Direct evidence that the glucocorticoid receptor binds to hsp90 at or near the termination of receptor translation in vitro.

We have translated the rat glucocorticoid receptor in both reticulocyte lysate and in wheat germ extract. Receptor synthesized in the reticulocyte lysate is immunoadsorbed by the 8D3 monoclonal antibody directed against the 90-kDa heat shock protein (hsp90) and it has a normal ability to bind glucocorticoid in a high affinity manner. Although the wheat germ extract synthesizes the full length receptor, the receptor is not immunoadsorbed by 8D3 and we cannot demonstrate high affinity steroid binding. Receptor synthesized by the reticulocyte lysate can be immunoadsorbed by antibody directed against hsp90 as soon as the translation product is full length, suggesting that the receptor becomes associated with hsp90 late during translation or immediately at the termination of translation. When newly synthesized receptor is bound with steroid and incubated at 25 degrees C, it is converted to a form that binds to DNA. This study provides direct evidence that association of hsp90 with the glucocorticoid receptor is a very early event and that the newly formed heteromeric receptor-hsp90 complex is fully competent to undergo transformation.

is stabilized by molybdate contains two molecules of hsp90 per molecule of glucocorticoid binding protein (7-10): There is good reason to think that molybdate exerts its effects on the receptor by interacting with the binding site for an ubiquitous, endogenous metal anion that stabilizes the association of the GR with hsp90 and produces the same effects on GR function as molybdate (11,12). Both molybdate and hsp90 appear to interact with the steroid binding domain of the receptor (13,14).
Recently, we have reported that hydrogen peroxide, through its ability to promote disulfide bond formation, produces all of the effects on the GR that are produced by molybdate (15,16). Since both peroxide and the transition metal oxyanions stabilize the association of hsp90 with unliganded receptors and since conditions that disrupt the heteromeric complex inactivate the steroid binding capacity of cytosols, we postulated that interaction of hsp90 with the GR might be required to generate and/or stabilize a competent steroid binding conformation of the receptor (16). This proposal was supported by the observations that the glucocorticoid binding capacity of immunopurified receptors correlates with the relative amount of hsp90 that is present and that immunopurified, hsp90-free GR does not bind steroid (17).
One of the major problems that has hampered the study of the receptor-hsp90 interaction is that no one has yet been able to reassociate hsp90 with any steroid receptor. It has also not been possible to reassociate hsp90 with the avian viral transforming protein pp60"". The failure to reassociate hsp90 with the receptor has made it impossible to study directly any role of hsp90 in generating the steroid binding conformation of the GR. Glucocorticoid receptors that are translated i n vitro by a reticulocyte lysate system are able to bind steroid (e.g. Refs. 18 and 19), and in this paper, we show that GR synthesized by the reticulocyte lysate i n vitro becomes associated with hsp90 during or immediately at the termination of translation. Although GR synthesized in the rabbit reticulocyte lysate binds glucocorticoid with high affinity, GR synthesized in wheat germ lysate, where there is no identifiable hsp90, does not bind steroid. While this work was in progress, Denis and Gustafsson (20) reported that rat glucocorticoid receptor translated in reticulocyte lysate has a sedimentation coefficient or 9 S and can be transformed to the DNA binding state by heating the lysate at 25 "C in the presence of steroid.  , radioinert dexamethasone, nonimmune mouse IgG, TES, Tris,  Hepes, protein A-Sepharose CL-4B, bovine serum albumin, and goat anti-mouse IgM were from Sigma. Immobilon P membranes were from Millipore Corp. Rabbit reticulocyte lysate, wheat germ lysate, and Riboprobe core system transcription kit were from Promega. BuGR2 monoclonal antibody prepared against the rat glucocorticoid receptor (21) was kindly provided by Dr. Robert Harrison I11 (Dept.

Materials
of Medicine, Univ. of Arkansas, Little Rock), and the AC88 monoclonal antibody against the 90-kDa heat shock protein (22) was kindly provided by Dr. David Toft (Dept. of Biochemistry and Molecular Biology, Mayo Medical School, Rochester, MN). The 8D3 monoclonal antibody (IgM) was originally prepared against partially purified Ah receptor (the dioxin receptor) and subsequently found to be specific for hsp90 (23). The EC1 monoclonal antibody, which was originally prepared against partially purified rabbit progesterone receptor and subsequently found to react with a 59-kDa receptor-associated protein (24), was kindly provided by Dr. Lee Faber (Depts. of Obstetrics/ Gynecology and Physiology, Medical College of Ohio, Toledo, OH). Plasmid T3.1118 containing the gene for the rat glucocorticoid receptor was kindly provided by Dr. Keith Yamamoto (Dept. of Biochemistry and Biophysics, University of California, San Francisco).

