsp90 binding occurs. When GR mutants lacking increasing portions of the carboxyl terminus were translated in reticulocyte lysate, Dalman et al. (12) identified a minimal region from 616 -671 (rat GR) 2 that was required for a high yield of high affinity hsp90 binding. With the same approach, Howard et al. (13) found that the region 568 -616 was sufficient to yield some hsp90 binding. Taken together, these reports suggested a minimal hsp90-binding site of about 100 amino acids (568 -671). However, Cadepond et al. (14) divided the human GR HBD into three subregions of roughly equal length and showed that the fusion of each segment to GR with carboxyl-terminal truncation at amino acids 550 or 568 was sufficient to confer hsp90 binding. One of these regions sufficient for conserving hsp90 binding was the carboxyl-terminal one-third of the HBD (697-777 human or 715-795 rat), which lies completely outside of the minimal hsp90-binding region determined from carboxyl-terminal truncations. In a similar study, Schowalter et al. (15) showed that three separate regions of the progesterone receptor HBD could confer hsp90 binding when fused to a mutant PR lacking the HBD. As with the GR (14), deletion of any of these regions did not abolish hsp90 binding (15).
Thus, for the last several years, it has been the conclusion that several regions throughout the GR and PR HBDs are involved in hsp90 complex formation. No critical sequences or motifs required for hsp90 binding have been identified, and no similarity has been established between steroid receptor HBDs and the catalytic domains of protein kinases, such as pp60 src and Raf, that are also assembled into hsp90-containing heterocomplexes by the same hsp90-based chaperone machinery (16,17). It is important to bear in mind that the previously identified hsp90 binding sequences may represent only a final product and bear little relationship to the region of the HBD, or even the whole receptor, required for the initial interaction with the multiprotein chaperone machinery. In any case, it is thought that some general property, such as exposure of hydrophobic residues in partially denatured regions (18 -20), must account for the ability of the steroid receptor HBDs to form relatively stable complexes with hsp90. Unfortunately, none of the above approaches could provide convincing evidence regarding the site of hsp90-GR interaction involved in ligand binding because steroid binding was not concomitantly determined. Using a series of such fusion proteins containing dihydrofolate reductase and HBD fragments with deletions at the amino-and carboxyl-terminal ends, Xu et al. (21) found that some sequence, which did not have to be that of the GR, fused to the amino terminus of the HBD was absolutely necessary for the production of stable protein, even when the resulting protein could not bind steroid. Furthermore, the boundaries of the steroid-binding domain were localized to 550 -795 of the rat GR (21). These results raised the question of whether the loss of steroid binding upon deletion of amino acids beyond 550 was due to the absence of sequences necessary for interaction with steroid or to the lack of residues required for hsp90 binding.
In this work, we decided to examine the steroid-binding and hsp90-binding properties in parallel for a series of fusion proteins containing very short amino-terminal truncations of, or before, the HBD that are otherwise intact to the carboxyl terminus. We show that the chimeric proteins lose all hsp90 binding activity and all steroid binding activity with the removal of a region amino-terminal to any sequences previously thought to be involved in hsp90 binding to GR. Additional mutations allowed us to discard structural explanations for this behavior. These results are discussed in terms of a model in which some or all of these 7 specific amino acids are required for hsp90 binding to GR. [6, H]Triamcinolone acetonide (TA, 42.8 Ci/mmol) and 125 I-conjugated goat anti-mouse IgG were obtained from NEN Life Sciences Products and [ 3 H]dexamethasone was from Amersham Corp. Goat antimouse IgM and the monoclonal anti-GST antibody (clone GST-2) were from Sigma. The AC88 monoclonal IgG against hsp90 was from Stress-Gen (Victoria, British Columbia), and the 3G3 monoclonal anti-hsp90 IgM was from Affinity Bioreagents (Golden, CO). Anti-GR antibody (aP1) was a gift from Dr. Bernd Groner (Friedrich Miescher-Institut, Basel, Switzerland). The sources of enzymes used in cloning were: EcoRI (Stratagene), SphI (New England Biolabs), T 4 DNA ligase (Life Technologies, Inc.), and Taq polymerase (Promega).

Methods
Construction of Plasmids-All enzymatic manipulations were performed according to the manufacturers' recommendations. The constructs were transformed into DH5␣ competent cells (Life Technologies, Inc.), selected on LB plates containing 50 g/ml ampicillin (Digene Diagnostics, Inc.), and grown in Superbroth (Quality Biologicals, Inc.). The plasmid DNAs were extracted and purified by the Qiagen Mini or Maxi Kits.
