Structural transitions modulate the chaperone activities of Grp94

Grp94 Collaborates with BiP in Protein Remodeling In Vitro.

Previous studies have shown that Grp94 has low basal ATP hydrolysis rates, which are stimulated by BiP (22, 27). It has also been shown that Grp94’s pre-N domain plays a regulatory role, as the full truncation of the pre-N domain (residues 1–72) accelerates ATP hydrolysis rates, while partial truncations have more modest effects on hydrolysis rates (7, 12). Moreover, a distal portion of the pre-N domain including residues 58–62 is hypothesized to provide a regulatory role for structural transitions (12). The presence of Grp94’s pre-N domain is essential for the maturation of some clients (12), although the pre-N domain promotes aggregation of another client (13). To explore the role of Grp94’s pre-N domain in protein remodeling and in conformational sampling, we created Grp94 truncation mutant proteins lacking all (residues 1–72) or part (residues 1–47) of the pre-N domain of Grp94, termed Δ72 and Δ47 (SI Appendix, Fig. S1A). The deletion mutants were purified alongside the wild-type (WT) protein. Truncation of the pre-N domain does not affect the purity, overall fold, or client binding ability of Grp94 (SI Appendix, Fig. S1 B–D). We tested the ATP hydrolysis activity of our Grp94 constructs alone and in the presence of BiP. In agreement with other studies, the results show that Grp94 alone has minimal ATP hydrolysis rates, which are enhanced in the presence of BiP (Fig. 1A) (22, 27).

J-proteins are cochaperones of Hsp70s (BiP) that stimulate Hsp70 ATP hydrolysis rates and promote the ADP-bound conformation for client/substrate trapping (28). We observed stimulation of BiP’s ATP hydrolysis rates by a J protein, DnaJB11, in this study (Fig. 1A) and in our previous work (26). Therefore, we tested the effect of BiP and DnaJB11 on ATP hydrolysis rates with our Grp94 constructs. DnaJB11 and BiP increased ATP hydrolysis rates of all the Grp94 constructs significantly above the stimulated rate by BiP alone (Fig. 1A). The most dramatic effect was observed for Grp94 FL, whose ATP hydrolysis increased by 14-fold in the presence of BiP and DnaJB11, followed by Grp94 Δ47 with a 11-fold increase and a more modest increase of sixfold for Grp94 Δ72. The results show that all Grp94 constructs, regardless of basal ATP hydrolysis levels, hydrolyze ATP at a similar rate when BiP and DnaJB11 are present. In control experiments, a strong stimulatory effect by DnaJB11 alone on the ATPase activity of Grp94 was not observed (Fig. 1A), which is in agreement with previous findings (29). To confirm that the increase in ATP hydrolysis rates was specific to Grp94 and resulted from direct interactions of Grp94 with BiP, we created ATP hydrolysis (E103A), ATP-binding (D149N), and BiP-binding (K467A) defective mutants of Grp94 (7, 22, 30). The ATP-binding and ATP hydrolysis–defective mutants in both FL and Δ72 Grp94 constructs showed reduced ATP hydrolysis rates compared to WT (Fig. 1B). Together, our results show that DnaJB11 enhances BiP’s ability to stimulate Grp94 ATP hydrolysis, possibly by stabilizing the BiP ADP conformation. Previous studies show that the Grp94 interacting residues in the nucleotide-binding domain (NBD) of BiP are exposed when BiP is in the ADP conformation (22). We tested the effect of the isolated BiP NBD on the ATP hydrolysis activity of the Grp94 constructs (SI Appendix, Fig. S2). Similar to the results obtained in the presence of BiP and DnaJB11 and in agreement with previous results, the BiP NBD alone efficiently enhanced the ATP hydrolysis activity of all the Grp94 constructs.

