Molecular Cell

Article Modulation of the Hsp90 Chaperone Cycle by a Stringent Client Protein Oliver Robin Lorenz,1 Lee Freiburger,1,2 Daniel Andreas Rutz,1 Maike Krause,1 Bettina Karolina Zierer,1 Sara Alvira,3 Jorge Cue´llar,3 Jose´ Marı´a Valpuesta,3 Tobias Madl,1,2,* Michael Sattler,1,2,* and Johannes Buchner1,* 1Center

for Integrated Protein Science Munich, Department Chemie, Technische Universita¨t Mu¨nchen, 85478 Garching, Germany of Structural Biology, Helmholtz Zentrum Mu¨nchen, 85764 Neuherberg, Germany 3Centro Nacional de Biotecnologı´a (CNB-CSIC), Darwin 3, 28049 Madrid, Spain *Correspondence: [email protected] (T.M.), [email protected] (M.S.), [email protected] (J.B.) http://dx.doi.org/10.1016/j.molcel.2014.02.003 2Institute

SUMMARY

Hsp90 is the most abundant molecular chaperone in the eukaryotic cell. One of the most stringent clients is the glucocorticoid receptor (GR), whose in vivo function strictly depends on the interaction with the Hsp90 machinery. However, the molecular mechanism of this interaction has been elusive. Here we have reconstituted the interaction of Hsp90 with hormone-bound GR using purified components. Our biochemical and structural analyses define the binding site for GR on Hsp90 and reveal that binding of GR modulates the conformational cycle of Hsp90. FRET experiments demonstrate that a partially closed form of the Hsp90 dimer is the preferred conformation for interaction. Consistent with this, the conformational cycle of Hsp90 is decelerated, and its ATPase activity decreases. Hsp90 cochaperones differentially affect formation of the Hsp90-GR complex, serving as control elements for cycle progression and revealing an intricate interplay of client and cochaperones as molecular modulators of the Hsp90 machine. INTRODUCTION Hsp90 is the most complex chaperone machinery in the eukaryotic cell. Together with a large set of cochaperones, it assists the structure formation of hundreds of proteins (Li et al., 2012; Picard, 2002). Further, Hsp90 has been implicated in quality control and the degradation of cytosolic proteins (McClellan et al., 2005). Besides kinases and E3 ligases, members of the steroid hormone receptor (SHR) family represent the most stringent Hsp90 client proteins (Taipale et al., 2012). The glucocorticoid and progesterone receptors, especially, have served as role models to elucidate the composition of the Hsp90 machinery (Pratt and Toft, 1997). The glucocorticoid receptor (GR) is a steroid hormoneactivated transcription factor that regulates numerous genes involved in various biological processes (Heitzer et al., 2007). GR activity in vivo is tightly regulated by Hsp90 (Picard et al., 1990). Early reconstitution experiments established a minimal system for GR maturation, which contains five proteins: Hsp40, Hsp70, Hop, Hsp90, and p23 (Pratt and Dittmar, 1998). Hsp40

and Hsp70 interact with the receptor at an early stage of maturation (Herna´ndez et al., 2002). Hop/Sti1 connects this assembly to the Hsp90 cycle (Chen and Smith, 1998). Binding of a PPIase (FKBP51, FKBP52, and Cyp40) to Hsp90 weakens the binding of Hop/Sti1, resulting in its release (Li et al., 2011). Finally, p23, which operates during late steps in receptor action, stabilizes the Hsp90-SHR-interaction (Freeman et al., 2000). Although believed for a long time, it seems that Hsp90 and associated cochaperones are not released from GR upon binding of the hormone. In this context, Hsp90 also plays a role in trafficking of the holoreceptor into the nucleus (Galigniana et al., 2001). In recent years, work on Hsp90 has mainly focused on its structure, dynamics, and cochaperone interactions (Jackson, 2013; Ro¨hl et al., 2013). Each subunit in the Hsp90 dimer consists of three domains: an N-terminal ATP-binding domain (Hsp90-N), a middle domain (Hsp90-M), and a C-terminal dimerization domain (Hsp90-C) (Ali et al., 2006). During its functional cycle, Hsp90 progresses from an N-terminal open conformation to a closed state via several intermediates (Hessling et al., 2009). These conformational transitions preceding ATP-hydrolysis are the rate-limiting steps of the cycle (Graf et al., 2009; Pullen and Bolon, 2011; Weikl et al., 2000). For eukaryotic Hsp90, they seem to be isoenergetic (Mickler et al., 2009; Southworth and Agard, 2008). Thus, Hsp90 will fluctuate randomly between different states. ATP hydrolysis provides directionality and cochaperones render the cycle deterministic (Li et al., 2013). It is unknown how clients fit in this picture, despite some studies on the effects of Hsp90 on model clients (Genest et al., 2013; Street et al., 2011). Here we used the GR-ligand binding domain (GR-LBD) as the physiological interaction partner of Hsp90 (Howard et al., 1990) for elucidating the interplay between Hsp90 and a stringent client. We show that the GR-LBD interacts with specific conformations of Hsp90 in a nucleotide-dependent manner. Binding to Hsp90 slows down the conformational transitions, and cochaperones differentially affect the GR interaction. These data provide a conceptual framework to integrate GR into the Hsp90 cycle. RESULTS Analysis of the Hsp90 Dependence of a GR Mutant The GR consists of three domains (Figure 1A): an N-terminal domain, a DNA-binding domain, and a ligand-binding domain, linked by a short-hinge region (Kumar and Thompson, 1999). Molecular Cell 53, 941–953, March 20, 2014 ª2014 Elsevier Inc. 941

Molecular Cell The GR-Hsp90 Interplay

Figure 1. Characterization of the Purified GR-LBD (A) Domain architecture of the human GR. NTD indicates N-terminal domain; DBD indicates DNAbinding domain; hinge indicates hinge region; LBD indicates ligand binding domain. Domain borders as indicated. (B) Hsp90 dependence of GR mutants; shown are the averaged b-galactosidase activities and SDs of three independent experiments with each GR variant. Black bars show b-galactosidase activities of DMSO-treated control cells and red bars show remaining activity after treatment with 20 mM Radicicol. (C) Secondary structure of apo- and holo-GRLBDm. (D) AUC sedimentation velocity analysis of apoand holo-GR-LBDm. (E) Thermal stability of GR-LBDm was followed by CD spectroscopy. (F) Hormone binding affinity of GR-LBDm. Data was fit by a ligand-depletion model; inset shows binding kinetics of apo-GR-LBDm to F-DEX.

GR function in vivo strictly depends on the interaction of the GRLBD with the Hsp90 chaperone machinery (Picard et al., 1990). As the wild-type (wt) GR-LBD is rather unstable (Seitz et al., 2010), we tested whether the stabilized GR mutant GR-LBDF602S/A605V/V702A/E705G/M752T (GR-LBDm) (Seitz et al., 2010) can be used to shed light on the Hsp90-GR interplay. To test its Hsp90 dependence in vivo, we expressed the respective full-length GR variant (GRm) in yeast and assayed its activity via a reporter assay (Nathan and Lindquist, 1995). For comparison, we also included wt GR and GR variants with only one stabilizing mutation. Upon activation with the ligand 11-deoxycorticosterone, all GR mutants showed increased hormone responsiveness compared to the wt protein (Figure 1B). In the presence of the Hsp90 inhibitor Radicicol, the activities of wt GR and GR mutants were significantly decreased, indicating that the mutants are Hsp90 dependent. Characterization of the Purified GR-LBD To determine the characteristics of a stringent Hsp90 client, we expressed the GR-LBDm in E. coli in the presence of the stabi942 Molecular Cell 53, 941–953, March 20, 2014 ª2014 Elsevier Inc.

