The function of the yeast molecular chaperone Sse1 is mechanistically distinct from the closely related Hsp70 family

The Sse1/Hsp110 molecular chaperones are a poorly understood subgroup of the Hsp70 chaperone family. Hsp70 can refold denatured polypeptides via a carboxyl-terminal peptide binding domain (PBD), which is regulated by nucleotide cycling in an amino-terminal ATPase domain. However, unlike Hsp70, both Sse1 and mammalian Hsp110 bind unfolded peptide substrates but cannot refold them. To test the in vivo requirement for interdomain communication, SSE1 alleles carrying amino acid substitutions in the ATPase domain were assayed for their ability to complement sse1 ∆ yeast. Surprisingly, all mutants predicted to abolish ATP hydrolysis (D8N, K69Q, D174N, D203N) complemented the temperature sensitivity of sse1 ∆ and lethality of sse1 ∆ sse2 ∆ cells, whereas mutations in predicted ATP binding residues (G205D, G233D) were nonfunctional. Complementation ability correlated well with ATP binding assessed in vitro . The extreme carboxyl-terminus of the Hsp70 family is required for substrate targeting and heterocomplex formation with other chaperones, but mutant Sse1 proteins with a truncation of up to 44 carboxyl-terminal residues not included in the PBD were active. Remarkably, the two domains of Sse1 when expressed in trans functionally complement the sse1 ∆ growth phenotype and interact by coimmunoprecipitation analysis. In addition, a functional PBD was required to stabilize the Sse1 ATPase domain, and stabilization also occurred in trans . These data represent the first structure-function analysis of this abundant but ill-defined chaperone, and establish several novel aspects of Sse1/Hsp110 function relative to Hsp70.


Introduction
Cells respond to heat shock by induction of a specific set of genes that allow them to cope with and recover from the stress. Many of these heat shock proteins (HSPs) function as molecular chaperones, binding unfolded proteins and preventing aggregation or facilitating their refolding (1). One of the most well-studied classes of chaperones is the Hsp70 family. Saccharomyces cerevisiae possesses 14 Hsp70 homologs, distributed in the cytosol, mitochondria, and endoplasmic reticulum (2). Hsp70s promote folding of nascent polypeptides, facilitate translocation of proteins across membranes, and protect the cell from protein-denaturing stresses. All Hsp70s share a common domain architecture consisting of an amino-terminal ATPase domain and carboxyl-terminal peptide binding domain (PBD 1 ) . The PBD is responsible for binding unfolded peptide substrates and is regulated by the nucleotide binding status of the ATPase domain (3).
The PBD of the E. coli Hsp70 homolog DnaK is composed of a series of eight β-strands (β ) that form a peptide-binding cleft, followed by a series of α-helices that form the "lid" (α subdomain) thought to regulate entry and exit of the substrate (4). 4 proteins in vitro but is unable to actively refold them and instead acts as a "holdase," maintaining substrate polypeptides in a folding-competent state (9,10). No endogenous substrates have been identified for the Hsp110s to date, however overexpression of Hsp110 increases thermotolerance in Chinese hamster ovary (CHO) cells (9). More recently overexpression of the Hsp110 family member Hsp105α was shown to suppress protein aggregation and subsequent apoptosis in COS-7 cells expressing the polyglutamine tract-containing truncated androgen receptor (tAR), suggesting a potential role for Hsp110 in the prevention of protein plaque-associated pathologies (11).
SSE1 and its close paralog SSE2 are the Saccharomyces cerevisiae members of the Hsp110 subfamily. SSE1 was identified biochemically as a calmodulin binding protein and genetically as a high copy suppressor of the hyperactive PKA mutant ira1∆ (12,13). Yeast lacking SSE1 are slow growing and slightly temperature sensitive.
