GR chaperone cycle mechanism revealed by cryo-EM: the GR-maturation complex


 Hsp90 is a conserved and essential molecular chaperone responsible for the folding and activation of hundreds of ‘client’ proteins. The glucocorticoid receptor (GR) is a model client that constantly depends on Hsp90 for activity. Previously, we revealed GR ligand binding is inhibited by Hsp70 and restored by Hsp90, aided by the cochaperone p23. However, a molecular understanding of the chaperone-induced transformations that occur between the inactive Hsp70:Hsp90 ‘client-loading complex’ and an activated Hsp90:p23 ‘client-maturation complex’ is lacking for GR, or for any client. Here, we present a 2.56Å cryo-EM structure of the GR-maturation complex (GR:Hsp90:p23), revealing that the GR ligand binding domain is, surprisingly, restored to a folded, ligand-bound conformation, while simultaneously threaded through the Hsp90 lumen. Also, unexpectedly, p23 directly stabilizes native GR using a previously uncharacterized C-terminal helix, resulting in enhanced ligand-binding. This is the highest resolution Hsp90 structure to date and the first atomic resolution structure of a client bound to Hsp90 in a native conformation, sharply contrasting with the unfolded kinase:Hsp90 structure. Thus, aided by direct cochaperone:client interactions, Hsp90 dictates client-specific folding outcomes. Together with the GR-loading complex structure (Wang et al. 2020), we present the molecular mechanism of chaperone-mediated GR remodeling, establishing the first complete chaperone cycle for any client.


Introduction 27
Hsp90 is required for the functional maturation of 10% of the eukaryotic proteome, 28 including signaling proteins, such as kinases, and steroid hormone receptors (SHRs), such as 29 GR 1,3,12 . We previously uncovered the biochemical basis for GR's Hsp90 dependence using in 30 vitro reconstitution starting with an active GR ligand binding domain (hereafter GR, for 31 simplicity) 10 . We demonstrated that GR ligand binding is regulated by a cycle of GR:chaperone 32 complexes (Fig. 3d). In this chaperone cycle, GR is first inhibited by Hsp70, then loaded onto 33 Hsp90:cochaperone Hop (Hsp70/Hsp90 organizing protein) forming an inactive 34 GR:Hsp90:Hsp70:Hop loading complex (Wang et al. 2020). Upon ATP hydrolysis on Hsp90,35 Hsp70 and Hop are released, and p23 is incorporated to form an active GR:Hsp90:p23 36 maturation complex, restoring GR ligand binding with enhanced affinity. Progression through 37 this cycle is coordinated by the ATPase activities of both Hsp70 and Hsp90, which dictate large 38 conformational rearrangements 2,13 . Particularly, Hsp90 functions as a constitutive dimer that 39 undergoes an open-to-closed transition upon ATP binding and this conformational cycle is 40 further regulated by cochaperones 14 . The cochaperone p23 specifically binds and stabilizes the 41 closed Hsp90 conformation 15 and p23 is required for full reactivation of GR ligand binding in 42 vitro 10 and proper function in vivo 16 . Altogether, the coordinated actions of Hsp70, Hsp90, and 43 cochaperones remodel the conformation of GR to control access to the buried, hydrophobic 44 ligand binding pocket. 45 folded conformation, as well as the highest resolution structure of full-length Hsp90 to date. In 161 the maturation complex, GR simultaneously threads through the closed Hsp90 lumen and adopts 162 a native, ligand-bound conformation that is extensively stabilized by both Hsp90 and the p23tail-163 helix. No GR apo complexes were identified during image analysis, suggesting GR apo is either too 164 dynamic or quickly released from the complex. The native GR conformation in the maturation 165 complex is in striking contrast with the loading complex, in which GR is partially unfolded and 166 unable to bind ligand (Wang et al. 2020). In both complexes, GR threads through the Hsp90 167 lumen and also interacts on the surface with Hsp90 F349,W320 , although different GR segments are 168 involved. Supporting a general role in client recognition, Hsp90 W320 is critical for client 169 activation in vivo [22][23][24] . The active, native GR in our complex also starkly contrasts with the only 170 other structure of a closed Hsp90:client complex, which stabilizes an unfolded kinase client 11 . 