Humans contain many HSP (heat-shock protein) 70/HSPA- and HSP40/DNAJ-encoding genes and most of the corresponding proteins are localized in the cytosol. To test for possible functional differences and/or substrate specificity, we assessed the effect of overexpression of each of these HSPs on refolding of heat-denatured luciferase and on the suppression of aggregation of a non-foldable polyQ (polyglutamine)-expanded Huntingtin fragment. Overexpressed chaperones that suppressed polyQ aggregation were found not to be able to stimulate luciferase refolding. Inversely, chaperones that supported luciferase refolding were poor suppressors of polyQ aggregation. This was not related to client specificity itself, as the polyQ aggregation inhibitors often also suppressed heat-induced aggregation of luciferase. Surprisingly, the exclusively heat-inducible HSPA6 lacks both luciferase refolding and polyQ aggregation-suppressing activities. Furthermore, whereas overexpression of HSPA1A protected cells from heat-induced cell death, overexpression of HSPA6 did not. Inversely, siRNA (small interfering RNA)-mediated blocking of HSPA6 did not impair the development of heat-induced thermotolerance. Yet, HSPA6 has a functional substrate-binding domain and possesses intrinsic ATPase activity that is as high as that of the canonical HSPA1A when stimulated by J-proteins. In vitro data suggest that this may be relevant to substrate specificity, as purified HSPA6 could not chaperone heat-unfolded luciferase but was able to assist in reactivation of heat-unfolded p53. So, even within the highly sequence-conserved HSPA family, functional differentiation is larger than expected, with HSPA6 being an extreme example that may have evolved to maintain specific critical functions under conditions of severe stress.
- heat-shock protein (HSP) 70 machine
The HSP (heat-shock protein) 70 family of chaperone proteins is one of the most conserved protein families in evolution. It is found in all organisms excluding some hyperthermophilic archaea . HSP70 proteins function in de novo protein folding and the suppression of protein aggregation under stress conditions [2–4]. After the sequencing of the human genome, it became apparent that the human HSP70 family (or HSPA family) consists of 13 different proteins (excluding the four homologous HSP110/HSPH proteins)  (for the nomenclature of the human HSP members and some of their characteristics, see Supplementary Table S1 at http://www.BiochemJ.org/bj/435/bj4350127add.htm). Although some of the gene expansion may have resulted to serve cellular compartmentalization, recent data obtained with protein localization prediction methods suggest that most of the HSPA proteins localize to the cytosolic and nuclear compartment [6,7], and it is currently unclear why eukaryotic cells need so many cytosolic/nuclear HSPA proteins.
In addition to 13 HSPA genes, over 49 different HSP40/DNAJ-encoding genes have been found in the human genome . DNAJ proteins have in common a conserved J-domain and are subdivided into three distinct groups (DNAJA, DNAJB and DNAJC) based on the presence of certain structural motifs  that, however, do not edify functional properties . All DNAJ proteins that have been investigated so far accelerate the ATPase activity of HSPA proteins; in conjunction with other cofactors such as BAG-1, HSPBP1, HSPH/HSP110, HIP [Hsc (heat-shock cognate) 70-interacting protein] and CHIP (C-terminus of HIP), they tightly regulate the activity of the HSPA machine. As for the HSPA proteins, most of the DNAJ proteins are predicted to localize to the cytosolic or nuclear compartment . A previous detailed study in yeast revealed that cytosolic and nuclear DNAJ proteins show complex overlapping as well as unique properties . However, for the mammalian situation, details on the regulation of HSPA proteins by DNAJ proteins are largely lacking.
HSP proteins in yeast have been divided into stress chaperones and ‘housekeeping’ or ‘CLIPS’ (chaperones linked to protein synthesis) . Indeed, chaperones have been found that are only expressed after (severe) stress conditions (e.g. HSPA6). In line with this, HSPA8 has been proposed as a ‘housekeeping’ or CLIPS chaperone. However, other chaperones, such as the human HSPA1A/B and DNAJB1 genes, show basal expression under normal conditions, as well as induction upon proteotoxic stress and the existence of a distinct CLIPS and HSP network in humans is currently unclear.
