Vol. 269, No. 15, Issue of April 15, pp. 11155-11161, 1994 Printed in U S A .

Ttre Joumu. OF BIOUXICALCHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc.

Characterization of the Protein-Protein Interactions Determining the Heat Shock Protein (hsp90*hsp70*hsp56) Heterocomplex* (Received for publication, July 16, 1993, and in revised form, December 17, 1993)

Michael J. Czars, Janet K. Owens-GrilloS,Kurt D. DittmarS, Kevin A. HutchisonS, Anthony M. ZacharekS, Karen L. Leach§, Martin R. Deibel, Jr.0, and William B. PrattSn From the $Department of Pharmacology, The University of Michigan Medical School, Ann Arbor, Michigan 48109 and the §Department of Cell Biology and Biochemistry, The Upjohn Company, Kalamazoo, Michigan 49001

We have reported previously that the three heat shock in part)for the abilityof the receptor hormone binding domain proteins hsp66, hsp70,and hap90 exist together in a het- to regulate other receptor functions (see Ref. 2 and references erocomplex in human lymphocyte cytosol (Sanchez, E. therein). Bindingof the glucocorticoid receptor (GR)to hsp90 is R, Faber, L. E., Henzel, W. J., and Pratt, W. B. (1990) required for its high affinity steroid binding conformation (3), acquiBiochemistry 29, 61456152). All three of these proteins and dissociation of the GR from hsp90 is correlated with also exist in the native glucocorticoid receptor hetero- sition of DNA binding activity in vitro (4). The steroidreceptorcomplex isolated from WCL2 cell cytosol and we have hsp heterocomplex can be formed under cell-free conditions by recently shown that the three heatshock proteins are incubating immunopurified unliganded receptors with rabbit present when immunopurified mouse glucocorticoid re- reticulocyte lysate (5, 6). Receptor heterocomplex assembly is ceptor is reconstituted into a heterocomplex by rabbit ATP-dependent (7,8) and has a strict requirementfor the presreticulocyte lysate (Hutchison, K. A, Scherrer, L. C., ence of a monovalent cation, such asK+(8).The association of Czar, M. J., Ning, Y.,Sanchez, E. R., Leach, K. L., Deibel, the GR with hsp9O and hsp56 is strictly hsp70-dependent(91, M.R., Jr., and Pratt,W. B. (1993)Biochemistry 32,3953- and itis thought that the proteinunfolding activity of hsp7O is 3967). In this work, we show that highly purified mouse required for hsp90 t o bind to the hormone binding domain of hsp90 binds in a reversible equilibrium to immunopuri- the receptor (for reviewof heterocomplex assembly, see Refs. 10 fied rabbit hsp66, but hsp56 does not bind to purified and 11). mouse hsp70. In contrast to the equilibrium binding of Several years ago, our laboratory reported that immunoadhap90 to hsp56, purified hap90 binds poorly or not at all to purified hap70 unless a third factor from reticulocyte sorption of hsp56 witha monoclonal antibody resulted incoimlysate is present to permit complexformation. This munoadsorption of hsp90, hsp70, and a small amount of a hsp70.hsp90 complex-formingfactor is heat-labile, and 23-kDa protein from human IM-9 lymphoblast cytosol (12). in thepresence of this factor and ATP, a heat shock pro- Because the amounts of hsp56, hsp70, and hsp90 thatcopuritein heterocomplex can be reconstituted from purified fled were in greatstoichiometric excess of the amountof GR in mouse hsp90and hap70 and rabbit hap66 that is present IM-9 cytosol, we concluded that the hsps must exist together in in the factor preparation. Our data areconsistent with a a heat shock protein complex independent of the presence of model in which hap66 and hap70 bind to different sites receptors (12). The concept that the three heatshock proteins on hsp90 but do not interact with each other. Thepres- may exist together in a heterocomplex was supported by the ence of hsp56 in the heat shock protein heterocomplex is observations of Perdew and Whitelaw (13) who immunoadnot stabilized by molybdatebut hap56 is stabilized if the sorbed hsp9O with a monoclonal antibody andfound five major glucocorticoid receptor is present in addition to hap90 proteins ofM, 68,000,63,000,56,000,50,000,and 188,000 to be and hsp70. coimmunoadsorbed from Hepa lclc7 cytosol. The 68- and 56kDa proteins wereidentified as hsp70 and hsp56, respectively. The 50-kDa protein was subsequently demonstrated (14) to be In molybdate-stabilized cytosols prepared from hormone-free the 50-kDaphosphoprotein that wasreportedintheearly cells, mammalian progesterone receptors exist in a heteropro- 1980s to be a component of thenative heterocomplex of Ref. 15 for review). This 50-kDa tein complex that contains three heat shock proteins: hsp90, pp60"-"" withhsp90(see hsp70, and hsp56.' Native glucocorticoid receptor heterocom- protein is also presentin pp60"'"~hsp9O.hsp70 heterocomplexes also contain hsp90 and hsp56, but depending upon the plexes that are formed in vitro by incubating immunopurified ' ~ reticulocyte lysate (16). cell type, they may or may not contain hsp7O (for review see p ~ 6 0 " -with shock proteins in the heterocomplex are chapRef. 1).The hormone-regulated bindingof the receptors to the The three heat hsp90 component of the complex is thought toaccount (at least erone proteins. hsp90 is an ubiquitous, essential, and highly conserved (see Ref. 17 for review) protein with ATPase activity * This investigation was supported by National Institutes of Health (18).The ability of hsp90 to chaperoneprotein folding is indiGrants DK31573 and CA28010. The costs of publication of this article cated by recent experiments in which it was found to bind to were defrayed in part by the payment of page charges. This article must and suppress the aggregation of proteins in vitro (19). hsp70 therefore be hereby marked "aduertisement" in accordance with 18 has protein unfoldase activity, and it binds to proteins during U.S.C. Section 1734 solely to indicate this fact. ll To whom correspondence should be addressed: Dept.of Pharmacol- their translation and during their passage across the memogy, 6448 Medical Science Bldg.I, The University of Michigan Medical branes of organelles (see Ref. 20 for review). hsp56 (also called School, Ann Arbor, MI 48109-0626. Tel.: 313-764-5414; Fax:313-763- p59 and FKBP52) was first described by the Faber laboratory 4450. (21) who established that it was a component of steroid receptor The abbreviations used are: hsp56, hsp70, and hsp90 refer to the 56-, 70-, and 90-kDa heat shock proteins; GR, glucocorticoid receptor; heterocomplexes. Recently, hsp56 was shown to be an immuTES, 2-([2-hydroxy-l,l-bis(hydroxymethyl)ethyllamino~ethanesulfonic nophilin of the FK506 binding class (22-24). The immunophiacid; PAGE, polyacrylamide gel electrophoresis. lins are proteins that bind immunosuppressive agents cylike