Deletion of SSE2 results in no observable growth defects and deletion of both SSE genes was reported to be equivalent to deletion of SSE1 alone (12,13). Mammalian Hsp110 does not complement the deletion of SSE1 in yeast (J. Subjeck, personal communication; data not shown), suggesting distinct cellular substrate or cofactor specificities. In vitro, Sse1 binds denatured luciferase and accelerates its refolding upon addition of yeast cytosol (14,15). Sse1 also participates in Hsp90 signal transduction --its absence leads to the derepression of the yeast heat shock transcription factor, Hsf1, and loss of glucocorticoid receptor (GR) signaling, two established roles for Hsp90 (16). Recently, SSE1 was isolated as a high copy suppressor of a mutated form of the cytosolic Hsp40, Ydj1, possibly due to the participation of both proteins in Hsp90-dependent functions (15). To learn more about this abundant yet ill-defined chaperone, we have undertaken a molecular genetic analysis of Sse1 in baker's yeast. Residues known to be involved in ATP binding and hydrolysis in the Hsp70s were mutagenized in Sse1 and assayed for complementation of sse1∆ phenotypes. Mutants that were able to bind ATP affinity resin also complemented sse1∆, while mutants that did not bind were non-functional in vivo.
Deletion analysis of the carboxyl-terminus revealed that Sse1 can tolerate removal of 44 residues with no effect on function, suggesting that this region is not involved in  β-Galactosidase Assay-Liquid β-galactosidase assays for Hsf1 activity were performed exactly as described (16).

Sse1 ATPase domain mutants are functional-Previous research has shown that the
Hsp70 ATPase activity is essential for function both in vivo and in vitro. Extensive biochemical and genetic studies have been carried out using point mutations in the ATPase domain to determine which residues are important in the ATP binding and hydrolysis cycle of Hsp70s (17)(18)(19)(20)(21). For example, several residues in bovine Hsc70 are important for ATP hydrolysis and mutations caused dramatic decreases in k cat and moderate increases in K m values (20,21). Equivalent mutations in the yeast cytosolic Hsp70, Ssa1, and ER lumenal Hsp70 Kar2/BiP, were shown genetically and biochemically to render the protein non-functional (18,19). Notably, the ATPase domain of Sse1 shares significant sequence homology with other Hsp70s (36% identity with the Ssa1 ATPase domain). In order to study the role of the N-terminal ATPase domain in Sse1 function, we mutagenized residues known to be involved in the ATP binding and hydrolysis cycle of these other Hsp70s. The residues we targeted for mutation were D8N (D10 in bHsc70), K69Q (K71), D174N (E175), D203N (D199), G205D (G201), and G233D (G229). It was reported previously that Hsf1-regulated genes are derepressed in sse1∆ (16). Because SSE1 is a known target gene of Hsf1 we decided to place wild type and mutant SSE1 constructs under control of the heterologous TEF1 promoter (12,24).
Immunoblotting experiments showed that expression levels of these proteins were approximately two-fold higher than that expressed from the native SSE1 promoter (data not shown).
Two different complementation assays were employed to assess function of the Sse1 mutants. In the first, the mutants were assayed for the ability to rescue the lethality 12 of an sse1∆ sse2∆ strain upon loss of a Yep24SSE1 plasmid on 5-FOA, shown in Fig.   1A. It was previously reported that deletion of SSE2 in a sse1∆ strain did not show any additional phenotypes (12). However, our lab recently found that SSE1 and SSE2 constitute an essential gene pair 2 . Surprisingly, we found that the D8N, K69Q, D174N, D203N, and G205D mutants all allowed for the efficient loss of pYep24SSE1. G233D was the only point mutant unable to permit loss of wild type SSE1 in the sse1∆ sse2∆ strain background. A construct consisting of the carboxyl-terminal peptide binding domain (PBD) lacking the ATPase domain was also unable to complement ∆sse1∆sse2 (data not shown). We then assayed for the ability of the mutants to complement the temperature sensitive phenotype of the sse1∆ mutant. Fig. 1B shows that the D8N, K69Q, D174N, and D203N mutants complemented this phenotype while the G205D and G233D mutants did not. All of the mutants are expressed at or near the levels of wild type Sse1 under control of the TEF1 promoter ( Fig 1C).
