Novel function of the C-terminal region of the Hsp110 family member Osp94 in unfolded protein refolding

Living cells protect against deleterious effects resulting from heat shock, osmotic shock and environmental heavy metal exposure by expressing members of a specific class of highly conserved stress proteins known as heat-shock proteins (HSPs; also known as molecular chaperones). To date, numerous HSPs have been identified in eukaryotes, eubacteria and some archaea. These HSPs have been implicated in a wide-ranging variety of cellular processes, including protein folding, prevention of protein aggregation, protein transport across membranes and protein degradation (Bukau et al., 2006). Hsp110 (also known as HSPH1 and HSPH2) (Lee-Yoon et al., 1995; Yasuda et al., 1995; Kaneko et al., 1997; Ishihara et al., 1999) and Grp170 (also known as ORP150, HYOU1 and HSPH4) (Lin et al., 1993; Chen et al., 1996; Ikeda et al., 1997), both of which have a molecular mass greater than that of Hsc70/Hsp70 family proteins (Hsc70 is also known as HSPA8), have been identified as Hsp70 superfamily members. Characterizations of functions of Hsp110 and Grp170 have revealed several unique biological and structural properties that are distinct from other Hsp70 family proteins. Similar to Hsp70, both Hsp110 and Grp170 inhibit protein aggregation and mediate the refolding of unfolded proteins (Oh et al., 1997, 1999). Intriguingly, however, Hsp110 and Grp170 each reportedly function as a nucleotide exchange factor in the refolding of unfolded proteins driven by Hsp70 in concert with Hsp40 (herein referring to DNAJB1) (Raviol et al., 2006b; Dragovic et al., 2006; Andréasson et al., 2008a,b, 2010; Bracher and Verghese, 2015). Hsp110 and Grp170 accelerate the dissociation of ADP from Hsp70, which enables ATP to re-bind to Hsp70 and facilitate release of the refolded protein (Dragovic et al., 2006; Raviol et al., 2006b; Andréasson et al., 2008a,b; Bracher and Verghese, 2015).

In terms of molecular structure, Hsp70 superfamily proteins, including the Hsp110 and Grp170 families, share a common architecture comprising an N-terminal ATP-binding domain/nucleotide-binding domain (NBD), an interdomain linker, a β-sheet substrate-binding domain (SBDβ) composed of a β-sandwich, a loop region inserted into SBDβ, and an α-helix domain (SBDα) with a C-terminal extension (Zhu et al., 1996; Liu and Hendrickson, 2007; Cabrera et al., 2019). The SBDβ domain binds unfolded proteins, the affinity of which is regulated by the ATP/ADP binding status in the NBD (Mayer and Gierasch, 2019). A distinctive structural feature of Hsp110 and Grp170 compared with Hsp70 is that the C-terminal region of Hsp110 and Grp170 is larger than that of Hsp70 due to the long acidic loop region inserted into the SBDβ domain (Oh et al., 1999; Easton et al., 2000; Liu and Hendrickson, 2007; Schuermann et al., 2008; Cabrera et al., 2019). The presence of these unique structural features suggests that the Hsp110 and Grp170 families may have evolutionarily acquired novel biological properties that differ from those of the Hsp70 family. Moreover, both Hsp110 and Grp170 exhibit cancer vaccine and immunoadjuvant activities, indicating potential roles for these proteins in cancer immunotherapy (Wang et al., 2001, 2003, 2006; Park et al., 2006; Gao et al., 2008). These vaccine and immunoadjuvant activities of Hsp110 and Grp170 reportedly involve the C-terminal region including the loop region, SBDβ posterior to the loop region, SBDα, and the C-terminal extension (Wang et al., 2001; Park et al., 2006), although the detailed molecular mechanisms remain to be elucidated.

Recent studies have indicated that dysregulated protein quality control plays a role in various neurodegenerative diseases, such as Alzheimer's disease (AD), Parkinson's disease (PD) and amyotrophic lateral sclerosis (ALS), leading to aggregation and accumulation of disease-specific misfolded proteins (Tanaka and Matsuda, 2014; Lim and Yue, 2015; Ciechanover and Kwon, 2017; Hetz and Saxena, 2017). Protein quality control components and systems, including molecular chaperones, endoplasmic reticulum (ER)-associated degradation, the unfolded protein response, the ubiquitin-proteasomal system and the autophagy-lysosome system, contribute to the degradation of harmful misfolded/unfolded proteins, suggesting that dysfunction in the systems has a significant effect on cell function and survival (Balchin et al., 2016; Pohl and Dikic, 2019). In the ubiquitin-proteasomal degradation system, unfolded proteins are initially recognized by molecular chaperones, such as Hsp70, in a process known as ‘molecular triage’. The unfolded proteins are then either refolded to their native conformation by folding co-chaperones or targeted for proteasomal degradation via ubiquitylation in a process mediated by co-chaperones such as C-terminus of Hsp70-interacting protein (CHIP; also known as STUB1) (Arndt et al., 2007; Kästle and Grune, 2012). The Hsp70-mediated protein refolding process requires Hsp40 and Hsp110 as cooperative folding co-chaperones. Minami et al. (2000, 2006) used a rabbit reticulocyte lysate (RRL) to study the refolding reaction of thermally denatured luciferase (Luc) bound by Hsp90 (herein referring to HSP90AA1) with cooperative involvement of Hsc70 (also known as HSPA8) and Hsp40. They found that the α and β subunits of PA28 (also known as 11S REG), a 20S immunoproteasome activator, mediate the refolding process by transferring thermally denatured Luc from Hsp90 to Hsc70. PA28 is induced by interferon (IFN)-γ and has three distinct isoforms, α, β and γ (also known as PSME1, PSME2 and PSME3, respectively), which bind to and activate the 20S immunoproteasome core (Rechsteiner et al., 2000; Cascio, 2014). The PA28α and PA28β isoforms have four α-helices (Knowlton et al., 1997) that form structural features similar to the SBDα of Hsp110 (Liu and Hendrickson, 2007; Schuermann et al., 2008). In a recent study using PA28α and PA28β-overexpressing model mice, Adelöf et al. demonstrated that PA28α and PA28β may exhibit a chaperone-like function rather than activation of the 20S immunoproteasome (Adelöf et al., 2018).

In our previous studies, we isolated Osp94 (osmotic stress protein 94 kDa, also known as Apg-1 or HSPA4L), a member of Hsp110 family, from mouse renal inner medullary collecting duct (mIMCD3) cells (Rauchman et al., 1993). Expression of Osp94 was found to be up-regulated by hypertonicity (Kojima et al., 1996) and heat shock (Santos et al., 1998), and we identified tonicity- and heat shock-response elements in the enhancer region of the Osp94 gene (Kojima et al., 2004). Osp94-knockout mice exhibited a defect in renal development, particularly in the inner medullary region, that was associated with increased susceptibility to osmotic stress and fertility defects related to reduced sperm motility (Held et al., 2006). However, details regarding other biological functions of Osp94, such as chaperone activity, remain unclear. Amino acid sequence similarities with Hsp110 suggest that Osp94 contains the conserved domains, such as the NBD, interdomain linker, SBDβ, loop region inserted into SBDβ, SBDα and C-terminal extension, but the functions of these domains in Osp94 are poorly understood. Members of the mammalian Hsp110 family have a longer C-terminal region than yeast homologs of these proteins, such as Sse1 (Easton et al., 2000). Regarding the function of the C-terminal region of Osp94, based on a number of reports, we hypothesized that the loop region with β-strand of SBDβ posterior to the loop region (L domain), and SDBα with C-terminal extension (H domain) (collectively termed the LH domain in this study) may function like PA28 in protein refolding processes. For example, several reports have indicated that the Hsp110 and Grp170 families have unique functional and structural features (Wang et al., 2001; Park et al., 2006; Liu and Hendrickson, 2007; Schuermann et al., 2008). This hypothesis is also influenced by the studies of Minami et al. (2000, 2006) and Adelöf et al. (2018) and structural similarity between PA28α and PA28β (Knowlton et al., 1997) and the SBDα of the yeast Hsp110 protein Sse1 (Easton et al., 2000; Liu and Hendrickson, 2007).