The construction of pMTGST537C, which contains a thrombin cleavage site between GST and the GR receptor sequences, was described previously by Xu et al. (21). pMTGST520C was constructed on the backbone of plasmid pdhfr537C (21). pdhfr537C was digested with EcoRI to generate two fragments (a 4.7-kilobase vector fragment and a 1.3-kilobase fragment containing DHFR and amino acids 537 to 781 of the rat GR). The sequence containing full-length GST was generated by PCR using pMTGST537C as the template. The PCR primers were as follows: 5Ј primer is 5Ј-GCCAGAATTCATGTCCCCTATACTAGG-3Ј; 3Ј primer is 5Ј-TATAGCATGCGGATCCACGCGGAA-3Ј. Amino acids 520 to 795 of GR were amplified by PCR using pSVLGR (22) as the template. The PCR primers were as follows: 5Ј primer is 5Ј-GTATAGCAT-GCCAGCAAGCCACTG-3Ј; 3Ј primer is 5Ј-CGGAATTCAACTTTCTT-TAAGGCAAC-3Ј. Both PCR products were digested with EcoRI plus SphI, and ligated to the 4.7-kilobase vector described above to generate pMTGST520C.
The pMTGST554C/ACA5 vector was used to convert the linker ACAAAAAC to APAAAAAC. Site-directed mutagenesis was performed according to the specification of the GeneEditor kit (Promega) using oligo ACA5/554PRO (5Ј-TGGATCCGCACCAGCTGCGGCCGCT-3Ј). A unique PvuII restriction site was created at the site of the Pro mutation, and PvuII digestion was used to confirm mutant.
Cell Growth and Transfection-Monolayer cultures of COS-7 cells were grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 5% heat inactivated fetal bovine serum. All the fusion proteins were expressed in transiently transfected COS-7 cells as described previously (23).
Assay of Steroid Binding-Cytosol of transiently transfected COS-7 cells containing the steroid-free receptors was obtained by the lysis of cells at Ϫ80°C and centrifugation at 15,000 ϫ g as described previously (24). [  Reconstitution of GST/GR⅐hsp90 Heterocomplexes-Aliquots (200 l) of undiluted cytosol from transfected COS-7 cells were immunoadsorbed with 15% anti-GST antibody or nonimmune IgG and subsequently incubated with 8-l pellets of protein A-Sepharose. Immunoadsorbed GST fusion proteins were stripped of associated hsp90 by incubating the immune pellets an additional 2 h at 4°C with 0.5 M NaCl, followed by one wash with 1 ml of TEG buffer (10 mM TES, pH 7.6, 50 mM NaCl, 4 mM EDTA, 10% glycerol) and a second wash with 1 ml of Hepes buffer (10 mM Hepes, pH 7.4). Immune pellets containing stripped fusion proteins were suspended in 50 l of rabbit reticulocyte lysate. Dithiothreitol (1 l) was added to each incubation to a final concentration of 5 mM, and 5 l of an ATP-regenerating system (50 mM ATP, 250 mM creatine phosphate, 2 mM MgOAc, and 100 units/ml creatine phosphokinase) were added to all assays to yield a final assay volume of 56 l. The assay mixtures were incubated for 20 min at 30°C with suspension of the pellets by shaking the tubes every 5 min. At the end of the incubation, the pellets were washed twice with 1 ml of ice-cold TEGM buffer (TEG buffer with 20 mM sodium molybdate), and GST fusions and hsp90 were assayed in each sample by Western blotting.
A portion of each immune pellet was assayed for steroid binding by incubation overnight in 100 l of TEGM buffer plus 4 mM dithiothreitol and 50 nM [ 3 H]TA. Samples were then washed twice with 1 ml of TEGM and counted by liquid scintillation spectrometry as described previously (25). The steroid binding is expressed as counts/minutes [ 3 H]TA bound/ anti-GST immune pellet prepared from 100 l of cytosol.