11155

11156

The Heat Shock Protein Complex

closporinA, FK506, and rapamycin (for review, see Refs. 25and 26). Passage of cell extracts over matrices of immobilized FK506 or rapamycin results in binding of hsp56 (22, 24) and coretention of hsp90 and hsp7O (24), confirming the existence of the heat shock protein heterocomplex by another method of isolation. As has been demonstrated for other immunophilins, hsp56 has peptidylprolyl isomerase activity (27). Like hsp9O and hsp70, it is thought that the immunophilins play major roles in protein folding and protein trafficking in the cell (26). The existence of the three heat shock proteins (hsp56, hsp70, and hsp9O) in a complex implies that they may act together in a coordinated manner, and it hasbeen proposed that the heat shock protein heterocomplex mayfunction as a protein folding/ trafficking structure (28). In this work we examine the effects of temperature, salt, and transition metal anions (molybdate, tungstate) on the native hsp heterocomplex isolated from rabbit reticulocyte lysate and on the same proteins in the native GR heterocomplex isolated from WCL2 cells. We show that purified hsp56 binds in a reversible equilibrium to purified hsp9O but not hsp70, and our data areconsistent with a model in which hsp70 in the native hsp heterocomplex is bound to hsp90 but not to hsp56. However, purified hsp90 and purified hsp7O bind to each other either very weakly or with very low efficiency unless a third factor in reticulocyte lysate is present to permit complex formation. Although it is thought that molybdate stabilizes receptor heterocomplexes by interacting directly with and altering theconformation of hsp90 (29,30), the stabilizing effect of molybdate on the presence of hsp56 in the heterocomplex is seen only when the GR is present.