In the present study, we first examined whether Osp94 exhibits chaperone activity. We then explored the novel role of the C-terminal region of Osp94 in the refolding of unfolded proteins. We report here for the first time that Osp94 inhibits the aggregation of thermally denatured proteins as a holdase and also functions in the refolding of denatured proteins in cooperative fashion with Hsc70 and Hsp40. As is the case with PA28α, the LH domain of Osp94 appears to be involved in moving unfolded proteins to Hsc70 during the refolding of Hsp90-bound unfolded proteins, a role that is distinct from the nucleotide exchange function previously described for Hsp110 family members.

The inhibitory effect of Osp94 on thermal aggregation of luciferase (Luc) was investigated using Osp94 expressed in Escherichia coli (Fig. 1A). As shown in Fig. S1, antibody against Osp94 was available for western blotting and immunoprecipitation experiments. As shown in Fig. 1B, western blotting revealed that most Luc remained in the supernatant after a centrifugation of the Luc solution after storage at 4°C (Sup, upper panel). However, after incubation of the Luc solution alone at 39°C, a large amount of Luc was detected in the pellet fraction, with only a minimal amount detected in the supernatant (Fig. 1B, pellet, middle panel). In contrast, Luc remained in the soluble fraction when subjected to heat treatment in the presence of Osp94 (Fig. 1B, Sup, lower panel).

To determine whether the observed Osp94-mediated inhibition of Luc aggregation was due to direct interaction with Luc, co-immunoprecipitation was carried out using anti-Osp94 antibody. As shown in Fig. 1C(I.), after incubation of a mixture of Luc and Osp94 at 39°C, Luc co-precipitated in greater abundance with Osp94 compared to proteins co-incubated at 4°C. In addition, re-probing of the same nylon membrane used for western blotting with anti-Osp94 antibody revealed a similar Osp94 signal intensity under both temperature conditions [Fig. 1C(II.)], indicating that the higher Luc signal shown in Fig. 1C(I.) was due to direct interaction between Luc and Osp94 rather than a significant difference in the amount of antibody added under the different temperature conditions.

Osp94-mediated inhibition of heat-induced aggregation of Luc and citrate synthase (CS) was assayed using a light-scattering method, which is generally used to evaluate the chaperone activity of heat-shock proteins. The optical density of samples of Luc alone (Fig. 1D,E) and CS alone (Fig. 1F,G) increased following heat-induced denaturation at 39°C. However, Hsc70 (positive control) and Osp94 suppressed the increase in optical density of Luc and CS by 25–70% (Fig. 1D,F) and 40–80% (Fig. 1E,G), respectively.

RRL in combination with Hsp70, Hsp40 and Hsp90 reportedly reactivates thermally denatured proteins; thus, 60% RRL is typically used as an appropriate material source in refolding assays (Nimmesgern and Hartl, 1993; Schumacher et al., 1994; Minami et al., 2000). We therefore employed an RRL system to assay refolding and determine whether Osp94 requires other heat-shock proteins to reactivate thermally denatured Luc. As shown in Fig. 2A, following incubation of thermally denatured Luc in the presence of RRL and ATP but without Osp94, no reactivation of Luc was observed (Luc alone). By comparison, thermally denatured Luc maintained in the soluble state by Osp94 exhibited a 30–50% reactivation in the presence of RRL and ATP (Luc:Osp94). In contrast, bovine serum albumin (BSA) had no significant effect on the in vitro reactivation of thermally inactivated Luc (Luc:BSA). These results suggest that the functional cooperation of other heat-shock proteins, such as Hsp70 and Hsp40, known to be present in RRL, is required for the reactivation of thermally denatured Osp94-bound proteins.

Hsc70 and Hsp40 reportedly mediate the refolding of denatured proteins in a cooperative manner (Freeman and Morimoto, 1996; Minami et al., 1996). To further evaluate the results shown in Fig. 2A, we examined Hsc70 and Hsp40 rather than RRL as cooperative factors in the protein refolding process. As shown in Fig. 2B, Osp94-bound Luc was reactivated in the presence of Hsc70, Hsp40 and ATP (Osp94+Hsc70+Hsp40+ATP), with refolding activity similar to that shown in Fig. 2A. No significant reactivation of Luc was observed in the absence of ATP (Osp94+Hsc70+Hsp40−ATP). Interestingly, thermally denatured Luc incubated in the absence of Hsc70 and Hsp40 was not reactivated (Osp94−Hsc70−Hsp40+ATP). These results indicate that Hsc70 or Hsp40, or both, are functionally required for refolding of denatured proteins retained in a folding-competent state by Osp94. These results also indicate that ATP is essential for refolding and that Osp94 functions as a holdase more similar to Hsp110 than Hsc70.

Binding of ATP by Osp94 was evaluated using ATP-agarose and compared with binding to other HSPs. Fig. 3 compares serial elution profiles of Osp94, Hsc70, Hsp40, Bip (also known as HSPA5) and Hsp90 following incubation of Neuro-2a cell lysate with agarose beads containing either ATP [ATP (+)] or no ATP [ATP (−)]. ATP at a high concentration [ATP (+)] excludes proteins that specifically bind to ATP on ATP-agarose. Comparing the eluates from both experiments, there are no significant differences in the eluates with the exception of column 5, indicating that proteins in the eluates of columns 1 to 4 non-specifically bound to ATP-agarose. Incubation of Neuro-2a cell lysate with ATP-agarose resulted in the binding of Osp94 to ATP-agarose under normal and heat shock conditions, as well as Hsc70 and Bip, whereas Hsp40 and Hsp90 did not bind ATP-agarose [ATP (−), lane 5]. Incubation of cell lysate in buffer containing ATP [ATP (+), lane 5] resulted in no obvious binding of the proteins to ATP-agarose, indicating that Osp94 binds ATP.

To identify the domains of Osp94 that mediate the observed disaggregation and reactivation of thermally denatured Luc, Osp94 deletion mutants were generated on the bases of similarity to the sequence of Hsp110 (Oh et al., 1999; Easton et al., 2000) (as determined using PredictProtein) and used to examine the inhibitory and reactivation activity of each Osp94 mutant in in vitro luciferase aggregation assays. Fig. 4A shows a schematic illustration of the Osp94 deletion mutants and SDS-PAGE of the purified Osp94 deletion mutants with Coomassie Brilliant Blue (CBB) staining and western blotting are shown in Fig. 4B. The aggregation assay (Fig. 4C) revealed that most Luc remained soluble in the supernatant when the BLH deletion mutant of Osp94 was present, whereas LH domains did not prevent aggregation as Luc was present in the pellet fraction. Incubation with the ABL deletion mutant of Osp94 resulted in an equal amount of Luc in both the supernatant and pellet fractions.