Next, we asked whether the Grp94 constructs could collaborate with BiP and its co-chaperones to fold a model client protein, luciferase, in vitro. We modified the luciferase assay from our previous studies to include BiP, BiP cochaperones, Grp94, ATP, and luciferase during the heat stress step to investigate collaboration (26). Chaperone-dependent refolding was monitored after removing heat stress and adding an ATP regenerating system. The BiP system was effective in refolding luciferase compared to a denatured luciferase control without any added chaperones (Fig. 1C). Interestingly, the addition of all Grp94 constructs led to enhanced BiP refolding rates by 12%, suggesting that all the constructs were capable of collaborating with the BiP system to fold luciferase. The similar refolding rates agree with our ATP hydrolysis results showing that all Grp94 constructs have similar ATP hydrolysis activity in the presence of BiP and DnaJB11 (Fig. 1A). The presence of BiP is required for refolding, as the Grp94 constructs alone with the BiP co-chaperones were unable to refold luciferase (SI Appendix, Fig. S3A). To determine the requirement for Grp94 chaperone activity, we tested Grp94 ATP hydrolysis (E103A) and ATP-binding (D149N) defective mutants in addition to the BiP-binding mutant of Grp94 (K467A), as the collaboration of the cytosolic Hsp70 and Hsp90 homologs requires direct protein–protein interactions between the two chaperones (23). In luciferase refolding experiments, all the mutants were defective and did not enhance luciferase reactivation beyond what is observed by the BiP system (Fig. 1D). These results suggest that Grp94’s collaboration with BiP in refolding requires ATP hydrolysis by Grp94 and direct interactions with BiP. Importantly, we observed that the ATP-binding mutant of Grp94 strongly inhibited the refolding activity of the BiP system (Fig. 1D). Similar results were obtained for the Grp94 Δ72 construct (SI Appendix, Fig. S3B). When we tested Grp94 alone without the BiP system, neither WT nor mutant Grp94 could facilitate luciferase refolding. Interestingly, the Grp94 ATP-binding mutants also inhibited the amount of luciferase that could spontaneously fold without BiP and its co-chaperones (SI Appendix, Fig. S3 C and D). Together, these results demonstrate that Grp94 and BiP collaborate in protein remodeling. The collaboration requires direct interactions between the two chaperones and ATP binding and hydrolysis by Grp94.

DnaJB11 Enhances the Interaction of Grp94 and BiP under ATP Conditions.

To further investigate the requirements for Grp94 chaperone activity in collaboration with BiP, we analyzed the interactions of Grp94 WT and mutant proteins with the BiP system and luciferase using a protein–protein interaction assay (Materials and Methods). Streptavidin beads were used to isolate biotinylated Grp94 complexes during the luciferase refolding assay, and bound proteins were separated on an SDS-Page gel. In experiments lacking BiP cochaperones, the association of BiP with Grp94 in complexes, both in the presence and absence of luciferase, could not be detected (SI Appendix, Fig. S4). The presence of DnaJB11 significantly enhanced the amount of BiP associated with Grp94 in complexes and did not require the presence of luciferase for complex formation (SI Appendix, Fig. S4 and Fig. 2A). When we tested the interaction of the Grp94 mutants, the BiP-binding mutant showed ~twofold less BiP associated with Grp94 in complexes, alluding to a direct BiP–Grp94 interaction (Fig. 2 A and B). Notably, the Grp94 ATP-binding and hydrolysis mutants were associated with a similar amount of BiP compared to the WT protein. We also observed that DnaJB11 was found in the protein complexes with Grp94 WT and mutants (Fig. 2 A and C). The amount of DnaJB11 loosely correlated with the amount of BiP in the complexes, although differences in the amount of DnaJB11 bound was not statistically significant (Fig. 2 A and C). Overall, our results suggest a crucial role for DnaJB11 in facilitating the collaboration of Grp94 and BiP. In addition to targeting substrates to BiP, DnaJB11 plays an active role in enhancing the interactions of BiP and Grp94 under ATP conditions, which is important for substrate transfer from BiP to Grp94. Similarly, cytosolic J-proteins can stimulate the Hsp70-Hsp90 interaction when ATP is present, and the interaction is enhanced when additional co-chaperones are present (23, 31).

Nucleotide Binding Reduces Grp94 Client Affinity.

Next, we examined Grp94’s interactions with client by quantifying the amount of luciferase in complex with Grp94 in the presence of BiP and its co-chaperones. Analysis of the complexes showed that the amount of luciferase found in complexes was similar for the Grp94 ATP hydrolysis–defective mutant when compared to WT, while a slightly reduced amount of luciferase was in complexes with the BiP-binding mutant (Fig. 2 A and D). Interestingly, 1.7-fold more luciferase was found in complexes with the ATP-binding defective mutant of Grp94 compared to Grp94 WT (Fig. 2 A and D). These results suggest that neither ATP binding nor hydrolysis is required for the complex formation of Grp94 with BiP and luciferase. Similarly, cytosolic Hsp90 does not require ATP binding for forming complexes with Hsp70 and clients when co-chaperones are present (9, 32). In control experiments, Grp94 WT and mutants bound similarly to luciferase in the absence of the BiP system (Fig. 2 A and E). Together, these results suggest that ATP binding in Grp94 is correlated with interactions with clients.

