Folding of the Glucocorticoid Receptor by the Heat Shock Protein (hsp) 90-based Chaperone Machinery

In cytosols from animal and plant cells, the abundant heat shock protein hsp90 is associated with several proteins that act together to assemble steroid receptors into receptor·hsp90 heterocomplexes. We have reconstituted a minimal receptor·hsp90 assembly system containing four required components, hsp90, hsp70, p60, and p23 (Dittmar, K. D., Hutchison, K. A., Owens-Grillo, J. K., and Pratt, W. B. (1996) J. Biol. Chem. 271, 12833–12839). We have shown that hsp90, p60, and hsp70 are sufficient for carrying out the folding change that converts the glucocorticoid receptor (GR) hormone binding domain (HBD) from a non-steroid binding to a steroid binding conformation, but to form stable GR·hsp90 heterocomplexes, p23 must also be present in the incubation mix (Dittmar, K. D., and Pratt, W. B. (1997)J. Biol. Chem. 272, 13047–13054). In this work, we show that addition of p23 to native GR·hsp90 heterocomplexes immunoadsorbed from L cell cytosol or to GR·hsp90 heterocomplexes prepared with the minimal (hsp90·p60·hsp70) assembly system inhibits both receptor heterocomplex disassembly and loss of steroid binding activity. p23 stabilizes the GR·hsp90 heterocomplex in a dynamic and ATP-independent manner. In contrast to hsp90 that is bound to the GR, free hsp90 binds p23 in an ATP-dependent manner, and hsp90 in the hsp90·p60·hsp70 heterocomplex is in a conformation that does not bind p23 at all. The effect of p23 in the minimal GR heterocomplex assembly system is to stabilize GR·hsp90 heterocomplexes once they are formed and it does not appear to affect the rate of heterocomplex assembly. Molybdate has the same ability as p23 to stabilize GR heterocomplexes with mammalian hsp90, but GR heterocomplexes with plant hsp90 are stabilized by p23 and not by molybdate. We propose that incubation of the GR with hsp90·p60·hsp70 forms a GR·hsp90 heterocomplex in which hsp90 is in an ATP-dependent conformation. The ATP-dependent conformation of hsp90 is required for the hormone binding domain to have a steroid binding site, and binding of p23 to that state of hsp90 stabilizes the GR·hsp90 heterocomplex to inactivation and disassembly.

Steroid receptors are recovered from hormone-free mammalian cells in multiprotein complexes that contain hsp90, 1 a 23-kDa protein (p23), one of several immunophilins, and, often, substoichiometric amounts of hsp70 (for review, see Refs. 1 and 2). Several protein kinases involved in signal transduction (e.g. Src, Raf) exist in similar cytosolic heterocomplexes with hsp90 (1,2), and genetic experiments in yeast show that hsp90 is an integral component of both steroid receptor and protein kinase signaling pathways (see Nathan and Lindquist (3) and references therein). Steroid receptor⅐hsp90 heterocomplexes can be created in vitro by incubating immunoadsorbed, hormone-free receptors with rabbit reticulocyte lysate (4,5). This cell-free heterocomplex assembly system also forms heterocomplexes between hsp90 and protein kinases, such as Src, Raf, and Mek (6 -8). In that concentrated lysates from mammalian, insect, and even plant cells are able to assemble the glucocorticoid receptor (GR) into a complex with hsp90 (9), it is likely the heterocomplex assembly system performs an ubiquitous and basic cellular function.
The hormone binding domain (HBD) of the GR must be bound to hsp90 for it to bind steroid (10), and the heterocomplex assembly system converts the GR HBD from a non-steroidbinding to a steroid-binding form (5,11). Physical studies suggest that the HBD of the hsp90-free GR is in a folded conformation in which the steroid-binding pocket is not accessible to ligand and that the hsp90-bound HBD may be partially unfolded, such that a hydrophobic binding pocket is accessible to the ligand (12). Using formation of steroid binding sites as a rapid assay of "folding" 2 as well as direct assay of receptor⅐ hsp90 complex assembly, several details of the assembly mechanism have been established. Assembly of receptor⅐hsp90 complexes requires ATP/Mg 2ϩ , a monovalent cation, such as K ϩ (11,13), and at least three proteins, hsp70 (13,14), p60 (15), and p23 (16,17). Although the mammalian DnaJ homolog, hsp40, has not been demonstrated in receptor heterocomplexes, it is possible that it also is required for heterocomplex assembly.
The p60 component of the assembly system was first identified by Smith et al. (13) in progesterone receptor heterocomplexes formed in reticulocyte lysate when ATP was limiting. The rabbit p60 (18) was then shown to be the homolog of a human protein that had been cloned by Honoré et al. (19) and a homolog of the nonessential yeast heat shock protein Sti1 * This investigation was supported in part by National Institutes of Health Grant DK31573 (to W. B. P.) and by a grant from the National Sciences and Engineering Research Council of Canada (to P. K.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Trainee under Pharmacological Sciences Training Program Training Grant GM07767 from the National Institutes of Health.
