hsp110 Protects Heat-denatured Proteins and Confers Cellular Thermoresistance*

The 110-kDa heat shock protein (hsp110) has long been recognized as one of the primary heat shock proteins in mammalian cells. It belongs to a recently described protein family that is a significantly diverged subgroup of the hsp70 family and has been found in organisms as diverse as yeast and mammals. We describe here the first analysis of the ability of hsp110 to protect cellular and molecular targets from heat damage. It was observed that the overexpression in vivo of hsp110 conferred substantial heat resistance to both Rat-1 and HeLa cells. In vitro heat denaturation and refolding assays demonstrate that hsp110 is highly efficient in selectively recognizing denatured proteins and maintaining them in a soluble, folding-competent state and is significantly more efficient in performing this function than is hsc70. hsp110-bound proteins can then be refolded by the addition of rabbit reticulocyte lysate or hsc70 and Hdj-1, whereas Hdj-1 does not itself function as a co-chaperone in folding with hsp110. hsp110 is one of the principal molecular chaperones of mammalian cells and represents a newly identified component of the primary protection/repair pathway for denatured proteins and thermotolerance expression in vivo.

The 110-kDa heat shock protein (hsp110) has long been recognized as one of the primary heat shock proteins in mammalian cells. It belongs to a recently described protein family that is a significantly diverged subgroup of the hsp70 family and has been found in organisms as diverse as yeast and mammals. We describe here the first analysis of the ability of hsp110 to protect cellular and molecular targets from heat damage. It was observed that the overexpression in vivo of hsp110 conferred substantial heat resistance to both Rat-1 and HeLa cells. In vitro heat denaturation and refolding assays demonstrate that hsp110 is highly efficient in selectively recognizing denatured proteins and maintaining them in a soluble, folding-competent state and is significantly more efficient in performing this function than is hsc70. hsp110-bound proteins can then be refolded by the addition of rabbit reticulocyte lysate or hsc70 and Hdj-1, whereas Hdj-1 does not itself function as a co-chaperone in folding with hsp110. hsp110 is one of the principal molecular chaperones of mammalian cells and represents a newly identified component of the primary protection/repair pathway for denatured proteins and thermotolerance expression in vivo.
It has been long recognized that the major heat shock proteins (hsps) 1 of mammalian cells are observed at 28, 70, 90, and 110 kDa (1)(2)(3) and other hsp families, e.g. hsp60 and hsp40, have been subsequently identified. All of these stress protein groups have been intensively studied, excluding the hsp110 species. The cloning of hsp110 from hamster, mouse, yeast, arabadopsis, and a variety of other species has been recently described (4 -11, 29, 30). Moreover, as is the case with the hsp70 family, multiple members of the hsp110 family have also been found in individual organisms (8 -11). These studies indicate that hsp110 is a significantly enlarged and diverged relative of the hsp70 family of proteins but also includes unique sequence components. The notable constitutive expression and stress inducibility of hsp110 is highly suggestive of a major role in unstressed cells as well as in the heat shock response and the expression of thermotolerance (1,2). A description of the heat shock response in eucaryotes is not possible without an understanding of the roles played by this major stress protein.
We describe here an analysis of the characteristics of hsp110, both in vivo and in vitro.