I n Vitro
Transcription and Translation-Transcription reactions (100 pl) contained 2 yg of PuuII-linearized plasmid T3.1118, 2.5 mM ATP, CTP, GTP, and UTP, 10 mM DTT, 1 unit/pl RNasin, 20 units SP6 polymerase in the buffer provided by Promega. After 2 h at 37 "C, the mixture was extracted with phenol, followed by an extraction with chloroform and precipitation of nucleic acids with ethanol. The nucleic acid pellet was dissolved in 10-30 pl of diethyl pyrocarbonate-treated water.
lysate or wheat germ lysate, 1 mM amino acids mix (minus methio-Samples were incubated at 30 "C for times up to 1 h, and reactions were stopped by cooling in ice or in one experiment by incubation with 50 p~ puromycin. For experiments in which hormone binding was measured and when translation products were analyzed by immunoblotting, 1 mM radioinert methionine was used instead of the [35S]methionine. Preparation of L Cell Cytosol-L929 murine fibroblasts were grown in monolayer culture in modified Eagle's medium supplemented with calf serum. Cells in log phase were harvested by scraping into Earle's saline and centrifuged at 600 X g for 5 min. Following a wash by resuspension into Earle's saline and centrifugation, cells were resuspended in 1.5 volumes of 10 mM Hepes, 1 mM EDTA, 10 mM molybdate, pH 7.35 at 4 "C, and ruptured by Dounce homogenization. The homogenate was centrifuged at 100,000 X g for 1 h, and after removal of the floating lipid layer the supernatant (referred to as cytosol) was frozen. Incubation with Antibodies and Adsorption to Protein A-Sepharose"BuGR2, AC88, and EC1 antibodies were prebound to 30 pl of protein A-Sepharose pellets in 0.1 ml of TEG buffer for 2 h at 4 "C with rotation and pellets were then washed once with TEG buffer to remove unbound antibody. The 8D3 monoclonal antibody against hsp90 (30 pg of IgM) was adsorbed to 30 pl of protein A-Sepharose that had been prebound with goat IgG directed against mouse IgM. Translation reactions (25 pl) were diluted to 100 pl with cold TEG buffer containing 20 mM molybdate and were added to protein A-Sepharose pellets with prebound antibodies. After rotation for 3 h a t 4 "C, Sepharose pellets were washed three times with 1-ml aliquots of TEG buffer containing 10 mM molybdate. Samples were then boiled in SDS sample buffer and proteins were resolved by SDS-PAGE.
Gel Electrophoresis and Imrnunoblotting-SDS-polyacrylamide gel electrophoresis was performed in 1.5-or 3-mm 8% slab gels according to Laemmli (25). Gels were cooled to 4 "C during electrophoresis. All samples were extracted from protein A-Sepharose by boiling in SDS sample buffer containing 10% P-mercaptoethanol. M, standards were: Immunoblotting was carried out by transferring proteins from polyacrylamide slab gels to an Immobilon P membrane, followed by overnight incubation with 1% BuGR antibody against the glucocor-ticoid receptor or 0.3% AC88 antibody against hsp90. Immunoblotted proteins were detected by reaction with 1251-conjugated goat antimouse IgG, followed by autoradiography.
Steroid Binding-Translation reactions prepared with nonradioactive methionine were diluted 1:l with Hepes buffer containing 40 mM sodium molybdate and 20 mM DTT and incubated for 4 h at 0 "C with either 100 nM [3H]dexamethasone mesylate or 25 nM [3H] triamcinolone acetonide in the presence or absence of either 50 or 25 p M radioinert dexamethasone, respectively. Receptors bound with [3H]dexamethasone 21-mesylate were resolved by SDS-PAGE. In the case of [3H]triamcinolone acetonide-bound receptors, free steroid was removed with charcoal and specific binding was determined by subtracting the nonspecific value obtained in the presence of nonradioactive dexamethasone from the total binding obtained in the presence of vehicle. In the rabbit reticulocyte lysate, nonspecific binding of [3H]triamcinolone acetonide is 15-20% of the total binding, and in the wheat germ lysate, all of the binding observed a t 25 nM [3H] triamcinolone acetonide is nonspecific. Twenty-five nM triamcinolone acetonide is well beyond the concentration required to occupy all receptor sites having a normal dissociation constant for this steroid in the range of 2 nM (26).