The effects of the Osp94 mutants on heat-aggregation of Luc were then examined in greater detail. As shown in Fig. 4D, LH did not protect Luc from heat-induced aggregation evident as increased light scattering. ABLH inhibited the heat-induced aggregation of Luc to a greater degree than ABL. In contrast, BLH and ABLH exhibited comparable disaggregation activity. These results suggest that the B domain is required to maintain denatured proteins in the soluble state, and the H domain is required to elicit the full disaggregation activity of Osp94. Our data also indicate that the A domain is not essential for Osp94-mediated inhibition of Luc aggregation. These results were consistent with the western blotting results shown in Fig. 4C.

As shown in Fig. 4E, BLH was able to reactivate thermally inactivated Luc. In contrast, neither ABL nor LH mutants could reconstitute Luc activity, indicating that the H domain is critical for full reactivation (i.e. refolding) of thermally denatured proteins bound by the B domain of Osp94 (Fig. 4E). In the case of Hsc70/Hsp70, the SBDα domain functions as a ‘lid’ to retain denatured substrates, in concert with the SBDβ domain, when ADP replaces ATP in the nucleotide binding site (Moro et al., 2004). As the Osp94 H domain harbors the SBDα domain, the H domain might hold denatured Luc with greater strength, leading to more efficient reactivation of Luc.

Our results demonstrated that the Osp94 H domain participates partially in the disaggregation of denatured Luc and fully in its reactivation (Fig. 4). We therefore examined the role of the L domain (i.e. the loop region and the SBDβ C-terminal to the loop region) in the chaperone activity of Osp94 using BLH deletion mutants. As shown in Fig. 5C, all of the mutants significantly suppressed the aggregation of thermally denatured Luc, indicating that the Osp94L domain is not necessary for retaining thermally denatured proteins in the soluble state. Interestingly, the refolding activity of the BLH domain appeared to decline as the L domain became shorter (Fig. 5D). These results indicate that the L domain is required for the full reactivation of denatured Luc bound by the BLH domain in the refolding machinery composed of RRL and ATP.

As described in the Introduction, we hypothesized that the L and H domains of Osp94 function in a similar manner to PA28. Prior to addressing this hypothesis experimentally, the following conditions were optimized: inactivation temperature, duration of Luc heat denaturation (Fig. 6A) and RRL concentration (Fig. 6B) for the refolding of Hsp90-bound Luc. Subsequent experiments on refolding of Luc bound by Hsp90 were then experimentally assessed under the conditions of Luc inactivation at 39°C for 5 min and RRL concentration of 5%, based on these results. Activation of denatured Luc increased linearly up to 80 min at 30°C (Fig. 6C; Hsp90+RRL+Hsc70+Hsp40+ATP). In contrast, omission of one of the three components (RRL, Hsc70 or Hsp40) resulted in either less or no recovery of Luc activity (Fig. 6C). These data indicate that RRL and Hsc70 are required for full refolding of thermally denatured Luc bound by Hsp90, and that Hsp40 plays at least some role in the reactivation process. The apparent requirement for RRL in the reactivation of Hsp90-bound Luc suggests that PA28 in the RRL functions as a third factor in the refolding process, as reported by Minami et al. (2000). We therefore further investigated the role of the LH domain under the same experimental conditions as those used in the experiment shown in Fig. 6C. Interestingly, adding the LH domain instead of RRL to the refolding reaction resulted in a significant increase in Luc activity compared to the reaction without RRL (Fig. 6D). As demonstrated by Minami et al. (2000), PA28α increased Luc activity (Fig. 6E). In terms of components likely related to the refolding process in RRL, western blotting identified PA28α, Hsc70, Hsp40 and Hsp90 (Fig. 6F).

Minami et al. (2000) demonstrated the involvement of PA28 in the refolding of Hsp90-bound Luc by analyzing the transfer of the biotin tag from biotin-labeled Luc to proteins that interact with Luc. We also examined whether the LH domain functions like PA28 in the refolding of Hsp90-bound Luc. PA28α was examined because both the α- and β-subunits are functional (Minami et al., 2000). As shown in Fig. 7, Hsp90-bound Luc interacted with Hsc70 during the early time points (∼5 min), as demonstrated by the increase in biotin labeling of Hsc70, and the earlier transfer of biotin to Hsc70 was consistent with the results reported by Minami et al. (2000). Interestingly, the LH domain was labeled with the biotin tag in addition to PA28α and Hsp40 (Fig. 7A,B). However, the H domain alone, lacking the L domain, did not interact with Luc (Fig. 7C). In contrast, ovalbumin as a control exhibited no clear biotin labeling (Fig. 7D). These results indicate that the LH domain binds to and/or interacts closely with Luc, suggesting that the LH domain plays a role similar to that of PA28α in the refolding of Hsp90-bound Luc. In addition, these results indicate that the L domain is required for the interaction with Luc. Minami et al. (2000) also identified PA28 in the refolding complex. To further verify the presence of the LH domain in the refolding complex formed by Luc, Hsc90, Hsc70 and Hsp40, we performed an immunoprecipitation analysis using anti-Osp94 antibody. The LH domain was detected in the refolding complex for up to 30 min after onset of the refolding reaction (Fig. 7E), suggesting that the LH domain functions in a similar manner to PA28α in the refolding of Hsp90-bound Luc, in conjunction with Hsc70 and Hsp40.

To investigate the PA28-like function of the LH domain, particularly the H domain, in more detail, we examined the disaggregation and reactivation of thermally denatured Luc using Osp94 chimeric proteins (ABL–PA28α and BL–PA28α) in which the Osp94 H domain was swapped with the full PA28α sequence. ABL–PA28α (1:2), but not PA28α, inhibited the aggregation of thermally denatured Luc by ∼50% (Fig. 8B). Moreover, ABL–PA28α, but not PA28α, reactivated thermally denatured Luc in the presence of Hsp70, Hsp40 and ATP to a level almost equal to that produced by the BLH domain (Fig. 8C). In addition, the effect of ABL–PA28α on Luc reactivation was comparable to that of wild-type Osp94, as demonstrated in Fig. 2B. Furthermore, the Hsp90-bound Luc refolding activity of ABL–PA28α and BL–PA28α was comparable to that of PA28α and the LH domain (Fig. 8D). These results confirm that the H domain of Osp94 functions in a similar manner to PA28α in the refolding of Luc bound by Hsp90.

Most HSPs, including Hsc70/Hsp70, reportedly inhibit the aggregation of denatured proteins by functioning as ‘holdases’. HSPs also play a critical biological role as ‘foldases’ in mediating the refolding of denatured proteins (Mogk and Bukau, 2017). Several HSPs and cofactors that function cooperatively in the complex machinery to ensure efficient refolding have been identified (Mogk et al., 2018). Indeed, Hsc70 and Hsp40, an Hsp70 co-chaperone, reportedly function in a cooperative manner in the refolding complex (Freeman and Morimoto, 1996; Minami et al., 1996). Among HSP molecular chaperones, the biological functions of the high-molecular-mass HSPs, including Hsp110 and Grp170, are poorly understood. In this study, we characterized the chaperone activity of Osp94, an Hsp110 family member, and explored the novel role of the Osp94 C-terminal region in the refolding process driven by Hsc70–Hsp40.