Gel Electrophoresis and Western Blotting-For assay of GST fusion proteins and associated hsp90, immune pellets were boiled in SDS sample buffer with 10% ␤-mercaptoethanol, and proteins were resolved on 7% SDS-polyacrylamide gels. Proteins were then transferred to Immobilon-P membranes and probed with 0.2% anti-GST or 0.01% aP1 for fusion proteins and 1 g/ml AC88 for hsp90. The immunoblots were then incubated a second time with 125 I-conjugated goat anti-mouse IgG to visualize the immunoreactive bands.
Cell-free Transcription and Translation-Cell-free transcription and translation of pSPGST537C and pSPGST547C was performed using a TNT kit (Promega) according to the procedure of the manufacturer. Aliquots of translation mixture containing [ 35 S]methionine-labeled fusion protein were immunoadsorbed with the 3G3 monoclonal IgM against hsp90 as described by Dalman et al. (11). The radiolabeled fusion protein co-adsorbed with hsp90 was detected by autoradiography. The immunoblot was cut just above the full-length fusion protein translation product and probed with AC88 for hsp90. The blot was incubated a second time with 125 I-labeled counterantibody and developed by autoradiography. The fusion protein and hsp90 bands were excised and counted for 35 S and 125 I radioactivity, respectively. Nonradioactive translation mixture was used for quantitation of [ 3 H]dexamethasone binding as described above.

Steroid Binding Properties of GST/GR HBD Fusion
Proteins-We previously utilized a series of DHFR/GR constructs with deletions at the amino-and carboxyl-terminal ends to determine the boundaries of the HBD of the rat GR to be amino acids 550 -795 (21). Attempts to examine the hsp90 binding properties of the DHFR/GR fusions adsorbed to methotrexateagarose were unsuccessful due to high nonspecific binding of hsp90 to the matrix. Thus, we constructed the series of GR HBD fusions with GST shown in Fig. 1. As noted previously, plasmids encoding GR segments 537-673 or 537-795 did not yield stable polypeptides but stable proteins were obtained when ␤-galactosidase or dihydrofolate reductase were fused upstream of these segments (21). As summarized in Fig. 1, the four GST/GR HBD fusions also produced stable proteins of similar abundance as determined by Western blotting (data not shown; see also below).
The GST537C fusion bound dexamethasone to the same extent and with the same affinity as the wild-type GR. The GST547C fusion also displayed the same steroid binding affinity as the wild-type GR but had only about 50% of its binding activity. Deletion of 5 additional amino acids from the amino terminus almost completely eliminated steroid binding activity. After minor correction for the amount of fusion protein present in cytosol from cells transfected with pMTGST552C, the steroid binding capacity was found to be ϳ5% of GST547C and ϳ2% of GST537C. Deletion of two more amino acids to yield GST554C further reduced even this trace of steroid binding activity.
Binding of GST/GR HBD Fusion Proteins to hsp90 in Cells-COS cell-expressed receptor chimeras were assayed for associated hsp90 by immunoadsorbing the fusion protein from transfected cell cytosol with anti-GST, washing the immune pellet, and Western blotting the pellet-bound fusion protein and the co-immunoadsorbed hsp90. As shown in lanes 1 of Fig.  2, both GST537C ( Fig. 2A) and GST547C (Fig. 2B) were recovered in native heterocomplexes with monkey (i.e. COS cell) hsp90. The immune pellet of GST547C contained less hsp90 than that of the GST537C pellet, which was expected in view of the lower steroid binding activity (Fig. 2B). The GST552C immune pellet (Fig. 2C) included only a very small amount of hsp90 (visible in the original photograph), consistent with the trace steroid binding, and no hsp90 was recovered with GST554C (Fig. 2D). The loss of hsp90 binding between 547 and 554 suggests that some feature of the HBD that is required for the hsp90-based chaperone machinery to form a stable HBD⅐hsp90 heterocomplex has been eliminated.