EXPERIMENTAL PROCEDURES
Purification of Recombinant His-tagged HSP110 -cDNA for hsp110 was cloned into pRSET vector (Invitrogen), resulting in introduction of a His 6 -(enterokinase recognition sequence)-Arg-Ser tag to the amino terminus of hsp110 (pRSET-hsp110). pRSET-hsp110 was transformed into Escherichia coli strain JM109(DE3) cells. The transformant containing pRSET-hsp110 was grown at 37°C in LB medium with ampicillin until the OD reached 0.6, when the expression of His-hsp110 was induced by the addition of 0.4 mM isopropyl-1-thio-␤-D-galactopyranoside during further incubation at 30°C for 5 h. Cells were lysed in 20 mM Tris-HCl, pH 7.9, 0.5 M NaCl, 5 mM imidazole, 0.1% Nonidet P-40, using lysozyme treatment and sonication. The lysate was centrifuged at 25,000 ϫ g for 30 min, and His-hsp110 was purified from the supernatant on Ni 2ϩ -nitrilotriacetic acid-agarose columns (QIAGEN, Inc.) following manufacturer's instruction. Briefly, the supernatant was loaded on the column, washed with sonication buffer, washed with wash buffer containing 20 mM Tris-HCl, pH 7.8, 0.5 M NaCl, 60 mM imidazole, 10% glycerol, and hsp110 was eluted with wash buffer containing 500 mM imidazole (instead of 60 mM). The eluent was dialyzed against 20 mM Tris-HCl, pH 7.8, 150 mM NaCl for 48 h and concentrated by sprinkling polyethylene glycol 8000 or by ultrafiltration using Centricon 50. The concentration of proteins was determined using the Bio-Rad protein assay kit.
hsp110 Expression in Tissue Culture Cells-Myc epitope (EEQKLI-SEEDLLR) was added to the COOH terminus of hsp110 using polymerase chain reaction amplification of the fragment from SstI site to the stop codon of hsp110 cDNA. Amplification was performed using primers containing the SstI site, the inserted Myc epitope, and the stop codon. The nucleotide sequence of the amplified region was verified by DNA sequencing. The Myc-tagged cDNA for hsp110 was cloned into pUHD10 -3 (gift from Dr. H. Bujard), which contains the tetracyclinedependent promoter (tet-hsp110). The plasmid tet-hsp110 was cotransfected with the thymidine kinase-hygromycin plasmid into Rat-1-R12 (ATCC) and tetracycline-dependent transactivator 1 (HtTA-1) (ECACC) cells, which are Rat-1 and HeLa cell lines, respectively, transformed with the tetracycline-dependent transactivator. Doxycycline (tetracycline derivative, 2 ng/ml) was added to the culture medium (Dulbecco's modified Eagle's medium ϩ 10% fetal bovine serum ϩ penicillin-streptomycin-neomycin antibiotic mixture). For stable clone selection, 150 g/ml hygromycin and 400 g/ml G418 were included in the culture medium. hsp110-Myc-expressing clones were screened by Western analysis with monoclonal antibody against Myc epitope in the absence of doxycycline. Selected clones were routinely maintained in medium containing doxycycline, hygromycin, and G418.
Thermal Aggregation Experiments-150 nM luciferase (Boehringer Mannheim) or 75 nM citrate synthase (Sigma) alone or with bovine serum albumin, ovalbumin, hsp110, or hsc70 (StressGen, Inc.) were equilibrated to room temperature in 25 mM Hepes, pH 7.9, 5 mM magnesium acetate, 50 mM KCl, 5 mM ␤-mercaptoethanol, and 1 mM ATP (as indicated) followed by incubation at 43°C in a thermostated cuvette. Light scattering by protein aggregation was determined by measuring the increase of optical density at 320 nm with a spectrophotometer. After measuring the light scattering, the samples were transferred to microcentrifuge tubes and centrifuged for 15 min at 16,000 ϫ g at 4°C, and the supernatant and pellet were separated. The samples taken before centrifugation were considered as total protein. Total, supernatant, and pellet fractions were run on SDS-polyacrylamide gel electrophoresis and probed with anti-luciferase (or anti-hsp110) antibody.
Detection of the Interaction between Luciferase and hsp110 by Immunoprecipitation-Luciferase (150 nM) was incubated with hsp110 (150 nM) in buffer used for aggregation experiments at room temperature or 43°C for 30 min and chilled on ice. Anti-hsp110 antibody was added to * This work was supported in part by Public Health Service Grant GM45994. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed. Tel.: 716-845-3147; Fax: 716-845-8389. 1 The abbreviations used are: hsp, heat shock protein; RRL, rabbit reticulocyte lysate; hsc, constitutively expressed hsp70; SSE, yeast stress seventy E family. the luciferase solution, and the buffer was adjusted to radioimmune precipitation buffer (phosphate-buffered saline, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS) containing 1 mg/ml ovalbumin and incubated for 1 h at 4°C followed by incubation with protein A-Sepharose for 18 h at 4°C. The protein A-Sepharose pellet was collected and washed six times, and the pellet was resolved in SDS-polyacrylamide gel electrophoresis and subjected to Western analysis with anti-hsp110 or anti-luciferase antibody.