quots of rabbit reticulocyte lysate with 8 pl of protein-A-Sepharose prebound with 4 pl of UPJ56 antiserum, and the immune pellets were washed three times with TEG buffer as above. hsp90 was immunoadsorbed from 100-plaliquots of reticulocyte lysate with 8 pl of proteinA-Sepharose preadsorbed with 3G3 antibody. For preabsorption, protein A-Sepharose pellets were suspended for 1h in 300 pl of HEG buffer (10 m~ Hepes, 1 m~ EDTA, 10% glycerol, pH 7.4) and 100 pl of goat anti-mouse IgM, centrifuged, aspirated, and the pellets were resuspended for1h in 300 pl of HEG with 20 pl of 3G3 (IgM)ascites fluid and washed with HEG after the 3G3 step. The 3G3-preadsorbed protein A-Sepharose pellet was suspended in 300 pl of HEG, rotated with 100 pl of reticulocyte lysate to prepare the immune pellet, and the immune pellet was washed three times with 1ml of HEG buffer. Gel Electrophoresis and Immunoblotting-For assay of GR, hsp90, hsp70, and hsp56, immunopellets were boiled in SDS sample buffer with 10%P-mercaptoethanol,and proteins were resolved on 7% or 10% SDS-polyacrylamidegels as describedpreviously (3). Proteins were then transferred to Immobilon-P membranes and probed with 2 pg/ml BuGR2 for the glucocorticoid receptor, 0.1% AC88 for hsp90, 1 pg/ml N27F3-4 for hsp70, or 0.1% UPJ56 for hsp56. The immunoblots were then incubated a second time with the appropriate lZ5I-Labeledcounterantibody, followed by horseradish peroxidase-conjugated counterantibody to visualize the immunoreactive bands. In some experiments, the bands visualized by the peroxidase reaction were excised from the membrane and quantitated by y counting as described previously(33). Purification ofhsp90, hsp70, and hsp56-hsp90 was purified from L cell cytosolessentially as described by Bresnick et al. (34). Briefly, L cell cytosol (25 ml) was chromatographed on a 2 x 20-emDE52 column equilibrated with HE buffer, and theproteins were eluted with a 300-ml gradient of 0-0.5 M KC1. hsp90 and hsp7O were detected by resolving an aliquot of each fraction by SDS-PAGE and immunoblotting. Fractions containing hsp9O were pooled and diluted with an equal volume of 20 m~ &HPO,, 1m~ EDTA, pH 7.5, and then chromatographed on a 2 x 8-cm hydroxylapatite column,which was equilibrated in the same %HPO, buffer, and proteins were eluted with a 300-ml gradient of EXPERIMENTALPROCEDURES 0-0.5 M qHPO,. Fractions containing hsp90 were pooled and passed Materials over ATP-agaroseto remove the last trace of contaminating hsp7O and the column drop-through containing purified hsp9O wasdialyzed [6,7-3HlTriamcinoloneacetonide (42.8 CUmmol) and lZ6I-conjugated against TEG bufferplus 5 m~ dithiothreitol. The final preparation was goat anti-mouse and anti-rabbit IgGs were obtained from DuPont NEN. flash-frozen and stored in small aliquots. FK506 was purified from fermentation broths of Streptomyces tsukuhsp7O was purified from the DE52-chromatographed L cell cytosol buenis as described previously (22) and custom labeled by tritium ex- fractions that did not contain hsp90 as described by Schlossman et al. change (49.1 Cilmrnol) byAmersham Corp. CyclosporinA was provided (35) with modifications. DE52 column fractions containing hsp7O but by Sandoz Research Institute (East Hanover, NJ). Untreated rabbit not hsp9O (detected by Western blotting) were pooled and applied to a reticulocytelysate was fromGreen-Hectares (Oregon,WI).Nonimmune 25-ml column of ATP-agarose. Nonspecific proteins were removed by mouse IgG, ATP-agarose (coupled a t the C8 positionof the purine ring), washing with HE buffer containing 500 m~ KCl, and bound proteins goat anti-mouse IgM, and goat anti-mouse and anti-rabbit IgG-horse- were eluted with wash buffer containing 5 m~ ATP. Eluted proteins radish peroxidase conjugates were from Sigma. The BuGR2 IgG mono- were pooled,concentrated, and precipitated with ammonium sulfate to clonal antibody against the GR and the 3G3 monoclonal IgM against concentrate the hsp7O and remove hsp70-boundATP. The protein pellet hsp9O were from Affinity Bioreagents (Neshanic Station, NJ). TheAC88 was solubilized in Hepes buffer, dialyzed against Hepes buffer plus 5 monoclonal IgG against hsp90 and the N27F3-4 anti-7Z73-kDa heat m~ dithiothreitol, and finally flash frozen in small aliquots. shock protein monoclonal IgG (anti-hsp7O) were fromStressGen (Vichsp56 was partially purified from L cell cytosol and rabbit reticulotoria, British Columbia). The UPJ56 rabbit antiserum against hsp56 cyte lysate by combining the hsp56-containing fractions (identified by has been described previously(31). UPJ56 immunoblotting)eluting just prior to hsp9O from DE52 and then passing the fraction pool through ATP-agarose to remove hsp7O. The Methods partially purified hsp56 was shown to be free of hsp9O and hsp70 by Cell Culture and Fractionation"L929 mouse fibroblasts (L cells) immunoblotting. weregrown in monolayer in Dulbecco'smodifiedEagle's medium Deatment of Reticulocyte Lysate-To prepare hsp90- and hsp7O-free supplemented with 10% bovineserum. The WCLB line of Chinese ham- reticulocyte lysate containing hsp70.hsp90 complex-forming activity,5 ster ovary cellsoverexpressing the mouse GR was established by Hirst ml of reticulocyte lysate was chromatographed on DE52 as described et al. (32), and the cells were grown in monolayer culture using Dul- above. The drop-through material and all fractions that did notcontain becco'smodifiedEagle'smedium plus 3 p~methotrexate, 40 pdml hsp90 were combinedand applied to a 25-ml column ofATP-agarose. All proline, and 10% iron-supplemented calf serum. Cells were harvested of the material that was not adsorbed to the matrix was passed through by scraping into Earle's balanced saline, suspended in 1.5 volumes of a second columnof ATP-agaroseto remove the last traces of hsp7O. The HE buffer (10m~ Hepes, 1m~ EDTA, pH 7.4),or HE buffer plus 20 m~ resulting hsp70.hsp90-free lysate was dialyzed against HE buffer plus sodium molybdate and ruptured by Dounce homogenization. Cell ho100 m~ KC1 and 5 m~ dithiothreitol, concentrated to one-half the origimogenates were centrifuged for 1h at 100.000 x g , and the supernatant nal volume of lysate, flash frozen, and stored in small aliquots. from this centrifugation is referred to as "cytosol." Zmrnunoadsorption-The GR heterocomplex was immunoadsorbed RESULTS from replicate (100-200 pl) aliquots of L cell or WCLB cellcytosol transiEffect of Molybdate on the hsp Heterocomplex-The prepared in HE plus molybdate by incubating for 2 h at 4 "C with 8 pl of protein-A-Sepharose preadsorbed with BuGr antibody and sus- tion metal oxyanions molybdate, vanadate, and tungstate stapended in 300 pl of TEG buffer(10 m~ TES, 50 m~ NaC1,4 m~ EDTA, bilize the association of hsp90 with steroid receptors (for re10%glycerol, pH 7.6). Antibody waspreadsorbed by suspending 8 pl of view, see Ref. 36) and pp60".sm(29), and molybdate has also protein-A-Sepharose in 300 pl of TEG buffer plus 2 pg of BuGR2 and rotating 30 min at 4 "C. ARerimmunoadsorptionof GR from cytosol,the been reported to stabilize the presence of hsp56 in steroid reimmune pellet was washed three times by suspension in 1 ml of TEG ceptor heterocomplexes (21,23). It is thought that stabilization buffer and centrifugation. hsp56 was immunoadsorbed from 100-plali- of the receptor heterocomplexes represents an effect of the

The Heat Shock Protein Complex

A

*'I

.

11157

B

I

100

75+MOO, /e-

50

50 -

*\

e

-- "-"

e

t , . . a o i

25 -0

30

60

90

120

0 0

"

I

I

I

60 90 TIME (minutes) 30

I

120

TIME (minutes) FIG.1. Effect of temperature and molybdate on hsp56-associatedhsp90. A, hsp56 was immunoadsorbed from reticulocyte lysate with UPJ56, and the immunopellets were washed, resuspended in HE buffer ( 0with )or without(0)20 m sodium molybdate, and incubatedat 30 "C for the indicated times. The relative amount of hsp90 in each immunopellet was determined by probing Western blots of each samplefirst with the inset, ACSS antibody and then with '2sI-labeled counterantibody. The'2sI-labeled bands, which are shown in the autoradiogram in thewere cut out of the Western blot and counted to obtain the valuesfor hsp56-associated hsp90 shownin the graph. The graph presented as inset an in A shows molybdate stabilization of GR-associated hsp90. Replicate GR immunopellets prepared from WCL2 cytosol with the BuGR antibody were suspended inbuffer with (0)or without (0)molybdate, incubated at 25 "C, and assayed for GR-associated hsp90 by Western blotting followed by counting the'*'I radioactivity in the bands. B , both isoforms of hsp90 are present in the immunopurified hsp56.hsp90complex, and they dissociate equivalently whenthe complex is heated. hsp56 was immunoadsorbed with UPJ56 from replicate aliquotsof WCL2cell cytosol, the immunopellets were washed three times in TEG buffer, resuspended in HE buffer, and incubated at 30 "C. At various times, the hsp56 and hsp90 in the immunopellet were assayedby Western blotting. For both panels, the numbers over the lanes in the Western blot autoradiogram insets represent the incubation time in minutes, andNI indicates a nonimmune sample.

metal anion on hsp90 (29), and both vanadate and molybdate have been shownto induce a conformational change in purified hsp90, apparently through binding to anATP site (30). If molybdate stabilizes a conformation of hsp90 that has higheraffinity for the GR, then it is conceivable that the metal-stabilized conformation of hsp90 hasa higher affinity for hsp56 as well. To determine if the hsp56.hsp90 complex is stabilized by molybdate in the absence of the GR, hsp56 was immunoadsorbed from reticulocyte lysate, the immunopellet was incubated at 30 "C, and hsp56-associated hsp90 was assayed. As shown in Fig. lA, about half the hsp90 dissociates from the heterocomplex, and thereis no stabilization by molybdate. For contrast, we show that virtually all of the hsp90 dissociates from the GR heterocomplex immunoadsorbed from WCL2 cytosol, and thecomplex is stabilized by molybdate (inset of Fig.