To whom correspondence should be addressed: Dept. of Pharmacology, 1301 Medical Science Research Bldg. III, University of Michigan Medical School, Ann Arbor, MI 48109-0632. Tel.: 313-764-5414; Fax: 313-763-4450. (20). Smith (21) showed that p60 associates with progesterone receptors incubated with reticulocyte lysate early in the heterocomplex assembly process and then exits the complex. Chen et al. (22) have shown that p60 binds to hsp70 via an Nterminal tetratricopeptide repeat region, and it binds to hsp90 via a central tetratricopeptide repeat region to form an hsp90⅐p60⅐hsp70 complex. Recently, we have demonstrated that the mixture of purified rabbit hsp90 and hsp70 with bacterial lysate containing human p60 results in spontaneous formation of an hsp90⅐p60⅐hsp70 complex that can be adsorbed with an anti-p60 antibody, and the resulting immune complex converts the GR HBD to a steroid binding state in an ATP-dependent and K ϩ -dependent manner (23). In this minimal reconstituted system, conversion of the GR HBD to the steroid binding state was detected by incubating the GR with the chaperones in the presence of [ 3 H]triamcinolone acetonide, which binds to the receptor as soon as GR⅐hsp90 complexes are formed. Thus, despite the fact that the heterocomplexes formed by this minimal system undergo rapid inactivation and/or disassembly, evidence of proper folding of the GR HBD to the steroid binding conformation could be preserved.
To demonstrate maximum formation of GR⅐hsp90 heterocomplexes directly by immunoblotting, p23 must be present in the heterocomplex assembly mixture along with hsp90, hsp70, and p60 (23). p23 is a 23-kDa acidic protein that was first coimmunoadsorbed from cell lysates with the chicken progesterone receptor (24) and the murine GR (25). Toft and his co-workers prepared the human cDNA and showed that p23 was a unique and ubiquitous, 160-amino acid protein with an aspartic acid-rich COOH-terminal domain (26). p23 binds directly to hsp90 by a process that requires ATP (27,28). Interestingly, when human p23 is added to wheat germ extract, it binds to plant hsp90 in an ATP-dependent manner (29), and it has the same effect on GR heterocomplex assembly by plant chaperones as it does when added to a heterocomplex reconstitution mixture containing purified mammalian hsp90⅐p60⅐ hsp70 (17). This suggests that the binding of p23 to hsp90 is a highly conserved protein interaction in eukaryotic cells.
A recent study of hsp90 binding to phenyl-Sepharose suggests that hsp90 can exist in two functional states; in one state, hsp90 is bound by ADP and has a high affinity for hydrophobic resin, and in the other state, hsp90 is bound by ATP and has a low affinity for hydrophobic resin (30). In direct experiments utilizing the purified proteins, it was shown that p23 binds to the ATP-dependent state of hsp90 and stabilizes it in the conformation with low affinity for hydrophobic resin (30).
Although p23 is required for the reconstituted assembly system to form stable GR⅐hsp90 heterocomplexes (23), it is not known whether p23 accelerates the rate of heterocomplex assembly or stabilizes assembled heterocomplexes, or both. It is also not known at what stage in the assembly process p23 can enter receptor heterocomplex. In this work, we show that p23 engages in a dynamic association with hsp90, not only when hsp90 is free as shown by Toft and his co-workers (27,30) but to hsp90 in GR⅐hsp90⅐p60⅐hsp70 complexes formed by the minimal assembly system and also to native GR⅐hsp90 complexes formed in intact cells. In contrast to the ATP dependence of p23 association with free hsp90, the association of p23 with hsp90 that is complexed with the GR does not require ATP. p23 stabilizes GR⅐hsp90 heterocomplexes once they are formed and does not appear to accelerate their rate of formation. Molybdate has the same effect as p23 in the reconstituted heterocomplex assembly system except when purified rabbit hsp90 is replaced with a purified, recombinant plant hsp90, which has equivalent heterocomplex forming activity and is stabilized by p23 but not by molybdate. Considering our observations together with Toft laboratory's observation that p23 binds an ATP-dependent state of hsp90, we would propose that the ATP-dependent conformation of hsp90 is required for the HBD to have a high affinity steroid binding site and that binding of p23 to that state of hsp90 stabilizes the GR⅐hsp90 heterocomplex against inactivation and disassembly. [6, H]Triamcinolone acetonide (42.8 Ci/mmol) and 125 I-conjugated goat anti-mouse and anti-rabbit IgGs were obtained from DuPont NEN. Untreated rabbit reticulocyte lysate was from Green Hectares (Oregon, WI), and wheat germ extract was from Promega. Protein A-Sepharose and goat anti-mouse and anti-rabbit IgG horseradish peroxidase conjugates were from Sigma. The BuGR2 monoclonal IgG antibody against the GR was from Affinity Bioreagents (Golden, CO). The AC88 monoclonal IgG against hsp90 and the N27F3-4 anti-72/73-kDa hsp monoclonal IgG (anti-hsp70) were from StressGen (Victoria, British Columbia). The JJ3 monoclonal IgG against p23 and Escherichia coli expressing human p23 were gifts from Dr. David Toft (The Mayo Clinic). The DS14F5 monoclonal IgG against p60 and E. coli expressing p60 were kindly provided by Dr. David Smith (University of Nebraska Medical School). Actigel-ALD (activated aldehyde agarose) affinity support for protein immobilization was from Sterogene Biochemicals (San Gabriel, CA). Hybridoma cells producing FiGR monoclonal IgG against the GR were generously provided by Dr. Jack Bodwell (Dartmouth Medical School).

