Search tips
Search criteria 


Logo of hhmipaabout author manuscriptssubmit a manuscriptHHMI Howard Hughes Medical Institute; Author Manuscript; Accepted for publication in peer reviewed journal
J Mol Biol. Author manuscript; available in PMC 2012 July 6.
Published in final edited form as:
PMCID: PMC3282117

Cross monomer substrate contacts reposition the Hsp90 N-terminal domain and prime the chaperone activity


The ubiquitous molecular chaperone Hsp90 plays a critical role in substrate protein folding and maintenance, but the functional mechanism has been difficult to elucidate. In previous work a model Hsp90 substrate revealed an activation process in which substrate binding accelerates a large open/closed conformational change required for ATP hydrolysis by Hsp90. While this could serve as an elegant mechanism for conserving ATP usage for productive interactions on the substrate, the structural origin of substrate catalyzed Hsp90 conformational changes are unknown. Here we find that substrate binding affects an intrinsically unfavorable rotation of the Hsp90 N-terminal domain (NTD) relative to the middle domain (MD) that is required for closure. We identify an MD substrate binding region on the interior cleft of the Hsp90 dimer and show that a secondary set of substrate contacts drive an NTD orientation change on the opposite monomer. These results suggest an Hsp90 activation mechanism in which cross-monomer contacts mediated by a partially structured substrate prime the chaperone for its functional activity.


Molecular chaperones confer stress resistance critical for survival under harsh environmental conditions and maintain protein homeostasis under normal conditions. Beyond their role in protein folding, chaperones affect protein activation and trafficking, facilitating the degradation of terminally misfolded proteins, and the formation and disassembly of macromolecular complexes. Hsp90 is a highly conserved member of the chaperone family, and plays a unique role by its regulatory influence in eukaryotes via the activation of specific classes of substrates (also known as clients), such as nuclear receptors and kinases 1. This broad regulatory influence is thought to underlie the potent influence of Hsp90 inhibitors on the growth of diverse cancer types 2. Despite its fundamental cell biological and clinical importance, the mechanism by which Hsp90 stabilizes and remodels client proteins is not understood.

One confounding problem is that Hsp90 is large, conformationally dynamic, and undergoes dramatic structural changes upon ATP binding and hydrolysis (Figure 1A)3; 4; 5. Small-angle x-ray scattering (SAXS) and electron microscopy measurements (EM) have revealed an underlying complexity of Hsp90’s conformational dynamics 6; 7; 8; 9; 10; 11. The Hsp90 monomer is composed of three stable domains (N-terminal domain, NTD; middle, MD; C-terminal, CTD), and conformational flexibility results from their rigid body-like rearrangement. Under apo conditions a weak MD/CTD interface allows for a wide range of arm-arm geometries that can be influenced by pH and osmolyte conditions 8; 12. This striking flexibility has been observed for highly diverse Hsp90 homologs 6; 9 and is postulated to be critically important to Hsp90’s ability to recognize a remarkably diverse set of client proteins.

Figure 1
Hsp90 conformational flexibility

Unlike other molecular chaperones, Hsp90 appears to prefer largely folded but non-native states. This poses an additional practical challenge as such states can be difficult to populate and are prone to aggregation. Previous work introduced a well-behaved model client protein, the partially folded, but non-aggregating protein, Δ131Δ, a fragment of Staphyloccocal Nuclease that has been studied extensively by the protein folding community 13. Using this model client revealed that (i) under apo conditions Hsp90 partially closes around Δ131Δ; (ii) Hsp90 binds a highly structured region of Δ131Δ; (iii) Δ131Δ accelerates a nucleotide-driven open/closed transition and stimulates ATP hydrolysis by Hsp90, effectively activating the chaperone by lowering a rate-limiting conformational barrier. Taken in the context that the ligand-binding domain of GR enhances the ATPase of the human Hsp90 14 and that the ribosomal subunit L2 enhances the ATPase of the bacterial Hsp90 15, this suggests that activation of the rate-limiting Hsp90 conformational transition is a conserved feature of bone fide Hsp90 clients, similar to Hsp70 activation by peptide substrates. However, the mechanism by which substrate binding can drive the dramatic Hsp90 open-closed transition is unknown. Indeed, a previous low resolution SAXS analysis 13 could not determine whether Δ131Δ makes cross-monomer contacts as has been observed for the activating cochaperone aha1 16, or solely intramonomer contacts as observed for an Hsp90-cdc37-cdk4 (chaperone-cochaperone-kinase substrate) EM reconstruction17.

The Hsp90 ATPase is slow, on the order of 0.1–1 hydrolysis events per minute depending on the homolog and conditions 18; 19; 20, and mirrors a slow conformational change from the open apo state to the closed ATP conformation 13; 21. This dramatic transition involves a large change in arm-arm proximity, a domain-level change in the NTD orientation, and local structural changes within the NTD (lid closure over the nucleotide binding pocket, strand exchange between NTDs) and the MD (restructuring of the catalytic loop) 3; 5; 19; 22. Although the relative importance of these structural changes to the closure rate is not known, the structures of the AMPPNP-bound canine Grp94 (the Hsp90 homolog specific to the ER) and the apo bacterial Hsp90 (HtpG) suggest that the NTD rotational state plays an important role. Both structures exhibit an open resting state in which the NTDs are diametrically opposed, requiring a significant conformational change to come into a closure-competent alignment 4; 5; 23. As illustrated in Figure 1A, the required movement involves a 90° rotation and a 25 Å translation of the NTD center of mass, rearranging ~2000 Å2 at the MD interface. This aligns closed-state contacts (Figure 1A, red and blue spheres) and also repositions ATP by ~20 Å allowing the γphosphate to contact a highly conserved arginine on the MD (R336 in HtpG) that is essential for both closure and the bound ATP hydrolysis 3; 20; 24 (Figure 1B). Importantly, a full lid closure over the nucleotide pocket, which appears to be necessary for closure 25, cannot occur in the NTD resting state due to a significant steric clash with the middle domain 4. These observations suggest that an NTD rotation may be involved in the timing and order of many critical steps in closure and subsequent ATP hydrolysis.

