Multiple Domains of the Co-Chaperone Hop Are Important for Hsp70 Binding

The Hop/Sti1 co-chaperone binds to both Hsp70 and Hsp90. Biochemical and co-crystallographic studies have suggested that the EEVD-containing C terminus of Hsp70 or Hsp90 binds specifically to one of the Hop tetratricopeptide repeat domains, TPR1 or TPR2a, respectively. Mutational analyses of Hsp70 and Hop were undertaken to better characterize interactions between the C terminus of Hsp70 and Hop domains. Surprisingly, truncation of EEVD plus as many as 34 additional amino acids from the Hsp70 C terminus did not reduce the ability of Hsp70 mutants to co-immunoprecipitate with Hop, although further truncation eliminated Hop binding. Hop point mutations targeting a carboxylate clamp position in TPR1 disrupted Hsp70 binding, as was expected; however, similar point mutations in TPR2a or TPR2b also inhibited Hsp70 binding in some settings. Using a yeast-based in vivo assay for Hop function, wild type Hop and TPR2b mutants could fully complement deletion of Sti1p; TPR1 and TPR2a point mutants could partially restore activity. Conformations of Hop and Hop mutants were probed by limited proteolysis. The TPR1 mutant digested in a similar manner to wild type; however, TPR2a and TPR2b mutants each displayed greater resistance to chymotryptic digestion. All point mutants retained an ability to dimerize, and none appeared to be grossly misfolded. These results raise questions about current models for Hop/Hsp70 interaction.


4
domains complexed with an EEVD-containing peptide (14). One co-crystal contained the N-terminal TPR domain of Hop, TPR1, and the octapeptide GPTIEEVD that corresponds to the Hsp70 C-terminus. The other co-crystal contained one of the central TPR domains, TPR2a, in complex with the pentapeptide MEEVD that corresponds to the C-terminal sequence of Hsp90. Similar to some other reported TPR domains, both of the Hop TPR domains consisted of an anti-parallel alpha helical stack that forms a large groove along one surface of the domain. The EEVD-containing peptides lodged within this groove, but in distinct orientations that related to TPR side-chain differences within the groove and unique amino acids in either peptide. In both co-crystal structures, basic TPR side chains within the groove formed salt bridges with acidic peptide side chains, forming what was termed the carboxylate clamp. Consistent with the TPR2a co-crystal structure, point mutation of a conserved carboxylate clamp position in any of several Hsp90-binding TPR co-chaperones has been shown to disrupt Hsp90 binding (13).  Samples were incubated 30 min at 30°C with brief   vortexing at 5 min intervals to resuspend resin. Resin-bound complexes were washed 3 X 1ml wash buffer, and bound proteins were separated by SDS-PAGE. Gels were stained with Coomassie Blue followed by autoradiography of dried gels. Bands on Xray film were quantitated by densitometry.
To generate the Sti1-minus yeast strain (sti1∆0), the entire coding region of STI1 was replaced with the Schizosaccharomyces pombe HIS5 gene flanked by loxP sites; the His marker was subsequently removed by transformation with a plasmid encoding the Cre recombinase (19). Gene deletion was confirmed by yeast colony PCR, and the absence of Sti1p was confirmed by Western blot analysis with the anti-Sti1 monoclonal antibody ST2 (provided by D. Toft).
As described previously (20), the GR expression strain contained a pG/N795 plasmid constitutively expressing rat GR and the GRE-lacZ reporter plasmid pUC∆SS-8 26X. Wildtype or mutant Hop cDNA was introduced into a yeast expression vector by ligating a HindIII/EcoRV fragment from pSPUTK into the constitutive expression plasmid p425GPD, which contains a LEU2 nutritional marker, a 2µ origin of replication, and a constitutive GPD transcriptional promoter.

Yeast assays for hormone-induced reporter gene expression
Hormone induction assays were conducted as previously described (20). Briefly, yeast strains were grown in selective media at 25°C to an optical density at 600 nm (OD 600 ) of 0.05 to 0.12 units. Growth was monitored by spectrophotometry for 30 min before hormone addition to ensure that the culture was in exponential phase.
Deoxycorticosterone (DOC) was added to the culture at 50 nM final concentration. To assay for β-galactosidase activity 100 µl of culture was withdrawn and immediately added to 100 µl of the Gal-Screen™ substrate (Tropix, Bedford, MA) in 96 well microtiter plates at room temperature. Samples were taken at 10 min intervals until 70 to 80 min after hormone addition. The plate was read in a luminometer 2 hr after the last sample was collected.
To determine the rate of reporter expression, β-galactosidase induction curves were first generated by plotting relative light units (RLU) against the OD 600 of the culture sample. Regression analysis of the linear portion of each data set yielded a best-fit line (typically, R 2 >0.98) whose slope is the growth-normalized rate of β-galactosidase expression.