1A 1. It is unclear why only about half of the hsp90 dissociates from hsp56 when the immunoadsorbed reticulocyte lysate hsp heterocomplex is incubated at 30 "C. Mammalian hsp90 is ena or the p coded by two genes, and it is possible that either the form of hsp90 is more tightly bound to hsp56. Rabbit a and p hsp90 migrate as a single band on SDS-polyacrylamide gel electrophoresis, but like the mouse where hsp9O consists of two isoforms of 84 and 86kDa (371,Chinese hamster cells contain two hsp9O isoforms that are present incytosol in roughly the same abundance and can readilybe resolved from each other by gel electrophoresis. To determine if one isoform of hsp9O dissociates selectively when the immunopurifiedhsp56.hsp90 complex is heated, WCLB cell cytosol was immunoadsorbed with UPJ56, theimmunopellet was heatedat 30 "C, and hsp90 dissociation from the immunopurified hsp56 was assayed. As shown in Fig. lB, half of the hsp90 dissociates, but there isno selective dissociation of one isoform versus the other. Fig. 2 shows that the native hsp56.hsp90 complex readily dissociates in the presence of salt. In this experiment, hsp56 was immunoadsorbed from reticulocyte lysate, theimmunopellets were incubated with various concentrations of salt, and hsp56-associated hsp90 was assayed. The effect of salt on the hsp56-bound hsp90 (Fig. 2, solid circles) is similar to the effect of salt on the GR.hsp9O complex (Fig. 2, open circles). In con-

NaCL CONCENTRATION

(mM)

FIG. 2. The hsp56.hsp90 complex is salt-labile.hsp56 was immunoadsorbed from reticulocyte lysate with UPJ56, and the immunopellets were washed in TEG buffer and resuspended in HE ice buffer for on 1h with various concentrations of NaCl. The pellets were washed again with TEG and pellet-associated hsp56 and hsp90 were by assayed Weste m blotting. The autoradiogramof the Western blot is shown above the graph where N I is a nonimmune sample, and the numbers over the lanes represent the NaCl concentration inm. The hsp90 bandswere excised and counted to obtain the hsp56-bound hsp90 values (0,solid line) of the graph. The values in the graph have been normalized to the relative amount of hsp56 recovered in each sample. The GR-bound hsp90 values (0,dashed l i n e ) are from an identical experiment performed with the GR heterocomplex immunoadsorbed from WCLBcytosol with the BuGR antibody, and the data have been normalized for GR recovery.

trast to the GR-hsp9O heterocomplex (381, however, salt-induced dissociation of hsp90 from the hsp56.hsp90 heterocomplex is not inhibited by molybdate or tungstate (Fig. 3). In fact, 20 m~ molybdate and tungstate promote additional dissociation of hsp90 from hsp56 in the immunoadsorbed hsp heterocomplex (Fig. 2, lanes 4 and 5) in the same manner as additional NaCl ( l a m 6 ) . Binding of Purified hsp90to Purified hsp56"h determine if hsp90 and hsp56 interact with oneanother directly, hsp56 was immunoadsorbed from reticulocyte lysate, and the immune pel-

--" -

The Heat Shock Protein Complex

11158 1 hsp9O hSP56

2

3

4

5

emp!gqP

Molybdate and tungstate do not inhibit salt-mediated dissociation of hsp90 from hsp56 in the immunoadsorbed hsp heterocomplex. hsp56 was immunoadsorbed from reticulocyte lysate with UPJ56, and the immune pellets were washed and suspended in HE buffer with or without 200 mM NaCl for 60 min at 0 "C in the presence or absence of 20 m~ molybdate or tungstate. hsp56 and hsp90 in the immunopellets were resolved by SDS-PAGE and Western blotting. Lane 1, nonimmune pellet incubated in buffer alone; lane 2, immune pellet incubated with buffer alone; lane 3 , immune pellet incubated with 200 m~ NaCl; lane 4 , immune pellet with NaCl and 20 m~ tungstate; lane 5 , immune pellet with NaCl and 20 rn molybdate; lane 6, immune pellet incubated with 220 m~ NaCl. FIG.3.

lets were washed several times with high salt buffer to remove coadsorbed proteins. The immunopurified hsp56 was then incubated with mouse hsp90 that was purified to -98% of homogeneity. As shown in Fig. 4, purified hsp90 binds to immunopurified hsp56 a t 0 "C (cf. lane 1 with lane 5). As ATP and elevated temperatures are not required (lanes 7 and 91,this appears to reflect a binding equilibrium that does not require a protein folding activity of hsp70, which is absolutely required for hsp90 binding to the GR (9). At this time, wedo not know how much of the hsp56 in reticulocyte lysate is bound to hsp90, but gel filtration experiments suggest that all of the hsp56 is ina large complex. In the experiment shown in Fig. 5, whole reticulocyte lysate was passed through a long column of Sepharose CL6B and hsp56 eluted with a Stoke's radius of 6.9 f 0.2 nm. This is to be compared with hsp56 partially purified (hsp90hsp70-free) from rabbit reticulocyte lysate, which elutes a t 5.6 5 0.3 nm. That thelarger size of hsp56 in whole reticulocyte lysate could reflect binding to hsp90 (in a complex that dissociates as it passes through the column) is suggested by the fact that purified hsp56 that hasbeen mixed with a stoichiometric excess of purified mouse hsp90 elutes from Sepharose CL6B with a Stoke's radius of -6.6 nm, similar to the elution of the endogenous hsp56 in whole reticulocyte lysate. Purified hsp70 Does Not Bind to hsp56-When hsp56 is immunoadsorbed from reticulocyte lysate, we find relatively little coadsorption of hsp7O (cf.lanes 1 and 2 in Fig. 6 A ) compared with cytosols prepared from other sources, such as IM-9 lymphocytes (12).In theexperiment of Fig. 6 A , the hsp56-containing UPJ56 immunopellet was washed with salt to remove coadsorbed heat shock proteins (Fig. 6 A , lane 3 ) , andthe immunopellet was then incubated with purified hsp7O. Some hsp7O binds to both preimmune and hsp56-containing pellets a t 0 "C in the absence of an ATP-generating system (lanes 4 and 51, and theamount of binding is increased by ATP(lanes 6 and 7), but, again, there is no difference between preimmune and hsp56-containing samples. As noted in studies of progesterone (7) and glucocorticoid (8) receptor heterocomplex reconstitution, hsp70 is rather "sticky: in that large amounts are recovered in both immune and nonimmune pellets incubated with reticulocyte lysate. We have noted before that the binding of hsp70 to nonimmune IgG-protein A-Sepharose pellets is increased byATP (8, 9, 391, and this may reflect the natural ability of hsp70 to bind to denatured proteins (40). From Fig. 6A as well as other experiments with the purified proteins in solution (data not shown), we see no evidence for binding of hsp56 to hsp70. Purifwd hsp70 Binds to hsp90 i f Lysate Factors Are Present-As shown in lanes 3 and 4 of Fig. 6B, immunoadsorption of hsp90 from reticulocyte lysate is accompanied by substantial