171 The Hsp90 conformation is nearly identical in both structures and both clients are threaded 172 through the Hsp90 lumen in a similar manner, suggesting a universal binding mode for Hsp90 173 clients (Extended Data Fig. 11a,b). Although the overall Hsp90:client interactions are similar, 174 the outcomes for folding and function of these two clients are opposing, demonstrating 175 evolutionarily determined, client-specific conformational remodeling by Hsp90 (Extended Data 176 Fig. 11c). 177 While previously thought to be a general cochaperone whose primary function is to 178 stabilize a closed Hsp90, our structure reveals that p23 also makes extensive contacts with GR 179 through a previously uncharacterized helix in the p23 tail. This p23tail-helix is necessary for the 180 observed enhanced GR ligand binding activity in vitro and may act by stabilizing ligand-bound 181 GR, securing GR within the complex to indirectly stabilize GRHelix1, and/or by allosterically positioning the dynamic GRHelix12. Thus, p23 not only serves as a cochaperone to stabilize the 183 closure of Hsp90, but also directly contributes to client maturation. In support of this essential 184 p23:GR interaction, the p23tail-helix and GR hydrophobic groove are well conserved. In fact, the 185 hydrophobic groove is conserved across SHRs, indicating the p23tail-helix may contribute to the 186 Hsp90-dependent chaperoning of all SHRs. Indeed, the activity of all SHRs is dependent on 187 p23 16 and the progesterone receptor (PR) requires the p23 tail for enhanced ligand binding 188 activity 25 . Intriguingly, NCoA3 contains a p23tail-helix -like motif, suggesting other GR 189 coregulators may utilize this novel helix motif to bind the hydrophobic groove on GR and 190 compete with p23, potentially facilitating GR release. Surprisingly, the p23tail-helix is conserved 191 among eukaryotes that lack SHRs; however, previous studies have demonstrated that the p23 tail 192 has general chaperoning activities 19,25 . To this point, we observed p23 alone could modestly 193 enhance the ligand binding activity of GR independent of Hsp90. Along with the discovery that 194 the Hop cochaperone interacts with the client in the loading complex (Wang et al. 2020), these 195 findings support an emerging paradigm in which Hsp90 cochaperones make specific, direct 196 contact with Hsp90 clients to aid in client recognition and function 11 . 197 Together with the structure of the GR-loading complex (Wang et al. 2020), we provide 198 for the first time, a complete picture of the chaperone cycle for any client (Fig. 3d). These two 199 structures reveal that GR transitions from a partially unfolded conformation in the loading 200 complex to an active, folded conformation in the maturation complex. In the loading complex, 201 GRpre-Helix1 is captured by Hsp70, GRHelix1 is stabilized by Hop, and GRpost-Helix1 is threaded 202 through the semi-closed Hsp90 lumen. First, Hsp70 releases and then Hop releases, which lets 203 GRpre-Helix1 slide into the Hsp90 lumen, allowing GRHelix1 to refold onto the GR core, thereby 204 generating a ligand binding capable, native GR, stabilized by the p23tail-helix (Fig. 3b). During 205 this transition, Hsp90 twists to the fully closed conformation, which likely facilitates client 206 sliding and helps rearrange the client binding site to fully enclose GRpre-Helix1 (Fig. 3a). Perhaps 207 most critical for GR function, GR becomes protected from Hsp70 rebinding and inhibition once 208 it is in the maturation complex. GRHelix1 has been proposed to function as a lid over the GR 209 ligand binding pocket 26 , thus ligand likely binds during the transition from the loading complex 210 to the maturation complex, just as GRHelix1 slides through the Hsp90 lumen to seal the ligand 211 binding pocket (Fig. 3c). In line with previous studies, our findings suggest that ligand-bound 212 GR may be translocated to the nucleus in the maturation complex 27,28 , perhaps with the aid of 213 FKBP52 29 , where it would be protected by Hsp90 from re-inhibition by Hsp70. Supporting this, 214 Hsp90 and p23 have been found in the nucleus colocalized with GR 30 , suggesting the maturation 215 complex may persist even after translocation. 