Methods
Cell Fractionation and Immunoadsorption-L929 mouse fibroblasts (L cells) were grown in monolayer in Dulbecco's modified Eagle's medium supplemented with 10% bovine serum. Cells were harvested by scraping into Earle's balanced saline, suspended in 1.5 volumes of HE buffer (10 mM Hepes, 1 mM EDTA, pH 7.4) and ruptured by Dounce homogenization. Homogenates were centrifuged for 1 h at 100,000 ϫ g, and the supernatant is referred to as "cytosol." Immunoadsorption of GR and p60 -Receptors were immunoadsorbed from 100-l aliquots of L cell cytosol by rotation for 2 h at 4°C with 8 l of Actigel-ALD precoupled to 80 l of FiGR ascites suspended in 300 l of TEG (10 mM TES, pH 7.6, 50 mM NaCl, 4 mM EDTA, 10% glycerol). Prior to incubation with reticulocyte lysate or with other additions as noted, immunoadsorbed receptors were stripped of associated hsp90 by incubating the immunopellet an additional 2 h at 4°C with 0.5 M NaCl followed by one wash with 1 ml of TEG and a second wash with 1 ml of Hepes buffer (10 mM Hepes, pH 7.4). For immunoadsorption of p60, 400-l aliquots of reticulocyte lysate or a mixture of purified proteins as noted were immunoadsorbed to 8 l of protein A-agarose prebound with DS14F5 antibody against p60 (5%) or nonimmune mouse IgG (5%). The samples were rotated at 4°C for 2 h, and immunopellets were washed twice with 1 ml of Hepes buffer.
Glucocorticoid Receptor Heterocomplex Reconstitution-FiGR immunopellets containing GR stripped of hsp90 were incubated with 50 l of rabbit reticulocyte lysate or with combinations of proteins (12 g of purified hsp90, 20 g of purified hsp70, 12.5 g of purified p23, 3 l of lysate from bacteria expressing p60) and adjusted to 50 l with HKD buffer (10 mM Hepes, 100 mM KCl, 5 mM dithiothreitol, pH 7.35). For reconstitution of GR by the immunoadsorbed p60 heterocomplex, stripped receptors were suspended in 50 l of an assay mix consisting of HKD buffer, and then the whole GR immunopellet suspension was pipetted onto the DS14F5 immunopellet containing the immunoadsorbed p60 and its associated proteins. 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, 20 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 for soluble protein conditions or every minute for the immunoadsorbed p60 condition. 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 assayed for steroid binding capacity and, in some experiments, receptor-associated proteins. To conserve the purified components of the reconstitution system, each experimental condition represents a single sample. The experimental observations have been replicated, and in most cases, the key observation from an experiment appears again as one of the conditions presented in another panel in the same figure or in one of the subsequent figures.
Assay of Steroid Binding Capacity-Immune pellets to be assayed for steroid binding were incubated overnight in 100 l of TEGM buffer plus 5 mM dithiothreitol and 50 nM [ 3 H]triamcinolone acetonide. Samples were then washed twice with 1 ml of TEGM and counted by liquid scintillation spectrometry as described previously. The steroid binding is expressed as counts/min [ 3 H]triamcinolone acetonide bound/FiGR immunopellet prepared from 100 l of cytosol. As noted previously (11), 100 l of L cell cytosol contains 60,000 -80,000 cpm [ 3 H]triamcinolone acetonide binding capacity and we immunoadsorb about 50% of the GR. Thus, reactivation of 100% of receptors to the steroid binding form represents 30,000 -40,000 cpm of binding activity.
Gel Electrophoresis and Western Blotting-For assay of GR and associated proteins or p60 and associated proteins, immune pellets were boiled in SDS sample buffer with 10% ␤-mercaptoethanol, and proteins were resolved on 7% SDS-polyacrylamide gels (12% for resolving p23). Proteins were then transferred to Immobilon-P membranes and probed with 2 g/ml BuGR monoclonal antibody for the GR, 1 g/ml AC88 for hsp90, 1 g/ml N27F3-4 for hsp70, 0.1% DS14F5 anti-p60 mouse ascites for p60, or 0.1% JJ3 mouse ascites for p23. The immunoblots were then incubated a second time with the appropriate 125 I-conjugated counterantibody to visualize the immunoreactive bands.