To further explore the effect of ATP on Grp94–client interactions, we analyzed the interactions of Grp94 with another model client, Δ131Δ, using a fluorescence polarization assay. Under apo conditions, Grp94 FL bound Δ131Δ weakly with a K_d of 7.2 ± 0.6 μM (SI Appendix, Fig. S5 A and E). Interestingly, adding ATP or AMP-PNP decreased the client binding affinity of Grp94 by ~twofold compared to apo conditions. Similar to the FL construct, Grp94 Δ72 showed reduced client binding in the presence of ATP and AMP-PNP, with reductions in binding affinity by ~fourfold to sixfold, respectively (SI Appendix, Fig. S5 B and E). As expected, the ATP-binding mutant of Grp94 FL did not show any change in binding affinity in the presence of ATP. In contrast, the ATP hydrolysis mutant behaved similarly to WT, suggesting that nucleotide binding but not hydrolysis was sufficient to alter client interactions of Grp94 (SI Appendix, Fig. S5 C–E). The results from the binding assays agree with our observations from the pulldown assays that abolishing nucleotide binding enhanced Grp94 client interactions. Such increased or stable interactions with the client will sequester the client from the solution and prevent productive folding, which explains the inhibitory effect of this mutant on both spontaneous and BiP-mediated refolding of luciferase. Additionally, since Grp94 forms complexes with BiP, preventing ATP binding will stall the ternary complex and prevent the recycling of the BiP chaperones. These results also explain our previous observations that the holding activity of Grp94 required ATP binding but not hydrolysis to release bound clients for subsequent folding (26).

The Grp94 pre-N Domain Suppresses ATP-induced Conformational Changes.

Hsp90s couple ATP and co-chaperone binding to large-scale conformational changes that regulate chaperone activity. The transition from an open Hsp90 to a closed state proceeds through several intermediates, with active client folding requiring passage through the catalytically active closed state (33). To determine the structural effects of ATP and BiP binding and the coupling of structural transitions and chaperone activity, we characterized conformational cycling of Grp94 using Double Electron Electron Resonance (DEER) EPR spectroscopy. DEER EPR spectroscopy measures distance distributions between labeled sites on a protein and can report on the structure and conformational dynamics (34). The experiments are performed on frozen solutions; thus, individual protein conformers captured in different conformations are retained, providing a clear picture of the conformational ensemble. We selected sites in Grp94 expected to undergo significant changes in interdomain distances between the resolved conformations (see Materials and Methods and SI Appendix, Fig. S6 for details). For each site, the MtsslWizard program was used to predict distance distributions while considering spin-label and protein flexibility. All EPR constructs had a similar level of purity, folded similarly to WT, and retained functional activity (SI Appendix, Fig. S1 E–G).

We first probed the NTD, which is expected to undergo large conformational transitions, by labeling position 161 (K161C) to glean information on dimer opening and closing. In the apo state, the NTD of Grp94 FL populates broad distance distributions reflecting a variety of conformations. The distance distributions can be categorized into shorter distances (25 Å to 45 Å) and a longer distance (55 Å) (Fig. 3A). Some shorter distances are within the expected range for the partially open crystal structure based on the computational predictions. However, the longer distances, representing the prominent peak, suggest that Grp94 may populate more open conformations characterized by a larger separation of the N-terminal domains. Our results are in agreement with SAXS measurements on apo Grp94 that reported more open conformations than observed in the crystal structures currently available (35). When ADP and ATP were added to Grp94 FL, no significant shift in the distance distributions compared to the apo state was observed. We only observed a shift of some of the longer distances (55 Å) to shorter distances (22 Å) closely matching the expected distances in the closed state when Grp94 was incubated with AMP-PNP (Fig. 3A).