216 The proposed sliding mechanism may be a general theme for Hsp90's client remodeling 217 mechanism. Our two other reconstructions, Hsp90:p23 and MBP:Hsp90:p23, have density in the 218 Hsp90 lumen, suggesting Hsp90 has bound regions in our construct other than GR. Given that 219 GRHelix1 slides through the Hsp90 lumen from the loading complex to the maturation complex, it 220 is possible that Hsp90 can act processively during open-to-closed transitions to remodel other 221 client domains beyond the one initially engaged. This would explain these other structural 222 classes, although this remains to be tested. Nevertheless, the mechanism of GRHelix1 sliding 223 explains how Hsp90 can provide protected refolding of client domains as they exit the lumen to 224 become directly stabilized by cochaperones. Our results suggest Hsp90 may use this mechanism 225 to allow domains to fold independently on either side of the lumen or uncouple annealed 226 misfolded regions to ensure folding fidelity. 227 yeast Ydj1 were expressed in the pET151 bacterial expression plasmid with a cleavable N-324 terminal, 6x-His tag. Human Bag-1 isoform 4 (116-345) was expressed in a pET28a vector with 325 a cleavable N-terminal, 6x-His tag. Proteins were expressed and purified by the following 326 procedure. Proteins were expressed in bacterial BL21 star (DE3) strain. Cells were grown in 327 either LB or TB at 37C until OD600 reached 0.6-0.8 and then induced with 0.5 mM IPTG 328 overnight at 16C. Cells were harvested and lysed in 50 mM Potassium Phosphate pH 8, 500 329 mM KCl, 10 mM imidazole pH 8, 10% glycerol, 6 mM ME, and Roche cOmplete, mini 330 protease inhibitor cocktail using an EmulsiFlex-C3 (Avestin). Lysate was centrifuged and the 331 soluble fraction was affinity purified by gravity column with Ni-NTA affinity resin (QIAGEN). 332

Main
The protein was eluted with 30 mM Tris pH8, 50 mM KCl, 250 mM imidazole pH 8, and 6 mM 333 ME. For Hsp90, Hsp70, and Ydj1, an extra wash step with 0.1% Tween20 and 2 mM 334 ATP/MgCl2 was added to the Ni-NTA resin before eluting. The 6x-His tag was removed with 335 TEV protease during the following overnight dialysis in 30 mM Tris pH 8, 50 mM KCl, and 6 336 mM ME.  Fig.1a). The elution was concentrated and purified by size 363 exclusion using a Shodex KW-804 on an Ettan LC (GE Healthcare)(Extended Data Fig.1b,c). 364 Fractions containing the full complex were concentrated to ~2 M. 2.5 L of sample was applied 365 to glow-discharged QUANTIFOIL R1.2/1.3, 400-mesh, copper holey carbon grid (Quantifoil 366 Micro Tools GmbH) and plunge-frozen in liquid ethane using a Vitrobot Mark IV (FEI) with a 367 blotting time of 15 seconds, at 10C, and with 100% humidity. 368 369

Cryo-EM data acquisition 370
The images were collected on a FEI Titan Krios electron microscope (Thermo Fisher 371 Scientific) operating at 300kV using a K3 direct electron camera (Gatan) and equipped with a 372 Bioquantum energy filter (Gatan) set to a slit width of 20 eV (example micrograph Extended 373 Data Fig. 1d). Images were recorded at a nominal magnification of 105,000, corresponding to a 374 physical pixel size of 0.835Å. A nominal defocus range of 0.8 m -2.0 m underfocus was used. 375 A total exposure of 5.9 seconds was used with 0.05 second subframes (117 total frames). The 376 total accumulated electron dose was 60 electrons/Å 2 and 0.5128 electrons/Å 2 /frame. Data was 377 acquired using SerialEM software v.3.8-beta 6 . 378 A small dataset on the GR-maturation complex was collected before the larger dataset 379 described above. The smaller dataset was collected from the same GR-maturation complex 380 sample preparation concentrated to 1.2 M with grids prepared in a similar manner. Images were 381 collected on a FEI Titan Krios electron microscope (Thermo Fisher Scientific) operating at 382 300kV using a K3 direct electron camera (Gatan). Images were recorded at a nominal 383 magnification of 105,000, corresponding to a physical pixel size of 0.835Å. A nominal defocus 384 range of 0.8 m -2.0 m underfocus was used. A total exposure of 3.0 seconds was used with 385 0.0255 second subframes (118 total frames). Data was acquired using SerialEM software. 