Survival Assay-Hamster hsp110-transfected Rat-1-R12 cells (described above) were grown in Dulbecco's modified Eagle's medium plus 10% fetal bovine serum and were induced to overexpress hsp110 by removal of 2 ng/ml doxycycline. Cells were cultured in the absence of doxycycline for 2 days with fresh changes of doxycycline-free media twice a day to remove residual drug. After indicated heat exposures, cells were counted and plated at different dilutions. After 10 -14 days, colonies were stained with methylene blue and counted. The uninduced cells were treated identically, except doxycycline was maintained in the culture media. The presence or absence of doxycycline had no effect on the heat sensitivity of the parental Rat-1 cells. HeLa tetracycline-dependent transactivator 1 cells (HtTA-1, described above) were cultured and treated as described for Rat-1 cells.

RESULTS
To assess the role of hsp110 in vivo, we determined the effect of its overexpression on the long-recognized phenomenon of thermotolerance. For this purpose, we established two tissue culture cell lines (HeLa and Rat-1) in which the expression of hsp110 can be selectively controlled by a tetracycline-regulated expression system ("tet-off") in which removal of tetracycline induces its gene expression (12). Fig. 1A shows the expression of hamster hsp110 (induced), with a Myc-tag added to the carboxyl end in the control and heated Rat-1 cells using this system. In the top panel, an antibody reactive with both hamster and rat hsp110 demonstrates the expression of these hsp110 proteins in control and heated Rat-1 cells, whereas in the bottom panel, an anti-Myc antibody demonstrates the expression of hamster hsp110 in control cells. In these in vivo expression studies, the inducible level of hamster plus rat hsp110 in control cells was approximately comparable to the level of induction of rat hsp110 alone after heat exposure (i.e. is physiological). The expression data from the HeLa line were comparable. Hamster hsp110-induced cells were then challenged with potentially lethal heat doses, following which, the number of surviving cells was determined by clonogenicity. The effect of overexpression of hsp110 on the resistance of Rat-1 cells and HeLa cells is presented in Fig. 1, B and C, respectively. It is evident that cells containing exogenous hsp110 were significantly more resistant to heat killing. For comparative purposes, a survival analysis of fully thermotolerant cells arising from a conventional pre-heat treatment is also presented. Overexpression of hsp110 alone is capable of achieving approximately 25-33% full thermotolerant effect (the survival for uninduced Rat-1 cells at the later time point fell below the level of detection). This indicates that in the absence of overexpression of other hsps, hsp110 is still effective in protecting cells against potentially lethal heat exposures.
To better understand how hsp110 may protect cells from thermal shock in vivo, we determined the characteristics of this protein in vitro by utilizing previously applied assays for the analysis of other heat shock proteins and molecular chaperones. For this purpose, we purified histidine-tagged hsp110 from E. coli to homogeneity. Since hsp110 shares sequence similarities with the hsp70 family and since hsp70 proteins have been studied for their abilities to inhibit protein aggregation and promote protein folding (13-15), we similarly examined the ability of hsp110 to perform these chaperoning functions. For aggregation studies, we used luciferase and citrate synthase as model proteins. It is seen in Fig. 2A that hsp110 is efficient in inhibiting the heat-induced aggregation of luciferase in vitro. Most notably, hsp110 was nearly totally effective in inhibiting aggregation as assayed by light scattering when present in a 1:1 molar ratio. This suggests that the interaction of one hsp110 to one denatured luciferase protein was sufficient to maintain solubility as measured in this way. For comparitive purposes, the efficiency of hsc70 in this process was examined in parallel. Consistent with some earlier studies (13) but not others (15), hsc70 was also capable of inhibiting the protein aggregation. However, in this case, total suppression of luciferase aggregation requires the association of 4 hsc70 proteins to 1 luciferase protein compared with the 1:1 ratio obtained with hsp110. Heat-induced aggregation studies using citrate synthase (shown in Fig. 2B) provide very similar data to that presented in Fig. 2A with luciferase. Therefore, in this important molecular chaperoning characteristic, hsp110 functions in a manner similar to hsc70 but is far more efficient in performing this molecular process than is hsc70.