The ability of the G205D mutant to function as the sole source of SSE1 in sse1∆ sse2∆ at 30˚C and the failure of this mutant to complement the temperature sensitivity of sse1∆ suggested that this was a temperature sensitive allele. This hypothesis was tested by assaying the growth of sse1∆ sse2∆ mutants harboring a plasmid expressing either wild type SSE1 or sse1-G205D at 30˚C and 37˚C. As shown in Fig. 1D, the inability of sse1∆ sse2∆ expressing G205D to grow at 37˚C illustrates that sse1- Wild type Sse1 or Ssa1 produced in E. coli efficiently bound ATP-agarose, and binding to ATP-agarose was significantly more efficient than the non-specific binding to nucleotide-free agarose (Fig. 3A, compare lanes 3 and 4 to 5 and 6, respectively). These results demonstrate that Sse1 binds ATP in the absence of yeast cytosolic components.
This assay was then utilized to determine if the point mutants had altered nucleotide binding capabilities (Fig. 3B and

Sse1 stability is dependent on a functional peptide binding domain-To test the
idea that the PBD may be required for Sse1 stability, a deletion mutant was constructed that disrupted the β-sheet region of the peptide binding domain by removing two of the eight predicted β-sheets. We used an HA-tagged version of Sse1 to aid in immunodetection of the deletion mutant protein, designated HA-Sse1 ∆394-419 (Fig. 6A).
As expected, this mutant was unable to complement sse1∆ phenotypes, whereas wild type HA-Sse1 functioned normally (data not shown). Importantly, like the ∆103 and ∆188 mutants, this mutant was found to be poorly expressed by Western blot analysis ( To begin to localize the stabilization determinants in Sse1, we attempted to express the ATPase domain alone in yeast and found it to be extremely unstable (Fig. 7).
However, we reasoned that co-expression of the peptide binding domain might stabilize the ATPase domain and therefore examined the stability of the two domains expressed in trans. The full length peptide binding domain (PBD) as well as several mutants that removed 25, 50, and 75 residues from the amino terminus of the PBD were co-expressed with the ATPase domain of SSE1 (Fig. 7A). Expression levels of the ATPase domain were very low in all cases except where the full length PBD was co-expressed (Fig. 7B).
In contrast, all of the PBD deletion mutant fragments were readily detectable. Several studies have shown evidence for such communication using intrinsic tryptophan fluorescence or intramolecular suppression of mutations in distinct domains (31)(32)(33)(34)(35). The data presented in Fig. 7 showing that stability of the ATPase domain depends on the peptide binding domain likewise suggest Sse1 interdomain communication and perhaps interaction. We tested this hypothesis by transforming sse1∆ cells with plasmids bearing one or both of the Sse1 domains and testing for complementation of the temperature sensitivity phenotype. The ATPase and peptide binding domains expressed in trans were able to complement the slow growth phenotype of sse1∆ yeast, indicating that these independent domains are able to function similarly to wild type, full-length protein when expressed separately (Fig. 8A). This complementation did not occur when either domain was expressed alone. In addition, trans complementation was poorer at 37˚C, indicating that the trans interaction may be weaker than the cis interaction (data not shown).
The dependence of a full length PBD on ATPase domain stability and the fact that these domains function when expressed in trans suggested an in vivo interaction. This was investigated by performing coimmunoprecipitation experiments. A full length PBD was co-expressed with or without a FLAG-tagged ATPase domain in sse1∆ yeast followed by immunoprecipitation with the M2 anti-FLAG antibody. We found that the PBD coimmunoprecipitated with FLAG-ATPase but did not precipitate when FLAG-ATPase was absent (Fig. 8B), indicating interaction between the two Sse1 domains.
There are two models to account for the functional interaction seen between the Sse1 domains; 1) Sse1 is a monomer and the two domains associate tightly in an intramolecular interaction or 2) Sse1 is predominantly dimeric, with the possibility of intermolecular domain interactions. To distinguish between these possibilities, HA-and FLAG-tagged versions of Sse1 were co-expressed in sse1∆ cells and coimmunoprecipitation experiments were attempted. We were unsuccessful in several attempts to detect HA-Sse1 coimmunoprecipitating with FLAG-Sse1 (data not shown).
Together, these data suggest that Sse1 does not function in a homo-multimeric state in yeast.