Protein Purification-The bacterial expression of human p23 and its purification have been described by Johnson and Toft (16). Briefly, p23 is soluble in bacterial lysates, and its abundance and high affinity for DEAE-cellulose allowed purification to 90% purity by chromatography on DEAE-cellulose. The protein was concentrated by precipitation with ammonium sulfate at 80% of saturation. It was dissolved and dialyzed into 10 mM Tris, 100 mM KCl, and 10% glycerol, pH 7.4, and stored at Ϫ70°C.
Rabbit hsp70 and hsp90 were purified from brain cytosol as described previously (14). Briefly, reticulocyte lysate was chromatographed on a DE52 column exactly as described by Dittmar et al. (15). Fractions containing hsp70 were chromatographed on an ATP-agarose column and eluted with ATP followed by ammonium sulfate precipitation, and DE52 fractions containing hsp90 were chromatographed on hydroxylapatite followed by chromatography over ATP-agarose exactly as described by Hutchison et al. (14). The purified hsp70 and hsp90 were dialyzed against HKD buffer, flash frozen, and stored at Ϫ70°C.
For expression and purification of plant hsp90, the Brassica napus hsp90-1 cDNA clone (31) was modified to add six histidines at the COOH terminus and expressed in Sf9 insect cells to prepare a recombinant, tagged hsp90. 3 The recombinant protein was purified over a Ni 2ϩ -nitrilotriacetic acid-agarose column and its identity confirmed with the plant hsp90-specific rabbit R2 antiserum (31).
Expression of p60 -The bacterial expression of p60 has been described previously by Johnson et al. (26). Control E. coli and bacteria expressing p60 were grown to an A 600 of 0.6, induced with isopropyl-1thio-␤-D-galactopyranoside for 3 h at 25°C, and harvested. Bacterial lysates were prepared by sonication in phosphate-buffered saline, and aliquots were flash frozen and stored at Ϫ70°C.

RESULTS
p23 Potentiates Reactivation of Steroid Binding Activity by hsp90⅐p60⅐hsp70 -In the experiment of Fig. 1, p60 was immunoadsorbed from rabbit reticulocyte lysate, and the immune pellet was washed and incubated with GR immune pellets that had been stripped of associated proteins by washing them with salt. Fig. 1A shows the composition of the washed p60 immune pellet. This native p60 heterocomplex contains hsp90 and hsp70, but not p23. In Fig. 1B it can be seen that the immune pellet containing p60 and its coadsorbed proteins generates only a modest number of steroid binding sites (lane 4) and that their formation is ATP-dependent (cf. lanes 3 and 4). In this experiment, steroid binding activity was assayed in the usual manner by incubating the immunoadsorbed GR with [ 3 H]triamcinolone acetonide after heterocomplex assembly at 30°C. When molybdate is present to stabilize the GR⅐hsp90 complexes as they are formed, then there are more steroid binding sites at the end of the incubation (cf. lane 6 with lane 4). In the presence of p23 (lane 8), GR reactivation is nearly to the level achieved with reticulocyte lysate (lane 2). It is interesting that addition of p23 to reticulocyte lysate increases the extent of GR reactivation by 20 -50% (data not shown), suggesting that p23 may be limiting in the lysate.
To examine the relationship between p23 potentiation of steroid binding activity and hsp90 association with the receptor, stripped GR immune pellets were incubated with the minimal assembly system consisting of purified rabbit hsp90 and hsp70 and bacterial lysate containing expressed human p60. As shown in Fig. 2, when stripped receptors (lane 1) are incubated with rabbit reticulocyte lysate (lane 2), they become associated with hsp90 and hsp70 and they are activated to the steroid binding state (it will be shown later (Fig. 7A) that GR⅐hsp90 complexes assembled in reticulocyte lysate also contain p23). Incubation with the minimal assembly system produces a GR⅐hsp90⅐p60⅐hsp70 complex (lane 3) with low steroid binding activity. It is not known why p60 is present in GR⅐hsp90 heterocomplexes formed with purified hsp90⅐p60⅐ hsp70 and not in complexes formed by reticulocyte lysate, but we have previously suggested that reticulocyte lysate must contain an as yet unidentified activity that facilitates the exit of p60 from the receptor heterocomplex and is not present in the reconstituted system (23).