Next, we investigated how the pre-N domain affects Grp94 conformational changes by repeating the experiments using the pre-N domain truncation construct labeled in the same position (Grp94 Δ72 K161C). The conformations of Grp94 Δ72 in the apo state were very similar to Grp94 FL and ADP did not induce any significant changes in the distance distributions (SI Appendix, Fig. S7A). However, a larger population of closed conformations was observed for Grp94 Δ72 with both ATP and AMP-PNP, while Grp94 FL required AMP-PNP to reach these conformations. These results suggest that Grp94 Δ72 is more dynamic and sensitive to nucleotide, which is in agreement with accelerated ATP hydrolysis rates of the pre-N domain truncated Grp94 (12).

We also probed Grp94’s conformations using a site in the middle domain, R395C, that is expected to have a similar separation in either of the available structures (SI Appendix, Fig. S6B). The observed distance distributions under apo and nucleotide conditions closely matched the predicted distances, while a small population of molecules sample longer distances (SI Appendix, Fig. S7B).

These results reveal important differences in the conformational ensemble of Grp94 compared to other Hsp90s. The NTD of Grp94 exhibits flexibility and can populate an ensemble of conformations. Despite this flexibility, Grp94 FL cannot sample the closed conformation under apo or ATP conditions, unlike other Hsp90 homologs in equilibrium between open and closed conformations (36). The presence of the pre-N domain does not change the conformational ensemble of Grp94 but instead constitutes a barrier to suppress nucleotide-induced dimer closure (11, 12). Together, the nucleotide insensitivity of Grp94 is in agreement with our ATPase assays showing that Grp94 FL alone has negligible ATP hydrolysis activity. The results with Grp94 FL R395C showed that a very small population of molecules exist in MD extended configurations like those captured for apo bacterial Hsp90 (37). Together, these results suggest a more compact and less flexible conformation at the MD/CTD interface of apo Grp94. As discussed later, the MD/CTD configuration is critical for client interactions (35, 38, 39). However, we cannot eliminate the possibility that very extended conformations outside of the distance range accessible to DEER EPR are sampled by Grp94.

BiP Stabilizes a Partially Closed Conformation of Grp94 that Is Primed for Closure.

Since nucleotide had minimal effect on large-scale conformational changes of Grp94, we tested the effect of BiP on Grp94 conformations using the isolated BiP NBD. The addition of BiP alone (apo) induced significant conformational changes in Grp94 FL K161C with a shift of the longer distances (55 Å) to relatively shorter distances centered at ~42 Å (Fig. 3B). In the presence of BiP NBD and ATP, there was a dramatic shift of all the distances to a major peak (23 Å) corresponding to the closed state. Remarkably, BiP NBD and ADP induced a similar shift in distance distribution toward the closed state. Similar results were obtained for Grp94 Δ72 (Fig. 3B and SI Appendix, Fig. S7C), in agreement with recent studies (40). In contrast to Grp94 FL, Grp94 lacking the pre-N domain sampled populations of more open configurations characterized by distances beyond 45 Å under conditions containing nucleotide and BiP NBD. This suggests the pre-N domain provides a stabilizing role in maintaining more closed Grp94 conformations. As expected, the MD of Grp94 appeared to be relatively stable in solution, as the observed conformations agree with the crystal structures (SI Appendix, Fig. S7D).

To better characterize the BiP stabilized conformations, we further probed the NTD dimerization interface of Grp94 by labeling residue K87. In the presence of BiP NBD alone, distances shorter than the expected distance for the closed NTD dimer interface were observed (Fig. 3C). When either ADP or ATP was added, the observed distances perfectly matched the expected distances for the closed NTD dimer interface. These results suggest that the presence of BiP overcomes rate-limiting steps in Grp94’s conformational cycle, and transition to the closed conformation from the Grp94 apo+BiP conformation can occur regardless of the nucleotide identity. We decided to explore this conformation in more detail.