386 Cryo-EM data processing 387 The smaller dataset consisted of ~1500 dose-fractionated image stacks, which were 388 motion corrected using UCSF MotionCor2 7 and analyzed with RELION v.3.0.8 8 . Motion 389 corrected images without dose weighting were used for contrast transfer function (CTF) 390 estimation using CTFFIND v.4.1 9 and template-based particle picking was done with 391 Gautomatch v.0.53 (http://www.mrc-lmb.cam.ac.uk/kzhang/) with the Hsp90:p23 crystal 392 structure (2CG9) as a reference to select a total of 718,080 particles. Multiple rounds of 3D 393 classification were performed with 2CG9 as a low pass filtered (40 Å) initial model until a 394 medium-resolution (~8 Å) GR:Hsp90:p23 reconstruction was obtained from 13,570 particles. 395 This reconstruction was used as a reference for the larger dataset. 396 The larger dataset consisted of 5,608 dose-fractionated image stacks, which were motion 397 corrected using UCSF MotionCor2 and analyzed with RELION v.3.0.8. Motion corrected 398 images with dose weighting were used for contrast transfer function (CTF) estimation using 399 CTFFIND v.4.1 and reference-free particle picking was done with RELION v.3.0.8 Laplacian-400 of-Gaussian auto-picking to select a total of 6,062,152 particles. The processing scheme is 401 depicted in Extended Data Fig. 2a. An initial round of three-dimensional (3D) classification 402 was performed without symmetry using a reference model from a previously collected smaller 403 dataset (see above). The class with clearly recognizable Hsp90 density was used for a second 404 round of 3D classification. In this second round, a class with only Hsp90:p23 density was 405 obtained (454,385 particles). This class was refined after per-particle CTF and beam-tilt 406 correction in RELION to a nominal resolution of 2.66 Å. Particles from two other classes, which contained GR density, were combined (~1 million particles) for a third round of 3D 408 classification. After the third round of 3D classification, particles from classes with the best GR 409 density were then combined (~340,000 particles) and refined. To improve the resolution of GR 410 and the p23 tail helix, these regions were further refined using focused classification with a mask 411 including GR and the p23 tail helix. The best focused classes were combined (140,217 particles) 412 and refined to a nominal resolution of 2.56 Å. Using the 2.56 Å reconstruction, per-particle CTF 413 and beam-tilt were refined using RELION. Although the FSC showed slightly improvement over 414 the pre-refined reconstruction at medium resolution range (5-10 Å), the nominal resolution at 415 0.143 FSC remained unchanged. Nevertheless, we used the CTF/beam-tilt refined particles for 416 the following focused refinement on GR:p23 tail helix and for the resulting reconstructions used 417 for model building. To further improve the resolution of GR and the p23 tail helix for model 418 building, these regions were refined using focused refinement with a mask including GR and the 419 p23 tail helix. 420 From the third round of 3D classification, particles from 3D classes with MBP density 421 were combined (~650,000 particles) and refined. To improve the resolution of MBP, the MBP 422 region was further refined using focused classification with a mask on MBP. The best focused 423 3D classes were combined (31,556 particles) and refined to a nominal resolution of 3.63 Å after 424 per-particle CTF and beam-tilt correction in RELION. 425 All final reconstructions were post-processed in RELION in which the nominal 426 resolution was determined by the gold standard Fourier shell correlation (FSC) using the 0.143 427 criterion (Extended Data Fig. 2b). Maps were sharpened and filtered automatically determined 428 by RELION according to an estimated overall map B-factor and filtered to their estimated 429 resolution. RELION was used to estimate the local resolution of each map (Extended Data Fig. 2a). For GR:Hsp90:p23, a composite map was generated by combining the overall refinement 431 map with the GR:p23 tail focused refinement map using vop maximum in Chimera. 432 433

Model building and refinement 434