Here we use our HtpG activating substrate to interrogate the open/closed transition and how this process is substrate-catalyzed. Key questions include (i) whether substrate contacts are within a single monomer or across monomers, (ii) defining the substrate binding region on HtpG in greater detail than could be achieved from our previous SAXS analysis, (iii) establishing whether a single set or multiple substrate contacts are utilized, (iv) determining how substrate binding affects HtpG structural dynamics, particularly at the NTD, and how this is related to the large energetic barrier to closure.


Monitoring NTD movement by FRET

HtpG can be substrate-activated by accelerating the kinetics of a slow open/closed structural transition required for maximal ATP hydrolysis. Previously, closure kinetics were measured by FRET, in which opposite monomers were labeled with donor and acceptor pairs 13. To monitor NTD/MD rotation, we designed a FRET pair within a single monomer. We identified residues S52/D341 on the NTD/MD that significantly change distance (22 to 39 Å) in the open/closed conformations, yet remain solvent exposed. These sites were mutated to cysteine and labeled with Alexafluor 647 and Alexafluor 555. Since this pair is on the same monomer, a 20x excess of unlabeled wild-type HtpG was added to ensure only one labeled monomer per dimer.

Figure 2A shows the apo and AMPPNP fluorescence spectrum for this FRET pair. There is significant acceptor signal in the apo state, and after a prolonged incubation with AMPPNP there is a decrease/increase in acceptor/donor signal. Similar to previous studies, these measurements were performed at pH 9 because for HtpG at this pH there is a complete conversion between the open/closed state 8. Upon adding AMPPNP there is a slow time-dependent loss in acceptor signal with single exponential kinetics (Figure 2B) and a rate (k=0.002 s−1) that is the same as the arm-arm closure rate measured previously 13. As a control, we designed a fixed-point FRET pair within the MD (residues 350 and 362), which does not change distance in the open/closed transition and confirmed that closure did not affect this FRET signal (Supplemental Figure 1A).

Figure 2
Substrate binding affects an intrinsically unfavorable NTD rotation required for closure

The similar rates for NTD rotation and arm-arm closure suggests simple two-state cooperativity, however, given the large number of structural motions that can occur in Hsp90 this observation does not rule out other intermediates. Cooperativity can only be established by the coincidence of a large number of structural probes. To examine this possibility we measured closure kinetics by SAXS. Previous work demonstrated that SAXS measurements can determine the conformational equilibrium of HtpG by linear combination fitting of the scattering spectra 7; 8. Given that closure is slow, it is possible to simultaneously initiate closure on multiple samples and sequentially measure scattering at different timepoints. The robotic sample loading system at the SIBYLS beamline at the Advanced Light Source (Berkeley, CA) allows for each measurement to take only ~2 minutes. An advantage of this method is that SAXS reports on all scattering positions, as opposed to the limited sites measured by FRET. Here we find that the kinetics of closure by SAXS match well with the arm-arm closure and NTD rotation measured by FRET (Supplemental Figure 1B). Although these measurements do not indicate whether local sequential conformational changes occur prior to the rate-limiting step, they do indicate that the gross conformational changes associated with HtpG closure obey simple two-state cooperativity.

Given this cooperativity, we wanted to know if NTD rotation was contributing to the high-energy barrier separating the open/closed states. In other words, we wanted to know whether NTD rotation could be significantly populated in isolation on a single monomer or whether this rotation is intrinsically unfavorable and requires stabilization from NTD dimerization contacts with the opposite monomer. To test this question we made the same NTD/MD FRET pair in the monomeric NM fragment of HtpG (residues 1–495). The acceptor/donor fluorescence spectrum on the NM fragment is similar to the corresponding spectrum on the full-length dimer (not shown). We find that the acceptor fluorescence on the NM fragment does not undergo any net change upon addition of AMPPNP even in the presence of Δ131Δ (Figure 2B), demonstrating that the NTD rotation required for closure requires stabilization from the opposite monomer. These results reveal a major energetic mismatch in the local and global energetics associated with Hsp90 closure. NTD rotation creates highly stabilizing dimer contacts in the closed state at the expense of a locally disfavored NTD/MD interface. This suggests that substrate binding could activate HtpG by relieving this rotational penalty.