In the DEER experiments, Grp94 K87C had a shallow modulation depth in the presence of BiP NBD alone compared to when nucleotides were added (Fig. 3 C, Left). The low modulation depth could indicate distances too short or too long to capture in the DEER experiments. The expected distance for Grp94 K87C apo is long and beyond the DEER limit; thus, no experimental distances were obtained (SI Appendix, Fig. S6C). We compared the local environment of K87C in the Grp94 apo+BiP conformation to the Grp94 apo conformation using CW-EPR line shape analysis. Adding BiP alone significantly broadens the line shape of the spectra and increases the rigid component’s rotational correlation time from 16 ns to 60 ns, suggesting significant differences in the local environment (SI Appendix, Fig. S8). This suggests that the low modulation depth in the Grp94 apo+BiP conformation may arise from shorter distances that were not captured by DEER experiments. We also observed differences in the local environment of Grp94 ADP+BiP vs. ATP+BiP conformation in the CW-EPR line shape analysis, although the DEER data showed similar distances (Fig. 3C and SI Appendix, Fig. S8). However, closer inspection of the DEER data shows a small population of shorter distances (15 to 25 Å) for K87C under ADP+BiP conditions that is absent under ATP+BiP conditions. This could indicate an alternate conformation of Grp94 ADP+BiP that is more compact than the Grp94 ATP+BiP conformation. A compact closed conformation has been observed for yeast Hsp90 and represents a conformation after ATP hydrolysis prior to the opening of the Hsp90 dimer (41). The shorter distances for K87C in Grp94 ADP+BiP conformation would also explain the observed cross-linking of this region when Grp94 Δ72 was incubated with BiP and ADP in other studies (40).

It was recently proposed that BiP stabilizes a closure intermediate of Grp94, similar to the coiled-coil structure of TRAP-1 (24, 42). To determine whether the coiled-coil state is populated in our experiments, we created a homology model of Grp94 based on the TRAP-1 structure. We then used the MtsslWizard program to predict the expected distance distribution of the homologous residues corresponding to Grp94 K87C and K161 (SI Appendix, Fig. S9A). Although the predicted distance for K87C matches the experimental distances, that of K161C (58.5 Å) is much longer than experimentally observed (42 Å) (Fig. 3D). Interestingly, the observed distances for K161C more closely match a recently reported partially closed structure for human Hsp90 (9) (SI Appendix, Fig. S9B and Fig. 3D). Additionally, the predicted distances for K87C in the partially closed structure of human Hsp90 suggest distances shorter than 20 Å, which could explain the low signals from our DEER experiments (Fig. 3D). Thus, we conclude that the Grp94 apo+BiP conformation is better represented by the partially closed structure of human Hsp90.

Grp94 Collaborates with BiP to Fold Proteins In Vivo.

Our functional studies indicate that Grp94 collaborates with BiP to fold proteins in vitro. Additionally, our structural studies provide insight into the mechanism for the collaboration of two proteins. The BiP co-chaperone DnaJB11 stabilizes the interaction of Grp94 and BiP under ATP conditions, similar to the role of Hop in the cytosolic Hsp90–Hsp70 collaboration. Once BiP binds, it serves as a co-chaperone of Grp94 to push Grp94 into the closed conformation through a partially closed intermediate (40). The effect of BiP on the conformational cycle of Grp94 suggests a strict dependence of Grp94 on BiP for chaperone activity. However, we have previously shown that Grp94 possesses an ATP-dependent chaperone activity independent of BiP interactions (26). To determine the impact of Grp94–BiP interaction on the folding and maturation of cognate client proteins of Grp94, we generated Grp94 K467A retrovirus. We then transduced a Grp94-null pre-B cell line with either wild type or K467A mutant virus (43). These cells were then analyzed for cell surface expression of several known Grp94 clients, including integrins and TLRs (44, 45). We showed previously that the bona fide client of Grp94 is trapped in the endoplasmic reticulum and cannot export to the cell surface without Grp94 (44, 46). By intracellular stain followed by flow cytometry, we found WT and K467A mutants were expressed at the same level (Fig. 4 A and B). Interestingly, surface expression of integrins but not Toll-like receptor 2 (TLR2) was significantly reduced in Grp94 K467A-expressing cells compared with WT cells (Fig. 4 A and B). These results confirm that Grp94 collaborates directly with BiP to fold client proteins in vivo. However, the details of the collaboration are client specific and do not represent a general mechanism of Grp94 chaperone activity.

Nucleotide Binding Alters the Conformation of Grp94 Lid Regions.