lizing ligand dexamethasone (DEX) and purified it to homogeneity. Far-UV-CD spectroscopy showed that GR-LBDm displays a high a-helical content (Figure 1C), in line with crystal structures of hormone-bound GR-LBD (Bledsoe et al., 2002; Seitz et al., 2010). Analytical ultracentrifugation (AUC) experiments revealed that GR-LBDm is monomeric (Figure 1D). The GR-LBDm is stable up to
50 C and then starts to unfold in a cooperative transition (Figure 1E). At higher temperatures, it is completely unfolded (Figure 1C). The apparent melting temperature of GR-LBDm was determined to be 59.6 C (Figure 1E). Hormone binding of GRLBDm was assayed after extensive dialysis of the GR-LBDm by fluorescence anisotropy using fluorescein-labeled DEX (F-DEX) (Figure 1F). Fitting of the titration data yielded an apparent dissociation constant (KD) of 77 nM. Competition experiments with excess unlabeled DEX after saturation with F-DEX demonstrated that the ligand can be readily exchanged (inset Figure 1F). Thus, hormone binding is dynamic, and the ligand binding pocket of GR-LBDm is accessible to hormone in the absence of Hsp90. In contrast to holo-GR, expression of the GR-LBDm in the absence of hormone (apo-GR-LBDm) yielded a partially structured and aggregation-prone protein (Figures 1C and 1D). Protein unfolding started already at 10 C and continued over a wide temperature range (Figure 1E). Interestingly, apo-GRLBDm was able to bind and exchange hormone with comparable affinity (KD = 97 nM) (Figure 1F). In line with previous studies (Bledsoe et al., 2002), the point mutant GR-LBD-F602S behaved similar to GR-LBDm in terms of secondary structure and hormone binding, supporting the notion that the additional

Molecular Cell The GR-Hsp90 Interplay

Figure 2. GR-LBDm Binds Preferentially to Closed Hsp90 Conformations (A) Nucleotide-induced Hsp90 conformations affect GR-binding; AUC sedimentation velocity experiments were performed using *GR-LBDm; normalized dc/dt values were plotted against s20,W/S values. See also Figure S1A for the specificity of the interaction, Figure S1B for binding of GR-LBD F602S to Hsp90, and Figure S1C for conservation in the human system. (B) GR-LBDm binds with different affinity to nucleotide-induced conformations of Hsp90; shown are AUC sedimentation velocity titrations with *GR-LBDm using increasing amounts of Hsp90. The normalized concentration of bound GR-LBDm was plotted against the Hsp90 concentration. Data was fit according to a single-site binding model. (C) SAXS data showing a comparison of the experimental radial density distributions of Hsp90ATP and with increasing stoichiometric ratios of GR-LBDm as indicated. (D) SAXS data showing a comparison of the experimental radial density distributions of Hsp90GR-LBDm complexes in the absence and presence of nucleotides. For experimental radial density distributions of Hsp90 alone with different nucleotides, see Figure S1D.

stabilizing mutations in GR-LBDm do not further affect its functional properties (data not shown). GR-LBDm Binds Preferentially to a Partially Closed Hsp90 Conformation As the apo-GR-LBDm turned out to be highly unstable in vitro, and since ligand binding to SHRs does not directly seem to stimulate Hsp90 release (Smith, 1993), we focused on the hormone-bound form of GR-LBDm to define the principles of the interaction of a client with Hsp90. We used AUC coupled to fluorescence detection to follow the association of Hsp90 with the holo-GR-LBDm labeled with a fluorescent dye (*GR-LBDm) (Figure 2A; Figure S1A available online). *GR-LBDm alone sedimented with an s value of 2.7 S, consistent with its monomeric state. Upon addition of yeast Hsp90, the s value increased to 5.6 S, which demonstrates complex formation between *GRLBDm and Hsp90. This interaction was observed both in the absence of nucleotides and in the presence of ADP. However, binding of GR was strongly enhanced in the presence of ATP and moderately increased with the slowly hydrolyzing ATP analog ATPgS (Figure 2A). Consistent with a gain in compaction of the Hsp90 dimer upon ATP or ATPgS binding (Li et al., 2013), the s value of the complex increased to 5.9–6.1 S. Addition of the nonhydrolysable ATP analog 50 -adenylyl-b, g-imido-diphosphate (AMP-PNP), which induces a fully closed Hsp90 conformation (Ali et al., 2006), was less efficient in promoting complex formation. The interaction was specific, as addition of the Hsp90 inhibitor Radicicol resulted in the release of a significant fraction of *GR-LBDm from Hsp90-ATP (Figure S1A). We also performed binding experiments with fluorescently labeled Hsp90 (*Hsp90) using unmodified GR-LBDm and GR-LBD-F602S (Figure S1B). Again, in the presence of ATP, significant peak shifts were

observed. *Hsp90-ATP-GR-LBD complexes sedimented with higher s values in these experiments (7.2 S), suggesting that Hsp90 can accommodate two GR molecules if saturating amounts of GR are present. The results above indicate that binding to a partially closed state of Hsp90 is preferred. To further test this notion, we titrated different amounts of Hsp90 to the *GR-LBDm in the absence and presence of various nucleotides and monitored complex formation by AUC (Figure 2B). GR-LBDm binds Hsp90 in the presence of ATP or ATPgS with a KD of 0.8 mM and 1.5 mM, respectively, whereas open Hsp90 bound more weakly to GR-LBDm (KD = 2.9 mM for apo-Hsp90 and KD = 2.5 mM for Hsp90-ADP). A completely closed conformation induced by AMP-PNP led to a further reduction of affinity (KD = 4.5 mM). This scenario is conserved between yeast and man, as complex formation of the GR-LBDm with human Hsp90b was nucleotide dependent as well (Figure S1C). SAXS experiments gave further insight into the Hsp90-GRLBDm complex. Consistent with recent literature (Krukenberg et al., 2011), the SAXS data of apo-Hsp90 indicate that it adopts an open conformation (high maximal dimension [Dmax] and radius of gyration [Rg]) (Figure S1D; Table S1). Binding of ATP, ATPgS, or AMP-PNP induces a more compact conformation of Hsp90, with Hsp90-AMP-PNP showing the highest degree of compaction (Figure S1D). The Hsp90-GR-LBDm complex was studied at different Hsp90-GR ratios in the presence of ATP. We found that the apparent molecular mass and the Rg of the Hsp90-GR-LBDm complex increased up to a 2:2 ratio (Hsp90 monomer:GR-LBDm) (Figure 2C; Table S1). Above that ratio, both the apparent molecular mass and the Rg decreased due to the presence of unbound GR-LBDm. Comparison of SAXS data obtained for free Hsp90 and in the presence of increasing Molecular Cell 53, 941–953, March 20, 2014 ª2014 Elsevier Inc. 943

Molecular Cell The GR-Hsp90 Interplay

Figure 3. NMR Analysis of the Hsp90-GR Interaction (A) Zoomed views of 1H,15N HSQC experiments for Hsp90-N (top) or Hsp90-M (bottom) domains free (black) or with one molar equivalent of GRLBDm in red. (B) Zoomed view of 1H,15N HSQC experiments for Hsp90-M in the presence of GR-LBDm spin labeled with IPSL oxidized (black) or reduced (red). (C) Zoomed view of 1H,15N HSQC experiments for Hsp90-C free (black) or bound to GR-LBDm (red). (D) CSPs and disappearing peaks observed in the titrations shown in (A) are highlighted as blue, green, and orange spheres for amides in the Hsp90-N, Hsp90-M, and Hsp90-C domains, respectively, and red spheres for peaks that experienced 20% bleaching (red) in the presence of oxidized spin label, mapped on a monomer in the structure of the Hsp90 closed dimer (PDB: 2CG9). (E) Zoomed views of 1H,15N correlations for segmentally isotope-labeled Hsp90-NM domain constructs, where either the -N (top) or -M (bottom) domain is detected free (black) or bound to GR-LBDm (red). (F) CSPs and disappearing peaks observed upon titration of the Hsp90-NM construct are indicated based on the NMR titrations shown in (E); color code as in (D). (G) Superposition of 1H,15N HSQC spectra of 15Nlabeled GR-LBDm free (black) or bound to the Hsp90-M (green) and Hsp90-NM (red) domains. N- and NM-domain constructs were measured in the presence of AMP-PNP. For full-spectra, CSP intensity ratios, and spectra of isolated Hsp90-N, Hsp90-M, Hsp90-C, and Hsp90-NM constructs and GFP, see Figure S2 and Table S2.

dependence of the molecular mass), and Hsp90 adopts an open conformation. Binding of ATPgS and AMP-PNP shifts the complex toward a more compact conformation (Rg: 54.9/51.3 A˚; Dmax: 220/220 A˚) (Figure 2D; Table S1).

amounts of GR-LBDm revealed a stepwise increase of the apparent molecular mass by approximately 25 kDa (from 160 kDa to 185 kDa to 208 kDa), which matches well the mass of GR-LBDm (29 kDa) (Figure 2C; Table S1). This result suggests that the two subunits of Hsp90 exhibit two independent binding sites for the GR-LBDm. The Rg and Dmax of Hsp90 and the Hsp90-GR-LBDm complex in the presence of ATP are very similar (Rg: 58.3 A˚; Dmax: 200/220 A˚), indicating that GR-LBDm binds to the central part of Hsp90. In the nucleotide-free complex, GR-LBDm binds weakly to Hsp90 (strong concentration 944 Molecular Cell 53, 941–953, March 20, 2014 ª2014 Elsevier Inc.