Yeast cell extracts and Western immunoblots
To prepare whole cell extracts, washed cell pellets were resuspended in cracking buffer (8 M urea, 5% w/v SDS, 40 mM Tris-HCl pH 7.5, 0.1 mM EDTA, 0.04% w/v bromophenol blue) at 4 mL/g of cells. Cells were homogenized with glass beads in a Mini Bead Beater (Biospec Products, Bartlesville, OK). Cell homogenates were centrifuged to remove insoluble material and heated at 95°C for 5 min. Lysate aliquots (10-15 µl) were separated by SDS-PAGE, transferred to PVDF membrane, and immunostained for Sti1p (mouse monoclonal IgG ST-2), Hop (mouse monoclonal IgG F5), or the yeast ribosomal protein L3 (mouse monoclonal IgG anti-L3). and sonicated to generate cell extracts.

Recombinant protein expression and purification
His-tagged proteins were purified by metal affinity chromatography according to manufacturer's instructions (Qiaexpressionist Kit, Qiagen, Valencia, CA). Extracts were applied to Ni 2+ -NTA affinity resin and incubated for 1 hour at 4°C. Resins were washed 10 3 X 1 ml wash buffer (50mM NaH 2 PO 4 , pH 8.0, 300 mM NaCl and 20 mM imidazole).
Protein was eluted at 250 mM imidazole.

GST-Hop pull down assays
Previously purified and quantitated GST-tagged Hop forms were incubated batch-wise with glutathione-Sepharose 4B (5 ug protein plus 20 ul bed volume resin) for 30 min at room temperature on a rocking platform and washed 3 X 1ml with incubation buffer (20 mM Tris, pH 7.5, 100 mM KCL, 5 mM MgCL 2 and 2 mM dithiothreitol). Resin pellets were resuspended in 150 µl incubation buffer plus 10 µg purified Hsp70 for 30 min at 30°C. Resin was subsequently washed 3 X 1 ml ice-cold incubation buffer containing .01% NP40. Bound proteins were eluted with SDS sample buffer, separated on gels and visualized by Coomassie blue staining.

Protease digestion of recombinant Hop forms
Purified His-tagged Hop forms were subjected to partial proteolytic analysis (21) using chymotrypsin (Type VIII, Sigma-Aldrich) or subtilisin Carlsberg (Type VIII, Sigma- to the corresponding Hsp90 domain, a central peptide binding domain (26), and a Cterminal 10-kDa domain (27). Hop contains three TPR domains plus a C-terminal DP domain and a putative central DP domain (further described below). As indicated, TPR1 is thought to bind the C-terminal PTIEEVD sequence of Hsp70 while Hop TPR2a is thought to bind the C-terminal MEEVD sequence of Hsp90 (14).
In Fig. 1B, the TPR2a/MEEVD co-crystal structure is depicted in two alternate views. The TPR motifs in anti-parallel alpha-helices form the peptide binding groove.
Positively charged side-chains of R305 and K301 that form the TPR2a carboxylate clamp are highlighted, as are the negatively charged side chains of the MEEVD peptide.
The TPR1/GPTIEEVD co-crystal shows a similar overall structure, although the carboxylate clamp residues and peptide side-chains are oriented in a unique manner.

Hsp70 C-Terminal Sequences Necessary for Hop Binding
To test directly which C-terminal sequences of Hsp70 are required for Hop binding, a series of C-terminal truncation mutants was prepared ( Fig. 2A), and each mutant was compared with full-length Hsp70 in a Hop co-immunoprecipitation assay