To investigate this possibility, we tested whether substrate binding is linked to NTD rotation in the HtpG dimer. We first investigated how Δ131Δ affects AMPPNP-driven closure kinetics as monitored by NTD rotation. Previously we observed that Δ131Δ binding accelerated arm-arm closure kinetics five-fold 13, and here a similar acceleration of the NTD rotation is observed (Figure 2B). The acceptor fluorescence for HtpG: Δ131Δ starts at a lower value than for HtpG, suggesting that substrate binding alone could affect an NTD rotation. Indeed under apo conditions Δ131Δ affects the NTD orientation, indicated by a loss of acceptor fluorescence (black circles, Figure 2C) directly coupled to an increase in donor fluorescence (not shown). These results support the idea that substrate binding in the chaperone apo state could prime Hsp90 for ATP driven closure by affecting an NTD rotation.

Given the large surface that would be rearranged by a substrate-driven NTD rotation, we reasoned that Δ131Δ could affect HtpG hydrogen exchange patterns. In particular, our results suggest that substrate binding may be altering the NTD/MD interface and potentially exposing previously buried surfaces, which should show increased hydrogen exchange. Also, substrate binding itself has the potential to protect regions of HtpG from exchange. To test these predictions we performed HX-MS measurements on HtpG and HtpG/Δ131Δ. The methodology, described previously for HtpG alone 26, involves rapid dilution into D2O and allowing exchange for 30 s. H-D exchange is quenched by lowering temperature and pH, and proteolytically digested fragments are then separated and analyzed by a combined HPLC-MS setup.

The effect of Δ131Δ on HtpG H-D exchange shows a striking pattern (Figure 3A). Regions at the MD/CTD become protected (blue spheres, Figure 3B), while regions at the NTD/MD interface, as well a patch at the MD/CTD interface, show increased exchange (red spheres). Two regions at the NTD/MD that become deprotected undergo large rearrangements and become significantly more exposed upon NTD rotation (residues 246–277 and 191–206). These results further support a model in which substrate binding results in an NTD rotation. Δ131Δ-induced HtpG protection (blue spheres, Figure 3B) is observed in both the CTD and MD, centered at the base of the dimer cleft. This area contains residues 543–565, which are disordered in the apo crystal structure (dashed lines, Figure 3B). In the isolated CTD structure this region adopts an amphipathic helix, postulated to be involved in substrate interactions 27. Although these results may suggest Δ131Δ binding extends to the CTD dimer cleft, it is not known whether the dominant source of Δ131Δ-induced protection is from an increase in structure of the amphipathic helices or by a direct interaction.

Figure 3
Substrate binding affects hydrogen-exchange patterns across the HtpG structure

To distinguish between these scenarios we measured an 15N HSQC of the isolated CTD alone and with Δ131Δ and observed a small number of binding-induced chemical shifts, suggesting the hydrogen exchange protection has a contribution from a direct interaction (Supplemental Figure 2). As described below we also identify a Δ131Δ binding region on the interior of the dimer cleft on the MD, which suggests that Δ131Δ binding may span both the MD and CTD.

Identifying a substrate-binding region on the middle domain

Previous SAXS measurements suggested an MD binding region, however the measurements were too low resolution to determine a residue-level surface. To gain this insight we next used NMR and mutagenesis. The HtpG middle domain (residues 231–495, 31 kDa) can be purified and 15N labeled for NMR studies, but the HSQC spectrum is rather crowded (Supplemental Figure 3A) and the residue assignments are unknown. Given that the structure of the middle domain is known we explored an approach utilizing selective amino-acid labeling to simplify the process of identifying a Δ131Δ binding location. Specifically, the Volker Doetsch lab introduced a method whereby multiple HSQC spectra are collected each one corresponding to a single type of amino acid being 15N labeled. By counting the number of peaks that are affected by the substrate it is possible to identify one or more patches on the structure that have the correct number of affected residues. Iterating this process with different labeled amino acids can identify a unique region that has correct surface residue composition of the binding site 28. Although this method is advantageous because single amino-acid labeling greatly simplifies the HSQC spectra, it only provides a predicted binding region so the results must be independently tested.

We produced four variants of the HtpG MD specifically 15N labeled on Asp, Phe, Tyr, and Gly residues. These residues have an asymmetric distribution over the MD suggesting the potential to uniquely identify a binding region. Δ131Δ affects both the chemical shifts and intensities for a subset of the labeled residues on the middle domain, an example with Asp is shown in Figure 4A–C. We measured chemical shifts and intensity changes, normalized them and defined their mean and standard deviation. The peaks that were significantly impacted were counted by defining a significance threshold for each amino-acid type (see Methods).

Figure 4
Mapping a substrate binding location on the HtpG middle domain

This process identified 2 Phe, 3 Tyr, 2 Asp, and 1 Gly, in the predicted binding region (Supplemental Figure 3 B–I). Inspection of the MD shows a patch facing into the HtpG dimer cleft in the apo state with this surface residue distribution (Figure 4D), whereas the opposite face shows no such site (Supplemental Figure 3J). As a test, we mutated three residues within this patch (positive predictions: W467A, F390A, D476K) and three analogous mutations outside of this patch (negative predictions: W224A, F257A, E369K). We included charge reversal mutations because the strong salt dependence of Δ131Δ binding suggested an electrostatic contribution.