The effect of hsp110 on luciferase aggregation as presented in Fig. 2A was also examined by Western blotting analysis using an antibody against luciferase. Fig. 2C demonstrates that whereas some freshly prepared luciferase is insoluble (pellet) under control conditions, most of the enzyme remains in the supernatant (Spt). However, when heated in the presence of bovine serum albumin, most luciferase is seen to become insoluble (-hsp110). If hsp110 is present during heating at a 1:1 molar ratio with luciferase (ϩhsp110), it is clear that the enzyme remains maximally soluble. In these studies, the presence or absence of ATP had no effect on the outcome, despite the fact that hsp110 contains the consensus sequences for ATP binding characteristic of this family of ATP binding proteins (4,16).
To verify that this protective effect of hsp110 was due to its direct interaction with luciferase, we precipitated hsp110 and examined the coimmunoprecipitation of luciferase. As seen in Fig. 2D, luciferase coprecipitates with hsp110 (bottom panel) only when heated but not when the two proteins were incubated at room temperature. Moreover, luciferase was not coimmunoprecipitated with hsp110 when it was incubated with pre-heated hsp110 at room temperature (data not shown). Therefore, these experiments demonstrate that hsp110 selectively recognizes denatured proteins and prevents their aggregation during heat shock in vitro. That these proteins are heat-denatured is indicated by their loss of enzymatic activity. Additionally, whereas hsp110 inhibited aggregation, it had no effect on the rate of loss of this enzyme activity, nor was activity

FIG. 2. The effect of hsp110 and hsc70 on the inhibition of protein aggregation in vitro.
It is seen that hsp110 and hsc70 inhibit luciferase (Luc, panel A) and citrate synthase (CS, panel B) aggregation and that hsp110 performs this function significantly more efficiently than does hsc70. Bovine serum albumin (BSA) and ovalbumin (OVA) were used as control proteins. Molar ratios are indicated in parentheses at right. Panel C indicates the Western blot analysis of luciferase in the supernatant (Spt) and pellet of control and heat-shocked luciferase with and without hsp110 present. Panel D shows a coimmunoprecipitation (IP) analysis using antibody against hsp110 followed by Western blotting and probing with antibody against luciferase. This demonstrates that antibody against hsp110 coprecipitates heated luciferase but not Luciferase incubated at room temperature.
hsp110 Is an Efficient Holder of Denatured Protein regained by further incubations of the hsp110⅐luciferase complex at room temperature (data not shown). Thus, hsp110 is a potent chaperone in inhibiting aggregation but is incapable of refolding of heat-denatured proteins on its own. This suggests that hsp110 would require the cooperation of other hsps and/or chaperones to refold denatured substrate proteins. To initially address this question, luciferase was heated in the presence of ovalbumin, hsc70, or hsp110 and then added to 60% RRL that has been shown to be an optimal refolding medium (17,18). As seen in Fig. 3A, luciferase heated in the presence of hsp110 regained 70% original activity, whereas luciferase heated with hsc70 regained only 20% original activity (at the same molar ratio, 20 ϫ hsp110 or hsc70 to 1 ϫ luciferase). It is clear from this data that optimal folding requires significant additional hsp110 or hsc70 compared with optimal maintenance of solubility (Fig. 2). This demonstrates that when bound to hsp110, denatured luciferase is maintained in a folding competent state.