Structural Features of the Hsp90GR-LBDm Complex To provide further details for the Hsp90GR interaction at residue resolution, we applied NMR spectroscopy using individual Hsp90-N, Hsp90-M, and Hsp90-C domains. As mainly the Hsp90-M domain seems involved in GR binding (Bohen and Yamamoto, 1993; Cadepond et al., 1993; Hawle et al., 2006), we recorded NMR spectra of the 15 N-, 13C-, and 2H-labeled Hsp90-M domain with and without GR-LBDm. The overall spectral similarity of 1H,15N correlation experiments indicates that GR-LBDm binding does not induce large conformational changes in the Hsp90-M domain (Figures 3A and S2E). However, several NMR signals show significant chemical shift perturbations (CSPs) or severe line broadening (Figures 3A and S2E; Table S2). Additionally, we recorded

Molecular Cell The GR-Hsp90 Interplay

Figure 4. GR-LBDm Modulates the ATPase and the Conformational Cycle of Hsp90 (A) GR-LBDm decreases the ATPase activity of Hsp90 in a concentration-dependent manner; The mean values of three independent measurements were plotted; error bars as indicated. See Figure S3A for the effect of GR-LBD-F602S on the Hsp90 ATPase, Figure S3B for modulation of the ATPase activity of a disulfide-bridged Hsp90-NM construct by GR-LBDm, and Figure S3C for ATPase activity in the presence of the nonclient protein GFP. (B and C) GR-LBDm decreases the nucleotideinduced closing reaction of Hsp90 in a concentration-dependent manner; closing kinetics upon addition of ATPgS in the presence and in the absence of GR-LBDm. The relative Hsp90 closing rate was plotted against increasing amounts of GR-LBDm. Please see Figure S3D for the influence of GR-LBD-F602S on the nucleotideinduced closing reaction of Hsp90. (D) GR-LBDm stabilizes a partially closed conformation of Hsp90; FRET-chase kinetics upon the addition of unlabeled Hsp90 or ADP in the presence or absence of GR-LBDm. The reason for its biphasic nature is not clear, but it likely results from conformational rearrangements in Hsp90, as they are also observed in the absence of GR-LBDm.

spectra using GR-LBDm spin labeled at C638 and 15N-labeled Hsp90-M. Here, specific peaks experienced paramagnetic relaxation enhancements (PREs) with a maximum signal reduction of 80% (Figure 3B), demonstrating direct interaction. We also obtained information on the binding orientation, as most of our PRE effects are located near the Hsp90-NM interface on the Hsp90-M domain. This suggests that C638 of the GRLBDm is orientated toward the Hsp90-N domain when interacting with Hsp90. Importantly, addition of the nonclient protein GFP to Hsp90-M did not result in any CSPs or intensity changes, confirming the specificity of the observations (Figure S2L). For the Hsp90-N domain, AMP-PNP induced significant CSPs similar to a previous report (Figure S2A) (Dehner et al., 2003). Upon addition of the GR-LBDm to the nucleotide-bound Hsp90-N domain, few NMR signals experience additional CSPs or reduced signal intensity (Figures 3A, S2D, S2I, and S2J; Table S2). The addition of GR-LBDm to the Hsp90-C domain resulted in a general reduction of NMR signal intensities (Figures 3C and S2F; Table S2). As this was also observed within the core of the Hsp90-C domain, we attribute these effects to a decrease in tumbling upon addition of GR-LBDm. For a few signals, such as I592, CSPs were observed near the Hsp90-MC domain interface and the dimerization interface. Mapping residues affected by GR-LBDm binding to the Hsp90-N, Hsp90-M and Hsp90-C domains on the Hsp90 crystal structure shows that the majority of CSPs face the internal cleft in the closed conformation of the Hsp90 dimer (Figure 3D). Since binding of GR-LBDm to Hsp90 is nucleotide dependent (Figures 2A and 2B), we also studied Hsp90-NM domain constructs. We employed segmental isotope labeling of the individual domains (Hsp90-N*M, -NM*) to reduce signal overlap due to

the large number of residues (>500 residues). In addition to the effects observed for AMP-PNP (Figures S2B and S2C), GRLBDm induced significant CSPs or intensity changes for several signals (Figures 3E, S2G, S2H, S2I, and S2J). Again, many of the residues are clustered at the inside cleft of the Hsp90 dimer in the closed conformation (Figure 3F). This suggests that GR-LBDm binds to a similar location in Hsp90-NM* and the Hsp90-M domain. Notably, the spectral changes observed in the Hsp90NM domain construct are more widespread and stronger than those seen upon binding to the individual Hsp90-N and Hsp90M domains (Figures 3A, 3E, S2D, S2E, S2G, and S2H). The residues affected represent a larger surface area compared to the isolated Hsp90-M domain, and many of the additional residues are located near the Hsp90-NM domain interface (Figure 3F; Table S2). To gain insight into the conformation of the GR-LBDm when interacting with Hsp90, we produced 15N-labeled GR-LBDm and compared NMR spectra of the protein free in solution and bound to either Hsp90-M or Hsp90-NM (Figure 3G). The NMR spectra of GR-LBDm are well dispersed, demonstrating that GR-LBDm is folded. In the presence of Hsp90-M, many signals of the GR-LBDm undergo CSPs or a reduction in intensity (Figure 3G). Notably, these effects are even more pronounced in the presence of the Hsp90-NM domain construct. Binding of the GR-LBDm Modulates the ATPase and Conformational Cycle of Hsp90 Next, we investigated the influence of GR-LBDm on the Hsp90 ATPase cycle. GR-LBDm decreased the ATPase activity of Hsp90 by 85% in a concentration-dependent manner (Ki = 7.1 mM) (Figure 4A). A comparable Ki of 8.7 mM was obtained for GR-LBD-F602S (Figure S3A). Consistent with the involvement Molecular Cell 53, 941–953, March 20, 2014 ª2014 Elsevier Inc. 945

Molecular Cell The GR-Hsp90 Interplay

(legend on next page)

946 Molecular Cell 53, 941–953, March 20, 2014 ª2014 Elsevier Inc.

Molecular Cell The GR-Hsp90 Interplay

of Hsp90-N and Hsp90-M in GR-LBDm-binding (Figure 3), the ATPase activity of a disulfide-bridged, dimeric Hsp90NM construct was also slowed down (Figure S3B). Notably, the Hsp90-ATPase was not affected by the nonclient GFP (Figure S3C). Our results imply that the rate-limiting conformational changes of Hsp90 are affected by binding of the GR-LBDm. To directly monitor the influence of the GR-LBDm on these rearrangements, we used our established Hsp90-FRET-system (Hessling et al., 2009). Monitoring the ATPgS-induced closing of the Hsp90FRET-complex, we observed a marked decrease in FRET efficiency in the presence of GR-LBDm (Figure 4B). In agreement with the ATPase data mentioned above (Figure S3A), GR-LBDF602S behaved strikingly similar to GR-LBDm (Figure S3D). Plotting the apparent half-time of the closing reaction against the GR-LBDm concentration (Figure 4C) yielded an apparent Ki of 1.9 mM, which is in excellent agreement with the observed binding affinity of GR-LBDm for Hsp90 under ATPgS conditions determined by AUC (Figure 2B). The stability of the FRET complex in the presence of the GR-LBDm and ATPgS was investigated by chase experiments using unlabeled Hsp90. Upon equilibration, the FRET signal was lost, as heterodimers between labeled Hsp90 and unmodified subunits form Figure 4D. This reaction is sensitive to the conformations of Hsp90 with open states exchanging more rapidly than closed ones (Hessling et al., 2009; Li et al., 2013). In the presence of the GR-LBDm, the dissociation of the FRET complex was almost completely inhibited, indicative of the presence of a stabilized, partially closed state. The reopening reaction of the Hsp90 dimer upon addition of excess ADP was not as strongly affected by GR-LBDm, which is in line with the notion that GR-LBDm has a lower affinity for an (open) Hsp90 posthydrolysis state (Figure 4D). Thus, our results suggest that the GR-LBDm favors a partially closed conformation of Hsp90 and that its binding decelerates the progression of the cycle. The GR-LBDm-Hsp90 Interaction Is Modulated by Cochaperones Cochaperones provide an additional level of Hsp90 regulation, as they shape the conformational transition and modulate the ATP turnover of Hsp90 (Li et al., 2012). As several cochaperones are important for GR processing in vivo (Chang et al., 1997; Dittmar et al., 1997; Galigniana et al., 2001), we determined their influence on the interaction of GR with Hsp90. Hop/Sti1 is a cochaperone that inhibits the Hsp90 ATPase by blocking the conformational transitions to the closed state (Hessling et al., 2009; Lee et al., 2012; Richter et al., 2003). We formed an Hsp90 complex with fluorescently labeled Sti1 (Sti1*) under ATPgS conditions and monitored the influence of GR-LBDm by