Oh et al. (1999) reported that Hsp110 inhibits Luc aggregation and reactivates Hsp110-bound Luc in the presence of RRL and/or a combination of Hsc70, Hsp40 and ATP. In the present study, we also found that Osp94 inhibits thermally denatured Luc and CS, and reactivates thermally denatured Luc bound by Osp94 under the same experimental conditions as those used by Oh et al. (1999) (Figs 1 and 2). In addition, thermally denatured Luc was not refolded in reactions lacking Hsc70, Hsp40 and/or ATP (Fig. 2B). These results indicate that Osp94 recognizes, binds to, and then maintains thermally denatured Luc in soluble and re-folding or folding-competent states in an ATP-dependent manner, and that Osp94 alone cannot directly refold thermally denatured Luc. Various structural studies have reported that yeast Hsp110 homolog Sse1 does not exhibit refolding (Liu and Hendrickson, 2007; Polier et al., 2010; Xu et al., 2012) or ATPase activities (Shaner et al., 2006), indicating that Hsp110 is a holdase but not a foldase like Hsc70. Thus, our results indicate that Osp94 is a holdase similar to Hsp110; therefore, it is Hsc70 and Hsp40 that predominantly drive the refolding of Osp94-bound denatured proteins. In the refolding process mediated by Hsc70–Hsp40–Hsp110 machinery, the ATPase activity of ATP-binding Hsp70 is accelerated by the binding of Hsp40 synergistically with substrate protein (Kityk et al., 2018), facilitating strong retention of the denatured protein in the substrate-binding cavity formed by the SBDβ and SBDα α-helical lid of ADP-binding Hsp70 (Moro et al., 2004). In turn, the nucleotide exchange factor Hsp110 binds Hsp70 via its NBD and SBDα domains (Andréasson et al., 2008a,b; Polier et al., 2008), which induces a conformational change in the NBD of Hsp70, leading to rebinding of ATP to the NBD, opening of the binding cavity and release of the substrate protein (Dragovic et al., 2006; Raviol et al., 2006b; Andréasson et al., 2008a,b, 2010; Bracher and Verghese, 2015). This proposed refolding mechanism is unable to explain the result shown in Fig. 2B, because Luc bound by Osp94 has to associate with the Hsc70–Hsp40 refolding machinery. Thus, our results further suggest that in the refolding shown in Fig. 2B, Luc bound on the B domain (SBDβ prior to the loop region) of Osp94 is transferred to Hsc70, which is followed by Hsp40 recruitment, ADP/ATP exchange mediated by Osp94 and the release of refolded Luc. The interaction between Osp94, Hsc70 and Hsp40 during the in vitro refolding reaction was supported by the in vivo interaction of these proteins in heat shock-stressed mIMCD3 and Neuro-2a cells, which accumulate denatured proteins intracellularly (Figs S2 and S3). The results of our immunocytochemical and immunoprecipitation analyses demonstrate that Osp94 binds directly to Hsc70, but not Hsp40, in vivo, consistent with previous studies reporting that Hsp110 binds to Hsc70 (Andréasson et al., 2008a,b; Polier et al., 2008) and that Hsp40 interacts with Hsc70/Hsp70 to promote the hydrolysis of ATP bound by Hsc70 (Kityk et al., 2018).

The binding of ATP to the NBD, a common structural feature of members of the DnaK/Hsp70 family, is crucial for the protein holding and refolding cycle driven by DnaK/Hsp70 (Zhu et al., 1996; Mayer and Gierasch, 2019). Like Hsc70, we found that Osp94 bound ATP under normal and heat stress. The binding of Osp94 to ATP-agarose in normal cells seemed to be weaker than that in heat-treated cells, as detected by the signal enhancer solution (Fig. 3). Raviol et al. (2006a,b) reported that Apg2 (also known as HSPA4 and HSPH2), a mammalian Hsp110 family member, showed temperature-dependent conformational changes as assessed by circular dichroism spectroscopy and the emission spectrum of tryptophan. Thus, high temperature may be a factor that augments the binding affinity of Osp94 to ATP through the structural alteration, although further studies are required to confirm this.

Here, we undertook analyses of Osp94 domain deletion mutants to functionally characterize Osp94 structural domains. The NBD can prevent the heat-aggregation of Luc (Fig. 4D). In contrast, the B domain was essential for maintaining denatured Luc in the folding-competent state (Fig. 4D), whereas the H domain was required for the full denatured-Luc holding activity of Osp94 (Fig. 4D). The H domain, by comparison, was found to be essential for refolding (Fig. 4E). Thus, it is likely that both the B and H domains of Osp94 mediate the efficient holding of denatured Luc. Furthermore, the H domain appears to participate in the subsequent refolding reaction mediated by Hsc70 and Hsp40.

Although several studies have examined the functional aspects of the Hsp70 interdomain linker as a conformational switch that facilitates allosteric communication between the NBD and SBDβ (Swain et al., 2007; Zhuravleva et al., 2012; English et al., 2017), the chaperone activity functional role of the acidic long loop region of Hsp110 family members, which is a prominent and distinctive structural feature compared with DnaK/Hsp70, remains poorly understood. A previous report showed that an Hsp110 mutant lacking the L domain retained the ability to inhibit Luc aggregation (Oh et al., 1999). The results of our study suggest that the L domain of Osp94 is not necessary for disaggregation of heat-denatured Luc (Fig. 5C), and this suggests that SBDβ posterior to the loop region is not largely involved in the disaggregation of heat-denatured Luc as opposed to the β-sheet bundle of SBDβ prior to the loop region (i.e. the B domain). Our results also show that there is reduced reactivation of heat-denatured Luc in Osp94 mutants lacking the L domain (Fig. 5D). Another recent study revealed that the regions of Apg2, including the loop and the SBDβ posterior to the loop, act as a conformational switch that facilitates optimal dissociation of the Hsc70 and Apg2 complex (Cabrera et al., 2019). Thus, the L domain of Osp94 may be required to assist in the efficient refolding mediated by Hsc70 and Hsp40, rather than in facilitating efficient retention of denatured proteins.