Using a previously described fluorescence polarization binding assay with IAEDANS-labeled Δ131Δ we measured the binding Kd of these variants (wild-type HtpG has a Kd of 9 μM). The hydrophobic truncations have significantly reduced binding (W467A: 42 μM, F390A: 38 μM) while the negative predictions are minimally affected (W224A: 11 μM, F257A: 10 μM), which confirms that Δ131Δ binds to the HtpG interior cleft at the MD. Both the positive and negative prediction charge reversal mutations show an intermediate reduction in binding (D476K: 19 μM, E369K: 21 μM), one explanation may be that long-range electrostatic interactions between HtpG and Δ131Δ contribute to binding. This would be consistent with the significant difference in pI between HtpG (5.1) and Δ131Δ (9.5) and also consistent with recent studies of unfolded citrate synthase binding to Hsp90 29. However, if long-range electrostatics are playing a role, then neither mutation (D476K and D369K) is a reliable test for direct binding. Therefore we used the W467A and F390A variants to investigate the relationship between substrate binding and HtpG conformational changes.

First, as a control, we tested the impact of these mutations on Δ131Δ-induced conformational changes in HtpG. For reference, our previous SAXS measurements revealed that substrate binding is coupled to large-scale conformational changes of HtpG both under apo and AMPPNP conditions 13, therefore we expected that by disrupting substrate binding these large conformational changes should be reduced. The Δ131Δ-induced conformational changes in HtpG are visibly evident in the contracted SAXS P(r) spectrum, which reflects the combined set of scattering distances within the complex. In contrast to wild-type HtpG, the W467A mutant has a significantly reduced conformational change from Δ131Δ under both apo (Figure 5A) and nucleotide conditions (inset), confirming that the reduction in substrate binding is directly translated in a reduction in the chaperone conformational response. The W467A mutation itself does not affect the HtpG conformation (Supplemental Figure 4A). Similar results were observed for the F390A mutation, although the mutation itself resulted in a change to the conformational state of HtpG (not shown).

Figure 5
Cross-monomer substrate contacts are coupled to HtpG conformational changes

A secondary set of cross-monomer substrate contacts activate HtpG

A crucial mechanistic distinction concerning Δ131Δ activation of HtpG is whether the substrate activates from within a monomer or across monomers and whether there are single or multiple contacting regions of the substrate. As discussed below, we addressed these questions in three ways, (i) a heterodimer analysis with the W467A variant and NM FRET measurements (ii) HtpG monomer exchange measurements and (iii) by studying different fragments of Δ131Δ.

Beyond identifying a substrate binding region in greater detail, the W467A mutation provides an opportunity to form heterodimers of HtpG where one arm contains the NM FRET pair, and the opposite arm is either wild-type HtpG or the W467A mutant. This type of heterodimer experiment has been used previously to identify cross-monomer determinants of Hsp90 hydrolysis rate 20; 25 and cochaperone activation 16. As discussed earlier, under apo conditions Δ131Δ affects an NTD rotation in the wild-type heterodimer (one arm NM FRET, second arm wild-type HtpG) as seen from a concentration-dependent loss of acceptor fluorescence (black circles, Figure 2C). In contrast, for the W467A heterodimer (one arm NM FRET, second arm W467A), Δ131Δ only has a modest impact on the NM FRET (blue squares, Figure 2C). This result shows that substrate binding at the MD of one arm is directly coupled to the NTD rotation on the opposite arm. Since the heterodimer has a wild-type MD on the FRET labeled monomer, Δ131Δ should make a modest acceleration of closure by its impact on the opposing NTD. Indeed, in contrast to the five-fold acceleration of AMPPNP-mediated closure observed for wild-type HtpG, Δ131Δ only accelerates the W467A heterodimer by a factor of two (Supplemental Figure 4B). W467A heterodimers have a similar intrinsic closure rate as the wild-type heterodimers (not shown).

A second test for cross-monomer contacts is that Δ131Δ should slow the rate of HtpG monomer exchange. Here we used a FRET-based assay developed in the Buchner lab 16; 21. In this experiment Hsp90 heterodimers are labeled with donor and acceptor fluorophores on opposite arms, with a resulting FRET signal that can be extinguished by adding an excess of unlabeled wild-type Hsp90 (shown schematically in Figure 5B). Here the loss of acceptor fluorescence occurs because monomer exchange randomizes fluorescently labeled monomers with unlabeled monomers. We observe a striking slowdown of monomer exchange kinetics in the presence of Δ131Δ (red circles, Figure 5B), corroborating that cross-monomer substrate contacts are formed.

There are two models that could explain cross-monomer substrate contacts. The first model is that there is a single dominant substrate-binding region that spans the Hsp90 monomers. The second model is that binding is predominantly contained within a monomer with secondary substrate contacts that span the monomers. To discriminate between these models we investigated a limited construct that only contains the dominant binding region of the substrate. For reference, previous measurements with Δ131Δ suggested that there was a dominant binding region of ~25 residues around position 100 in Δ131Δ, therefore we synthesized a 30 residue peptide corresponding to residues 87–116 in Δ131Δ. If cross-monomer contacts are due to secondary substrate contacts, then this limited construct will not rotate the NTD or slow monomer exchange. Indeed, although the peptide binds (Kd of 40 μM) it is unable to rotate the NTD (green diamonds, Figure 2C), does not change the monomer exchange rate (Figure 5C) and has a minimal impact on the closure kinetics (not shown). SAXS measurements under apo conditions show that HtpG still contracts upon binding the peptide (Supplemental Figure 4C), indicating an alteration of the MD/CTD interface.