Since hsc70 and Hdj-1 alone have been shown to be a folding ensemble, we repeated the above RRL refolding assay using only these specific molecular chaperones. Although less efficient than RRL, Fig. 3B demonstrates that hsp110 functionally interacts with hsc70⅐Hdj-1 to refold heat-denatured luciferase to 25% original activity. Whereas ATP is not required for inhibition of luciferase aggregation, it is necessary for the refolding functions of RRL or hsc70⅐Hdj-1 as has been demonstrated previously (13)(14)(15). Lastly, when the refolding step employs hsp110 (and not hsc70) and Hdj1, no recovery of luciferase activity is obtained when the holding step also uses hsp110 (Fig. 3B). However, when the holding step (i.e. during heating) employs hsc70, a small amount of refolding is achieved by adding hsp110 and Hdj-1, probably reflective of the interaction of hsc70 (already present) and Hdj-1. This indicates that (i) hsp110/hsc70⅐Hdj-1 is a relevant protein folding machine, (ii) that based on Figs. 2, A and B,  and Fig. 3A, hsp110 is notably more efficient than is hsc70 in protecting the denatured protein during the initial heating phase for subsequent refolding and iii), that Hdj-1 does not function as a folding co-chaperone with hsp110. DISCUSSION hsp110 was one of the earliest heat shock proteins described in mammalian cells and has been noted in numerous studies (e.g. Refs. 1-3). In Chinese hamster cells, hsp110 accounts for 0.7% total cell protein after heat shock compared with 3.2% for hsc70 ϩ hsp70 and 1.2% for hsp90 (1). Its expression levels in the Rat-1 and HeLa cells are less distinctive than in Chinese hamster cells; however it remains as one of the major heatinducible proteins in these cell lines. Its constitutive expression in different murine tissues varies widely, with highly significant levels of expression in liver and brain (4,6). Indeed, the expression of hsp110 in brain is comparable to heat-shocked Chinese hamster cells (i.e. making it approximately 0.7% total brain mass). Additionally, like the hsp70 family, the hsp110 family possesses at least three distinct members, each of which is approximately 60% identical in amino acid sequence to the other. These are hsp110, apg-1 (osp94), and apg-2 (9 -11). The cell and tissue expression of apg-1 and apg-2 is less well characterized than hsp110, although in testis and in renal medullary duct cells, apg-1/osp94 has been shown to be highly expressed (9,11). The level of expression of hsp110 and its family members in different cell lines and in murine tissues speaks to a significant role for this class of stress proteins in both the stress response and in the normal functioning of the nonstressed cell.
The recent cloning of the cDNA for this protein from a variety of organisms as diverse as yeast and mammals has shown it to be a large and highly diverged relative of the hsp70 family (4 -11, 29, 30). Secondary structure analysis of hsp110 demonstrates that it exhibits significant similarity to the secondary structure of hsp70 and DnaK, whose structures are well studied by several methods including crystallography. hsp70/DnaK is composed of 1) an amino-terminal ATP binding domain followed by 2) a 100-amino acid ␤-sheet region that has been identified as the peptide binding domain and 3) a carboxylterminal ␣-helical region that is involved in regulation of hsp70/DnaK function and appears as a lid covering the peptide binding domain (19). Analysis of the sequence and secondary structure of hsp110 demonstrates that it appears to have the same basic 1) ATP binding domain, 2) ␤-sheet (peptide binding) domain, and 3) carboxyl end ␣-helical region. However, the ATP binding domain of hsp110 binds and hydrolyzes ATP poorly relative to hsc70 2 (6,16). The ␤-sheet configuration (i.e. the predicted peptide binding region of hsp110) shows some 2 M. Murawski, H. Oh, and J. Subjeck, unpublished data.  Hdj-1 (panel B). The protein present during the heating period and its molar ratio to luciferase is indicated. In B, the molar ratios during heating were 20:1 in each case. Ovalbumin (OVA) was used as a control protein. Whereas hsp110 sustains luciferase in a folding-competent state after heat shock, it cannot refold this protein either alone or in combination with Hdj-1.