Similar to Hsp40, Hsp90 also plays a role in the refolding reaction as a co-chaperone of Hsp70 (Genest et al., 2011, 2015; Nakamoto et al., 2014; Kravats et al., 2018). In addition, Hsp90 binds to ADP-bound Hsp70 (Kravats et al., 2018) via the Hsp40-binding region of Hsp70 (Kravats et al., 2017). When examining the refolding of Hsp90-dependent bound proteins, Minami et al. (2000) found that PA28 is essential for the refolding of Hsp90-bound Luc in conjunction with Hsc70 and Hsp40. They also found that PA28 mediates the transfer of Luc from Hsp90 to Hsc70 in the refolding machinery complex with Hsp40, based on the transfer of biotin from biotin-tagged Luc bound by Hsp90 to PA28 and Hsc70 in the Hsc70–Hsp40-mediated refolding reaction (Minami et al., 2000). To address our question as to how the Osp94 LH domain facilitates efficient refolding, we evaluated the function of the LH domain under the same experimental conditions utilized by Minami et al. (2000). Interestingly, our study revealed that the LH domain of Osp94, like PA28α, is involved in the refolding and reactivation of thermally denatured Luc bound by Hsp90 (Fig. 6D,E). Cross-linking analyses further demonstrated an association between the LH domain and Luc as well as PA28α in addition to the inability of the H domain to interact with Luc during the refolding of Hsp90-bound Luc (Fig. 7). These results indicate that the LH domain of Osp94 functions like PA28, which transfers an Hsp90-bound misfolded protein molecule to Hsc70 in the chaperone machinery with Hsp40 (Minami et al., 2000), and that the Osp94 L domain plays a pivotal role in mediating the interaction with Luc. Taken together, these results suggest that in the reactivation of Luc demonstrated in Fig. 2B, Luc bound by the B domain of Osp94 was transferred to Hsc70 in the Hsc70–Hsp40 refolding machinery via the LH domain, which is why the L domain has a critical function. Importantly, these results also suggest that there is functional similarity between the LH domain of Osp94 and PA28α in the refolding of Hsp90-bound Luc. Crystallographic analyses of yeast Hsp110 (Schuermann et al., 2008) and mammalian PA28α (Rechsteiner et al., 2000) indicate that there are structural similarities between SBDα in the H domain of Osp94 and PA28α. Therefore, we further evaluated the disaggregation and refolding activities of chimeric proteins in which PA28α, rather than the H domain, was connected to the C-terminus of the Osp94 L domain. Interestingly, our results indicate that PA28α functioned in the chimeras in a similar manner to the Osp94 H domain in the refolding process. The molecular mechanism underlying the reactivation of denatured Luc as examined using the chimeric proteins can be explained as follows – Luc bound by the B domain of ABL–PA28α (Fig. 8C) and by Hsp90 (Fig. 8D) is passed to Hsc70 (possibly to the SBDβ) via PA28α in each chimera. Additional experiments also suggested that in the Hsp90-dependent Luc refolding process, ABLH (i.e. full-length wild-type protein) itself acts like PA28α, like BLH in Fig. 8D (Fig. S4). These results indicate that the H domain of Osp94 is functionally equivalent to PA28α; therefore, the H domain functions to transmit misfolded proteins to Hsp70 during Hsc70–Hsp40-mediated refolding when misfolded proteins are bound by Osp94 via its SBDβ. Importantly, the Osp94 L domain, possibly via the interaction with Hsc70, is essential for this process. The transfer of an unfolded protein molecule from one chaperone to another has been previously demonstrated by the transfer of Hsp70-bound Luc from Hsp70 to Hsp90 in the Hsp70–Hsp90-mediated refolding process including Hsp40 and Hop as co-factors (Wegele et al., 2006). In addition, analyses of the structure of Hsp110 revealed that SBDα does not function as a lid to retain an unfolded protein (Schuermann et al., 2008); thus, the role of Hsp110 SBDα remains to be fully elucidated. The C-terminal extension that follows the SBDα of mammalian Hsp110 is longer than that of yeast Hsp110 (Easton et al., 2000), but its role is poorly understood. The C-terminal extension of the small molecular chaperone αB-crystalline, which is unstructured and highly flexible, reportedly plays a critical role in mediating oligomeric assembly via intermolecular interactions with the α-crystallin domain (Jehle et al., 2010). Thus, the results of the present study suggest a protein-transmitting role for the H domain bearing the SBDα and the C-terminal extension, although further studies are needed to confirm this possibility.

In the protein quality control process within cells, the functional association between chaperones and the proteasomal system is critical for the degradation of potentially harmful unfolded proteins via specific proteins, such as CHIP ubiquitin ligase, which interacts with Hsc70/Hsp70–Hsp90 via its C-terminal tetratricopeptide repeat (TPR; EEVD) domain (Arndt et al., 2007; Kästle and Grune, 2012). Dysfunction of these systems, which leads to the accumulation of unfolded proteins, has been linked to the development of various neurological diseases, including AD, PD and ALS (Tanaka and Matsuda, 2014; Lim and Yue, 2015; Ciechanover and Kwon, 2017; Hetz and Saxena, 2017). However, the mechanism underlying the molecular recognition and transfer of unfolded proteins to the proteasome system is poorly understood. Thus, further explorations of the relationship between chaperones and the proteasome system in proteostasis are of significance to the development of pharmacological therapies to prevent or reverse neurodegeneration. Recent research has demonstrated that nucleotide exchange factors play a pivotal role in the proteasomal degradation of misfolded proteins (Abildgaard et al., 2020). For example, the nucleotide exchange factor BAG-1 reportedly interacts with the proteasome via its ubiquitin-like domain (Alberti et al., 2002). Moreover, direct binding of the yeast Hsp110 protein Sse1 to the 19S regulatory particles of the 26S proteasome plays a role in promoting the targeting of Hsp70-bound misfolded protein substrates for proteasomal degradation via recruitment of ADP-binding Hsp70 with substrates by proteasome-bound Sse1 (Kandasamy and Andréasson, 2018). These reports indicate that the direct structural interaction between nucleotide exchange factors and the proteasome effectively accelerates the proteasomal degradation of misfolded proteins. Recent structural studies using X-ray crystallography and NMR spectroscopy have enriched our knowledge regarding the molecular mechanisms underlying ATP hydrolysis in the Hsc70 NBD (Kityk et al., 2018), the structural dynamics of ATP- and ADP-binding Hsp70 (Bertelsen et al., 2009; Kityk et al., 2012), and molecular interactions between chaperones (Andréasson et al., 2008a,b; Polier et al., 2008; Schuermann et al., 2008; English et al., 2017; Genest et al., 2011; Nakamoto et al., 2014; Genest et al., 2015; Kravats et al., 2017, 2018). Although Osp94 has a TPR-like motif (MEVD) in the C-terminus, whether this motif plays a role in the interaction with Hsc70 remains unclear. Additional structural studies could provide insights into the functional significance of the transfer function of the Osp94 LH domain, including the TPR-like motif, in the Hsc70-Hsp40 refolding machinery during proteostasis. Further functional analyses of the Osp94L and H domains could also shed light on the molecular mechanisms underlying the immunomodulatory activity of the L and H domains of Hsp110 (Wang et al., 2001; Park et al., 2006).

In conclusion, the results of the present study indicate that Osp94 acts as a holdase to inhibit the aggregation of thermally denatured proteins and plays a role in the refolding of denatured proteins as part of a protein complex that includes Hsc70 and Hsp40. Similar to PA28α, the C-terminally located LH domain of Osp94 functions in the transmitting of misfolded proteins bound by Hsp90 and/or Osp94 to Hsp70 in the Hsc70–Hsp40 refolding machinery. In addition, the Osp94 L domain, which is a unique interdomain distinct from that of Hsc70, appears to play a pivotal role in the transmitting of denatured proteins via interaction with Hsc70. The present study therefore sheds light on a novel function of the C-terminal region of the mammalian Hsp110 family member Osp94.

mIMCD3 cells (RRID: CVCL_0429; Rauchman et al., 1993) and Neuro-2a cells (RRID: CVCL_0470, ATCC CCL-131, ATCC, Manassas, VA, USA) were grown in plastic dishes (10-cm diameter) in Dulbecco's modified Eagle's Medium/Ham's F12 (DMEM/F12, Wako, Tokyo, Japan) (1:1) and Eagle's minimum essential medium (EMEM, Wako), respectively, supplemented with 10% fetal bovine serum (HyClone, GE Healthcare Life Science, Chicago, IL) and 2% penicillin-streptomycin (Wako), and non-essential amino acids (Thermo Fisher Scientific, Waltham, MA) for Neuro-2a cells.