Role of Amino Acids 547-553 in Steroid Binding to Receptor in Whole Cells-Two possible explanations were considered for the decreased binding of steroid to receptors lacking the 7 amino acids 547-553. First, the further deletion could bring the GST molecule too close to the GR HBD and thus sterically interfere with hsp90, and consequently, steroid binding to the chimera. Second, the deleted sequences might be required for either a structural role and/or specific protein-protein interactions. To investigate these questions, an 8-amino acid ␣-helical linker (ACAAAAAC) was placed just amino-terminal to the receptor sequence of 554 -795 to give GSTACAAAAAC554C (GST554C/ACA5C; Fig. 3A). This chimera approximates GST547C as an 8-amino acid ␣-helix will now separate the GST and GR HBD domains. For a positive control, a similar 11amino acid ␣-helical linker (ACAAAAAAAAC) was inserted just amino-terminal to the receptor sequences to give GSTACAAAAAAAAC547C (GST547C/ACA8C), thus approximating GST537C (Fig. 3A). Interestingly, the ␣-helical linker did not increase the binding activity of the chimera containing just receptor sequences of 554C (Fig. 3A). This was not due to lack of expression of the fusion protein (Fig. 3B). Nor was the absence of binding due to some detrimental effect of an ␣-helical linker insert, as seen by the unaltered steroid binding affinity and only modest reduction in binding capacity for the control construct GST547C/ACA8C (Fig. 3A). On the basis of a general model based on the x-ray structures of retinoic acid receptor and 9-cis-retinoic acid receptor (RXR) HBDs (26), the three-dimensional structure of amino acids 547-553 of the rat GR is thought to be that of an ␣-helix. However, the presence of two prolines in this region suggests that a kink may be present near the beginning of this helix (helix 1). Such a kink could move the rest of the receptor (or GST in the fusion protein) out of the way of a surface required for hsp90 binding to GR. To test this hypothesis, a proline was introduced into GST554C/ACA5C to cause a break near the beginning of the introduced ␣-helix (Fig. 3A). This new chimera (GST554C/APA5C) contains the linker APAAAAAC. As indicated in Fig. 3A, the presence of this kinked linker still did not cause any increase in steroid binding over that seen for GST554C. Again, this was not due to the absence of expressed proteins in the transfected COS cells (Fig. 3B).
Cell-free hsp90 Binding to Stripped Chimeric Receptors-Rabbit reticulocyte lysate contains the multicomponent chap-erone system that forms stable GR⅐hsp90 heterocomplexes and reactivates steroid binding activity (4, 6). As shown in Fig. 2A, when the monkey hsp90 was stripped from the GST537C fusion protein, steroid binding activity was lost (lane 2). Incubation of the stripped immune pellet with reticulocyte lysate resulted in reassociation of a portion of the GST537C with rabbit hsp90 and there was ϳ30% reactivation of steroid binding activity (lane 4). The reticulocyte lysate system is dynamic in that receptor-hsp90 heterocomplexes are constantly being assembled and disassembled (27). We have recently shown that incubation of GR with reticulocyte lysate in the presence of [ 3 H]TA allows detection of complexes that would otherwise have been disassembled and thus not be in a steroid binding state by the end of the incubation (28). Thus, in lane 5 of Fig.  2A, when [ 3 H]TA was present during the incubation to bind the fusion protein as soon as complexes with rabbit hsp90 were formed, about 52 Ϯ 9% (ϮS.D., n ϭ 4) of the GST537C was restored to the steroid binding state. Surprisingly, the reactivation of steroid binding activity for stripped GST547C in the reticulocyte lysate system was much lower (8.9 Ϯ 0.5%, n ϭ 2) (Fig. 2B, lanes 4 and 5). Little or no hsp90 heterocomplexes were formed with GST552C or GST554C in either COS cells or the reticulocyte lysate reconstitution system (Fig. 2, C and D,  respectively).
Stability of GST/GR HBD⅐hsp90 Heterocomplexes-The decrease in recovery of reconstituted fusion protein-hsp90 heterocomplexes with truncation of the amino terminus of the HBD from 537 to 547 could reflect a progressive decrease in the cell-free stability of the heterocomplexes that were formed, rather than a progressive impairment of heterocomplex formation. In the experiment of Fig. 4, cytosols were prepared in molybdate-free buffer from cells transfected with either pMTGST547C or pMTGST537C. Aliquots of each cytosol were incubated for various times at 25°C and steroid binding activity was assayed. It can be seen that GST547C (open circles) loses its ability to bind steroid at the same rate as GST537C (closed circles).