Receptor Transformation and DNA Binding-Reticulocyte lysate containing [%]methionine-labeled translation products was diluted with an equal volume of 10 mM Hepes buffer, pH 7.4, with 10 mM DTT and 10% glycerol, with or without 40 mM sodium molybdate, and incubated 2 h on ice with 100 nM triamcinolone acetonide. This mixture containing steroid-bound receptors was diluted with 2.5 volumes of Hepes buffer with or without molybdate and incubated 45 min a t 25 "C to transform receptors. An equal volume of 12.5% DNAcellulose was added and the mixture was rotated at 4 "C for 45 min. After washing the DNA-cellulose pellet three times, DNA-bound proteins were eluted by boiling in 2 X SDS sample buffer.
Purification of h p 9 0 from L Cells-hsp9O was purified by a modification of the method of Welch and Feramisco (27). L cell cytosol (20 ml a t 10.7 mg of protein/ml) was chromatographed on a 2 X 20cm DEAE column equilibrated in 10 mM Tris, 0.1 mM EDTA, pH 7.1, and proteins were eluted with a 200-ml gradient of 0-0.4 M KC1. Samples of eluted fractions were resolved by SDS-PAGE and hsp90 was detected by immunoblotting with the AC88 antibody. Fractions containing hsp90 were pooled and concentrated by adsorption with polyethylene glycol. The concentrated material was diluted with an equal volume of 20 mM K2HPOa, 1 mM EDTA, pH 7.5, it was chromatographed on a 2 X 8-cm hydroxylapatite column (which was equilibrated in the same K2HP04 buffer), and proteins were eluted with a 200-ml gradient of 0-0.4 M K2HP0,. Fractions containing hsp90 were pooled, concentrated by polyethylene glycol adsorption, and frozen in aliquots at -70 "C. Prior to addition to the translation reactions, the hsp90 preparation was diluted in Hepes buffer and concentrated by Centricon filtration to remove salt. hsp90 prepared in this manner is about 98% pure.2 The protein concentration of the hsp90 preparation was determined by amino acid analysis.

Properties of GR Synthesized by Reticulocyte and Wheat
Germ Lysates- Fig. 1 shows the [35S]methionine-labeled proteins synthesized by rabbit reticulocyte and wheat germ lysates after addition of Brome mosaic viral RNA or GR RNA.
With 60 min of translation, both systems produce the full length rat GR. As with a large number of other mRNAs, the wheat lysate is less efficient than the reticulocyte lysate at synthesizing both the viral proteins and the GR. For this reason, twice as much GR RNA and two to three times as much wheat germ lysate was added to subsequent incubations as to those with reticulocyte lysate. In the wheat germ system, the specific activity of the [35S]methionine is not diluted by endogenous methionine, but according to the manufacturer, the reticulocyte lysate contains about 5 pM endogenous methionine; thus, the difference in translation efficiency is greater than that indicated by the [35S]methionine autoradiograms presented in this paper.