It is interesting that, in this experiment, the amount of GR-associated hsp90 is roughly the same in complexes made in reticulocyte lysate (Fig. 2, lane 2) as in complexes made by the minimal assembly system (lane 3), yet the steroid binding activity of the two complexes is very different. This is consistent with the possibility (which will be developed later) that the hsp90-bound GR HBD can be in a nonbinding conformation or in a steroid binding conformation. When the stripped GR was incubated with purified hsp90, p60, hsp70, and p23, the p23 caused about a doubling in the GR-associated hsp90 and an increase in steroid binding activity that is greater if p23 is present for the entire incubation (lane 5) than if p23 is added after 10 min of the 20-min incubation (lane 4). Thus, the pres- ence of p23 increases the number of GR⅐hsp90 complexes that are recovered, and it seems to stabilize the steroid binding state of the GR⅐hsp90 complex.
p23 Inhibits GR⅐hsp90 Complex Disassembly in a Dynamic Heterocomplex Assembly/Disassembly System-It has been shown previously that receptor⅐hsp90 heterocomplex assembly in reticulocyte lysate is very dynamic in the sense that both heterocomplex assembly and disassembly are occurring simultaneously (9,21). To determine the effect of p23 in a dynamic system, the mouse GR⅐hsp90 heterocomplex was incubated with wheat germ extract, a system where both heterocomplex assembly and disassembly can be assayed because the wheat hsp90 migrates faster than mouse hsp90 on gel electrophoresis. As shown in Fig. 3 (lane 2), the native GR⅐hsp90 heterocomplex immunoadsorbed from L cell cytosol contains hsp90 and p23. When the complex is incubated in buffer at 30°C, much of the hsp90 and the p23 dissociates from the GR (lane 4), and dissociation of both components is inhibited by molybdate (lane 5). When the native mouse GR⅐hsp90 heterocomplex is incubated in wheat germ extract, all of the mouse hsp90 is lost and a small amount of the faster migrating wheat hsp90 becomes associated with the GR (lane 7). Addition of purified recombinant human p23 to the wheat germ extract results in both a partial inhibition of GR⅐mouse hsp90 heterocomplex disassembly and increased recovery of GR⅐wheat hsp90 heterocomplexes (lane 9).
It seems clear from the data of Fig. 3 that p23 can associate with native GR⅐hsp90 complexes formed in the intact L cells to inhibit disassembly of these native heterocomplexes in wheat germ extract. The ability of p23 to increase recovery of GR⅐wheat hsp90 heterocomplexes could reflect the ability of p23 to inhibit their disassembly in this dynamic assembly/ disassembly system. To determine if p23 alone was sufficient to stabilize the native heterocomplex, we asked whether immunopurified, native GR⅐hsp90 complexes suspended in buffer could be stabilized.
p23 Stabilizes Native GR⅐hsp90 Heterocomplexes in a Non-ATP-dependent Manner-In the experiment of Fig. 4A, immunoadsorbed native GR⅐hsp90 heterocomplexes were incubated in the HKD buffer we use for heterocomplex assembly with or without an ATP-generating system. Because ATP is required for binding of p23 to free hsp90 (27)(28)(29)(30), we expected p23 stabilization of GR⅐hsp90 heterocomplexes to be ATP-dependent. However, as shown in Fig. 4A, p23 stabilized the steroid binding activity of native GR⅐hsp90 complexes better in the absence of ATP than in the presence of an ATP-generating system, and in the absence of ATP, the stabilization by p23 was concentration-dependent. It has been known for many years that ATP can promote steroid receptor transformation (see Pratt and Toft (1) and references therein), and it can be seen in Fig. 4B that dissociation of hsp90 from immunoadsorbed native GR⅐hsp90 heterocomplexes is complete when receptors are incubated in the presence of ATP (lane 4) but dissociation of hsp90 is not complete in the absence of ATP (lane 3). The basis for this modest ATP enhancement of dissociation of the native GR⅐hsp90 heterocomplex is unknown, but it may explain why complete dissociation of mouse hsp90 is observed in Fig. 3 when GR heterocomplexes are incubated in wheat germ extract (Fig. 3, lane  7), which contains an ATP-generating system, whereas dissociation of heterocomplexes incubated in buffer (Fig. 3, lane 4) is incomplete. In Fig. 4B, it can be seen that p23 stabilizes both native GR⅐hsp90 heterocomplexes and steroid binding activity much better in the absence of ATP (lane 6) than in the presence of ATP (lane 7). In contrast, we have consistently observed that the stabilization produced by molybdate is essentially the same in the presence or absence of ATP (cf. lanes 8 and 9 of Fig. 4B).
As shown in Fig. 5, GR⅐hsp90 heterocomplexes formed by the minimal hsp90⅐p60⅐hsp70 assembly system are also stabilized better by p23 in the absence of ATP (lane 4) than in the presence of ATP (lane 5). The data of Figs. 4 and 5 show that ATP is not required for p23 to bind to hsp90 that is bound to the GR, at least when the HBD is in the steroid binding conformation. The interaction of p23 with the GR⅐hsp90 complex in its steroid binding form must be dynamic and p23 alone is sufficient for producing stabilization. Also, we have observed that p23 usually stabilizes GR⅐hsp90 heterocomplexes somewhat more than the maximally effective concentration of molybdate.