1

6

hsp9O hsp56

2

3

4

5 "

'w

W

6

7

8

9

0

I,

F I G . 4. Purified hsp90 binds to immunopurified hsp56. hsp56 was immunoadsorbed from reticulocyte lysate with UPJ56, and the immune pellets were washed four times with HE buffer containing 0.5 M KC1 to strip off any coadsorbed proteins. The salt-stripped pellets were washed with HE and incubated for 15 min a t 0 or 30 "C with either rabbit reticulocyte lysate and an ATP generating system or with punfied mouse hsp90 in the presence of 100 m~ KC1 with or without the ATP-generating system. ARer the incubation, the immune pellets were washed and assayed for hsp90 and hsp56 by Western blotting. Lane 1, immune pellet stripped with salt; lanes 2 and 3, preimmune and immune pellets, stripped and incubated a t 30 "C with rabbit reticulocyte lysate; lanes 4 and 5 , preimmune and immune pellets, stripped and incubated with hsp9O at 0 "C; lanes 6 and 7, stripped pellets incubated with hsp90 and ATP a t 0 "C; lanes 8 and 9, stripped pellets incubated with hsp90 and ATP a t 30 "C.

coadsorption of hsp70 as well as coadsorption of some hsp56. This yield of hsp70 stands in contrast to the lower amount of hsp7O that is coadsorbed from the same lysate sample with hsp56 by the UPJ56 antibody (cf. lanes 2 and 4 in Fig. 6B). As there isa substantial yield of endogenoushsp70-hsp90 complex when hsp90 is immunoadsorbed from reticulocyte lysate, we mixed purified hsp70 with purified hsp90 to determine if heterocomplexes would form. When we immunoadsorbed hsp90, we found almost no coimmunoadsorptionof hsp70. However, as shown in Fig. 7, an hsp70.hsp90 complex can form if purified hsp7O and hsp90 are incubated with reticulocyte lysate from which the endogenous hsp7O and hsp90 have been eliminated. To prepare hsp70hsp90-free lysate, we first chromatographed reticulocyte lysate on DE52, combined all of the material that did not contain hsp90, and dropped that pool through ATPagarose to eliminate hsp7O. Samples of this hsp70hsp90-free lysate, purified mouse hsp90, and purified mouse hsp7O are shown in lanes 1,2, and 3, respectively, of Fig. 7. When purified hsp90 and purified hsp70 are incubated together and hsp90 is immunoadsorbed with the 3G3 antibody, there is very little coimmunoadsorption of hsp7O (lanes 4 and 5).However, when the hsp70hsp90-free lysate is present, there is significant coimmunoadsorption of hsp7O (lanes 6 and 7). This increase in coimmunoadsorption of hsp7O is not seen if the hsp70hsp90free lysate was boiled prior to the incubation with hsp70 and hsp90 (lanes 8 and 9),indicating that the complex-forming activity is heat-labile. In addition to mouse hsp7O and hsp90, the complex shown in lane 7 contains rabbit hsp56, which was provided by the hsp70hsp90-free lysate. hsp56 in the GR Heterocomplex Is Stabilized by MolybdateSteroid receptor heterocomplexes are dissociated by salt aswell as by heating, and molybdate, vanadate, and tungstateinhibit salt-induced GR transformation to the DNA binding state (38). In the only study of salt effects on the hsp56 component of receptor heterocomplexes, Renoiret al. (41)reported that tungstate inhibits salt-induced dissociation of hsp90, but not hsp56 from glucocorticoid and progesterone receptors in calf uterine and human MCF7 cell cytosols.As shown in Fig. 8, under our experimental conditions, both molybdate and tungstateinhibit salt-induced hsp90 dissociation from the GR (Fig. &I), and there is substantial inhibition of hsp56 dissociation as well (Fig. 8B). We have noted that, in contrast to hsp90, hsp56 is a loosely associated component of the GR heterocomplex. This can be seen in lane 1 of Fig. 8B where much of the hsp56 dissociated during incubation in salt-free buffer unless it was stabilized in thecomplex by molybdate or tungstate. The experiment of Fig. 9 demonstrates that hsp9O must be bound to the GR for purified hsp56 to bind to the receptor

The Heat Shock Protein Complex

40

30

50

60

11159

80

70

90

Fraction Number

forms a similar complex with partially FIG.5. All of the hsp56in reticulocytelysate is in a largeM,complex, and purified hspM purified hsp56. Rabbit reticulocyte lysate (1.00 ml) was chromatographed through a column (1.5 x 120 cm) of Sepharose CL6B in HE buffer. Fractions were lyophilized and dissolved by boiling in SDS-sample buffer prior to resolution of proteins by SDS-PAGE and Western blotting. Partially purified (hsp90hsp70-free) hsp56 from rabbit reticulocyte lysate and highly purified mouse hsp9O were chromatographed on the same Sepharose CG6B column alone and after they had been incubated with each other in HE buffer for 15 min at 0 "C. Column markers are thyroglobulin ( M , 669,000, R, 8.6 nm), apofemtin ( M , 443,000, R, 7.8 nm), alcohol dehydrogenase ( M , 150,000, R, 4.5 nm), purified hsp70 ( M , 72,000, R, 3.9 nm), and cytochrome c ( M , 12,000, R, 1.9 nm).The inset shows a plot of the Stoke's radii of the markers, with A indicating the R, of purified rabbit hsp56 and B the R, of hsp56 in whole reticulocyte lysate.