Although these results strongly suggest that multiple regions of the substrate are required to make cross monomer contacts, rotate the NTD, accelerate closure, and subsequently activate HtpG, a potential confounding factor could be that the peptide is either misfolded or does not have a sufficient level of structure to activate the chaperone. Given that previous studies on Δ131Δ demonstrated that the region around residue 100 has significant structure 30; 31; 32, this was a possibility we wanted to explore in detail. Therefore we performed NOESY measurements (Supplemental Figure 5A) on the peptide alone. We assigned the peptide using standard methods involving TOCSY and NOESY comparisons and a natural abundance 13C-1H HSQC (see Methods). Inspection of the pattern of non-local NOEs shows many i-i+3 and i-i+4 NOEs in the peptide region of the native α-helix in the wild-type structure (residues 98–107, Supplemental Figure 5B). Weak long-range NOEs suggest modest tertiary organization. Using NOE distance constraints and dihedral constraints based on Cα/Cβ chemical shifts with the DANGLE program33, we determined an ensemble of compatible structures with the ARIA program 34 (Supplemental Figure 5C). The peptide structure ensemble reveals a helical region centered on the native α-helix and adjacent N-and C-terminal loops that loosely interact. The central helical region is not present in all members of the ensemble, indicting that the helix is a folding equilibrium, consistent with early Δ131Δ studies 30. This result shows that the dominant substrate region recognized by Hsp90 has a moderate level of structure and that the peptide is not misfolded.

Finally, we wanted to investigate the location of the secondary contacts on the substrate. As shown here Hsp90 binds to a locally structured region of Δ131Δ around residue 100, and our results show that secondary contacts from this region are required for cross-monomer contacts that affect an NTD rotation, which primes the chaperone for ATP-driven closure. These findings are consistent with our previous NMR measurements that suggested a secondary binding site at the Δ131Δ C-terminus 13. To test whether the C-terminal region indeed makes secondary contacts, we investigated a Δ131Δ variant in which the C-terminal 30 residues are removed (residues 111–141). Indeed, this construct only results in a modest slowdown of HtpG monomer exchange (Figure 5B), and has a minimal impact on the closure kinetics (not shown). Although these results do not exclude a synergistic contribution from the Δ131Δ N-terminal region, secondary contacts from C-terminal region of Δ131Δ clearly play a central role in Hsp90 activation.


The Hsp90 ATPase is required for in-vivo function 35; 36 and is regulated by numerous co-chaperones in eukaryotes 37. Previous work with the model client Δ131Δ demonstrated that substrate binding can also regulate the activity of HtpG 13, similar to reports of human Hsp90 activation by the ligand binding domain of the glucocorticoid receptor 14 and an E. coli ribosomal protein L2 that activates HtpG 15. Elucidating the Hsp90 functional mechanism requires an understanding of how the chaperone can be activated by substrates. Here we have focused on NTD structural motions in the HtpG conformational cycle, identifying a client binding region on the MD, and establishing that multiple regions of the substrate make cross-monomer contacts required for HtpG activation (Figure 6).

Figure 6
Model of NTD rotation in substrate activation of Hsp90 ATP hydrolysis cycle

This proposed activation of HtpG by Δ131Δ has a parallel with the mechanism of the activating cochaperone Aha1 on the yeast Hsp90 16. Those authors demonstrate that cross-monomer Aha1 contacts prime Hsp90 for closure and subsequent hydrolysis, while FRET measurements suggested that Aha1 could affect NTD orientation 21. There is a second interesting parallel between our HtpG/Δ131Δ findings and cochaperone-stabilized Hsp90 conformations. Recent electron microscopy measurements of human Hsp90 demonstrated that the cochaperone Hop (involved in substrate loading from Hsp70) induces a partial closure of Hsp90 and fully rotates both NTDs into a closure-competent orientation 38. Previous measurements on Δ131Δ showed that substrate binding induces a partial closure of HtpG in the apo state 13, similar to the Hop-stabilized conformation. These similarities suggest that Hop stabilizes an Hsp90 conformation that is naturally predisposed for substrate binding, subsequent chaperone closure, and ATP hydrolysis.

Comparison of our results with an electron microscopy reconstruction of an Hsp90-cdc37-cdk4 complex 17, reveals differences between these substrates. Modeling suggested the kinase substrate, cdk4, binding both the NTD and MD on a single monomer, in contrast to the cross-monomer contacts we observe. In the Hsp90-cdc37-cdk4 complex the substrate-bound NTD is in a closure-competent conformation while the other monomer, bound to cdc37, is hinged away. While the primary Δ131Δ interaction is with the MD, one possibility is that secondary substrate contacts are made to the NTD. This would be consistent with early studies indicating that Hsp90 has two substrate binding sites with different specificities 39, an NTD site that can bind short unstructured peptides and a CTD/MD site that can bind partially folded substrates 40. Given that our HX-MS measurements do not show significant Δ131Δ-induced protection at the NTD, if Δ131Δ is contacting the NTD these contacts are likely transient.

Recent studies have identified that the ribosomal subunit L2 activates HtpG, similar to the effect from Δ131Δ, however L2 is unable to activate the yeast Hsp90 homolog 15. Interestingly, we find that Δ131Δ does not accelerate closure for the yeast Hsp90 but does accelerate closure for human TRAP1, the mitochondria-specific Hsp90 homolog (TOS and LAL unpublished observations). Although there is very strong evidence that the conformational states of Hsp90 are highly conserved, the degree to which substrates accelerate conformational transitions for Hsp90 homologs appears to be variable.