From the experiment shown in Fig. 2, one can obtain a general impression of the relative amount of rat GR synthesized in the reticulocyte lysate. In this experiment, the rat GR was immunoadsorbed from the lysate, resolved by SDS- PAGE, and immunoblotted with the BuGR anti-GR antibody, followed by "'I-conjugated anti-mouse IgG. The autoradiogram shows the in uitro translated rat GR in lune 4 and the GR contained in 10 p1 of mouse L cell cytosol in lune 5. For unknown reasons, the cytosolic mouse GR routinely migrates at 98-100 kDa, whereas the rat GR migrates at 94 kDa (see Fig. 5 in Ref. 16), despite the fact that the rat GR has a slightly longer amino acid chain. Fig. 2 shows that the in uitro translated rat GR migrates in a similar more rapid fashion with respect to the cytosolic mouse GR. Our standard L cell cytosol contains about 10 pmol of GR/ml. By excising and counting the radioactivity in the receptor bands in lunes 4 and 5, it was estimated that the lysate reaction synthesized an amount of rat GR equivalent to that present in 0.64 p1 of L cell cytosol or 180 fmol of rat GR/ml of lysate (assuming equivalent reaction of the BuGR antibody with the rat and mouse GR). The experiment of Fig. 3 examines the steroid binding capacity of the GR translated in both systems. Fig. 3A shows the [3'S]methionine-labeled receptor and Fig. 3B shows the translation products after site specific affinity labeling with [3H]dexamethasone 21-mesylate. The [3'S]methionine-labeled bands (the full length product only) and the [3H]dexamethasone 21-mesylate-labeled band were excised and counted. The wheat germ lysate synthesized 87% as much full Consistent with the [3H]dexamethasone 21-mesylate labeling, only GR synthesized by the reticulocyte lysate bound [3H]triamcinolone acetonide as determined by charcoal absorption assay (Fig. 3C). In this particular synthesis, there were 2.7 pmol of [3H]triamcinolone acetonide bound specifically per ml of reticulocyte lysate. This estimate of the concentration of synthesized rat GR is roughly an order of magnitude higher than the estimate made from radioactive immunoblotting in the experiment of Fig. 2. It should be noted, however, that the amount of GR synthesized from one experiment to the next varies over about a &fold range, both on the basis of [35S]methionine labeling and [3H]steroid binding. Fig. 3 0 shows the concentration dependence of [3H] triamcinolone acetonide binding to the reticulocyte lysate translation product. The apparent KD from one half struration is about 3 nM, which is a normal binding affinity for this ligand (26).
GR Translated in the Reticulocyte Lysate Is Bound to hp90"The reticulocyte lysate contains a 90-kDa protein that reacts with the AC88 monoclonal antibody against hsp90 on immunoblotting, but the wheat germ lysate does not (data not shown). The amount of hsp90 in the reticulocyte lysate can be estimated by the method shown in Fig. 4. In this method, aliquots of reticulocyte lysate or L cell cytosol were resolved on SDS-PAGE along with standardized amounts of purified hsp90. The relative amount of hsp9O in each lane was determined by quantitative immunoblotting with AC88 followed by ['251]anti-mouse IgG. After excision of the bands and counting, the amount of hsp90 is determined from a standard curve as shown in the figure. From two such experiments, we determined that the reticulocyte lysate contains about 0.15 mg of hsp90/ml of lysate, or about 1.6 pg of hsp90/mg protein. This is comparable to the concentration of hsp90 in our standard L cell cytosol, which we calculate to be 0.09 mg of hsp90/ml or 4.5 pg/mg protein (L cell cytosol has a lower protein content than reticulocyte lysate). Thus, L cell cytosol contains roughly 1 ~L M hsp90, with the ratio of hsp90 to GR being approximately 1OO:l. The reticulocyte lysate contains approximately 2 p~ hsp90, and because of the relatively low and variable amounts of GR produced from one experiment to another, the hsp9O to GR ratio ranges from about 300:l to more than 1500:l. Because the absolute amount of GR synthesized in the 25-50 pl of reticulocyte lysates is so small, we cannot employ our usual technique of detecting receptor-associated hsp90, which involves immunoadsorbing receptors with the BuGR antireceptor antibody and then detecting the immune-specific presence of hsp90 by immunoblotting using AC88 as a probe for hsp90 (5,6,17). Thus, we used the method shown in Fig.  5, which takes advantage of the strong [35S]methionine signal produced by the limited number of GR molecules that are present. As shown in lane 6 of Fig. 5, the GR can be immunoadsorbed with the 8D3 monoclonal (IgM) and absorption is immune-specific as suggested by the protein A-Sepharosebound anti-IgM control of lane 5. In contrast, the EC1 monoclonal antibody against the 59-kDa receptor-associated protein does not immunoadsorb the newly synthesized receptor.
In the experiment of Fig. 6, the amount of protein A- Quantitation of the hsp90 concentration in rabbit reticulocyte lysate and L cell cytosol. Aliquots of L cell cytosol (7.5 pl) and rabbit reticulocyte lysate (5 pl) were resolved by SDS-PAGE on a gel that also contained standardized amounts of purified L cell hsp90. The proteins were transferred to an Immobilon P membrane and the blot was probed with the AC88 antibody, followed by '"I-conjugated goat anti-mouse IgG and peroxidase-conjugated rabbit anti-goat IgG. The peroxidase-stained hsp9O bands were excised and counted for ' "I radioactivity. The inset shows an autoradiogram of the gel: lane 1, L cell cytosol; lane 2, rabbit reticulocyte lysate; lanes 3-7,0.08, 0.16, 0.32, 0.63, and 0.95 pg of purified hsp90. The solid circles in the main figure represent the radioactivity from the purified hsp90 standards and the triangle and the square represent the radioactivity from L cell and reticulocyte hsp90, respectively, plotted on the standard curve.