A

1 hsp90 hsp70 hsp56

B 1 hsp9O hsp70 hsp56

2

-

3

4

0 -

0 U

FIG.6. A, purified hsp70 does not bind to immunopurified hsp56. hsp56 was immunoadsorbed from reticulocyte lysate with UF'J56 or preimmune serum. Two of the immune pellets were washed three times with TEG, and the native heterocomplex was resolved by SDS-PAGE and Western blotting. The remaining pellets were washed four times with HE buffer containing 0.5 M KC1 to strip off associated proteins. The salt-stripped pellets were washed once with HE and incubated for 15 min at 0 "C with purified mouse hsp70 with or without an ATP-generating system. The pellets were then washed and assayed for hsps by Western blotting. In this instance, the colored hsp56 bands obtained with peroxidase-conjugated counterantibody were photographed directly. Lanes 1 and 2, respectively, preimmune and immune samples washed with TEG and receiving no further treatment;lane 3 , immune pellet stripped with salt; lanes 4 and 5,preimmune and immune pellets, stripped and incubated with purified hsp70; lanes 6 and 7, stripped pellets incubated with hsp70 and ATP. B , relative coimmunoadsorption of the other two hsps with hsp56 versus hsp90. Replicate aliquots of reticulocyte lysate were immunoadsorbed wit.h preimmune serum (lane 1 ), UPJ56 (lane 2), nonimmune IgM ( l a m 3 ) . or the 3G3 IgM against hsp90 (lane 4 ) . The pellets were washed three times with TEG, and the native heterocomplex was resolved by SDS-PAGEand Western blotting.

2 0

-

3

4

5

6

7

8

9

"(I

-Po"-

PT-

FIG.7. Factors in hsp70-depleted reticulocyte lysate promote binding of hsp70 to hsp90.hsp90- and hsp70-free rabbit reticulocyte lysate was prepared by chromatography of lysate on DE52 followed by ATP-agarose as described under "Methods." Purified mouse hsp7O and purified mouse hsp90 were incubated together in HE buffer containing an ATP-generating system for 15 min a t 0 "C in the presence or absence of hsp70hsp90-free lysate. Samples were immunoadsorbed with nonimmune IgM or the 3G3 antibody, washed, and assayed for hsp56, hsp70, and hsp90 by immunoblotting. Lane I , sample of hsp70hsp90free lysate; lane 2, sample of purified hsp90; lane 3, sample of purified hsp7O; lanes 4 and 5,respectively, nonimmune and 3G3 immunoadsorption of incubation containing purified hsp70 and purified hsp90; lanes 6 and 7, nonimmune and 3G3 immunoadsorption of incubation containing purified hsp70, hsp90, and hsp70hsp90-free lysate; lanes 8 and 9, nonimmune and 3G3 immunoadsorption of incubation containing purified hsp70, hsp90, and boiled hsp70hsp90-free lysate.

and theimmunopellet was resuspended in HE buffer for 1h, a treatment that results loss in of the ratherweakly bound hsp56 but not the tightly bound hsp90 (lane 3). When the hsp90bound GR immunopellet is incubated with partially purified mouse hsp56, the hsp56 binds to the GR heterocomplex (lane 51, but if hsp90 is stripped from the GR heterocomplex with salt (lane 6),then purified hsp56 does not bind (lane 8). DISCUSSION

The UPJ56-immunoadsorbed native heat shock protein heterocomplex is thermolabile, but only half of the hsp90 dissociates from the immunoadsorbed complex, with molybdate having no effect on dissociation (Fig. lA). The dissociation of only half of the hsp90 could be explained if hsp56 were bound selectively to one isoform of hsp9O. hsp90 is present in cytosols predominantly as a dimer, and there is some evidence that heterocomplex with a n affinity that is high enough to permit its mouse hsp90 is presentas d a and PIP homodimers (42). Fig. LB detection by our method. In this experiment, the GR hetero- shows that both isoforms of hsp90 are associated with immucomplex was immunoadsorbed from WCL2 cell cytosol (lane 2 1, noadsorbed hsp56. If only homodimers of hsp90 exist, then

The Heat Shock Protein Complex

11160

B N I I

2

3

4

5

FIG.8. Molybdate inhibits dissociation of hspS6 from salttreated G R Effect of molybdate and tungstateon dissociation of hsp90 (A) and hsp56 (€3) in the presence of salt. GR heterocomplexes were immunoadsorbed from WCL2 cell cytosol using the BuGR antibody, and the immunopellets were washed three times in TEG buffer alone or with 20 m~ sodium molybdateor sodium tungstate. Samples werethen incubated 1h on ice in HE buffer containing various concentrations of salt, supplemented where noted with molybdate or tungstate. Proteins in the immunopellets were resolved by SDS-PAGE and Western blotting. NI,nonimmune sample; lane 1, HE buffer alone; lane 2, 100 m~ NaCI; lane 3,200 m~ NaCl; lane 4,300 m~ NaCI; lane 5,400 m~ NaCl. 1

2

-3

4

5

6

-m

ma

hSP9O

m

.3,

hSP7O

1,m

GR

hsp56

0

7

0

9

m

I ,

'