p23 Does Not Appear to Accelerate GR⅐hsp90 Heterocomplex Assembly-Unlike p23 inhibition of GR⅐hsp90 heterocomplex disassembly, any potential effect of p23 on heterocomplex assembly must be detected in the dynamic system where both assembly and disassembly are occurring simultaneously. Fig. 6 presents the time course of reactivation of steroid binding activity by hsp90, p60, and hsp70 with (open circles) and without (closed circles) p23. In this experiment, [ 3 H]triamcinolone acetonide was present during the incubation with chaperones at 30°C, and steroid binding is used as an indirect assay of GR⅐hsp90 heterocomplex assembly. There is an initial lag of 2-3 min in reactivation of steroid binding activity, which likely reflects the interval required for samples kept in ice to warm to 30°C. Both in the presence and absence of p23, the assembly reaches a plateau in about 20 min, although the plateau for the assembly mixture that contains p23 is about double that for the mixture without p23. We do not know why the time course of assembly turns over before all of the receptors are reactivated to the steroid binding form. But, as shown in the inset in Fig. 6, when the initial reactivation values are plotted as a fraction of the maximal steroid binding achieved in each condition, there is no apparent effect of p23 on reactivation rate. Thus, we suggest that p23 may not affect the rate of GR⅐hsp90 heterocomplex assembly; rather, it may act to stabilize hsp90 in the receptor-bound conformation that yields the steroid binding state of the HBD.
Binding of p23 to hsp90 Alone and in Heterocomplex with GR, p60, and hsp70 -In Fig. 3 (lane 2), it was shown that native GR⅐hsp90 heterocomplexes isolated from L cell cytosol contain p23, and in Fig. 7A, it is shown that GR⅐hsp90 heterocomplexes contain p23 after assembly with either reticulocyte lysate (lane 2) or with the mixture of purified hsp90, hsp70, p60, and p23 (lane 4). Sullivan et al. (30) have shown that binding of purified p23 to purified hsp90 requires elevated temperature and ATP/Mg 2ϩ and is strongly promoted by the nonionic detergent Nonidet P-40 (0.02%). We also find that binding of the purified p23 to purified hsp90 is ATP-dependent and temperature-dependent (data not shown) and that the binding is increased by 0.02% Nonidet P-40 (Fig. 7C, cf. lane 4  with lane 2). However, addition of 0.02% Nonidet P-40 to the assembly mixture containing hsp90, hsp70, p60, and p23 elim- inates GR⅐hsp90 heterocomplex assembly activity (data not shown), and we find that p23 stabilization of GR⅐hsp90 complexes formed by hsp90, p60, and hsp70 does not require ATP (Fig. 5). This suggests that free hsp90 and hsp90 that are in the GR⅐hsp90 heterocomplex are in different states.
We have also found that native hsp90⅐p60⅐hsp70 heterocomplexes immunoadsorbed from rabbit reticulocyte lysate with anti-p60 do not contain any p23 (Fig. 1A) (32). In the experiment of Fig. 7B (lane 2), we incubated hsp90, p60, hsp70, and p23 with an ATP-regenerating system under the same conditions used for GR⅐hsp90 heterocomplex assembly, and at the end of the incubation, p23 was immunoadsorbed. Some hsp90 was coimmunoadsorbed with p23, demonstrating the formation of p23⅐hsp90 complexes, but no p60 or hsp70 was coadsorbed. The number of p23⅐hsp90 complexes formed was increased by Nonidet P-40 (cf. lanes 2 and 4), but again, p23⅐hsp90⅐p60⅐ hsp70 complexes were not formed. Thus, we suggest that hsp90 in the hsp90⅐p60⅐hsp70 complex is not in a p23-binding conformation, but that hsp90 in a GR⅐hsp90⅐p60⅐hsp70 complex formed under identical conditions with the same proteins is in a p23-binding conformation.
Stabilization of GR⅐Plant hsp90 Heterocomplexes-In Fig.  8A, stripped GR immune pellets were incubated with purified rabbit hsp70, human p60 and p23, and various concentrations of either purified rabbit hsp90 (closed circles) or purified plant hsp90 from Brassica napus (open circles). It can be seen that the plant hsp90 is nearly equivalent to the rabbit hsp90 at generating steroid binding activity in a mixed folding system where all of the cochaperones are from mammalian sources. However, as shown in Fig. 8B, the Brassica hsp90 differs from the rabbit hsp90 in that the GR⅐plant hsp90 complex is stabilized by p23 (lane 9) but not by molybdate (lane 10). The ability of plant hsp90 to function with mammalian hsp70, p60, and p23 underlines the fundamental nature of the hsp90-based chaperone system. But the failure of molybdate to stabilize the GR heterocomplex with Brassica hsp90 as opposed to its ability to stabilize all heterocomplexes with hsp90s of animal and fungal origin may suggest a fundamental difference in plant versus animal and fungal hsp90s. DISCUSSION We previously showed that hsp90, hsp70, and p60 are sufficient to produce the folding change in the GR HBD that yields a high affinity steroid binding site (23), and in this study, we show that p23 must also be present to produce a stable GR⅐hsp90 heterocomplex. We suspect that p23 somehow locks hsp90 in a conformation that is required for proper folding of the HBD, a folding state that is achieved only transiently in the absence of p23. As we show (Figs. 1 and 5), the overall effect of p23 in yielding a stable steroid-binding GR⅐hsp90 complex can be replaced by having molybdate present during the incubation with purified hsp90, hsp70, and p60.