Sepharose-bound 8D3 anti-hsp90 antibody was tripled to permit immunoadsorption of most of the hsp90 in the reticulocyte lysate. The relative amount of hsp90 (Fig. 6A) recovered in the pellet and supernatant after immunoadsorption was determined by immunoblotting samples and probing them with the AC88 antibody against hsp90 followed by '251-~~njugated goat anti-mouse IgG. The relative amount of GR in each sample was determined after excising and counting the complete translation product labeled with [35S]methionine (Fig. 6B). When this larger amount of antibody was used, we immunoadsorbed 65% of the hsp90 and 51% of the [35S] methionine-labeled full length translation product. This suggests that nearly all of the newly transformed GR may be bound to hsp9O and would be immunoadsorbed if all of the hsp90 could be immunoadsorbed.
Evidence That Binding to hsp90 Occurs Very Late in the Translation Process-To determine if the GR was undergoing cotranslational association with hsp90, we performed time course experiments like that of Fig. 7. As shown in Fig. 7A, full length receptor produced after 20 or 30 min of translation is coadsorbed in an immune-specific manner by the 8D3 monoclonal antibody against hsp90. In some experiments, some full length receptor has been synthesized by 10 min and this nascent receptor is immunoadsorbed by 8D3 (see inset, Fig. 7A), suggesting that the GR binds to hsp90 either late during translation or as soon as translation is complete.
We had hoped that we might be able to map the hsp90 binding site by determining if there was a critical length at which the translation product could be adsorbed by the 8D3 antibody. However, we are unable to absorb anything shorter than the full length product in an immune-specific manner. This is illustrated in Fig. 7B, where a ladder of translation products up to an apparent M, of about 90,000 is shown in lane 3. Lanes 1 and 2 show the [35S]methionine radioactivity adsorbed by nonimmune mouse IgM and 8D3, respectively. For comparison, lane 4 shows the full length GR produced after 20 min of incubation, which is readily adsorbed by 8D3.
The Newly Translated Receptor Can Be Transformed-As shown in Fig. 8, heating the newly translated, steroid-bound receptor at 25 "C increases its ability to bind to DNA-cellulose and this transformation is inhibited by molybdate. Transformation is accompanied by conversion of the receptor to a form that is no longer immunoadsorbed by the 8D3 antibody (not shown); thus, the 8D3 antibody does not immunoadsorb the GR because of a nonspecific interaction with the GR protein itself.

DISCUSSION
In light of the observation that there is a correlation between the amount of hsp9O bound to the GR and its steroid binding activity (17), it is interesting that the full length GR synthesized in the wheat germ lysate does not bind more than trace amounts of steroid. A heat shock response similar to that observed in animal cells has been demonstrated in cultured plant cells (e.g. Ref. 28) and high M, heat shock proteins in the 80,000-94,000 range have been demonstrated (29). However, we have not been able to determine from the literature if plants, such as wheat, produce a protein that is in any way analogous to hsp90. Although there are a variety of potential explanations for our failure to demonstrate steroid binding activity on the part of the full length translation product synthesized in wheat germ lysate, absence of hsp90 or of an hsp90-like homologue could be an important factor. The trace amounts of specific [3H]dexamethasone 21-mesylate labeling of the wheat germ translation product that we have occasionally observed could represent binding to a low affinity state of the steroid binding site with consequent very low covalent labeling by [3H]dexamethasone 21-mesylate.
We have added purified mouse hsp90 to the wheat germ lysate to see if the translation product was converted to a form that could bind steroid and be immunoadsorbed with the 8D3 monoclonal antibody. The GR translated in the wheat germ lysate in the presence of purified hsp90 does not bind [3H]triamcinolone acetonide or more than trace amounts of [3H]dexamethasone 21-mesylate and it is not immunoadsorbed by 8D3. Thus, it appears that the wheat germ lysate lacks some property that permits the GR to become associated with the heat shock protein. We have added the same amount of hsp9O that we determined was normally present in the reticulocyte system (Fig. 4). As recently reported by Rose et al. (30), we find that addition of the purified hsp90 inhibits translation in the rabbit reticulocyte system, but addition of the same preparation of hsp90 to the wheat germ lysate does not inhibit GR synthesis. This may again suggest that the two systems are intrinsically different in their ability to respond to hsp90.