. I

FIG.9. Purified hspS8 binds to the hspSO-bound, butnot to the hspOO-free,G R GR was immunoadsorbedfrom WCL2 cytosol, and the immunopellets were resuspended inHE buffer alone ( t o dissociate hsp56) or in HEbuffer with 0.5 M KC1 (to strip off hsp56 and hsp90) for 1h a t 0 "C. The pellets were then washed and incubated with partially purified L cell hsp56 in HEbuffer containing 20 m~ sodium molybdate. The pellets were washed again, and proteins were resolved by SDSPAGE and Western blotting. Lanes 1 and 2, respectively, nonimmune and BuGR immunoadsorption of WCL2 cytosol; lane 3, BuGR pellet resuspended in HE; lanes 4 and 5,respectively, nonimmune and BuGR immunopellets resuspendedin HE, washed,then incubated with hsp56; lane 6, BuGR immune pellet resuspended in HE plus 0.5 M KCl; lanes 7 and 8,respectively, nonimmune and BuGR immune pellets resuspended in HE plus0.5 M KCl, washed, then incubated with hsp56;lane 9, the BuGR antibody alone adsorbed to protein A-Sepharose. The diffuse shadow extending below the hsp56 in lanes 3, 6, and 8 represents reaction of the second antibody with the BuGR heavy chain (seelane 9 which contains only BuGR).

hsp56 may bind equivalently to each hsp90 isoform, and the thermal dissociation of only half of the hsp90 shown in Fig. 1, A and B , may reflect thermal lability of the hsp90 dimer itself, with selective dissociation of the unit that is not directly bound to hsp56. However, recent experiments with antibodies selective for the a and /3 forms suggest that hsp90 exists as alp heterodimers2as well as homodimers. In thatcase, hsp56 could form a thermostable complex with one hsp90 isoform and a more thermolabile complex with the other. Although the stoichiometry of hsp56 and hsp90 in the hspheterocomplex is not defined, the GR heterocomplex is known to contain 1molecule of GR and 1molecule of hsp56 per dimer of hsp90 (43, 44). It is clear that hsp90 and hsp56 bind directly to each other G. Perdew, personal communication.

(Figs. 4 and 5 and Ref. 18).Unlike the binding of hsp90 to the GR, this binding is not ATP-dependent and does not require hsp70. When GR that isstripped free of associated proteins is reconstituted into a heterocomplex by reticulocyte lysate, hsp7O is absolutely required for the presence of both hsp90 and hsp56 in thereconstituted heterocomplex (9).As hsp56 appears to bind to hsp90 in a reversible equilibrium, it seems likely that hsp70 is required for the original hsp90 binding to the GR, and hsp56 then binds in equilibrium fashion to the GR-associated hsp90, much as we show it can in Fig. 9. It is important to note that all of the hsp56 in reticulocyte lysate is present in a large complex that can be mimicked by mixing purified hsp9O with hsp56 (Fig. 5). There is only a small amount of hsp7O that is coadsorbed from reticulocyte lysate with hsp56, and there is no stable binding between purified hsp7O and hsp56 (Fig. 6). Whenreticulocyte lysate is immunoadsorbed with the 3G3 antibody against hsp90, much more hsp70 is coimmunoadsorbed (Fig. 6B). As hsp56 is present in cytosols a t much lower abundance than hsp7O and hsp90 (121, only a portion of the hsp90 can be bound to hsp56. Our results are consistent with the proposal that some of the hsp90 in lysate is bound to hsp7O but not to hsp56, some hsp90 is bound to hsp56 but not hsp70, and, at any particulartime, a relatively small portion of the hsp90 is present in a heterocomplex containing all threehsps. This leads us to the notion that thehsp heterocomplex in reticulocyte lysate is dynamic, as opposed to a three-component stable complex. In our work on GR heterocomplex reconstitution (9), we have found that although hsp7O is required for heterocomplex formation, it can subsequently leave the complex. This apparent dynamic nature of the hsp heterocomplex in reticulocyte lysate may contribute ultimately to our understanding of why reticulocyte lysate can efficiently direct the formation of steroid receptor heterocomplexes but other cytosols cannot. In contrast to the interaction of hsp56 with hsp90, the binding of hsp70 to hsp90 is not a simple bimolecular equilibrium. An activity in reticulocyte lysate is required for substantial formation of an hsp70.hsp90 complex (Fig. 7). This activity is heat-labile, and we are calling it an hsp70.hsp90 complex-forming factor. Whether this factor involves onlyone protein or more than one protein is unknown. As shown in Fig. 7, when the factor is present, a heterocomplex between purified mouse hsp90 and hsp70 and rabbit hsp56 can be assembled under cell-free conditions. The transition metal oxyanions molybdate and vanadate inhibit dissociation of hsp56 from the GR heterocomplex (Fig.8). The binding of hsp9O to hsp56 in theLJF'J56-immunoadsorbed hsp heterocomplex is destabilized by salt, much like the binding of hsp90 to the GR in the BuGR-immunoadsorbed complex (Fig. 2), but thehsp90-hsp56 interaction in the absence of the GR is not stabilized by molybdate (Fig. 3). Molybdateinduces a conformationalchange in hsp90 (30)that likely accounts for the ability of the metal anion to stabilize the association of hsp90 with both steroid receptors and viral protein kinases (29). Renoir et al. (41) have proposed that hsp56 interacts with hsp90 and not with the hormone-binding protein in the steroid receptor heterocomplex. It is difficult to envision how molybdate would stabilize the presence of hsp56 in theGR heterocomplex unless the GR-bound hsp90 is in a different conformation from hsp90 in the heat shock protein heterocomplex or unless hsp56 contacts both the receptor and hsp90 in theGR heterocomplex. It seems clear from Fig. 9 that theglucocorticoid receptor protein is not itself sufficient for stable binding of hsp56 that can be detected with our method.However, an additional weak interaction of hsp56 with the receptor could add substantially to the overall affinity of hsp56 for the GR-hsp9O complex.

Heat

The

Shock Protein Complex

Acknowledgments-We are very grateful to Gordon Ringold and Margaret Hi& for providing the wcL2 h e of Chinese hamster ovary cells and to Paul Housley who recloned them and kindly provided the recloned line.