The inset of Fig. 4 shows Western blots prepared from cytosol samples that were incubated for 60 min at 25°C and immunoadsorbed with anti-GST. The bands in the immunoblots were  3. Role of amino acids 547-553 in steroid binding. A, the various constructs of GST547C and GST554C Ϯ ␣-helical linkers. The highlighted sequence corresponds to the C-terminal region, starting at 554, that is common to each construct. The sequence between GST (not shown) and amino acid 554 of the wild-type receptor is shown for each construct, including the thrombin cleavage site of LVPRGS. The specific binding of [ 3 H]dexamethasone in cytosols of COS-7 cells that had been transiently transfected with receptor cDNA was determined by Scatchard analysis as described under "Methods." The disintegrations/min of specific binding is the amount above that of mock transfected COS-7 cells (1780 Ϯ 30 dpm, n ϭ 2). The K d is given when the listed specific binding was greater than twice that of mock transfected cells (i.e. total binding was twice that of mock transfected cells). The number of experiments is given in parentheses. B, Western blot of cytosols containing the receptors listed in A was performed as described under "Methods." excised and the radioactivity from the 125 I-labeled counterantibody was counted. Sixty-nine percent of the hsp90 disassociated from GST537C heterocomplexes and 74% from GST547C heterocomplexes during the 60-min incubation. Thus, we conclude that the two fusion proteins formed hsp90 complexes of similar stability.
Assembly of hsp90 Heterocomplexes with Fusion Proteins-The ability of hsp90 to associate with wild-type receptors is similar for receptors synthesized in intact cells and for in vitro translated receptors (11). Thus, when immunoadsorbed, stripped, wild-type GR from whole cell cytosols is incubated with reticulocyte lysate, we reactivate 70 -100% of the steroid binding activity of the immunoadsorbed native GR⅐hsp90 heterocomplex (29). When the wild-type GR is translated in reticulocyte lysates, all of the full-length [ 35 S]methionine-labeled translation product is in a heterocomplex with hsp90 (11). The very low efficiency of hsp90 binding to GST547C in the reculocyte lysate (lanes 4 and 5 versus lane 1 in Fig. 2B) thus raises the question of whether the reticulocyte lysate is deficient in some component that is required for the reconstitution of at least the GST547C. This question was examined by determining the ability to observe steroid binding in the various in vitro translated chimeric receptors. At the end of the reaction, molybdate was added to stabilize heterocomplexes and hsp90 was immunoadsorbed with the 3G3 monoclonal (IgM) antibody. In the experiment shown in Fig. 5, pSPGST537C and pSPGST547C were each co-immunoadsorbed with hsp90 in roughly the same proportion as hsp90 itself. The data from three separate translations are presented in the bar graph of Fig. 5C. In that immunoadsorption of ϳ70% of the hsp90 yielded co-immunoadsorption of roughly the same percentage of fusion protein, we conclude that all of the GST537C and GST547C translated in reticulocyte lysate was bound to hsp90. Thus, heterocomplex assembly during in vitro translation (Fig.  5) was more efficient than reconstitution of heterocomplexes with hsp90 by reticulocyte lysate (Fig. 2, A and B). This issue was further examined by determining whether the 8-amino acid ␣-helical linker of ACAAAAAC, which was unable to increase the steroid binding activity of the GST554C receptors expressed in intact cells, might have a different effect with in vitro translated receptors. The presence of the ␣-helical linker in GST554C/ACA5C did not cause any increase in the steroid binding activity of GST554C, despite the fact that comparable amounts of GST554C/ACA5C and GST547C receptor protein were made (data not shown).

DISCUSSION
Using GST/GR HBD fusion proteins, we have identified a region at the amino terminus of the GR HBD that is required for assembly of a stable heterocomplex with hsp90. Deletion of 10 amino acids from GST537C to form GST547C results in a reduction of both hsp90 binding and steroid binding activity (perhaps reflecting a reduced amount of fusion protein), but only minor effects on affinity, while deletion of the next 7 amino acids to form GST554C results in complete loss of stable HBD⅐hsp90 heterocomplex formation and steroid binding activity.
Cadepond et al. (14) and Schowalter et al. (15) have shown that segments of the GR or PR HBD lying carboxyl-terminal to helix 1 are sufficient for conferring stable hsp90 binding onto truncated receptors lacking the HBD. These observations led to the notion that several regions of the HBD are involved in binding hsp90 through some undetermined quality of the HBD tertiary structure (15). However, the approach used in those studies may have led to the formation of chimeras containing HBD segments that are incompletely folded or partially denatured. Thus, much like the carboxyl-terminal deletion approach of Dalman et al. (12) and Howard et al. (13), these experiments may have led to hsp90 heterocomplex formation in a manner that is quite different from the manner in which assembly occurs when the entire HBD assumes its normal, steroid-binding tertiary structure.