Discussion
In this report we present evidence that the yeast Hsp110 homolog, Sse1, functions in vivo by a mechanism distinct from its Hsp70 relatives. Several mutations known to abrogate ATP binding and hydrolysis in Hsp70 were introduced into the ATPase domain of Sse1 and had no effect on in vivo function as assayed by the ability to complement the growth and Hsf1-repression phenotypes of sse1∆ yeast. Non-complementing ATPase point mutants were also unable to bind ATP-agarose affinity resin while functional mutants retained the ability to bind, suggesting that nucleotide binding is critical for Sse1 function. We also showed that the stability of Sse1 depends on an intact peptide binding domain. Mutations in the peptide binding domain significantly destabilized the protein, suggesting a novel auto-chaperoning activity. Unexpectedly, the ATPase domain and PBD expressed in trans were able to complement sse1∆ yeast and were found to interact directly in vivo by coimmunoprecipitation, providing evidence for interdomain communication.
Extensive studies have been performed with Hsp70 ATPase domain point mutants, showing that the cycle of ATP binding and hydrolysis is essential for function.
Mutation of residues in Hsp70 corresponding to Sse1 D8, K69, D174, D203, G205, and G233 were previously demonstrated to be nonfunctional in vivo, in vitro or both.
However, Sse1 mutants D8N, K69Q, D174N, and D203N remained functional in vivo as judged by complementation of multiple sse1∆ phenotypes. An interesting correlation was observed in which mutants that complemented sse1∆ phenotypes were able to bind ATPagarose, albeit less efficiently than wild type. Only mutants that were unable to bind ATP-agarose did not function in vivo, suggesting that Sse1 must bind nucleotide to function. To date there are no published reports of a member of the Hsp110 family hydrolyzing ATP. Our genetic evidence suggests that if Sse1 has the ability to hydrolyze ATP, it likely plays a minor functional role. The potential lack of, or exceedingly weak, ATPase activity is consistent with the Hsp110/Sse1 family's inability to actively refold denatured protein substrates. Hsp110/Sse1 instead acts as a "holdase," allowing for more efficient refolding in the presence of Hsp70 (14). Indeed, Hsp110 binds more efficiently to denatured peptides than does Hsp70 (9). This cooperativity with Hsp70 could reflect the in vivo role of the Hsp110/Sse1 subfamily of chaperones. Hsp110 has been found in a multi-chaperone complex with Hsp70, but the significance of the Hsp110-Hsp70 interaction in vivo remains to be investigated (36). Together these data support a model wherein Hsp110 quickly recognizes a denatured protein substrate, binds it, and acts cooperatively with the "foldase"-competent Hsp70 to effect renaturation.
One of the most intriguing aspects of this study is the observed trans showed that a lethal mutation in the PBD of the yeast mitochondrial Hsp70, Ssc1, could be suppressed by a compensatory mutation in the ATPase domain (32). The same intragenic suppression was found to occur in DnaK, suggesting conservation through evolution (32). However, physical contact has not been directly demonstrated between the ATPase and PBD. In addition, there is no evidence that Ssa1 stability depends on the peptide binding domain. It is feasible that the 28-residue loop region in Ssel, which separates the β and α subdomains and is absent in Hsp70, allows the peptide binding domain to fold back in order to make contact with and "chaperone" the ATPase domain, as depicted in Fig. 9 (10). The in vivo role for this novel type of interdomain communication in a molecular chaperone is unclear, but it may function as a regulatory mechanism to control access of the PBD to substrate. The inherent instability of the ATPase domain raises the possibility that substrates could be targeted for degradation via association with Sse1. However, we do not have data to support such a mechanism, and the stability of intact Sse1 is inconsistent with a significant population of the chaperone being degraded with faster kinetics.
The interdomain contact suggests that Sse1 possibly exists in multimeric complexes, not unlike mammalian Hsc70, BiP, and yeast Kar2, which have all been demonstrated to form homomultimers in the cell (19,37,38). In support of Sse1 existing in one or more cytosolic protein complexes, the PBD was unable to stabilize the ATPase           This study pFLAG-Spe-SSE1G233D This study