REFERENCES 1. Pratt, W. B. (1993) J. B i d . Chem. 268,21455-21458 2. Scherrer, L. C., Picard, D., Massa, E., Harmon, J. M., Simons, S . S., Jr., Yamamoto, K R., and Pratt, W.B. (1993) Biochemistry 32,5381-5386 3. Bresnick, E. H., Dalman, F. C., Sanchez, E. R., and Pratt, W.B. (1989) J. Biol. Chem. %,49924997 4. Sanchez, E. R., Meshinchi, S . , Tienrungroj, W., Schlesinger, M. J., Toft,D. O., and Pratt, W.B. (1987) J. B i d . Chem. 262,6986-6991 5. Smith, D. F., Schowalter, D. B., Kost, S . L., and Toft, D. 0.(1990) Mol. Endocrinol. 4,1704-1711 6. Scherrer, L. C., Dalman, F. C., Massa, E., Meshinchi, S., and Pratt,W. B. (1990) J. Biol. Chem. 266,21397-21400 7. Smith, D. F., Stensgard, B.A.,Welch.W. J., and ToR, D. 0.(1992) J. Biol. Chem. 267, 1350I-1356 8. Hutchison, K. A,, Czar, M. J., Scherrer, L. C., and Pratt, W.B. (1992) J. Biol. Chem. 267, 14047-14053 9. Hutchison, K A,, Dittmar, K. D., Czar, M. J., and Pratt, W.B. (1994) J. Biol. Chem. 269,5043-5049 10. Pratt, W. B., Hutchison, K. A., and Scherrer, L. C. (1992) l k n d s Endocrinol. Metab. 3, 326-333 11. Smith, D. F., and Toft, D. 0.(1993) Mol. Endocrinol. 4, 1704-1711 12. Sanchez, E. R., Faber, L. E., Henzel, W. J.,andPratt, W. B. (199O)Biochemistry 29,5145-5152 13. Perdew, G . H., and Whitelaw, M. L. (1991) J. B i d . Chem. 286,6708-6713 14. Whitelaw, M. L., Hutchison, K., and Perdew, G . H. (1991) J. Biol. Chem. 266, 1643G16440 15. Brugge, J. S . (1986) Curr. Top. Microbial. Immunol. 123, 1-22 16. Hutchison, K. A,, Brott, B. K., De Leon, J. H., Perdew, G. H., Jove, R., and Pratt, W.B. (1992) J. B i d . Chem. 267,2902-2908 17. Lindquist, S., and Craig, E. A. (1988)Annu. Reu. Genet. 22,631-677 18. Nadeau, K., Das, A., and Walsh, C. T.(1993) J . Biol. Chem. 268,1479-1487 19. Wiech, H., Buchner, J., Zimmerman, R., and Jakob, U. (1992) Nature 388, 169-170 20. Rothman, J. E. (1989) Cell 59,591-601 21. Tai, P. K., Maeda, Y., Nakao, K., Wakim, N. G., Duhring, J. L., and Faber, L. E. (1986) Biochemistry 25,52695275 22. Yem, A. W., Tomasselli,A. G . , Heinrikson, R. L., Zurcher-Neely, H., Ruff,V.A,,

11161

Johnson, R.A,, and Deibel, M.R., Jr. (1992) J. Bwl. Chem. 267,2868-2871 23. Lebeau, M. C., Massol, N., Herrick, J., Faber, L. E., Renoir, J. M., Radanyi, c., and Baulieu, E. E. (1992) J. B i d . Chem. 267,42814284 24. Tai, P. K., Albers, M. W., Chang, H., Faber, L. E., and Scbreiber, S . L. (1992) Science 258. 1315-1318 25. Schreiber, S . L. (1991) Science 251,283-287 26. Walsh, C. T.,Zydowsky, L. D., and MeKeon, F. D. (1992) J. B i d . Chem. 267, 13115-13118 27. Peattie, D. A,, Harding, M.W., Fleming, M. A,, DeCenzo, M. T., Lipke, J. A., Livingston, D. J., and Benasutti, M. (1992)Proc. Natl. Acad. Sci. U.S. A. 89,10974-10978 28. Pratt, W. B. (1992) BioEssays 14, 841-848 29. Hutchison, K. A, Stancato, L. F., Jove, R., and Pratt, W.B. (1992) J. B i d . Chem. 267, 13952-13957 30. Csermely, P., Kajtar, J., Hollosi,M., Jalsovszky, G., Holly, S., Kahn, C.R., Gergely, P., Jr., Soti, C., Mihaly, K., and Somogy, J. (1993) J. Biol. Chem. 268,1901-1907 31. Rufl', V.A,, Yem, A. W., Mums, P. L., Adams, L. D., Reardon, I. M., Deibel, M. R., Jr., and Leach, K. L. (1992) J . Biol. Chem. 267,21285-21288 32. Hirst, M. A, Northrup, J. P., Danielson, M., and Ringold, G . M. (1990) Mol. Endocrinol. 4, 162-170 33. Meshinchi, S., Sanchez, E. R., Martell, K J., and Pratt, W. B.(1990) J. Biol. Chem. 266,4863A870 34. Bresnick, E. H., Dalman, F. C., and Pratt, W. B. (1990) Biochemistry 29, 520-527 35. Schlossman. D. M., Schmid, S. L., Braell, W. A., and Rothman, J. E. (1984) J. Cell Biol. 99, 723-733 36. Dahmer, M. K., Housley, P. R., and Pratt, W. B. (1984)Annu. Reu. Physiol. 48, 6741 .. 37. Ullrich, S . J., Robinson, E. A., Lau, L. W., Willingham, M., and Appella, E. (1986) Proc. Natl. Acad. Sci. U.S. A. 83,31213125 38. Leach, K L., Dahmer, M. K., Hammond, N. D., Sando, J. J., and Pratt, W.B. (1979) J. Biol. Chem. 264, 11884-11890 39. Scherrer, L. C., Hutchison, K A., Sanchez, E. R., Randall, S . IC, and Pratt, W. B. (1992) Biochemistry 31, 7325-7329 40. Pelham, H. (1988)Nature 332, 77G777 41. Renoir, J. M., Radanyi, C., Faber, L. E., and Baulieu, E. E. (1990) J. B i d . Chem. 266,10740-10745 42. Minami, Y., Kawasaki, H., Miyata, Y., Suzuki, K, andYahara, I. (1991) J. B i d . Chem. 286,10099-10103 43. Rexin, M., Busch, W., and Gehring, U. (1991) J. Biol. Chem. 286,24601-24605 44. Alexis, M. N., Mavridou, I., and Mitaiou, D. J. (1992) Eur. J . Biochem. 205, 7m4 "