In our DEER experiments, ATP did not induce significant changes in the NTD or MD conformation of Grp94. However, our client interaction assays show an effect of ATP binding on client interactions. Importantly, ATP binding but not hydrolysis also modulates Grp94 interactions with co-chaperones in vivo and is sufficient for folding some client proteins (19, 20). Recent studies in yeast report that ATP hydrolysis by Hsp90 is dispensable for viability (47, 48). To better define the role of nucleotide in modulating structural transitions in Grp94, we examined local conformational changes in Grp94 under different nucleotide conditions. Within the NTD, the lid region is an essential regulatory element known to undergo significant rearrangements (49). We probed the conformational dynamics of the lid region using CW-EPR line shape analysis. Several lid structures have been observed for Grp94, including open, extended open, partially closed, and closed structures (12, 49–51). For simplicity in data interpretation, we assign a more mobile lid as the closed lid and a more rigid lid as the open lid due to proximity and stabilizing interactions with helix 1 (SI Appendix, Fig. S10). In the apo state, the CW spectra show two components, 84% of the rigid component and 16% of the mobile component, suggesting the lid is dynamic and can populate multiple lid conformations (Fig. 5 and SI Appendix, Fig. S10C) (52). Upon adding nucleotide, we observed an increase in the population of the mobile component to 30%, suggesting that Grp94 is moving toward a closed lid configuration (Fig. 5 and SI Appendix, Fig. S10C). Concomitantly, the rotational correlation time of the rigid component decreased from 37.3 ns to 20.1 ns. We also tested the effect of BiP in the presence and absence of nucleotides on the mobility of the ATP lid. Adding BiP NBD alone did not significantly affect the lid conformation compared to Grp94 apo, while BiP and nucleotides had a more pronounced effect on lid closure than when nucleotides were present alone. With BiP and ATP, the rotational correlation time of the rigid component further reduced to 14.9 ns compared with 20.1 ns when ATP was present alone (SI Appendix, Fig. S10). Note that we did not see a complete disappearance of the rigid components in the lid region under all conditions tested. Conformational heterogeneity in the lid region of Hsp90, including the presence of several intermediate conformations and rapid conversion between open and closed states at physiological temperatures, has been recently characterized by NMR experiments (53, 54). A unique five-amino acid insertion in the Grp94 lid region also predisposes it to preferentially adopt an open conformation even when a nucleotide is bound (49). Nonetheless, our results suggest that while nucleotides do not initiate large-scale conformational changes in Grp94 FL, they induce significant rearrangements in the lid region, pushing the lid toward more closed conformations. Such local conformational changes may translate into structural remodeling within the subdomain to affect client interactions. These findings on lid mobility are in agreement with our previous computational study where varying degrees of lid mobility were observed in the catalytically active conformation, which was dependent on nucleotide (55).

ATP Modulates Client Interactions of Grp170.

The effect of ATP on Grp94 client interactions and client refolding was an interesting finding and presented additional detail into Grp94’s chaperone mechanism. ATP binding reduced client interactions and promoted client refolding, while abolishing ATP binding enhanced client interactions but inhibited client refolding. We questioned whether other ER chaperones like BiP and Grp170 can function similarly since they are ATP-dependent ER chaperones that can bind directly to unfolded and misfolded proteins to prevent aggregation (56–58). To this end, we took advantage of the luciferase refolding assays with BiP, DnaJB11, and Grp170 and performed an order of addition experiment (SI Appendix, Methods). The highest luciferase reactivation was achieved when all three chaperones and ATP are present during heating, suggesting a holding and prevention of aggregation activity for the ER chaperones (Fig. 6A, lanes 1-3). In reactions containing DnaJB11 only during denaturation, high refolding rates were observed, in agreement with the role of J proteins as effective holding chaperones (Fig. 6A, lane 4) (59). The presence of BiP during denaturation did not offer any superior advantage compared to buffer alone (Fig. 6A, lane 5). Interestingly, the presence of Grp170 during denaturation significantly inhibited luciferase refolding (Fig. 6A, lane 6). This inhibition of luciferase refolding by Grp170 persisted even when BiP was present during heating, but refolding was restored when DnaJB11 and ATP were present with Grp170 during heating (Fig. 6A, lanes 6 to 8). When ATP was omitted during denaturation, DnaJB11 again offered a superior advantage when present alone during denaturation, while low refolding rates were obtained when all three chaperones are present during denaturation (Fig. 6B, lanes 1 to 4). BiP had minimal effect on reactivation compared to buffer alone. Importantly, the presence of Grp170 significantly inhibited luciferase refolding, and this inhibition could not be restored by the addition of either BiP or DnaJB11 (Fig. 6B, lane 5 to 8). The inhibitory effect of Grp170 explains the low reactivation rates observed when all the three chaperones are present without ATP during heating (Fig. 6B, lane 3). The results suggest that Grp170 inhibits luciferase refolding when present during denaturation without ATP, possibly through enhanced substrate interactions. However, in the presence of ATP and DnaJB11, luciferase refolding is restored, suggesting that nucleotide affects Grp170 substrate interactions. This means that DnaJB11 can only outcompete Grp170 interactions with luciferase when nucleotide is present and commit luciferase to productive folding pathways. We did not observe a significant holding role for BiP; however, we noticed that the presence of BiP with DnaJ11 during denaturation caused a slight decrease in refolding rates compared to when DnaJB11 was present alone (Fig. 6 A and B, lane 9). Overall, our results suggest that similar to Grp94, Grp170 substrate interactions are modulated by nucleotide, which may represent a general chaperone mechanism for a subset of ER chaperones.