For the GR-maturation complex atomic model, the dexamethasone-bound human GR 435 crystal structure (1M2Z) and the human p23 crystal structure (1EJF) were used as a starting 436 model for model building. A homolog model of human Hsp90α was derived from human 437 Hsp90 from the Hsp90:Cdk4:Cdc37 cryo-EM structure (5FWK) with the sequence alignment 438 (86% sequence identity) obtained from HHpred server 10 and this was also used as a starting 439 model for model building (Extended Data Table 1). Models were refined using Rosetta v.3.11 440 throughout. Following the split map approach 11 to prevent and monitor overfitting, the Rosetta 441 iterative backbone rebuilding procedure was used to refine models against one of the half maps 442 obtained from RELION, with the other half map only used for validations. The structurally 443 uncharacterized p23tail-helix was first de novo built into the density using RosettaCM 12 and then 444 was further refined using the same Rosetta iterative backbone rebuilding procedure. With a 445 proper density weight obtained using the half maps, the final model of the GR:Hsp90:p23 446 complex was refined against the full reconstruction allowing only sidechain and small-scale 447 backbone refinement. The final refinement statistics are provided (Extended Data Table 1). For 448 the Hsp90:p23 and MBP:Hsp90:p23 map densities (Extended Data Fig. 9a and Extended Data 449  For Fig. 2d, the ConSurf server 17,18 was used to select and align 87 GR sequences. The human 486 GR crystal structure 19 (4P6X) was used to select sequences from UNIREF90 with maximal 487 percent ID at 95% and minimal percent ID at 65%. Conservation scores were calculated and 488 provided by the server. The conservation scores calculated by ConSurf were mapped onto GR 489 from the maturation complex atomic model using Chimera. 490 For Extended Data Fig. 7c, the sequences were aligned in Clustal Omega and mapped 491 onto GR from the maturation complex using Chimera. Sequences in the alignment are the human 492 steroid hormone receptors: glucocorticoid receptor, mineralocorticoid receptor, androgen 493 receptor, progesterone receptor, estrogen receptor α and β (Uniprot accession numbers: P04150, 494 P08235, P10275, P06401, E3WH19, Q92731, respectively). Conservation was calculated using 495 AL2CO 20 parameters (unweighted frequency estimation and entropy-based conservation 496 measurement). Relating to Fig. 2c and Extended Data Fig. 7d, the p23tail-helix motif search was 497 performed using ScanProsite 21 . The motif "FXXMMN" was used to search the UniProtKB 498 sequence database with taxonomy restricted to Homo Sapiens. There were 10 total hits on the 499 motif, which included p23 and NCoA3/SRC-3.  , Composite cryo-EM map of the GR-maturation complex. Hsp90A (dark blue), Hsp90B (light blue), GR (yellow), p23 (green). This color scheme is maintained in all figures that show the structure. b, Atomic model built into the cryo-EM map. c, Interface 1 of the Hsp90:GR interaction depicting the GR pre-Helix 1 region (GR 523-531 ) threading through the Hsp90 lumen. Hsp90A/B are in surface representation. Hydrophobic residues on Hsp90A/B are colored in pink. d, Interface 2 of the Hsp90:GR interaction depicting GR Helix 1 (GR 532-539 ) packing against the entrance to the Hsp90 lumen. Hsp90A/B are in surface representation. Hydrophobic residues on Hsp90A/B are colored in pink. e, Interface 3 of the Hsp90:GR interaction depicting residues on the Hsp90A MD loops (Hsp90A N318,W320,R346,F349 ) and Hsp90B amphi-α (Hsp90B T624,Y627,M628 ) packing against GR, which is in surface representation.

Hsp90A Hsp90B
Hsp90A Hsp90B Figure 1: Architecture of the GR-maturation complex  Fig. 7d). The alignment is colored according to the ClustalW convention. d, Sequence conservation mapped onto GR in surface representation using ConSurf, colored from most variable (white) to most conserved residues (maroon). 87 GR sequences were used for the calculation. The p23 tail-helix (light green) is overlaid to indicate the p23:GR interface. e, Equilibrium binding of 20nM fluorescent dexamethasone to 250nM GR with chaperone components and p23 tail mutants measured by fluorescence polarization (±SD). Assay conditions:5mM ATP, 2uM Hsp40, 15uM Hsp70, Hsp90, Hop, and p23 or p23 tail mutants. Individual data points shown in Extended Data Fig. 8b. Polarization (