AUC (Figure 5A, panels 1–3). Upon addition of the GR-LBDm, no significant change in the s value was observed, indicating that GR-LBDm did not bind. Consistent with this, we found that addition of Sti1 to a preformed *GR-LBDm-Hsp90 complex resulted in the release of *GR-LBDm (Figure 5A, panels 12–14). These data suggest that Sti1 reduces the affinity of Hsp90 for GR-LBDm, presumably by keeping Hsp90 in a fully open conformation even though ATPgS is present. In the absence of clients, Cpr6, Sti1, and Hsp90 form an asymmetric intermediate complex that allows further progression of the Hsp90 cycle (Li et al., 2011). We speculated whether binding of Cpr6 would promote GR binding to a Sti1*-Hsp90 complex. When the GR-LBDm was added to a ternary Sti1*-Hsp90-Cpr6 complex, the fraction of free Sti1* increased. Additionally, the s value of complexed Sti* shifted from 7.9 S to 8.4 S, indicating the formation of a quaternary complex of Sti1, Hsp90, Cpr6, and GR-LBDm (Figure 5A, panels 3–5). Notably, this quaternary complex could also be observed when Cpr6 or the GR-LBDm was labeled (Figure 5A, panels 10 and 13–16). In sum, these findings suggest that the Cpr6-Hsp90-Sti1 complex provides a GR-binding-competent Hsp90 conformation. ATPase measurements revealed that Cpr6 could compensate the loss in ATPase activity induced by GR-LBDm (Figure 5B). Even in the presence of the ATPase-inhibiting cochaperone Sti1, Cpr6 slightly increased Hsp90-ATP turnover, reflecting its important role in cycle progression. Of note, we further tested the ATPase activator Aha1 (Panaretou et al., 2002; Retzlaff et al., 2010) for complex formation with Hsp90 and GR-LBDm. Unexpectedly, Aha1 and GR-LBDm could not interact simultaneously with Hsp90, both in the presence and absence of Cpr6 (data not shown). As the conformational transition of Hsp90 progresses, p23 enters the cycle and promotes the exit of Sti1 in a concerted action with Cpr6 (Li et al., 2011). Using AUC, we found that p23 can interact with *GR-LBDm-Hsp90 complexes (7 S) and ternary GRLBDm-Hsp90-Cpr6 (8.4 S) complexes (Figure 5A, panels 17 and 18). In line with previous reports (Richter et al., 2004; Siligardi et al., 2004), p23 inhibited the ATPase rate of Hsp90 by roughly 40%. Interestingly, this effect was additive to that of GR-LBDm, as deduced from the reduction of 66% (Figure 5B). Addition of Cpr6 could only partly restore Hsp90 ATPase activity. Thus, the formation of late GR-chaperone assemblies lead to a prolonged interaction of Hsp90 with GR as ATP hydrolysis is slowed down. Taken together, these results show that the kinetics of the Hsp90 cycle are determined by agonistic and antagonistic effects of the client and cochaperones. Next, we wondered whether the GR binding stoichiometry might depend on Hsp90 cochaperones. To investigate this, we focused on the Hsp90-GR-LBDm-p23 complex and established

Figure 5. Cochaperones Alter the GR-LBDm-Hsp90 Interaction (A) Cochaperones influence the affinity of GR-LBDm for Hsp90. Shown are AUC sedimentation velocity experiments for different cochaperone combinations. The differences in the s values observed for identical complexes do not reflect differences in their composition. (B) Hsp90 ATPase activity in the presence of GR-LBDm and various cochaperones determined from three independent measurements. Error bars as indicated. (C) SAXS data showing a comparison of the experimental radial density distributions of Hsp90-ATPgS at increasing stoichiometric ratios of GR-LBDm and p23 as indicated. For complex formation of Hsp90 with p23, see Figure S4. (D) Excess ADP disrupts GR-LBDm-Hsp90 and cochaperone complexes. The fractions of complexed *GR-LBDm were calculated from AUC sedimentation velocity experiments and plotted for each combination.

Molecular Cell 53, 941–953, March 20, 2014 ª2014 Elsevier Inc. 947

Molecular Cell The GR-Hsp90 Interplay

Figure 6. Structural Models for Hsp90-GRLBDm and Hsp90-p23-GR-LBDm Complexes (A) EM reconstruction of the Hsp90-GR-LBDm complex. See also Figure S5 for further information. (B and C) Representation of Hsp90-GR-LBDm and Hsp90-GR-LBDm-p23 complexes. For lowest energy structures from the rigid body modeling calculations of free Hsp90, Hsp90-GR-LBDm, Hsp90-p23-GR-LBDm, and comparison of experimental radial density distribution with theoretical radial density distributions, see Figure S6. (D and E) Surface representation of Hsp90 showing the overlap with reported mutations (Bohen and Yamamoto, 1993; Fang et al., 2006; Genest et al., 2013; Nathan and Lindquist, 1995) and comparison of binding sites for the cochaperones Aha1 and Sti1 with the GR-LBDm binding site (Retzlaff et al., 2010; Schmid et al., 2012). The GR-LBDm binding site is shown in orange, surface exposed mutations are shown in magenta.

the GR binding stoichiometry by SAXS. In agreement with previous studies (Ali et al., 2006), p23 binds to Hsp90-ATPgS in a 2:2 stoichiometry (Figure S4) and shifts the ATPgS- and AMP-PNPbound equilibrium states further toward a closed conformation. However, for the Hsp90-GR-LBDm-p23 complex we observed a stable, compact complex at a 2:1:1 stoichiometry. Above that stoichiometry, no further increase in the molecular mass corresponding to an additional binding of p23 or GR-LBDm could be observed (Figure 5C). Similarly, we could not form a complex with two GR molecules bound to Hsp90 in the presence of Cpr6 (data not shown). These results imply that once Hsp90 cochaperones enter the Hsp90 cycle, the two binding sites are no longer independent of each other or that asymmetry is induced (Figures 2C and 2D). GR Release from Hsp90 Studies on the SHRs argue against the notion that ligand binding induces Hsp90 release from GR (Galigniana et al., 2001; Smith, 1993). However, client release from Hsp90 may be stimulated by ATP hydrolysis (Smith, 1993). Indeed, whereas roughly 50% of GR-LBDm was bound to Hsp90 under ATPgS conditions, the addition of excess ADP decreased the affinity for GR-LBDm and resulted in the release of Hsp90-bound *GR-LBDm (Figure 5D). This finding is consistent with the FRET experiments in which the reopening of the Hsp90 dimer by ADP was influenced by GR-binding (Figure 4D). The release in the presence of ADP was not affected by the nature of GRHsp90 complexes studied, suggesting that mainly the restoration of the open conformation after ATP hydrolysis is important for the dissociation of GR. 948 Molecular Cell 53, 941–953, March 20, 2014 ª2014 Elsevier Inc.