Plasmids and Proteins.

Grp94 truncation mutants Δ47 and Δ72 (His₆ -sumo-TEV-Grp94, aa 48-803 or 73-803) were generated by gene splicing from the human Grp94 full-length (FL) plasmid. A human BiP-NBD construct (aa 27-418) was generated by inserting a stop codon after position 418 using site-directed mutagenesis. Plasmid containing the Δ131Δ gene with a K16C mutation was a generous gift from Sue Wickner (84). All mutations in pet-15 Grp94 full-length and truncation mutants were generated by site-directed mutagenesis using the QuikChange Lightning Kit and confirmed by sequencing (GeneWiz). Purification of human FL Grp94, BiP, DnaJB11, and Grp170 was performed as previously described (26). Grp94 truncation mutants and BiP NBD were purified similarly to full-length proteins. Δ131Δ was purified similarly to previously described methods with modifications (SI Appendix, Methods) (68, 85).

Assays.

ATP hydrolysis assays, streptavidin pull-down assays, western blotting, and luciferase reactivation assays were performed as previously described with modifications (SI Appendix, Methods) (26). Fluorescence polarization methods are described in SI Appendix, Methods.

Site-directed Spin Labeling.

Native cysteine residues in Grp94 FL or Δ72 proteins (C138, C576, C645A) were replaced with alanine. Computational analysis of the Grp94 crystal structures [2O1V (partially open) and 5ULS (closed)] was used to select labeling sites with the criteria of solvent accessibility, little conservation among Hsp90 isoforms, and site–site distance within the EPR detection limit (20 to 70 Å) (12, 86). Selected labeling sites were mutated to cysteine residues, and protein constructs were purified similarly to WT proteins. Purified EPR constructs were reduced with 1 mM TCEP and buffer exchanged into labeling buffer (20 mM Hepes–KOH 7.5 and 100 mM KCl) before incubating with a 10-fold molar excess of MSL (3-maleimido proxyl) label. Labeling was allowed to proceed overnight at 4 °C in the dark with gentle shaking, and excess MSL label was removed by gel filtration on a Superdex 200 column (Cytiva) or 0.5-mL Zeba protein desalting columns (ThermoScientific) equilibrated with EPR buffer (20 mM potassium phosphate pH 7.0, 50 mM KCl, and 2 mM MgCl₂).

EPR Sample Preparation.

For both CW-EPR and DEER experiments, Grp94 was buffer exchanged with a 10-KDa cutoff centrifugal filter (Amicon) to a final concentration of 50 μM (dimer) in EPR buffer. Protein samples were incubated with 10 mM nucleotide (ADP, ATP, or AMP-PNP) and 10 mM MgCl₂ at 37 °C for 1 h (ATP or ADP) or 3 h (AMP-PNP). BiP NBD was buffer exchanged into EPR buffer and added at a final concentration of 100 μM (monomer). For DEER experiments, samples were supplemented with 30% glycerol-d₃ and frozen by immersion into liquid nitrogen before DEER data collection. Samples with Grp94 and BiP NBD with MgCl₂/ATP were supplemented with an ATP regenerating system (20 mM phosphocreatine and 60 μg/mL creatine kinase). Buffers for CW-EPR experiments were prepared with H₂O, and buffers for DEER experiments were prepared with D₂O.

Continuous-Wave (CW) EPR Measurements.

All EPR experiments were conducted at the Ohio Advanced EPR Laboratory, Miami University. CW-EPR spectra were collected at room temperature on a Bruker EMX X-band spectrometer with an ER041xG microwave bridge and an ER4119-HS resonator. The spectra shown represent the average of 45 scans with the following parameters: 3318 G central field, 150 G sweep width, 100 KHz modulation frequency, 1.0 G modulation amplitude, 41.53 s field sweep, and 10 mW microwave power. EPR spectral simulations are described in SI Appendix, Methods.