Subconfluent cells were washed twice with ice-cold phosphate-buffered saline (PBS), scraped from the dish, and centrifuged at 4000 g for 4 min. Pelleted cells were treated with lysis buffer consisting of 20 mM Tris-HCl pH 7.5, 150 mM NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA, 50 mM NaF, 1 mM Na₃VO₃, 0.5% Nonidet P-40, 20 mM β-glycerophosphate, 1 mM PMSF, 5 mM dithiothreitol (DTT) and protease inhibitor (Roche Diagnostics, Indianapolis, IN), placed on ice for 10 min, and then centrifuged at 4000 g for 5 min. The supernatant was collected, and protein content was measured using the Bradford assay (Bio-Rad, Hercules, CA) with BSA as the standard. The cell lysate was subjected to 7.5% or 10% SDS-PAGE under denaturing conditions and electroblotted onto a PVDF membrane (Millipore, Bedford, MA). The membrane was blocked for 2 h in 3% skimmed milk in Tris-buffered saline (TBS; 20 mM Tris-HCl pH 7.0, 150 mM NaCl) and then incubated overnight at 4°C in blocking solution with appropriate antibodies, as described below. The membrane was then washed three times with TBS with 0.1% Tween 20 (TBST), incubated for 1 h with a horseradish peroxidase-conjugated donkey anti-rabbit IgG (RRID: AB_772191, 1:3000, GE Healthcare Life Science), and washed with TBST. Signals were detected using enhanced chemiluminescence reagent (GE Healthcare Life Science). For intensifying of protein signal, immunoreaction enhancer solution (‘Can Get signal’, TOYOBO, Osaka, Japan) was also used.

A cell lysate was prepared as described above, and after centrifugation, the supernatant (cytoplasm fraction) was collected, divided into two portions, and an appropriate antibody (3 μg) was added to one portion. The antibody-containing cytoplasm fraction was incubated at 4°C overnight, after which Protein G beads (Dynabeads, Thermo Fisher Scientific) were added to each fraction, incubated at 4°C for 2 h, and then collected using a magnetic apparatus. The beads were then washed five times with lysis buffer, centrifuged (4000 g for 2 min), and the resultant pellet was boiled for 5 min after addition of 1× SDS sample buffer, followed by SDS-PAGE and western blotting using appropriate antibodies. Another portion of the cytoplasm fraction containing non-immune IgG (3 μg) instead of specific antibody was used as a negative control for nonspecific signals. For heat-shock experiments, mIMCD3 and Neuro-2a cells were incubated at 42°C for 2 h and then 37°C for 4 h, after which the cytosol fraction was obtained and prepared for immunoprecipitation as described above.

Osp94 cDNA (Gene ID: 18415, ACC#: D49482; Kojima et al., 1996) was used as a template for PCR assays to generate glutathione-S-transferase (GST)-tagged Osp94. PCR amplification to prepare wild-type Osp94 was performed using the following primers: sense, 5′-GTCGACTCATGTCGGTGGTGGGCATTGA-3′ (SalI); and antisense, 5′-GCGGCCGCTTAGTCCACTTCCATCTCTC-3′ (NotI). The resulting PCR fragment was ligated into the TA-cloning vector pCRII-TOPO (Thermo Fisher Scientific) and sequenced. The insert was digested using SalI and NotI, and the resulting fragment was inserted directionally into the pGEX-6P-1 vector (GE Healthcare Life Science) upstream of the GST coding sequence to generate GST-tagged Osp94 fusion protein. In separate experiments, possible structural domains of Osp94 were predicted based on the Hsp110 secondary structure and proposed folding (Oh et al., 1999; Easton et al., 2000; Liu and Hendrickson, 2007). These analyses suggested that the Osp94 residues 1–374, 375–505, 506–600 and 601–838 constitute the NBD with interdomain linker, SBDβ, loop region inserted into SBDβ, and SBDα with C-terminal extension, respectively. As shown in Fig. 4A, we named these the A, B, L and H domains for NBD with the interdomain linker, SBDβ prior to the loop region, the loop region with SDBβ posterior to the loop region, and SBDα with C-terminal extension, respectively.

Constructs to generate ABL, BLH and LH domain Osp94 deletion mutants were prepared by PCR using the primers shown in Table S1. Each respective PCR fragment was inserted into the pGEX-6P-1 vector after TA-cloning, sequencing, and restriction enzyme digestion. For additional experiments, the L domain portion of the BLH domain was partially (BΔLH, BΔΔLH) and fully deleted (BH) using the primers shown in Table S2, and each construct was processed as described above. All GST-tagged constructs were used for the purification of the respective recombinant proteins as described below.

Plasmid constructs were transformed into E. coli BL21 (DE3) cells, which were cultured at 37°C until reaching an optical density at 600 nm of 0.6, at which time 0.5 mM isopropyl-1-thio-β-D-galactopyranoside was added, and the cells were cultured for an additional 5 h at 30°C. The cells were harvested by centrifugation at 5000 g for 15 min, suspended in PBS containing 0.5 mM DTT and protease inhibitor cocktail (Roche Diagnostics), and lysed using a French press (FRENCH Pressure Cells, Thermo ELECTRON, Waltham, MA) at 8400 psi. The resulting cell extract was centrifuged at 100,000 g (Type 50.2 Ti, Beckman Coulter) for 1 h at 4°C, and the supernatant was subjected to further purification. In accordance with the manufacturer's instructions, the supernatant was incubated with glutathione Sepharose 4B resin (GE Healthcare Life Science), washed with PBS, incubated with PreScission protease (GE Healthcare Life Science) to cleave the GST tag, and then recombinant Osp94 protein was eluted using PBS. The eluent was concentrated using an Amicon Ultra-4 30K ultracentrifugal filter (Millipore), desalted using a Micro Bio-Spin column (Bio-Rad), and the protein content was determined as described above. Purified proteins were subjected to SDS-PAGE, and the gel was stained with Coomassie Brilliant Blue G-250 (CBB, Bio-Safe Coomassie, Bio-Rad). The molecular mass and purity of purified recombinant proteins were confirmed by western blotting.

Thermal aggregation was assessed according to the method reported by Oh et al. (1997). Briefly, 160 nM Luc (Sigma, St Louis, MO) or 75 nM CS (Sigma) alone or in combination with BSA, Hsc70 (bovine Hsc70, Sigma), wild-type Osp94, and/or the respective Osp94 domain deletion mutants were incubated at the indicated molar ratios for 30 min at 39°C in buffer composed of 25 mM HEPES-KOH (pH 7.9), 5 mM magnesium acetate, 50 mM KCl, 5 mM β-mercaptoethanol and 1 mM ATP. Protein aggregation was monitored by measuring the apparent absorption at 320 nm resulting from light scattering using an Ultrospec 3100 pro UV/visible spectrophotometer (GE Healthcare Life Science) equipped with a temperature-controlled cuvette holder.