Mutagenic Analysis of Hop Domains
To further explore Hop-Hsp70 interactions, we generated domain-specific point mutants of Hop to verify that TPR1 is required and to test whether other Hop domains participate in Hsp70 binding. Careful sequence analysis of Hop revealed a previously overlooked symmetry between the N-terminal and C-terminal regions of Hop. Figure 3 shows an alignment of TPR1 plus downstream sequences compared with TPR2b and downstream sequences. We have previously shown that sequences downstream from TPR2b contain a DP-repeat motif (underlined sites) that contributes to an independent structural domain (17). Truncation or point mutations within the so-called DP domain 14 lessens binding of Hop to Hsp70 without affecting Hop binding to Hsp90 . As the alignment in Fig. 3 reveals, there is similarity between the C-terminal DP domain and sequences correspondingly positioned downstream from TPR1. We refer to this recently recognized N-terminal DP region as DP1 and the original DP domain as DP2.
Point mutants were generated for each of the three TPR domains as well as the two DP regions. Within the TPR domains, point mutations targeted basic residues in the carboxylate clamp. For TPR2b, whose structure has not been solved, basic amino acids corresponding to the carboxylate clamp positions in TPR1 were mutated (double underlines in Fig. 3). In other Hsp90-binding TPR proteins, point mutation of basic amino acids that form the carboxylate clamp have been shown to efficiently disrupt Hsp90 binding (13,(28)(29)(30). In the present study, two alternative rationales were used for point mutagenesis of carboxylate clamp positions. First, either K or R was converted to glutamic acid, thus reversing the charge at this position. Since the charged side chain projects into the solvent space of the binding groove, we reasoned that reversing the Wildtype and mutant cDNAs were expressed in vitro to generate radiolabeled products. Molar equivalents of each radiolabeled form were added to rabbit reticulocyte lysate (RL) and tested for co-immunoprecipitation with Hsp90 or Hsp70 or tested for assembly with progesterone receptor (PR) complexes ( Figure 4) The ability of Hop mutants to assemble with PR complexes (lower set of panels) mostly corresponded with deficiencies in Hsp90 or Hsp70 binding. Exceptions are TPR2b mutants that fail to bind Hsp70 yet appear to assemble normally with the PR complex. Note that TPR1 mutants retain a partial ability to assemble with PR complexes despite our observation that these mutants bind poorly to Hsp70. Therefore, it is still possible for Hop mutants that are defective in Hsp70 binding to assemble with receptor complexes, presumably reflecting interactions with receptor-associated Hsp90.
Finally, in this particular data set the DP1 mutant D140A appears to assemble less well with PR, but this apparent defect was not reproducible in replicate experiments.
Due to the unexpected finding that point mutants of TPR2a and TPR2b, as well as the TPR1 mutants, failed to co-precipitate with Hsp70 complexes from RL, these mutants were directly tested for Hsp70 binding by a GST-pulldown approach ( Figure 5).
A series of GST fusion proteins was prepared in which glutathione S-transferase was fused to the N-terminus of wildtype Hop or Hop mutants. Either GST alone or one of the GST-Hop constructs was purified on glutathione beads and incubated with purified Hsp70. In a typical experiment (Fig. 5A), Hsp70 associated weakly with GST alone, reflecting the level of non-specific Hsp70 binding. Specific binding of Hsp70 to each Hop form was estimated from densitometric analysis of gel bands from 4 separate experiments after subtracting the level of non-specific Hsp70 (GST alone) from each sample. As seen in a summary of these data (Fig. 5B), Hsp70 bound R305E (TPR2a mutant) equally as well as wtHop, in contrast with reduced co-immunoprecipitation from RL (Fig. 4). However, consistent with the RL co-precipitation results, K73E (TPR1) and K429E (TPR2b) were both defective for binding Hsp70. Therefore, it consistently by guest on March 24, 2020 http://www.jbc.org/ Downloaded from appears that TPR2b along with TPR1 is somehow involved in Hsp70 binding. TPR2a, on the other hand, impacts Hsp70 binding in a context-dependent that is evident only in the more complex RL environment. Since reduced function in this assay could be due to differences in protein expression levels rather than inherent differences in activity, yeast culture extracts were assayed for relative Hop protein levels by Western immunoblotting (Fig. 6C). K73E and R305E were present at levels equal to or greater that wildtype Hop, so the reduced activity of these mutants cannot be attributed to limiting protein. K429E and AP1 mutants also accumulated at wildtype levels, thus the elevated activity of AP1 is not due to overexpression in yeast. AP2 accumulates to about half the level of wildtype Hop, but this modest difference is unlikely to account for the complete loss of activity in the AP2 strain.