In another approach to directly determine if hsp90 is required for synthesis of a GR with a high affinity steroid binding site, we have tried to eliminate the endogenous hsp90 from the reticulocyte lysate prior to using it for the translation reaction. We have found that this approach is simply not possible, because any manipulations involved in immunoadsorption (even preabsorption with protein A-Sepharose without any antibody) result in a drastic reduction in translation activity.
Our inability to demonstrate immune-specific isolation of newly synthesizing GR shorter than the full length translation product could be explained in a couple of ways. It is possible that hsp90 binds to the GR prior to completion of translation but that only the full length translation product-hsp9O complex accrues in sufficient amount to allow detection with our methods. One could argue that the complex is not accessible to the 8D3 antibody until it has dissociated from the translation machinery, but we have terminated the translation by puromycin release and still cannot detect immune-specific absorption of incomplete translation products. Another possibility is that hsp90 binds only at the end of translation.
Rusconi and Yamamoto (31) have shown that deletion of the five carboxyl-terminal amino acids reduces the affinity of the GR for dexamethasone by two orders of magnitude. Car-   (lunes 4-6) of molybdate, and binding to DNA-cellulose was assayed as described under "Methods." Lunes 1 and 4 , total translation products; lunes 2 and 5, products bound to DNA-cellulose; lunes 3 and 6, supernatant after DNA-cellulose absorption.
lstedt-Duke et al. (32) have demonstrated that the A ring of [3H]triamcinolone acetonide forms a covalent adduct with Cys-754 on UV irradiation, making it clear that residues within 40 amino acids of the carboxyl terminus at Lys-795 are participating directly in the formation of the steroid binding pocket. Thus, the carboxyl terminus would seem to be critical for formation of a high affinity steroid binding conformation of the GR and it is possible that it is critical for association with hsp90.
After longer periods of translation (20 min or more), we can see the formation of degradation products (or possibly downstream initiation products) which can be immunoadsorbed with 8D3. The major band at M , -47,000 in Fig. 6B, for example, is bound to hsp90, binds steroid (see faint band in Fig. 3B) and is transformed to the DNA binding state (Fig.  8). Mendel et al. (33) have shown that the rat GR can undergo considerable degradation and still be maintained in a molybdate-stabilized heteromeric complex, and Denis et al. (14) have cleaved the receptor to the M, -27,000 carboxyl-terminal fragment containing the steroid binding domain and have maintained the 9 S complex. When taken together with the results of experiments analyzing the physical state of receptors produced in cells transfected with GR cDNAs containing various domain deletions (13), these data strongly argue that the steroid binding domain of the GR contains the major sites of association with hsp90. We have recently shown that the hormone-free GR can be cleaved to the 27-kDa meroreceptor fragment with retention of both hsp90 binding and of high affinity steroid binding activity (17). Thus, it is not surprising that fragments smaller than the full length translation product can bind steroid and are associated with hsp90. But as far as we can tell at the moment, the GR must be essentially completely synthesized before there is detectable hsp90 binding, and it is likely that some hsp90-bound fragments are subsequently generated by proteolysis.
The observation that newly synthesized GR is associated with hsp90 is consistent with a model in which hsp90 is required for repression of receptor function (13). Thus, the GR would be in an inactive state in the cell from the time of its synthesis until it was subsequently exposed to hormone. It has been speculated that the binding of hsp9O to the GR may serve to stabilize the receptor in an unfolded state that is necessary for the high affinity steroid binding conformation (17,34). We would speculate that the attachment of hsp90 to the GR at or near the termination of the receptor translation process may reflect some type of chaperone function of this ubiquitous, abundant, and conserved stress protein.
Association of hsp90 with a variety of proteins may play important general roles in maintaining protein stability and in protein attachment to systems responsible for intracellular protein movement and transmembrane passage. The association of hsp90 with newly synthesized pp60"" may reflect such a primitive and necessary cellular function, facilitating its passage to the plasma membrane. One can envision, however, that during the process of evolution, the steroid receptor family and perhaps other transcription factors whose activity is regulated by small ligands, could have evolved a regulatory center which allows the subsequent dissociation of the protein-hsp90 complex to be directly controlled by the ligand as the primary event in signal transduction.
Such a model predicts that other as yet unidentified proteins that act in the nucleus and do not possess such a regulatory center may also be attached to hsp90 in a transient