It is easiest to discuss the findings of this work within the context of an evolving model of heterocomplex assembly, the latest version of which is presented as Fig. 9. p23 is shown in the model as binding to the final heterocomplex as well as to intermediates in the assembly pathway. The final GR heterocomplex may be either the native heterocomplex immunoadsorbed from L cell cytosol or the GR⅐hsp90 heterocomplex formed in rabbit reticulocyte lysate. These final GR⅐hsp90 heterocomplexes no longer contain p60 (23) but they contain an immunophilin, such as FKBP52 or CyP-40, which binds to hsp90 at the same site where p60 binds (32). A potential assembly intermediate that is p60-free but not yet bound by an FIG. 6. p23 does not appear to affect the rate of GR⅐hsp90 heterocomplex formation. Stripped GR immune pellets were incubated at 30°C for various times with an assembly mixture containing hsp90, p60, and hsp70 with (E) or without (q) p23. [ 3 H]Triamcinolone acetonide was also present during the 30°C incubation, and at the indicated times, samples were diluted with 1.0 ml of ice-cold buffer, centrifuged, and rapidly washed twice prior to assaying GR-bound radioactivity. The inset shows the 5-, 10-, and 15-min values replotted as a fraction of the maximum steroid binding achieved with that condition, yielding an indirect assay of the initial rate of GR⅐hsp90 heterocomplex assembly. FIG. 7. Binding of p23 to hsp90 in different states. A, presence of p23 in reconstituted GR⅐hsp90 heterocomplexes. Stripped nonimmune or GR immune pellets were incubated for 20 min at 30°C with rabbit reticulocyte lysate or with an assembly mixture containing purified hsp90, hsp70, p60, and p23. GR and associated proteins were assayed by Western blotting. Lanes 1 and 2, nonimmune and GR immune pellets, respectively, incubated with reticulocyte lysate; lanes 3 and 4, nonimmune and GR immune pellets, respectively, incubated with hsp90, hsp70, p60, and p23. B, p23 does not bind to the hsp90⅐p60⅐hsp70 complex. Purified hsp90, hsp70, p60, and p23 were incubated at 30°C with an ATP-generating system under the same conditions as for GR⅐hsp90 heterocomplex reconstitution except that no GR immune pellet was present. In the samples of lanes 3 and 4, 0.02% Nonidet P-40 was also present during the incubation. Samples were immunoadsorbed with nonimmune IgG or with the JJ3 monoclonal IgG against p23. Lanes 1 and 2, nonimmune and p23 immune pellets, respectively, from mixtures incubated without Nonidet P-40; lanes 3 and 4, nonimmune and p23 immune pellets, respectively, from mixtures incubated with 0.02% Nonidet P-40. C, binding of purified p23 to purified hsp90. Purified hsp90 and p23 were incubated at 30°C with an ATP-generating system, p23 was immunoadsorbed, and the two proteins were assayed by Western blotting. Lanes 1 and 2, nonimmune and p23 immune pellets from mixtures incubated without Nonidet P-40; lanes 3 and 4, nonimmune and p23 immune pellets from mixtures incubated with Nonidet P-40.
immunophilin is represented by the complex within brackets in the model (Fig. 9). We have shown here that p23 can bind to the native GR⅐hsp90 heterocomplex and stabilize it to both disassembly and loss of steroid binding activity (Figs. 3 and 4). p23 also stabilizes the intermediate GR⅐hsp90⅐p60⅐hsp70 complex assembled by purified hsp90, hsp70, and p60 (Fig. 5). In the case of both the intermediate and the final complex, stabilization by p23 does not require ATP (Figs. 4 and 5).
In previous work in which we showed that all of the components of reticulocyte lysate required to produce a stable GR⅐hsp90 heterocomplex were coimmunoadsorbed with hsp90, we assumed that p23 was a weakly associated component of the foldosome complex (17). It is clear now that p23 was coadsorbed with hsp90⅐p60⅐hsp70 complexes as a separate hsp90⅐p23 complex and that p23 is not itself a component of the foldosome (Figs. 1A and 7B). It is also clear that p23 alone is sufficient to stabilize the immunopurified GR⅐hsp90 complex (Fig. 4). Thus, it is not necessary that p23 enter an intermediate stage of receptor⅐hsp90 heterocomplex as a preformed p23⅐hsp90⅐ immunophilin unit as suggested in earlier models of the assembly mechanism derived from studies in reticulocyte lysate (22,33,34).