Conformational Analysis of Hop Mutants
We have previously reported that AP2 and other mutations in the DP2 region display heightened sensitivity to proteolytic digestion and altered partial proteolytic patterns. As shown in Figure 7, Hop domain mutants were analyzed by partial proteolysis for indications that mutant conformation differs from wildtype Hop.
Purified recombinant proteins were digested with chymotrypsin, which cleaves at positions with large hydrophobic side-chains, or with subtilisin Carlsberg, which cleaves more non-specifically. Trypsin, which cleaves at basic amino acids, was not employed since mutation of basic amino acids in the TPR mutants would confound interpretation of digestion patterns. For subtilisin and chymotryptic digests, 2 to 4 replicate digestions were analyzed for pattern reproducibility and, in the case of TPR mutants, both glutamic acid and alanine substitutions were tested. The fragment patterns obtained were consistent between replicates and between the alternative TPR mutants.
In comparing chymotryptic digestion patterns (Fig. 7A), there were only a few qualitative differences in band patterns that we consistently observed. The most notable difference is the resistance of certain mutant forms to initial cleavage. This is perhaps best illustrated by comparing half-lives of full-length proteins (Fig. 7B). Bands corresponding to full-length protein were quantitated by densitometry and values plotted versus duration of digestion. From these data, the half-life (i.e. time at which one-half the original amount of full-length protein remains) was calculated for each Hop form.
Full-length wildtype was degraded with a half-life of approximately 40 sec. In contrast, the TPR2a and TPR2b mutants (R305E and K429E, respectively) had half-lives of 100-120 sec, indicating that these mutants are more resistant to chymotryptic cleavage. The TPR1 and DP mutants varied less from wildtype.
Hop forms were similarly subjected to limited digestion with subtilisin, but, in contrast to chymotryptic digests, full-length wildtype and mutant proteins were cleaved at similar rates. As an example, the rates of subtilisin digestion for wildtype and R305E were identical as were the major fragment patterns (Fig. 7C) although differences in less abundant fragments were observed. The subtilisin results suggest that overall conformations between mutants are similar while the chymotryptic results suggest a change in the TPR2a and TPR2b mutant conformations such that enzyme access is restricted at one or more particularly sensitive cleavage sites.
Hop is reported to exist in a dimeric state (33,34), although the structural basis and functional significance of dimerization is unknown. To test whether TPR or DP1 mutants might affect the Hop dimeric state, purified proteins were analyzed by gel filtration (Figure 8). All forms eluted with similar retention volumes corresponding to the anticipated dimeric size of approximately 120 kDa. None of the forms displayed a tendency toward aggregation, yet a minor fraction of each sample migrated as a small peak that corresponds in predicted size to a tetrameric complex. The similarity of profiles among all Hop forms suggests that a dimeric state is retained and gross misfolding is not occurring with any mutant.

DISCUSSION
These studies were intended to experimentally address the nature of interactions between Hsp70 and the co-chaperone Hop. Previous findings have suggested that the TPR1 domain of Hop binds to C-terminal sequences in Hsp70 (35,36), and a co-crystal structure for TPR1 complexed with an octapeptide that corresponds to the EEVDcontaining C-terminus of Hsp70 further supported this view (14). However, several observations in the present study suggest that Hop-Hsp70 interactions are more complex and involve additional sequences in both Hop and Hsp70.

The Hop Binding Site in Hsp70
Contrary to expectations from earlier studies, truncation of the Hsp70 EEVD sequence plus as many as 34 upstream amino acids (N608) failed to significantly reduce the ability of mutant Hsp70 forms to bind Hop (Fig. 2) Unfortunately, the GPTIEEVD (639-646) and GGMP-repeat (615-635) portions of the crystallized peptide were not resolved in X-ray diffractions, suggesting at least that these motifs do not form a stable interaction with other C-terminal sequences.
In an attempt to refine localization of minimal C-terminal sequences necessary for Hop binding, we generated N-terminal truncations of Hsp70 that lacked the welldefined ATPase and peptide binding domains, but none of these constructs displayed significant binding to Hop (not shown). The ADP-bound state of Hsp70 is required for efficient binding to Hop (37,38). Therefore, the ATPase domain of Hsp70 somehow communicates with C-terminal sequences to which Hop binds. Although we can conclude that the GGMP-EEVD region is not strictly required for Hop binding, we nonetheless suspect, as discussed below, that direct interactions between the Hsp70 EEVD tail and the Hop TPR1 domain are physiologically relevant.