Luc (160 nM) was incubated with wild-type Osp94 and/or each Ops94 domain deletion mutant in the buffer used for the thermal aggregation assay at 39°C or 4°C for 30 min. Before centrifugation, aliquots were removed as total protein, and the samples were centrifuged at 16,000 g for 15 min at 4°C to generate soluble (supernatant) and aggregate (pellet) fractions. The total protein, supernatant, and pellet fractions for each sample were then subjected to SDS-PAGE and western blotting using rabbit polyclonal anti-Luc antibody (RRID: AB_2335880, 1:1000, Promega, Madison, WI).

Luc refolding was analyzed according to the methods described by Nimmesgern and Hartl (1993) and Schumacher et al. (1994). Briefly, Luc (160 nM) alone and in combination with BSA, wild-type Osp94, and/or the respective Osp94 domain deletion mutants were incubated at the indicated molar ratios for 30 min at 39°C in the same buffer used for the thermal aggregation assay. For the refolding reaction, aliquots of each solution were diluted 10-fold with refolding buffer A, composed of 25 mM HEPES-KOH (pH 7.5), 5 mM MgCl_2, 2 mM DTT, 2 mM ATP and 60% RRL (Promega). In a separate experiment, Luc and wild-type Osp94 were incubated for 30 min at 39°C, diluted with refolding buffer A lacking ATP and RRL, and then incubated with Hsc70 (1.6 µM, Enzo, Farmingdale, NY), human Hsp40 (3.2 µM, Enzo) and ATP (2 mM, Sigma) at the indicated combinations. The refolding assay was performed at 30°C for 2 or 5 h. Reaction mixtures were further diluted 5-fold with 25 mM HEPES-KOH buffer (pH 7.6) containing 1 mg/ml BSA, and then 10 μl of the mixture was added to 100 μl of Luc assay reagent (Promega). Luc activity was then measured using a MiniLumat 9506 luminometer (Berthold, Oak Ridge, TN) and expressed as the percentage of Luc activity stored at 4°C during the experiment.

ATP binding was assayed using a modification of a previously described method (Oh et al., 1999). In brief, a suspension of ATP-agarose (Sigma) in a 50% slurry was prepared in buffer B (20 mM Tris-HCl pH 7.2, 20 mM NaCl, 0.1 mM EDTA, 2 mM DTT). Neuro-2a cells prepared from normal and heat stress (42°C, 2 h) conditions were lysed in 250 µl of the lysis buffer used for western blotting. The lysate was centrifuged at 16,000 g for 5 min, and the resulting supernatant (120 μl) was incubated with 200 μl of ATP-agarose suspension for 16 h at 4°C. In a separate experiment designed to evaluate the specificity of binding to ATP on the agarose beads, 10 mM ATP and 3 mM MgCl₂ were added to a suspension of cell lysate and ATP-agarose and incubated as described above. After centrifugation at 2000 g for 2 min, the supernatant was removed and set aside (unbound). The resulting ATP-agarose beads were washed with buffer B (200 μl) containing 500 mM NaCl and 3 mM MgCl₂, centrifuged at 2000 g for 2 min, and the supernatant was stored 4°C (Buffer B+500 mM NaCl+3 mM MgCl₂). This process was repeated twice, and each resulting supernatant was combined for subsequent analyses described below. The resulting ATP-agarose beads were then treated with buffer B containing 5 mM GTP and/or lacking EDTA as described above, and each resulting supernatant was stored as Buffer B+5 mM GTP and Buffer B without EDTA samples, respectively. Proteins bound to the ATP-agarose were eluted by incubation for 4 h at room temperature (RT) with EDTA-free buffer B (400 μl) containing 1 mM ATP and 1 mM MgCl₂ (ATP-MgCl₂), centrifuged at 2000 g for 2 min, and then the supernatant was stored (Buffer B+1 mM ATP+1 mM MgCl₂). Each resulting supernatant was treated with acetone (sup.: acetone; 3:10), placed on ice for 20 min, and then centrifuged at 16,000 g for 10 min. The resulting pellet was dried for 20 min at RT and subjected to western blotting as described above. After treatment with 1 mM ATP-MgCl₂, to ensure protein elution, the resulting ATP-agarose beads were further incubated for 16 h at 4°C in EDTA-free buffer B (400 μl) containing 10 mM ATP and 3 mM MgCl₂. The resulting eluate (Buffer B+10 mM ATP and 3 mM MgCl₂) and ATP-agarose beads (Beads) were also subjected to western blotting using antibodies against Osp94 (1:1000, TransGenic, Kobe, Japan), Hsp70 (RRID: AB_10013742, 1:500, Enzo, New York, NY), Hsc70 (RRID: AB_1193540, 1:10,000, Enzo), Bip (RRID:AB_398291, 1:500, BD Biosciences, San Jose, CA), Hsp90 (RRID: AB_397798, 1:500, BD Biosciences), and Hsp40 (RRID: AB_10982482, 1:1000, Thermo Fisher Scientific). Protein signals were detected by re-probing with each antibody as described in the western blotting section above.

Protein interactions were assessed using a cross-linking assay with a biotin tag as described in a previous study (Minami et al., 2000). A conjugate composed of Luc and sulfo-N-hydroxysuccinimidyl-2-[6-(biotinamido)-2-(p-azidobenzamido)-hexanoamido]ethyl-1,3′-dithiopropionate (Sulfo-SBED; Thermo Fisher Scientific) was prepared at RT for 30 min in accordance with the manufacturer's instructions and then concentrated using a MicroconYM-30 centrifugal filter unit (Millipore). Sulfo-SBED–labeled Luc (40 nM) was inactivated at 42°C for 5 min in the presence of Hsp90 (0.8 μM), diluted with refolding buffer C containing Hsc70 (1 μM, Enzo), Hsp40 (0.6 μM, Enzo) and ATP (3 mM, Sigma) in combination with PA28α (1 μM, AFFINITI, Mamhead, UK), the LH domain (1 μM) or ovalbumin (1 μM, Sigma), and then incubated at 30°C for the refolding reaction. Aliquots were removed at the indicated time points, and after photolysis at 254 nm (Cross-Linker, Bio-Rad) for 3 min, the reaction mixtures were incubated for 30 min with 100 mM DTT to allow transfer of the biotin tag from Luc to the test proteins. The reaction samples were then subjected to western blotting using HRP-conjugated streptavidin (ImmunoPure, 1:5000, Thermo Fisher Scientific) to detect biotinylated proteins. To identify proteins to which biotin was transferred, the same nylon membrane was re-probed with antibodies against Hsp90, Hsc70, Hsp40, the LH domain, and ovalbumin (43 kDa) after stripping off antibodies already bound to the membrane.

Luc (160 nM) was heat-denatured (42°C, 5 min) in the presence of Hsp90 as described in the Luc refolding assay section above. The solution was diluted 10-fold with refolding buffer C, and then Hsc70 (1 μM), Hsp40 (0.55 μM), ATP (3 mM), and LH domain (1 µM) were added, and the refolding reaction was allowed to proceed at 30°C for 1 h. Aliquots obtained at the indicated time points were diluted with PBS and immunoprecipitated with anti-Osp94 antibody (1 μg) as described in the immunoprecipitation section above. Western blotting was performed using anti-Luc (1:1000, Promega), anti-Hsp90 (1:1000, BD Biosciences), anti-Hsp40 (1:5000, Thermo Fisher Scientific), anti-Hsc70 (1:3000, Enzo), and anti-Osp94 (1:400, TransGenic) antibodies. The same nylon membrane was re-probed with different antibodies as described in the western blotting section above.