Q-band DEER Experiments.

DEER experiments were performed using a Bruker ELEXSYS E580 (~34.0 GHz) spectrometer equipped with a SuperQ-FT pulse Q band system with either a 300 W or 150 W amplifier and EN5107D2 resonator. DEER samples were loaded into a 3 mm Precision Quartz EPR sample tube (Wilmad-LabGlass) and assembled into a sample holder before inserting into the resonator. DEER data were collected at 80 K using a four-pulse sequence [(π/2)_ν1 – τ₁ – (π)_ν1 – t – (π)_ν2 – (τ₁ + τ₂ – t) – (π)_ν1 – τ₂ – echo] with a 16-step phase cycling step. The probe pulse width was optimized depending on the amplifier power. Routine pulse widths were 8, 16, 16 ns or 10, 20, 20 ns for the 300 W amplifier and 20, 40, 40 ns or 22, 44, 44 ns for the 150 W amplifier. The pump pulse width was 70 ns frequency-swept pulse spanning 65 or 85 MHz. The frequency separation between the probe and pump pulse at the largest separation was 120 MHz. The shot repetition time was determined by the spin-lattice relaxation time T₁. Transverse relaxation data (T₂) were collected using a Hahn echo pulse sequence [(π/2) – τ₁ – (π) – τ₁ – echo] with optimized pulse width, τ₁ of 200 ns with 16 ns increment and two-step phase cycling. The T₂ was determined by fitting the data using a single exponential decay. DEER data were collected out to ~3.0 to 4.5 µs with signal averaging for 12 to 24 h. The data were analyzed with the DEERAnalysis program (2019) by transforming the time-domain data using Tikhonov regularization. The regularization parameter in the L curve was selected by examining the fit of the time-domain data. Background signals were corrected with a homogenous three-dimensional model. Samples to be directly compared were run using similar instrument settings.

Retroviral Packaging and Infection.

Plasmids containing WT or mutant Grp94 expression constructs were transfected into Plat-E cells using lipo2000 in serum-free DMEM (Gibco #11965-092) supplemented with 10 mM HEPES (Gibco #15630-080). After 6 h, the medium was replaced with complete DMEM with 10% FBS (Gibco #10082-147) and 1% penicillin–streptomycin (Gibco #15140-122), and 48 h posttransfection, the retroviral particles in the cell culture supernatant were collected. Grp94 null pre-B cells were then spinfected with the viral supernatant and polybrene for 90 min. The infected pre-B cells were cultured in complete RPMI (Gibco #11875-093) for another 48 h before proceeding for flow cytometry analysis.

Flow Cytometry.

Cells were collected and washed twice with PBS. Live/Dead Fixable Blue (Invitrogen #L23105) viability dye in PBS was used to stain dead cells for 15 min at 4 °C. Afterward, the cells were washed twice with FACS buffer. For analysis of CD11a, CD49d, and TLR2 expression, cells were incubated with biotin-conjugated anti-mouse CD11a (eBioscience #13-0111-85), CD49b (eBioscience #13-0492-85), or TLR2 (eBioscience #13-9021-82) antibodies separately at 4 °C for 1 h. After washing twice with FACS buffer, the cells were incubated with APC–streptavidin (eBioscience #14-4317-82) at 4 °C for 30 min. Following secondary antibody staining, cells were washed twice with FACS buffer and analyzed immediately with Cytek Aurora. To analyze Grp94 expression, cells were fixed with 4% PFA at room temperature for 10 min, followed by permeabilization with methanol at −20 °C for 10 min. Cells were then incubated with PE-conjugated anti-Grp94 antibody (Enzo #ADI-SPA-850PE-D) in FACS buffer at room temperature for 1 h. After washing with FACS buffer, cells were analyzed by Cytek Aurora. The mean fluorescence intensity was analyzed using FlowJo v10.7.

Molecular Modeling.

Spin label rotamers were modeled using MtsslWizard (http://www.mtsslsuite.isb.ukbonn.de) with the indicated crystal structures. Models of the semi-closed and coiled-coil conformations (PDB ID 5F3K (9) and 7KW7 (42)) were generated with I-TASSER (87) and SWISS-MODEL (88), respectively.