Using the approach of truncating from the amino-terminal side and leaving the rest of the HBD intact, we have shown that GST554C does not form a heterocomplex with hsp90 or have steroid binding activity. As shown in Fig. 5, all of the GST547C that is translated in reticulocyte lysate was bound to hsp90; thus, the loss of ability of the fusion protein to form a heterocomplex with hsp90 occurred between amino acids 547 and 554 and was almost complete with removal of just 5 amino acids to give GST552C (Fig. 2C). The loss of steroid and hsp90 binding activity appears to be sequence specific as replacement of the wild-type sequence of TPTLVSL with CAAAAAC did not regenerate any of the steroid binding activity lost with GST554C (Fig. 3A).
Several features of this 7-amino acid sequence that could influence hsp90 binding have been considered. As shown in Fig.  6, amino acids 547-554 of rat GR reside within what has been proposed to be helix 1 of the HBD of all steroid receptors (26) and overlap the amino-terminal one-third of an autonomous transactivation domain, AF2-a (30 -32) at the carboxyl terminus of the 2 domain (33). However, no conserved sequence is evident in an alignment with the other steroid-binding receptors (Fig. 6). The sequence of 547-553 of rat GR does contain four of the seven hydrophobic amino acids common to all steroid receptors (boxed in Fig. 6). The L550A/V551A mutations in GST547C/ACA5C and GST/APA5C would not be expected to dramatically alter the hydrophobic environment. The mutation of Leu-553 to the weakly polar cysteine is not likely to be of major importance as this residue is unchanged in GST552C, which retained only marginal binding activity (Fig. 2C). The invariant Leu-554, plus the other hydrophobic residues of 554 -562, are not sufficient in that GST554C has lost essentially all binding activity (Figs. 2D and 3A). A minimal physical separation between the GST and GR HBD domains, to avoid possible steric interference in hsp90 binding, appears to have been eliminated by the inactivity of the GST554C/ACA5C with an ␣-helical linker (Fig. 3A). To the extent that this region in the wild-type GR is predicted to be ␣-helical (26), the presence of an ␣-helix should not be detrimental. The presence of a polyalanine linker per se was also found not to present problems (Fig.  3A). Finally, the importance of a possible kink in the predicted helix 1 can be discounted by the inability of a kinked ␣-helix to restore binding activity in the context of the proline containing linker of GST554C/APA5C (Fig. 3A). Thus, we conclude that features of the amino acid sequence TPTLVSL of 547-553, such as tertiary structure or contributions to a larger surface for protein-protein interactions, are involved in hsp90 binding to the GR.
The marked decrease in reticulocyte lysate mediated reassociation of hsp90 with stripped GST547C, compared with GST537C, suggests that the 10 amino acids of 537-546 also have an important role under some conditions. In particular, the sequence of 537-546 is not required for hsp90 association in intact cells but is necessary for high levels of hsp90 reconstitution with stripped GR in the reticulocyte lysate. The molecular basis for this behavior is not known, but it may simply reflect variations in the efficiency of hsp90 association in the two systems. Alternatively, these results may reflect different initial, or intermediary, attachment points of hsp90 with newly translated GR (i.e. amino acids 547-554) versus pre-folded and then stripped GR (i.e. amino acids 537-547). In this case, the data would argue for hsp90 association with GR being a dynamic process with at least one intermediate being formed prior to the final product.
In conclusion, a combined assaying of steroid binding and hsp90 binding to GR chimeras has directly determined that the sequence upstream of, and including the amino-terminal region of, the GR HBD is required for hsp90 binding. This observation stands in opposition to the accepted model of steroid receptor-hsp90 heterocomplex assembly based on experiments with fragments of steroid receptor HBDs (14,15). Although binding of the HBD to hsp90 may ultimately require a general property such as exposure of hydrophobic residues in partially denatured regions as determined in in vitro reactions with other chaperones (18 -20), the requirement of a specific sequence for hsp90 binding to the intact HBD implies an initial event that enables interaction of the chaperone with more carboxyl-terminal hydrophobic residues not otherwise exposed. A structure defined by the 7-amino acid required sequence may be involved in an initial opening or unfolding of the steroid binding pocket by the multiprotein hsp90-based chaperone machinery (39 -41), an unfolding that exposes hydrophobic residues in the hsp90 contact region, permitting stable HBD⅐hsp90 heterocomplex assembly.