sequence homology to corresponding regions in the structure of DnaK and appears to be of similar overall size and organization. 3 The regulatory, ␣-helical "lid" region of hsp110 is similar to that of DnaK but is larger than the corresponding region in DnaK. The predominant structural difference between hsp110 and DnaK/hsp70 is that the ␣-helical lid region of hsp110 is connected by a 100-amino acid "loop" to the ␤-sheet peptide binding domain, a structure that is virtually absent in hsp70 (4) and that is responsible for much of the increased size of hsp110. If the ␣-helical lid plays an important role in peptide binding as suggested (cf. Ref. 19), the interposition of this 100-amino acid loop between it and the peptide binding domain may have significant functional implications that could be related to the differences between hsp110 and hsc70 described here. It is also in the predicted ␣-helical lid region of hsp110 that the conserved sequences, which act as a signature for members of the hsp110/SSE family, reside (4).
The data presented here demonstrate that hsp110 must be added to the list of identified molecular chaperones that can inhibit the aggregation of denatured cellular proteins, a list that includes the small heat shock proteins, hsp90, the immunophilins, p23, as well as members of the hsp70 family (13, 20 -24). However, it is evident from both the aggregation studies of luciferase and citrate synthase (Fig. 2) and from the refolding studies of luciferase with reticulocyte lysate (Fig. 3) that hsp110 is significantly more efficient in maintaining heatdamaged proteins in a soluble, refoldable state than is its evolutionary relative, hsc70. Interestingly, earlier reports are contradictory concerning the ability of mammalian hsp70 to hold denatured protein in vitro. Minami et al. (15) report that hsc70 alone does not inhibit aggregation of luciferase, whereas hsc70 plus Hdj-1 does. Freeman and Morimoto (13) show that hsp70 and hsc70 maintain the solubility of ␤-galactosidase as indicated by native gel electrophoresis, an observation in agreement with the present report. It is also shown here that hsp110 cannot refold luciferase on its own or with the addition of Hdj-1, a well known co-chaperone for folding with hsc70. However, the refolding obtained with rabbit reticulocyte lysate was much faster and higher than by hsc70 and Hdj-1, suggesting that other components in this lysate are required for efficient refolding of the denatured proteins. It is likely that other chaperones and co-chaperones exist, one or more of which could interact in protein folding with hsp110, possibly including a DnaJ homolog other than Hdj-1. Whereas the structural and sequence similarities between hsp110 and hsc70 indicate a similarity in function between these proteins, hsp110 could also play a primary role in the folding pathway without being directly involved in the final folding step. Furthermore, a molecular interaction between hsc70 and hsp110 has been observed in all cell lines and mouse tissues examined as indicated by coimmunoprecipitation studies and can be reproduced in vitro. 4 This suggests that in many instances at least, hsp110 may not function independently of hsc70 (and vice versa) and that the true in vivo physiologic function of these chaperones requires a cooperative interaction with substrate proteins, perhaps utilizing the different peptide binding capacities and ATP binding and hydrolysis characteristics of each. Indeed, since hsp110 is a prominent mammalian hsp (as it also appears to be in other organisms), it may play a primary role for holding the unfolded peptide chain, with hsc70 playing a secondary function in this initial chaperoning step.
That these in vitro chaperoning functions have a physiologic role in vivo is indicated by the thermoresistance/thermotoler-ance studies described in this report. A partial but significant degree of heat resistance is obtained by the overexpression of hsp110 in both HeLa and Rat-1 cells as determined by the cell ability to continue to proliferate and form colonies after severe heat exposures. This strongly suggests that the in vitro properties described above are operative in vivo.
Lastly, the hsp70 family has been shown to contain different cellular compartmental members with grp78 (or BiP) representing the endoplasmic reticulum representative. Likewise, hsp110 has a strong structural homolog resident in the endoplasmic reticulum known as grp170 (16,(25)(26)(27)(28), although this stress protein may also exhibit its own independent properties. Multiple compartmentalization of these large and greatly diverged hsp70-related proteins further underscores the physiological importance of this family, and further studies of hsp110 and grp170 are necessary to understand the full complement of major molecular chaperones and how they interact in mammalian cells.