Numerous studies have established a conserved Hsp90 mechanism in which structural rearrangements resulting in NTD dimerization organizes the catalytic machinery required for ATP hydrolysis 3; 9; 21; 25; 41. SAXS and EM studies have shown that rigid-body motions between the NTD/MD and the MD/CTD can account for the wide range of Hsp90 conformational states 7; 8; 9. The extreme apo state flexibility arises from a weak coupling between the MD/CTD, allowing for a wide range of arm-arm geometries. In contrast, here we find that the NTD/MD rotation required for closure is a high-energy state. The closed conformation, although stable, comes with a cost of adopting an unfavorable NTD/MD interface. However, our measurements do not indicate whether the NM rotation has a large kinetic barrier in addition to an unfavorable equilibrium constant, nor do our measurements assess the degree to which the NM rotation is unfavorable, in terms of kcal/mol. Also, it is not known whether Δ131Δ affects a discrete NTD rotation or whether substrate binding weakens the NTD/MD interface resulting in an ensemble of domain orientations. Further studies are needed to address these questions.


HtpG, variants of HtpG, and Δ131Δ were purified as described previously 7; 30. A peptide corresponding to residues 87–116 in Δ131Δ was synthesized (Genemed Synthesis), HPLC purified and confirmed by mass-spectrometry. A similar peptide was synthesized with a C-terminal cysteine and labeled with IAEDANS for fluorescence anisotropy measurements. Hydrogen exchange mass-spectrometry measurements 26 were performed at pH 7.5, 25 mM TRIS, 25 mM KCl, 5 mM MgCl2. The closure, monomer exchange, and NM FRET measurements, were performed at pH 9.0, 25 mM TRIS, 50 mM KCl, 5 mM MgCl2, 25 C.

Fluorescence measurements

Fluorescence anisotropy on IAEDANS-labeled Δ131Δ was measured on a Jobin Yvon fluorometer. Excitation and emission monochromator slits were both set to 5 nm, an integration time of 2 seconds, and excitation/emission wavelengths of 340/480 nm. The NTD/MD FRET pair (S52C/D341C) was labeled with a five-fold molar excess of Alexafluor 647 and Alexafluor 555 (Invitrogen) for three hours at room temperature and quenched with BME. Measurements were performed with the resulting mixture of labeled species with FRET occurring between HtpG labeled with both fluorophores. HtpG heterodimers with 250 nM labeled monomer and 5 μM wild-type HtpG were formed by incubation for 30 minutes at 30 °C. Closure was initiated by 5 mM AMPPNP, either in isolation or with 50 μM Δ131Δ. Excitation and emission monochromator slits were set at 2 and 3 nm, respectively. Monomer exchange measurements were performed with a previously described FRET pair at positions 62 and 341 on HtpG 13. Cross-monomer FRET was measured by forming heterodimers (250 nM of each monomer) by incubation for 30 minutes at 30 °C. A 20x excess of unlabeled HtpG was added and the loss of acceptor fluorescence was measured at 664 nm. Addition of 25 μM Δ131Δ resulted in slower monomer exchange kinetics.

SAXS measurements

SAXS measurements, as described previously, were performed at the Advanced Light Source in Berkeley 7; 8; 20. The concentrations of HtpG, variants of HtpG, and Δ131Δ were 50 μM. To measure closure kinetics by SAXS, closure was initiated by simultaneously adding 10 mM AMMPNP to multiple samples with a multichannel pipette. SAXS measurements were taken at varying timepoints on different samples to avoid radiation damage. The linear combination fitting used to determine the population of closed state has been described previously 7; 8.

NMR measurements

HSQC measurements were performed on a Bruker Avance800. Fully 15N labeled HtpG MD was produced by a 10 mL overnight starter culture, washed in M9 minimal medium, and resuspended in M9 with 1 g/L 15N ammonium chloride and 0.5 g/L isogrow supplement (Sigma). Selectively labeled MD samples were produced by supplying all 14N amino-acids except the labeled 15N amino-acid. These were added in the following quantities in each liter of minimal media (A:500 mg, R:400 mg, D:400 mg, C:50 mg, Q:400 mg, E:650 mg, G:550 mg, H:100 mg, I:230 mg, L:230 mg, K:420 mg, M:250 mg, F:130 mg, P:100 mg, S:210 mg, T:230 mg, Y:170 mg, V:230 mg, N:300 mg, 500 mg of tryptophan was added after autoclaving the media). NMR buffer conditions were 25 mM MES pH 6.0, 25 mM KCl, 5 mM MgCl2.

For each labeled MD sample, chemical shifts and peak intensities were measured in ccpNMR ( For changes in the chemical shifts, shifts in the 1H dimension were normalized in magnitude to the shifts in 15N dimension by multiplying each shift in 1H dimension by a ratio of mean shift changes in 15N over mean shift changes in 1H. After this an overall change in chemical shifts was determined in 2 dimensions, and an overall mean change was found for each spectrum. This mean was subtracted from the chemical shift change for each particular peak, divided by the standard deviation and plotted to generate Supplemental Figure 2, E–H. For changes in the peak intensities, a ratio of intensities of bound versus unbound for each peak were calculated, a mean and standard deviation of all ratios for a pair of spectra were found. To generate Supplemental Figure 2 A–D, the mean ratio was subtracted from the ratio for each particular residue, and then divided by the standard deviation. Residues over a 1.5 σ threshold from the mean for either chemical shift or intensity changes were counted. Although the choice of 1.5 σ was an adjustable parameter, values significantly above and below yielded surface residue compositions for the Δ131Δ binding site that were incompatible with the MD surface.