A Model for the Hsp90-GR-LBDm (-p23) Complex In addition to SAXS measurements, we also performed negative-stain electron microscopy (EM) to gain structural information on the GR-Hsp90 complex. Hsp90-GR-LBDm complexes were formed in the presence of AMP-PNP, isolated using GraFix (Kastner et al., 2008), and negatively stained (Figures S5A and S5B). From the electron micrographs, a total of 20,016 particles were selected, classified, and used for the 3D reconstruction, as described in Experimental Procedures (Figures S5C–S5F). The reconstructed complex (20 A˚ resolution) has an elongated shape (135 A˚ long and 110 A˚ wide) expected for the Hsp90 dimer with two protrusions located in the middle of the long sides of the structure, which we attribute to two GR-LBDm molecules bound to Hsp90 (Figure 6A). Our SAXS- and NMR-data-derived structural model shows that, similar to Hsp90 alone, the overall conformation of the Hsp90-GR-LBDm complex in the presence of ATP is extended. Two GR-LBDm molecules are positioned in the central part of Hsp90 in between the Hsp90-NM arms and the C-terminal domain (Figures S1D and S6A–S6C). In the ‘‘open’’ complex, Hsp90 and the GR-LBDm form a large interaction interface (1500–2000 A˚2) with
90% of the interface being located within Hsp90-M. ATPgS and AMP-PNP induce large conformational rearrangements that shift the complex toward the ‘‘closed’’ state (Figures S6D and S6E). To generate a structural model for the ‘‘closed’’ state, we docked the GR-LBDm onto the ‘‘closed’’ crystal structure of Hsp90 (PDB 2CG9) using EM, SAXS, and NMR data for the binding interfaces (Figure 6B). In agreement with the biochemical data, the model reveals independent binding sites for GR-LBDm on the two subunits of the Hsp90 dimer. Consistent with the distribution of NMR CSPs and the PRE data, additional contacts between the GR-LBDm and the Hsp90-N domain of the opposite Hsp90 monomer can be observed (Figures 6B and S6F).

Molecular Cell The GR-Hsp90 Interplay

Binding of p23 to Hsp90-GR-LBDm changes GR binding stoichiometry (Figure 5C). The structures of the ternary Hsp90-GRLBDm-p23 complexes in the presence of ATPgS were modeled using the high-resolution crystal structure of the Hsp90-p23 complex and GR-LBD, as well as NMR data of the Hsp90-GR-LBDm binding interface and SAXS data as input (Figures 6C, S6G, and S6H). Docking calculations suggest that the position of the GRLBDm is identical to the ‘‘closed’’ Hsp90-GR-LBDm complex (Figures S6F and S6G). p23 binds to the opposite side of the Hsp90-N-N interface. Simultaneous binding of p23 and the GRLBDm to one Hsp90 monomer seems disfavored (Figure 5C). DISCUSSION In the cell, there seems to be a large, but defined, set of Hsp90 clients under physiological conditions (McClellan et al., 2007; Taipale et al., 2012). Among the best-studied clients of Hsp90 are the SHRs, and especially the GR. Here, we reconstituted the GR-Hsp90 complex from purified components. The recombinant apo-GR-LBDm expressed in E. coli is only partly folded and very aggregation prone. Only in the presence of hormone were we able to obtain stable, monomeric GR-LBDm, which formed specific complexes with Hsp90 in a conformation-sensitive manner. The binding site for clients on Hsp90 is still enigmatic, with evidence in the literature for an involvement of all three domains of Hsp90 (Ro¨hl et al., 2013). We found that binding of GR-LBDm to Hsp90 does not induce large conformational changes in the individual domains. Instead, specific residues located mainly in the Hsp90-M domain are affected, which cluster in the cleft between the two Hsp90 subunits formed in its closed state. Also, additional residues in the Hsp90-N and Hsp90-C domain are involved. Previous interaction analysis with clients had already indicated the M-domain as an important client binding interface (Hagn et al., 2011; Vaughan et al., 2006). For GR, we found several residues, including E381, E431, E507, and T511 in the Hsp90-M domain (Figure 6D), whose mutation had resulted in decreased GR function in vivo (Bohen and Yamamoto, 1993; Genest et al., 2013; Nathan and Lindquist, 1995). Further, one of the residues with a notable CSP in the M-/C-interface is I592, which had been suggested to be involved in Hsp90-dependent GR activation (Fang et al., 2006). This residue is part of a short helix in the C-terminal domain of Hsp90 (helices 2 and 24 in full-length Hsp90, respectively), close in space to helix 8 in the Hsp90-M domain (helix 19 in full-length Hsp90), which was proposed to be involved in client binding (Harris et al., 2004). Consistent with the observation that all three Hsp90 domains contribute together to the interaction, the affinities of the GRLBDm for individual domains were much lower than for fulllength Hsp90. This is also supported by experiments with the Hsp90-NM construct. The fact that spectral changes are more pronounced and affect larger regions in the Hsp90-NM construct compared to the single domains indicates a more extended binding surface or increased affinity, or may reflect in part allosteric effects. Interestingly, the Hsp90 binding region for GR-LBDm defined here (Figures 6D and 6E) matches closely to the previously reported Aha1-N binding site (Panaretou et al., 2002; Retzlaff

et al., 2010). This may explain why we did not observe Aha1 and GR-LBDm interacting simultaneously with Hsp90 in our studies. Our data also indicate that Sti1 negatively affects GRLBDm binding to Hsp90. The proposed binding site for the Sti1 TPR2A-TPR2B domains (Schmid et al., 2012) is in close proximity to that of the GR-LBDm (Figure 6E) and could thus interfere with GR binding. As one Sti1 molecule is able to inhibit an Hsp90 dimer (Li et al., 2011), it could regulate the number of GR molecules binding to Hsp90. This is in line with our AUC data where the addition of Sti1 to a preformed GR-LBDm-Hsp90 complex resulted in the release of GR-LBDm. Concerning the conformation of the client, we could show that the GR-LBDm interacts with Hsp90 in a folded conformation and that it did not unfold upon interaction with Hsp90-M. This fits the notion that Hsp90 does not bind to nascent proteins (Nathan et al., 1997) but does recognize already folded conformations. The interaction of Hsp90 with structured parts of a protein was observed for model clients (Jakob et al., 1995; Street et al., 2011) and for the DNA-binding domain of p53 (Hagn et al., 2011), although in the latter case there were also reports that binding of unfolded conformations is preferred (Park et al., 2011; Ru¨diger et al., 2002). For the GR-LBDm, and consistent with its folded structure, charge complementarity seems to play a role in the interaction. How a client affects the Hsp90 chaperone cycle has been largely uncharted territory. Since the fluctuations between the different conformational states leading to a closed conformation are rate limiting, their formation is affected by nucleotides and cochaperones (Graf et al., 2009; Ro¨hl et al., 2013). Our results establish that the client plays an important role in this scheme: it influences the conformational equilibrium together with modulating cochaperones, which select certain Hsp90 conformations. Binding of the GR-LBDm to a partially closed conformation of the Hsp90 dimer is preferred, and this complex is stabilized against exchange of Hsp90 monomers compared to Hsp90 alone. These structural traits have profound functional consequences, as they slow down the ATPase reaction and prolong the dwell time of the GR in Hsp90 complexes. For establishing the interaction between Hsp90 and GR-LBDm, ATP binding seems to be important, while ATP hydrolysis, which results in the formation of open Hsp90 (Hsp90-ADP), may trigger GR release. It should be noted that in a previous study on the interaction between Hsp90 and refolded wt GR-LBD, a stimulatory effect on the Hsp90 ATPase was reported (McLaughlin et al., 2002). However, in this case, a detergent was required, and this may affect the conformational dynamics of Hsp90. Apart from that, also for a model client, the ATPase of Hsp90 was reported to be stimulated (Genest et al., 2013; Street et al., 2011). At the moment, it is not clear whether the interactions with model clients may follow a different mechanism compared to the stringent client studied here. Intriguingly, binding of the holo-GR-LBDm to Hsp90 and further recruitment of GR-specific cochaperones contradicts concepts where GR dissociates from Hsp90 upon ligand binding (Pratt and Toft, 1997). However, there is an increasing body of evidence that liganded GR binds Hsp90 in vivo and that those chaperone complexes seem to regulate nuclear import and transcriptional activity of GR (Freeman and Yamamoto, Molecular Cell 53, 941–953, March 20, 2014 ª2014 Elsevier Inc. 949

Molecular Cell The GR-Hsp90 Interplay

Figure 7. Integration of the GR into the Hsp90-Cochaperone Cycle Two scenarios exist for feeding GR into the Hsp90 machinery. In the case of apo-GR, the client assembles first with Hsp70 and Hsp40. After transfer to the Hsp90 chaperone machinery via Hop/Sti1, GR binds hormone. In the next step, Hsp70 release, together with the entry of a PPIase such as Cpr6, accelerates the cycle, resulting in the formation of a partially closed asymmetric holo-GR-Hsp90-PPIase-Hop/Sti1 complex. Then Hop/Sti1 is released by the concerted action of Cpr6 and p23, forming the late complex consisting of Hsp90, PPIase (Cpr6), p23, and holo-GR. Binding of p23 induces further closure of the Hsp90 dimer and reduces the ATPase activity of Hsp90, prolonging the interaction between the client and its chaperone. In the second scenario, holo-GR is bound directly by Hsp90. Whether the conformation of GR bound directly or delivered via Hsp70 differs is an open question at the moment. In the absence of cochaperones, Hsp90 is able to accommodate two GR molecules via two independent binding sites. Binding of p23 or Cpr6 to the Hsp90-holo-GR complex changes GR binding stoichiometry. Hydrolysis of ATP leads to the release of cochaperones and client.