When hsp90, hsp70, p60, and p23 are incubated together, we cannot create a complex that contains p60 and hsp70 as well as p23 and hsp90 (Fig. 7B). Also, p23 is not present in the native hsp90⅐p60⅐hsp70 foldosome heterocomplex isolated from reticulocyte lysate with anti-p60 (Fig. 1A). Thus, it seems that hsp90 in the foldosome complex (hsp90⅐p60⅐hsp70) is not in a conformation that binds p23. We have shown previously that the stripped GR combines with the proteins of the foldosome (step 2 in Fig. 9) in a process that is both ATP-dependent and K ϩ -dependent to produce a GR⅐hsp90⅐p60⅐hsp70 complex with steroid binding activity (23). The hsp90 in the GR⅐hsp90⅐p60⅐ hsp70 complex is in a conformation that can be bound and stabilized by p23 (Fig. 5). In so far as we can determine, the sole role of p23 in the minimal assembly system is to stabilize the GR⅐hsp90⅐p60⅐hsp70 complex once it is formed and not to accelerate the rate of heterocomplex assembly (Fig. 6).
In their study of hsp90 binding to phenyl-Sepharose, Sullivan et al. (30) showed that hsp90 that is either bound by ADP or nucleotide-free has a high affinity for binding the hydrophobic resin. However, this hydrophobic binding state of hsp90 is altered by ATP such that the hsp assumes a conformation with low hydrophobic binding activity. This ATP-dependent conformation is the one that is bound by p23 (30). It is reasonable to suggest that hsp90 in the foldosome is in the hydrophobic binding conformation that does not bind p23 and that hsp90 in the steroid binding form of the GR⅐hsp90⅐p60⅐hsp70 complex is in the ATP-dependent conformation that binds and is stabilized by p23.
It seems clear that hsp90 in the GR⅐hsp90⅐p60⅐hsp70 complex can exist in either a steroid binding or nonbinding comformation. When GR⅐hsp90⅐p60⅐hsp70 complexes are assembled by hsp90⅐p60⅐hsp70 in the presence of the hsp90-binding antibiotic geldanamycin, they do not have steroid binding activity (23). Thus, hsp90 can be bound to the GR without the HBD being properly folded to the steroid binding state. It seems reasonable that the proteins of the foldosome may bind to the GR to yield first a nonsteroid-binding GR⅐hsp90⅐p60⅐hsp70 complex in which hsp90 is in the conformation with hydrophobic affinity and that an ATP-induced change in hsp90 conformation is required for the folding change in the GR HBD that yields the steroid binding site. In this conformation, hsp90 can now be bound by p23, which essentially stabilizes a conformation of hsp90 that is required for opening the steroid-binding pocket in the GR HBD.
Although p23 binds to free hsp90 in an ATP-dependent manner (27)(28)(29)(30), we have not shown that bimolecular interaction in the model of GR⅐hsp90 heterocomplex assembly in Fig. 9, because we cannot demonstrate the presence of p23 in either native (Fig. 1A) or reconstituted (Fig. 7B) foldosome (hsp90⅐ p60⅐hsp70) complexes. The study by Sullivan et al. (30) of p23⅐hsp90 complex formation utilizing purified proteins was extremely useful in proposing that the ATP requirement for forming the bimolecular complex reflects the ATP-dependent conversion of hsp90 from a conformation with hydrophobic affinity to a conformation with low hydrophobic affinity that can complex with p23. It is interesting that the ATP-dependent binding of p23 to free hsp90 is increased by the presence of the detergent Nonidet P-40 (30) (Fig. 7C). It is thought that the detergent promotes a change in the conformation of hsp90 that is stabilized by the binding of ATP (30). The effect of the detergent may mimic a similar change in the conformation of hsp90 that takes place with unfolding of the GR HBD by the proteins of the foldosome complex. It is reasonable to propose that the hsp90 in native GR⅐hsp90 heterocomplexes isolated from L cell cytosol is in this ATP-dependent conformation; thus, as shown in the model of Fig. 9, p23 binds to the receptor heterocomplex in a reversible manner that does not require ATP (Fig. 4). FIG. 9. The heterocomplex assembly pathway with sites of p23 interaction. The association of purified hsp90, p60, and hsp70 in step 1 to form an hsp90⅐p60⅐hsp70 complex (foldosome) and the ability of the proteins in this unit to associate in step 2 with the stripped, nonsteroid binding state of the GR to assemble a GR⅐hsp90⅐p60⅐hsp70 complex with a steroid binding site (indicated by the steroid structure) was reported by Dittmar and Pratt (23). IMM stands for an immunophilin, such as FKBP52 or CyP-40, which is present in the final heterocomplex formed in reticulocyte lysate or isolated from cytosols. The immunophilins and p60 bind to hsp90 via tetratricopeptide repeat domains, which are indicated by the solid black crescents on these proteins. Native GR⅐hsp90 heterocomplexes isolated from cells usually do not contain hsp70, but native receptor complexes from some cells, as well as complexes assembled in reticulocyte lysate, do contain hsp70. Thus, the hsp70 with dashed lines indicates its optional presence in the final product. p23 associates with hsp90 in the GR⅐hsp90⅐p60⅐hsp70 assembly intermediate formed by the minimal assembly system as well as with native GR⅐hsp90 heterocomplexes isolated from cells. In both cases p23 binding is reversible, does not require ATP, and yields stabilization of the complex.