NOESY measurements on 800 μM peptide were performed on a Bruker Avance800 with a 120 ms mixing time. The spectra were processed with NMRpipe 42 and analyzed using ccpNMR. Structural ensembles were calculated using ARIA 34.

Supplementary Material



We thank Mark Kelly for help with NMR. Funding for this project was provided by the Howard Hughes Medical Institute. TOS was supported by a Damon Runyon Cancer Research Foundation fellowship. Many thanks to members of the Agard lab for helpful discussions.


Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


1. Young JC, Moarefi I, Hartl FU. Hsp90: a specialized but essential protein-folding tool. J Cell Biol. 2001;154:267–73. [PMC free article] [PubMed]
2. Workman P. Combinatorial attack on multistep oncogenesis by inhibiting the Hsp90 molecular chaperone. Cancer Lett. 2004;206:149–57. [PubMed]
3. Ali MM, Roe SM, Vaughan CK, Meyer P, Panaretou B, Piper PW, Prodromou C, Pearl LH. Crystal structure of an Hsp90-nucleotide-p23/Sba1 closed chaperone complex. Nature. 2006;440:1013–7. [PubMed]
4. Dollins DE, Warren JJ, Immormino RM, Gewirth DT. Structures of GRP94-nucleotide complexes reveal mechanistic differences between the hsp90 chaperones. Mol Cell. 2007;28:41–56. [PMC free article] [PubMed]
5. Shiau AK, Harris SF, Southworth DR, Agard DA. Structural Analysis of E. coli hsp90 reveals dramatic nucleotide-dependent conformational rearrangements. Cell. 2006;127:329–40. [PubMed]
6. Krukenberg KA, Bottcher UM, Southworth DR, Agard DA. Grp94, the endoplasmic reticulum Hsp90, has a similar solution conformation to cytosolic Hsp90 in the absence of nucleotide. Protein Sci. 2009;18:1815–27. [PubMed]
7. Krukenberg KA, Forster F, Rice LM, Sali A, Agard DA. Multiple conformations of E. coli Hsp90 in solution: insights into the conformational dynamics of Hsp90. Structure. 2008;16:755–65. [PMC free article] [PubMed]
8. Krukenberg KA, Southworth DR, Street TO, Agard DA. pH-Dependent Conformational Changes in Bacterial Hsp90 Reveal a Grp94-Like Conformation at pH 6 That Is Highly Active in Suppression of Citrate Synthase Aggregation. J Mol Biol 2009 [PMC free article] [PubMed]
9. Southworth DR, Agard DA. Species-dependent ensembles of conserved conformational states define the Hsp90 chaperone ATPase cycle. Mol Cell. 2008;32:631–40. [PMC free article] [PubMed]
10. Bron P, Giudice E, Rolland JP, Buey RM, Barbier P, Diaz JF, Peyrot V, Thomas D, Garnier C. Apo-Hsp90 coexists in two open conformational states in solution. Biol Cell. 2008;100:413–25. [PubMed]
11. Onuoha SC, Coulstock ET, Grossmann JG, Jackson SE. Structural studies on the co-chaperone Hop and its complexes with Hsp90. J Mol Biol. 2008;379:732–44. [PubMed]
12. Street TO, Krukenberg KA, Rosgen J, Bolen DW, Agard DA. Osmolyte-induced conformational changes in the Hsp90 molecular chaperone. Protein Sci. 2010;19:57–65. [PubMed]
13. Street TO, Lavery LA, Agard DA. Substrate binding drives large-scale conformational changes in the hsp90 molecular chaperone. Mol Cell. 2011;42:96–105. [PMC free article] [PubMed]
14. McLaughlin SH, Smith HW, Jackson SE. Stimulation of the weak ATPase activity of human hsp90 by a client protein. J Mol Biol. 2002;315:787–98. [PubMed]
15. Motojima-Miyazaki Y, Yoshida M, Motojima F. Ribosomal protein L2 associates with E. coli HtpG and activates its ATPase activity. Biochem Biophys Res Commun 2010 [PubMed]
16. Retzlaff M, Hagn F, Mitschke L, Hessling M, Gugel F, Kessler H, Richter K, Buchner J. Asymmetric activation of the hsp90 dimer by its cochaperone aha1. Mol Cell. 2010;37:344–54. [PubMed]
17. Vaughan CK, Gohlke U, Sobott F, Good VM, Ali MM, Prodromou C, Robinson CV, Saibil HR, Pearl LH. Structure of an Hsp90-Cdc37-Cdk4 complex. Mol Cell. 2006;23:697–707. [PubMed]
18. Prodromou C, Panaretou B, Chohan S, Siligardi G, O'Brien R, Ladbury JE, Roe SM, Piper PW, Pearl LH. The ATPase cycle of Hsp90 drives a molecular 'clamp' via transient dimerization of the N-terminal domains. Embo J. 2000;19:4383–92. [PubMed]
19. Richter K, Muschler P, Hainzl O, Buchner J. Coordinated ATP hydrolysis by the Hsp90 dimer. J Biol Chem. 2001;276:33689–96. [PubMed]