2002; Galigniana et al., 2001). Our study is in agreement with these findings and suggests chaperoning of GR by Hsp90 after hormone binding as an important aspect of SHR signaling. Since there are several events downstream of hormone binding and before activation of transcription, it is tempting to speculate that additional signals are required for release. The picture emerging from this and previous studies on the chaperone cycle for GR is summarized in Figure 7. Based on the current evidence, there seem to be two routes to feed GR into the Hsp90 system. Apo-GR can be delivered to Hsp90 by Hsp70 and its cochaperone Hsp40 with Hop/Sti1 mediating client transfer. In this context, our finding that Hop/Sti1 negatively affects GR-LBDm binding seems counterintuitive at first glance. However, as Hsp90 has to undergo conformational rearrangements from an open conformation (Sti1 bound) to a closed conformation (and subsequent Hsp70 release), Hop/ 950 Molecular Cell 53, 941–953, March 20, 2014 ª2014 Elsevier Inc.

Sti1 has to be displaced during the cycle. The asymmetric Cpr6-Hsp90-Sti1 complex relieves the inhibitory potential of Hop/Sti1 and accelerates the cycle in the presence of GR. This complex plays a key role for the recruitment of p23, which binds exclusively to closed Hsp90. Conceptually, the synergistic and antagonistic effects observed here for different cochaperones reinforce the idea of the asymmetric nature of active Hsp90-cochaperone complexes (Retzlaff et al., 2010). Importantly, binding of cochaperones during the Hsp90 cycle also changes GR-binding stoichiometry. While the binding sites on the two subunits of the Hsp90 dimer seem to be independent in the absence of cochaperones, only one client interacts with Hsp90 in the presence of p23 or Cpr6. Release is achieved as Hsp90 resumes an open conformation. It should be noted that hormone-bound GR can bind directly to Hsp90 and does not require the Hsp70 system. As the concentration of Hsp90

Molecular Cell The GR-Hsp90 Interplay

exceeds that of its cochaperones by a factor of ten in vivo, this scenario is physiologically relevant. EXPERIMENTAL PROCEDURES Strains, buffers, concentrations, suppliers, and additional experimental procedures are provided and described in the Supplemental Experimental Procedures section. Yeast Reporter Assay GR expression in S. cerevisiae and the b-galactosidase-based reporter assay were performed as described previously (Nathan and Lindquist, 1995). Protein Expression and Purification Hsp90, Hsp90 fragments, p23/Sba1, Sti1, and Cpr6 were expressed and purified as described earlier (Buchner et al., 1998; Schmid et al., 2012; Wegele et al., 2003). GFP was a kind gift from T. Dashivets. Human GR-LBD variants were expressed at 18 C overnight and purified according to (Seitz et al., 2010) with minor modifications. For NMR experiments, uniformly labeled samples were isotopically enriched with 13C, 15N, and 2H as described for partially deuterated samples (Marley et al., 2001) and for perdeuterated samples (Rosen et al., 1996). Segmental labeling was performed by expressed protein ligation using a variant of Sortase A (Chen et al., 2011). ATPase Measurements ATPase assays were performed at 30 C utilizing an ATP-regenerating system as described elsewhere (Richter et al., 2003). Background ATPase activity was determined by the addition of Radicicol. Fluorescence Measurements Steady-state hormone binding affinity and kinetics were determined by fluorescence anisotropy using F-DEX. Measurements were performed at 25 C in a Spex Fluoromax 3 supplied with polarizers. Excitation and emission wavelengths were set to 484 nm and 520 nm, respectively. FRET experiments of Hsp90, labeled at C61 with donor and acceptor dye, were carried out as described previously (Hessling et al., 2009). For chase experiments, unmodified yeast Hsp90 or excess ADP was added.

microscope, operated at 100 kV on Kodak SO-163 film at 60.000 nominal magnification, and digitized in a Zeiss SCAI scanner to a final sampling resolution of 4.66 A˚/pixel. A total of 20,016 particles were selected. For data analysis and image processing, see Supplemental Information. SUPPLEMENTAL INFORMATION Supplemental Information includes six figures, two tables, and Supplemental Experimental Procedures and can be found with this article online at http:// dx.doi.org/10.1016/j.molcel.2014.02.003. AUTHOR CONTRIBUTIONS O.R.L. and D.A.R. performed biochemical experiments, and M.K. did the AUC experiments. B.K.Z. helped with FRET experiments. L.F. performed NMR experiments. T.M. performed SAXS analysis and structural modeling. EM reconstitution and data analysis were done by J.C., S.A., and J.M.V. O.R.L., M.S., T.M., and J.B. designed experiments and participated in data analysis. O.R.L., D.A.R., L.F., M.S., T.M., J.M.V., and J.B. wrote the manuscript. ACKNOWLEDGMENTS We thank A. Liebscher for practical assistance and E. Boczek, A. Rehn, and M. Rosam for discussions and reading the manuscript. We acknowledge C. Go¨bl and C. Hartlmu¨ller for help with the SAXS measurements. This work was supported by The Deutsche Forschungsgemeinschaft (grant SFB 1035 A3 to J.B. and M.S.; Emmy Noether program MA 5703/1-1, to T.M), the Bavarian Ministry of Sciences, Research and the Arts (Bavarian Molecular Biosystems Research Network, to T.M.), the Austrian Academy of Sciences (APART-fellowship, to T.M.), the Center for Integrated Protein Science Munich (CIPSM), and the Deutsche Forschungsgemeinschaft (Emmy Noether program MA 5703/1-1, to T.M.). L.F. is supported by an EMBO Longterm Fellowship and Marie Curie FP7 International Incoming Fellowship. D.R. is supported by a fellowship of the ‘‘Fonds der Chemischen Industrie.’’ J.M.V is supported by the Spanish Ministry of Innovation grant BFU2010-15703. The authors acknowledge the Leibniz Supercomputing Centre (LRZ; www.lrz.de) for providing computing time on the Linux Cluster.

Analytical Ultracentrifugation Sedimentation analysis of GR-LBDm was performed in an Optima-XL-I Beckman centrifuge with absorbance optics. Hsp90-interaction studies were performed as described previously (Li et al., 2013). Data analysis was performed with dcdt+v. 2.4.0 (Philo, 2006), SEDFIT v. 14.1 (Schuck, 2000), and OriginPro 8. Solvent density and viscosity were calculated with SEDNTERP (Laue et al., 1992).

Received: September 3, 2013 Revised: December 19, 2013 Accepted: January 28, 2014 Published: March 6, 2014

CD Spectroscopy Experiments were performed in a Jasco J715 CD Spectrometer with a PTC348 WI Peltier unit. FAR-UV-CD spectra were recorded at 10 C and 85 C. Thermal transitions were monitored at 220 nm from 10 C to 85 C.

Ali, M.M., Roe, S.M., Vaughan, C.K., Meyer, P., Panaretou, B., Piper, P.W., Prodromou, C., and Pearl, L.H. (2006). Crystal structure of an Hsp90-nucleotide-p23/Sba1 closed chaperone complex. Nature 440, 1013–1017.