Refolding of Hsp90-bound thermally denatured Luc was assayed using the method reported by Minami et al. (2006). Luc (160 nM) was inactivated by incubation with Hsp90 (3.2 µM, Enzo) at 39°C for 5 min in buffer consisting of 30 mM MOPS-KOH (pH 7.2) and 2 mM DTT, followed by 10-fold dilution in refolding buffer C (10 mM MOPS/KOH pH 7.2, 50 mM KCl, 3 mM MgCl_2, 2 mM DTT) containing RRL (5%, Promega), Hsc70 (3.9 μM, Enzo), Hsp40 (2.2 μM, Enzo), ATP (3 mM), PA28 (5 μM, AFFINITY) or the Osp94 LH domain (1 μM), as indicated. For the refolding assay, the above mixture was incubated at 30°C for 1 h, after which Luc activity was assayed as described in the Luc refolding assay section above.

PA28α was chosen to prepare the chimera based on the report of Minami et al. (2000), who showed that PA28α and PA28β function equivalently in assisting the refolding of Hsp90-bound thermally denatured Luc. Mouse PA28α (Gene ID: 19186, ACC#: D87909) was cloned from mIMCD3 cells treated with IFN-γ (10 ng/ml, R&D Systems, Minneapolis, MN) for 24 h using RT-PCR with the following primers: 5′-ATGGCCACACTGAGGGTCCA-3′ (sense) and 5′-TCAATAGATCATTCCCTTGG-3′ (antisense). The coding region of the mouse PA28α gene was ligated into the pGEX-6P-1 vector at the SalI and NotI sites, as described in the plasmid construction section above. Chimeric proteins consisting of Osp94 and PA28α (ABL–PA28α and BL–PA28α) were generated by substituting PA28α into the C-terminus of the L domain of Osp94 at residues 601–838 in place of the H domain. Briefly, the ABL domains of Osp94 and PA28α were amplified by PCR using the following primer pairs: 5′-GTCGACTCATGTCGGTGGTGGGCATTGA-3′ (sense, with SalI site) and 5′-GTAACTATTGAGAAGGTCTT-3′ (antisense), and 5′-CTTCTCAATAGTTACATGGCCACACTGAGGGTCCA-3′ (sense, with 15 nucleotides of Osp94) and 5′-GCGGCCGCTCAATAGATCATTCCCTTGG-3′ (antisense, with NotI site), respectively. Each resulting PCR fragment was re-amplified using the primer pair 5′-GTCGACTCATGTCGGTGGTGGGCATTGA-3′ (sense) and 5′-GCGGCCGCTCAATAGATCATTCCCTTGG-3′ (antisense) and then ligated into the pGEX-6P-1 vector. The GST-ABL-PA28α recombinant plasmid was then amplified by PCR using the primer pair 5′-TCTAGAATGTCCCCTATACTAGGTTC-3′ (sense) and 5′-TCTAGATCAATAGATCATTCCCTTGG-3′ (antisense) and inserted into the pCold III vector (TAKARA, Osaka, Japan) at the XbaI site. Recombinant GST–ABL–PA28α fusion protein was expressed in pKJE7/BL21 E. coli (TAKARA) according to the manufacturer's instructions, and then the ABL–PA28α chimera protein lacking GST was prepared as described above. In addition, BL-PA28α was expressed in pKJE7/BL21 E. coli cells as described above. Purified PA28α, ABL-PA28α and BL-PA28α proteins were analyzed by SDS-PAGE followed by western blotting using anti-PA28α (RRID: AB_10207625, 1:1000, MBL, Nagoya, Japan) and anti-GST (RRID: AB_2279558, Cell Signaling Technology, Danvers, MA) antibodies.

The effect of ABL–PA28α, BL–PA28α and PA28α on disaggregation and reactivation of heat-denatured Luc was assessed as described in the thermal aggregation assay and the Luc refolding assay sections, respectively above. To evaluate the effect of ABL–PA28α and BL–PA28α on Hsp90-dependent refolding of Luc compared with RRL and the LH domain, Hsp90-bound Luc was reactivated by incubation at 30°C for 1 h in refolding buffer C [Hsc70 (3.9 μM), Hsp40 (2.2 μM), and ATP (3 mM)] containing RRL (5%), LH (1 μM), PA28α (1 μM), BLH (1 μM), ABL-PA28α (1 μM) or BL-PA28α (1 μM).

Polyclonal anti-Osp94 antibody (TransGenic) was raised in rabbits against a synthetic Osp94 C-terminal peptide fragment (residues 826-838, SSQHTDSGEMEVD). The specificity of the Osp94 antibody was determined as follows. Osp94 was synthesized using a cell-free expression system (TNT Quick-Coupled Transcription/Translation System, Promega) in accordance with the manufacturer's instructions. Synthetic Osp94 (an aliquot of the reaction mixture) and a sample of the cytoplasm (30 μg) of mIMCD3 cells were subjected to western blotting using the anti-Osp94 antibody (TransGenic), as described in the western blotting section above. To verify the suitability of the anti-Osp94 antibody for use in immunoprecipitation experiments, Osp94 expressed in mIMCD3 cells was immunoprecipitated using the Osp94 antibody (TransGenic). In brief, the cytoplasmic fraction of mIMCD3 cells was divided into two portions, and anti-Osp94 antibody (1 μg, TransGenic) was added to only one portion. Cytoplasm fractions with or without anti-Osp94 antibody were incubated at 4°C overnight, after which the antibody–protein complex was immunoprecipitated using protein G-agarose (Thermo Fisher Scientific) as described in the immunoprecipitation section above, followed by western blotting using anti-Osp94 antibody (3 μg, TransGenic).

To demonstrate in vivo interaction between Osp94 and Hsc70, cells grown in a chamber slide (Chamber cover, Matsunami, Tokyo, Japan) were heat treated at 42°C for 2 h and then incubated at 37°C for 4 h. The cells were then washed twice with ice-cold PBS, fixed with cold methanol (−20°C) for 2 min, and washed three times with ice-cold PBS. The cells were incubated in blocking solution containing 5% non-fat milk and 1.5% goat serum in PBS for 1 h, followed by treatment with anti-Osp94 antibody (1:1000, TransGenic) and anti-Hsc70 antibody (1:3000, Enzo) overnight at 4°C. After washing with PBS-0.5% Triton X-100, the cells were treated with secondary antibodies for 1.5 h at room temperature at the following dilutions: Alexa Fluor 488- and Alexa Fluor 594-conjugated goat anti-rabbit- and mouse-IgG (RRID: AB_2556544 and RRID: AB_2534073, respectively, 1:1000, Thermo Fisher Scientific) for Osp94 and Hsc70 staining, respectively. The cells were washed with PBS with 0.5% Triton X-100 and viewed under a confocal laser microscope (Zeiss LMS 880, Oberkochen, Germany).

Data were analyzed by one-way ANOVA and Kruskal–Wallis test. Data are expressed as mean±s.d., and a P value of <0.05 was considered statistically significant.

We thank Misaki Ogura, Hayato Mori, and Mitsumasa Haruna for technical assistance.

The peer review history is available online at https://journals.biologists.com/jcs/article-lookup/doi/10.1242/jcs.258542.