20. Cunningham CN, Krukenberg KA, Agard DA. Intra- and intermonomer interactions are required to synergistically facilitate ATP hydrolysis in Hsp90. J Biol Chem. 2008;283:21170–8. [PMC free article] [PubMed]
21. Hessling M, Richter K, Buchner J. Dissection of the ATP-induced conformational cycle of the molecular chaperone Hsp90. Nat Struct Mol Biol. 2009;16:287–93. [PubMed]
22. Richter K, Moser S, Hagn F, Friedrich R, Hainzl O, Heller M, Schlee S, Kessler H, Reinstein J, Buchner J. Intrinsic inhibition of the Hsp90 ATPase activity. J Biol Chem. 2006;281:11301–11. [PubMed]
23. Huai Q, Wang H, Liu Y, Kim HY, Toft D, Ke H. Structures of the N-terminal and middle domains of E. coli Hsp90 and conformation changes upon ADP binding. Structure. 2005;13:579–90. [PubMed]
24. Meyer P, Prodromou C, Hu B, Vaughan C, Roe SM, Panaretou B, Piper PW, Pearl LH. Structural and functional analysis of the middle segment of hsp90: implications for ATP hydrolysis and client protein and cochaperone interactions. Mol Cell. 2003;11:647–58. [PubMed]
25. Vaughan CK, Piper PW, Pearl LH, Prodromou C. A common conformationally coupled ATPase mechanism for yeast and human cytoplasmic HSP90s. Febs J. 2009;276:199–209. [PMC free article] [PubMed]
26. Graf C, Stankiewicz M, Kramer G, Mayer MP. Spatially and kinetically resolved changes in the conformational dynamics of the Hsp90 chaperone machine. Embo J. 2009;28:602–13. [PubMed]
27. Harris SF, Shiau AK, Agard DA. The crystal structure of the carboxy-terminal dimerization domain of htpG, the Escherichia coli Hsp90, reveals a potential substrate binding site. Structure. 2004;12:1087–97. [PubMed]
28. Reese ML, Dotsch V. Fast mapping of protein-protein interfaces by NMR spectroscopy. J Am Chem Soc. 2003;125:14250–1. [PubMed]
29. Wayne N, Bolon DN. Charge-rich regions modulate the anti-aggregation activity of Hsp90. J Mol Biol. 2010;401:931–9. [PMC free article] [PubMed]
30. Alexandrescu AT, Abeygunawardana C, Shortle D. Structure and dynamics of a denatured 131-residue fragment of staphylococcal nuclease: a heteronuclear NMR study. Biochemistry. 1994;33:1063–72. [PubMed]
31. Alexandrescu AT, Shortle D. Backbone dynamics of a highly disordered 131 residue fragment of staphylococcal nuclease. J Mol Biol. 1994;242:527–46. [PubMed]
32. Wang Y, Shortle D. The equilibrium folding pathway of staphylococcal nuclease: identification of the most stable chain-chain interactions by NMR and CD spectroscopy. Biochemistry. 1995;34:15895–905. [PubMed]
33. Cheung MS, Maguire ML, Stevens TJ, Broadhurst RW. DANGLE: A Bayesian inferential method for predicting protein backbone dihedral angles and secondary structure. J Magn Reson. 2010;202:223–33. [PubMed]
34. Linge JP, Habeck M, Rieping W, Nilges M. ARIA: automated NOE assignment and NMR structure calculation. Bioinformatics. 2003;19:315–6. [PubMed]
35. Obermann WM, Sondermann H, Russo AA, Pavletich NP, Hartl FU. In vivo function of Hsp90 is dependent on ATP binding and ATP hydrolysis. J Cell Biol. 1998;143:901–10. [PMC free article] [PubMed]
36. Panaretou B, Prodromou C, Roe SM, O'Brien R, Ladbury JE, Piper PW, Pearl LH. ATP binding and hydrolysis are essential to the function of the Hsp90 molecular chaperone in vivo. Embo J. 1998;17:4829–36. [PubMed]
37. Pearl LH, Prodromou C. Structure and mechanism of the Hsp90 molecular chaperone machinery. Annu Rev Biochem. 2006;75:271–94. [PubMed]
38. Southworth DR, Agard DA. Client-Loading Conformation of the Hsp90 Molecular Chaperone Revealed in the Cryo-EM Structure of the Human Hsp90:Hop Complex. Mol Cell. 2011;42:771–81. [PMC free article] [PubMed]
39. Young JC, Schneider C, Hartl FU. In vitro evidence that hsp90 contains two independent chaperone sites. FEBS Lett. 1997;418:139–43. [PubMed]
40. Scheibel T, Weikl T, Buchner J. Two chaperone sites in Hsp90 differing in substrate specificity and ATP dependence. Proc Natl Acad Sci U S A. 1998;95:1495–9. [PubMed]
41. Richter K, Soroka J, Skalniak L, Leskovar A, Hessling M, Reinstein J, Buchner J. Conserved conformational changes in the ATPase cycle of human Hsp90. J Biol Chem. 2008;283:17757–65. [PubMed]
42. Delaglio F, Grzesiek S, Vuister GW, Zhu G, Pfeifer J, Bax A. NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J Biomol NMR. 1995;6:277–93. [PubMed]