NMR Spectroscopy NMR spectra were recorded on a Bruker AV900 spectrometer at 25 C. NMR experiments where processed using NMRPipe (Delaglio et al., 1995), and the data were analyzed using CcpNMR Analysis (Vranken et al., 2005). For PRE experiments, GR-LBDm was spin labeled using 3-MaleimidoPROXYL at the surface-exposed cysteine residue 638 as described in the Supplemental Information. SAXS Measurements SAXS data were recorded on an in-house Anton Paar SAXSess mc2 SAXS instrument. Data analysis, modeling, and further details are provided in the supplement. EM The Hsp90:GR-LBD complex was assembled and purified using GraFix (Kastner et al., 2008). Images were recorded in a JEOL 1200EX-II electron

REFERENCES

Bledsoe, R.K., Montana, V.G., Stanley, T.B., Delves, C.J., Apolito, C.J., McKee, D.D., Consler, T.G., Parks, D.J., Stewart, E.L., Willson, T.M., et al. (2002). Crystal structure of the glucocorticoid receptor ligand binding domain reveals a novel mode of receptor dimerization and coactivator recognition. Cell 110, 93–105. Bohen, S.P., and Yamamoto, K.R. (1993). Isolation of Hsp90 mutants by screening for decreased steroid receptor function. Proc. Natl. Acad. Sci. USA 90, 11424–11428. Buchner, J., Weikl, T., Bu¨gl, H., Pirkl, F., and Bose, S. (1998). Purification of Hsp90 partner proteins Hop/p60, p23, and FKBP52. Methods Enzymol. 290, 418–429. Cadepond, F., Binart, N., Chambraud, B., Jibard, N., Schweizer-Groyer, G., Segard-Maurel, I., and Baulieu, E.E. (1993). Interaction of glucocorticosteroid receptor and wild-type or mutated 90-kDa heat shock protein coexpressed in baculovirus-infected Sf9 cells. Proc. Natl. Acad. Sci. USA 90, 10434–10438. Chang, H.C., Nathan, D.F., and Lindquist, S. (1997). In vivo analysis of the Hsp90 cochaperone Sti1 (p60). Mol. Cell. Biol. 17, 318–325.

Molecular Cell 53, 941–953, March 20, 2014 ª2014 Elsevier Inc. 951

Molecular Cell The GR-Hsp90 Interplay

Chen, S., and Smith, D.F. (1998). Hop as an adaptor in the heat shock protein 70 (Hsp70) and hsp90 chaperone machinery. J. Biol. Chem. 273, 35194–35200. Chen, I., Dorr, B.M., and Liu, D.R. (2011). A general strategy for the evolution of bond-forming enzymes using yeast display. Proc. Natl. Acad. Sci. USA 108, 11399–11404. Dehner, A., Furrer, J., Richter, K., Schuster, I., Buchner, J., and Kessler, H. (2003). NMR chemical shift perturbation study of the N-terminal domain of Hsp90 upon binding of ADP, AMP-PNP, geldanamycin, and radicicol. ChemBioChem 4, 870–877. Delaglio, F., Grzesiek, S., Vuister, G.W., Zhu, G., Pfeifer, J., and Bax, A. (1995). NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J. Biomol. NMR 6, 277–293. Dittmar, K.D., Demady, D.R., Stancato, L.F., Krishna, P., and Pratt, W.B. (1997). Folding of the glucocorticoid receptor by the heat shock protein (hsp) 90-based chaperone machinery. The role of p23 is to stabilize receptor.hsp90 heterocomplexes formed by hsp90.p60.hsp70. J. Biol. Chem. 272, 21213– 21220. Fang, L., Ricketson, D., Getubig, L., and Darimont, B. (2006). Unliganded and hormone-bound glucocorticoid receptors interact with distinct hydrophobic sites in the Hsp90 C-terminal domain. Proc. Natl. Acad. Sci. USA 103, 18487–18492. Freeman, B.C., and Yamamoto, K.R. (2002). Disassembly of transcriptional regulatory complexes by molecular chaperones. Science 296, 2232–2235. Freeman, B.C., Felts, S.J., Toft, D.O., and Yamamoto, K.R. (2000). The p23 molecular chaperones act at a late step in intracellular receptor action to differentially affect ligand efficacies. Genes Dev. 14, 422–434. Galigniana, M.D., Radanyi, C., Renoir, J.M., Housley, P.R., and Pratt, W.B. (2001). Evidence that the peptidylprolyl isomerase domain of the hsp90-binding immunophilin FKBP52 is involved in both dynein interaction and glucocorticoid receptor movement to the nucleus. J. Biol. Chem. 276, 14884–14889. Genest, O., Reidy, M., Street, T.O., Hoskins, J.R., Camberg, J.L., Agard, D.A., Masison, D.C., and Wickner, S. (2013). Uncovering a region of heat shock protein 90 important for client binding in E. coli and chaperone function in yeast. Mol. Cell 49, 464–473. Graf, C., Stankiewicz, M., Kramer, G., and Mayer, M.P. (2009). Spatially and kinetically resolved changes in the conformational dynamics of the Hsp90 chaperone machine. EMBO J. 28, 602–613. Hagn, F., Lagleder, S., Retzlaff, M., Rohrberg, J., Demmer, O., Richter, K., Buchner, J., and Kessler, H. (2011). Structural analysis of the interaction between Hsp90 and the tumor suppressor protein p53. Nat. Struct. Mol. Biol. 18, 1086–1093.

Jakob, U., Lilie, H., Meyer, I., and Buchner, J. (1995). Transient interaction of Hsp90 with early unfolding intermediates of citrate synthase. Implications for heat shock in vivo. J. Biol. Chem. 270, 7288–7294. Kastner, B., Fischer, N., Golas, M.M., Sander, B., Dube, P., Boehringer, D., Hartmuth, K., Deckert, J., Hauer, F., Wolf, E., et al. (2008). GraFix: sample preparation for single-particle electron cryomicroscopy. Nat. Methods 5, 53–55. Krukenberg, K.A., Street, T.O., Lavery, L.A., and Agard, D.A. (2011). Conformational dynamics of the molecular chaperone Hsp90. Q. Rev. Biophys. 44, 229–255. Kumar, R., and Thompson, E.B. (1999). The structure of the nuclear hormone receptors. Steroids 64, 310–319. Laue, T.M., Shah, B.D., Ridgeway, T.M., and Pelletier, S.L. (1992). Computeraided interpretation of analytical sedimentation data for proteins. In Analytical Ultracentrifugation in Biochemistry and Polymer Science), pp. 90–125. Lee, C.-T., Graf, C., Mayer, F.J., Richter, S.M., and Mayer, M.P. (2012). Dynamics of the regulation of Hsp90 by the co-chaperone Sti1. EMBO J. 31, 1518–1528. Li, J., Richter, K., and Buchner, J. (2011). Mixed Hsp90-cochaperone complexes are important for the progression of the reaction cycle. Nat. Struct. Mol. Biol. 18, 61–66. Li, J., Soroka, J., and Buchner, J. (2012). The Hsp90 chaperone machinery: conformational dynamics and regulation by co-chaperones. Biochim. Biophys. Acta 1823, 624–635. Li, J., Richter, K., Reinstein, J., and Buchner, J. (2013). Integration of the accelerator Aha1 in the Hsp90 co-chaperone cycle. Nat. Struct. Mol. Biol. 20, 326–331. Marley, J., Lu, M., and Bracken, C. (2001). A method for efficient isotopic labeling of recombinant proteins. J. Biomol. NMR 20, 71–75. McClellan, A.J., Scott, M.D., and Frydman, J. (2005). Folding and quality control of the VHL tumor suppressor proceed through distinct chaperone pathways. Cell 121, 739–748. McClellan, A.J., Xia, Y., Deutschbauer, A.M., Davis, R.W., Gerstein, M., and Frydman, J. (2007). Diverse cellular functions of the Hsp90 molecular chaperone uncovered using systems approaches. Cell 131, 121–135. McLaughlin, S.H., Smith, H.W., and Jackson, S.E. (2002). Stimulation of the weak ATPase activity of human hsp90 by a client protein. J. Mol. Biol. 315, 787–798. Mickler, M., Hessling, M., Ratzke, C., Buchner, J., and Hugel, T. (2009). The large conformational changes of Hsp90 are only weakly coupled to ATP hydrolysis. Nat. Struct. Mol. Biol. 16, 281–286.

Harris, S.F., Shiau, A.K., and Agard, D.A. (2004). The crystal structure of the carboxy-terminal dimerization domain of htpG, the Escherichia coli Hsp90, reveals a potential substrate binding site. Structure 12, 1087–1097.