What roles do Hop domains play in Hsp70 binding?
Consistent with predictions from current models, point mutation of carboxylate clamp residues in TPR1 disrupts Hsp70 binding (Figs. [4][5] and Hop function (Fig. 6) point mutants ( Fig. 4 and 3,17). Furthermore, we report here that carboxylate clamp mutations in TPR2a and TPR2b can also inhibit Hsp70 binding in vitro (Fig. 4-5). The TPR2a mutants are defective in vivo, but this may largely reflect a primary role for TPR2a in Hsp90 binding. The function of TPR2b and a potential ligand for this domain have not been identified, so it was intriguing to observe that carboxylate clamp mutations diminish Hsp70 binding in both complex (Fig. 4) and purified (Fig. 5) systems.
On the other hand, the TPR2b mutants were fully competent for supporting GR function in vivo (Fig. 6) and assembled with receptor complexes in vitro similar to wildtype Hop (Fig. 4).
The newly recognized DP1 domain (Fig. 3) was a compelling target for mutagenesis since the related DP2 domain is important for Hsp70 binding and Hop function. Mutations in this domain did not alter Hsp70 binding (Fig. 4); curiously, though, DP1 mutants displayed a reproducible, albeit modest increase in Hop function in vivo (Fig. 6). These findings do not suggest a critical function for DP1 in Hsp70 or Hsp90 binding, yet neither do they exclude an influence of DP1 on Hop function at some level.
Is there a second Hsp70 binding site in Hop apart from TPR1? Hernandez et al. (38) reported that Hsp70 binds Hop in a 2:1 molar ratio in the absence of Hsp90 and a 1:1 ratio in the presence of Hsp90, which might indicate that Hsp90 binding to Hop occludes a second Hsp70 binding site. In our GST pull-down assay, we did not observe a molar excess of Hsp70 binding to Hop. On the other hand, none of the Hop point mutations reduced Hsp70 binding to background levels, suggesting that there might be separate sites for Hsp70 binding (Fig. 5). In an attempt to detect directly an alternative Hsp70 binding site, various truncation mutants lacking TPR1 were tested, but with each truncated construct we observed only increased levels of non-specific Hsp70 binding, presumably due to misfolding of the truncation mutants (results not shown). We also tested a triple point mutant in which carboxylate clamp positions in each of the three TPR domains were mutated, but the triple mutant bound Hsp70 at the same residual level as either TPR1 or TPR2b single mutants (comparison not shown). Therefore, if there are two Hsp70 binding sites, they are somehow interdependent or one of the sites does not involve carboxylate clamp interactions.

Conformational roles for Hop domains
Based on structural studies of various TPR domains, we did not anticipate that  (Fig. 8), each of the Hop mutants migrated as a homogeneous peak that closely resembled wildtype Hop. From this we conclude that mutants retain their propensity for dimer formation and do not display gross misfolding that would favor aggregation. Still, we did infer conformational differences based on limited chymotryptic digests (Fig. 7). TPR1 and DP1 mutants yielded fragment patterns much like wildtype Hop. In contrast, TPR2a and TPR2b mutants consistently displayed a similar resistance to proteolysis compared with other Hop forms (Fig. 7B). This finding is suggestive of domain-domain interactions involving TPR2a and TPR2b, either between dimeric pairs or within a single Hop polypeptide. Perhaps apparent defects in these mutants for Hsp70 binding in vitro are a consequence of conformational alterations that can be reversed in vivo.

Hsp70/Hop interactions and progressive client maturation
During the progressive assembly of steroid receptor complexes, Hop plays an important, transient role, but little is known about the molecular mechanisms that direct receptor transit from intermediate to mature complexes. ATP hydrolysis and changes in the nucleotide-bound states of Hsp70 and Hsp90 are important, likely because these lead to changes in Hsp conformation and co-chaperone interactions. Hop, whose binding to either Hsp is nucleotide-sensitive and can affect ATPase activities (37)(38)(39), is in position to stimulate and coordinate Hsp structural changes and help translate these into remodeling of receptor complexes.
In order to reconcile our present studies of Hop-Hsp70 interactions with previous reports and to place these findings in the context of receptor maturation, the following speculative model is proffered. We propose that a region of Hsp70 upstream from the EEVD terminus binds TPR1 and that this non-EEVD interaction is critical for initial Hsp70 binding to Hop. Subsequently, within the context of the receptor heterocomplex and ATP hydrolysis, the EEVD tail displaces the upstream Hsp70 site from TPR1. This putative rearrangement of Hop-Hsp70 interactions would in turn stimulate a change in Hsp70 or Hsp90 behavior that promotes re-engineering of the receptor heterocomplex. In this conceptual model, EEVD-TPR1 interaction serves as a secondary regulatory signal rather than a primary binding interface.
Freeman et al. (40) published evidence that the EEVD tail can interact with other regions of Hsp70, so there may be intramolecular interactions of EEVD that change with Hsp70 status such that the EEVD tail becomes available for TPR1 binding in a regulated manner. Structural data on the peptide binding and C-terminal domains of Hsp70 have also suggested domain-domain interactions (27). Recall also that Hop binding is sensitive to the nucleotide-bound state of Hsp70. Therefore, the ATPase domain of Hsp70 communicates with C-terminal sequences of Hsp70 to which Hop binds. In addition to Hop's adaptor role in bringing together Hsp70 and Hsp90 in a common complex, we think our results add a novel structural dimension to a model (38) in which Hop also monitors change in Hsp70 status and stimulates additional change that bears on the chaperone machinery and modeling of steroid receptor complexes.