Nathan, D.F., and Lindquist, S. (1995). Mutational analysis of Hsp90 function: interactions with a steroid receptor and a protein kinase. Mol. Cell. Biol. 15, 3917–3925.

Hawle, P., Siepmann, M., Harst, A., Siderius, M., Reusch, H.P., and Obermann, W.M. (2006). The middle domain of Hsp90 acts as a discriminator between different types of client proteins. Mol. Cell. Biol. 26, 8385–8395.

Nathan, D.F., Vos, M.H., and Lindquist, S. (1997). In vivo functions of the Saccharomyces cerevisiae Hsp90 chaperone. Proc. Natl. Acad. Sci. USA 94, 12949–12956.

Heitzer, M.D., Wolf, I.M., Sanchez, E.R., Witchel, S.F., and DeFranco, D.B. (2007). Glucocorticoid receptor physiology. Rev. Endocr. Metab. Disord. 8, 321–330.

Panaretou, B., Siligardi, G., Meyer, P., Maloney, A., Sullivan, J.K., Singh, S., Millson, S.H., Clarke, P.A., Naaby-Hansen, S., Stein, R., et al. (2002). Activation of the ATPase activity of hsp90 by the stress-regulated cochaperone aha1. Mol. Cell 10, 1307–1318.

Herna´ndez, M.P., Chadli, A., and Toft, D.O. (2002). HSP40 binding is the first step in the HSP90 chaperoning pathway for the progesterone receptor. J. Biol. Chem. 277, 11873–11881. Hessling, M., Richter, K., and Buchner, J. (2009). Dissection of the ATPinduced conformational cycle of the molecular chaperone Hsp90. Nat. Struct. Mol. Biol. 16, 287–293. Howard, K.J., Holley, S.J., Yamamoto, K.R., and Distelhorst, C.W. (1990). Mapping the HSP90 binding region of the glucocorticoid receptor. J. Biol. Chem. 265, 11928–11935. Jackson, S. (2013). Hsp90: Structure and Function. In Molecular ChaperonesMolecular Chaperones (Heidelberg: Springer Berlin Heidelberg), pp. 155–240.

952 Molecular Cell 53, 941–953, March 20, 2014 ª2014 Elsevier Inc.

Park, S.J., Borin, B.N., Martinez-Yamout, M.A., and Dyson, H.J. (2011). The client protein p53 adopts a molten globule-like state in the presence of Hsp90. Nat. Struct. Mol. Biol. 18, 537–541. Philo, J.S. (2006). Improved methods for fitting sedimentation coefficient distributions derived by time-derivative techniques. Anal. Biochem. 354, 238–246. Picard, D. (2002). Heat-shock protein 90, a chaperone for folding and regulation. Cell. Mol. Life Sci. 59, 1640–1648. Picard, D., Khursheed, B., Garabedian, M.J., Fortin, M.G., Lindquist, S., and Yamamoto, K.R. (1990). Reduced levels of hsp90 compromise steroid receptor action in vivo. Nature 348, 166–168.

Molecular Cell The GR-Hsp90 Interplay

Pratt, W.B., and Dittmar, K.D. (1998). Studies with Purified Chaperones Advance the Understanding of the Mechanism of Glucocorticoid Receptorhsp90 Heterocomplex Assembly. Trends Endocrinol. Metab. 9, 244–252. Pratt, W.B., and Toft, D.O. (1997). Steroid receptor interactions with heat shock protein and immunophilin chaperones. Endocr. Rev. 18, 306–360. Pullen, L., and Bolon, D.N. (2011). Enforced N-domain proximity stimulates Hsp90 ATPase activity and is compatible with function in vivo. J. Biol. Chem. 286, 11091–11098. Retzlaff, M., Hagn, F., Mitschke, L., Hessling, M., Gugel, F., Kessler, H., Richter, K., and Buchner, J. (2010). Asymmetric activation of the hsp90 dimer by its cochaperone aha1. Mol. Cell 37, 344–354. Richter, K., Muschler, P., Hainzl, O., Reinstein, J., and Buchner, J. (2003). Sti1 is a non-competitive inhibitor of the Hsp90 ATPase. Binding prevents the N-terminal dimerization reaction during the atpase cycle. J. Biol. Chem. 278, 10328–10333. Richter, K., Walter, S., and Buchner, J. (2004). The Co-chaperone Sba1 connects the ATPase reaction of Hsp90 to the progression of the chaperone cycle. J. Mol. Biol. 342, 1403–1413.

Seitz, T., Thoma, R., Schoch, G.A., Stihle, M., Benz, J., D’Arcy, B., Wiget, A., Ruf, A., Hennig, M., and Sterner, R. (2010). Enhancing the stability and solubility of the glucocorticoid receptor ligand-binding domain by high-throughput library screening. J. Mol. Biol. 403, 562–577. Siligardi, G., Hu, B., Panaretou, B., Piper, P.W., Pearl, L.H., and Prodromou, C. (2004). Co-chaperone regulation of conformational switching in the Hsp90 ATPase cycle. J. Biol. Chem. 279, 51989–51998. Smith, D.F. (1993). Dynamics of heat shock protein 90-progesterone receptor binding and the disactivation loop model for steroid receptor complexes. Mol. Endocrinol. 7, 1418–1429. Southworth, D.R., and Agard, D.A. (2008). Species-dependent ensembles of conserved conformational states define the Hsp90 chaperone ATPase cycle. Mol. Cell 32, 631–640. Street, T.O., Lavery, L.A., and Agard, D.A. (2011). Substrate binding drives large-scale conformational changes in the Hsp90 molecular chaperone. Mol. Cell 42, 96–105.

Ro¨hl, A., Rohrberg, J., and Buchner, J. (2013). The chaperone Hsp90: changing partners for demanding clients. Trends Biochem. Sci. 38, 253–262.

Taipale, M., Krykbaeva, I., Koeva, M., Kayatekin, C., Westover, K.D., Karras, G.I., and Lindquist, S. (2012). Quantitative analysis of HSP90client interactions reveals principles of substrate recognition. Cell 150, 987–1001.

Rosen, M.K., Gardner, K.H., Willis, R.C., Parris, W.E., Pawson, T., and Kay, L.E. (1996). Selective methyl group protonation of perdeuterated proteins. J. Mol. Biol. 263, 627–636.

Vaughan, C.K., Gohlke, U., Sobott, F., Good, V.M., Ali, M.M., Prodromou, C., Robinson, C.V., Saibil, H.R., and Pearl, L.H. (2006). Structure of an Hsp90Cdc37-Cdk4 complex. Mol. Cell 23, 697–707.

Ru¨diger, S., Freund, S.M., Veprintsev, D.B., and Fersht, A.R. (2002). CRINEPTTROSY NMR reveals p53 core domain bound in an unfolded form to the chaperone Hsp90. Proc. Natl. Acad. Sci. USA 99, 11085–11090.

Vranken, W.F., Boucher, W., Stevens, T.J., Fogh, R.H., Pajon, A., Llinas, M., Ulrich, E.L., Markley, J.L., Ionides, J., and Laue, E.D. (2005). The CCPN data model for NMR spectroscopy: development of a software pipeline. Proteins 59, 687–696.

Schmid, A.B., Lagleder, S., Gra¨wert, M.A., Ro¨hl, A., Hagn, F., Wandinger, S.K., Cox, M.B., Demmer, O., Richter, K., Groll, M., et al. (2012). The architecture of functional modules in the Hsp90 co-chaperone Sti1/Hop. EMBO J. 31, 1506–1517. Schuck, P. (2000). Size-distribution analysis of macromolecules by sedimentation velocity ultracentrifugation and lamm equation modeling. Biophys. J. 78, 1606–1619.

Wegele, H., Muschler, P., Bunck, M., Reinstein, J., and Buchner, J. (2003). Dissection of the contribution of individual domains to the ATPase mechanism of Hsp90. J. Biol. Chem. 278, 39303–39310. Weikl, T., Muschler, P., Richter, K., Veit, T., Reinstein, J., and Buchner, J. (2000). C-terminal regions of Hsp90 are important for trapping the nucleotide during the ATPase cycle. J. Mol. Biol. 303, 583–592.

Molecular Cell 53, 941–953, March 20, 2014 ª2014 Elsevier Inc. 953