Previously, we have cloned the major part of the cytosolic/nuclear HSPH, HSPA, DNAJA and DNAJB family . In the present study, we used two potential chaperone clients with fundamentally different biochemical properties to compare the functionalities of these chaperones in living mammalian cells: (i) heat-inactivated luciferase, known to be refoldable after heat denaturation both in vitro and in vivo [13–15]; and (ii) the genetically unstable, expanded polyQ (polyglutamine)-containing Huntingtin, an aggregation-prone protein . Our results demonstrate that some HSP members that can stimulate luciferase cannot suppress polyQ aggregation. Inversely, several other members that suppress polyQ aggregation cannot stimulate luciferase refolding. This clearly suggests some degree of functional differentiation between the cytosolic/nuclear HSP. The most extreme example of this concerns HSPA6. This most strongly heat-inducible member of the human HSP70 family that shows high sequence homology with HSPA1A lacks the canonical properties of HSP70s as a rather non-selective chaperone. Given its exclusive expression under conditions of heat shock and its high intrinsic, merely DNAJ-independent, ATPase activity, we speculate that HSPA6 might serve to chaperone certain critical substrates under extreme stress conditions.
MATERIALS AND METHODS
Cell lines, cell culture and transient transfections
Both the hamster O23 cell line and HEK (human embryonic kidney)-293 cells stably expressing the tet (tetracycline) repressor (Flp-In T-Rex HEK-293; Invitrogen) were grown in DMEM (Dulbecco's modified Eagle's medium; Gibco) supplemented with 10% FBS (foetal bovine serum; Sigma) and 100 units/ml penicillin and 100 μg/ml streptomycin (Invitrogen). Flp-In T-Rex HEK-293 cells were kept under selective pressure for the tet repressor and the Flp-in recombination site by adding 5 μg/ml blasticidine (Sigma) and 100 μg/ml zeocin (Invitrogen) to the culture medium. Cultures were maintained at 37 °C and 5% CO2 in a humidified incubator. For transient transfections, cells were grown to 50–60% confluence in 35 mm diameter dishes coated with 0.001% poly-L-lysine (Sigma) and/or on coated coverslips for confocal microscopy analyses. Cells were transfected with a total of 1 μg of DNA using Lipofectamine™ (Invitrogen) according to the manufacturer's instructions.
Detailed information about the primers and plasmids used in the present study can be found in Supplementary Tables S2 and S3 at http://www.BiochemJ.org/bj/435/bj4350127add.htm. Briefly, tet-inducible HSP-expression plasmids were constructed as follows. First, the GFP (green fluorescent protein) and the V5 tag, harbouring a Kozak consensus ATG initiation codon and lacking a stop codon, were cloned into the pcDNA5/FRT/TO vector. Subsequently the coding sequence of the different chaperones was amplified using the primers listed in Supplementary Table S2. As a template source, cDNA was made from total RNA as described previously . As a source of total RNA, QPCR (quantitative PCR) human reference total RNA (Stratagene) was used. DNAJB4, DNAJB5 and DNAJB8 were amplified from cloned full-length cDNAs purchased from Open Biosystems (clone ID: DNAJB4, 4340658; DNAJB5, 4684829; and DNAJB8, 5296554). The fragments were cloned in pcDNA5/FRT/TO GFP no stop, and the presence of the correct gene was sequence-verified. Protein expression was verified by Western blotting. Subsequently, fragments were subcloned into pcDNA5/FRT/TO V5 and pcDNA5/FRT/TO.
peGFP-HDQ23 and peGFP-HDQ74 were kindly provided by David Rubinsztein (Cambridge Institute for Medical Research, Cambridge, U.K.).
siRNA (small interfering RNA)
HSPA1A- and HSPA6-gene knockdown was performed using siGENOME SMARTpool siRNA from Dharmacon. siRNA was transfected using Lipofectamine™ 2000 at a final concentration of 50 nM. Cells were transfected with siRNA 96 h before heat treatment and were again transfected with siRNA in combination with DNA 48 h prior to heat treatment. Pre-heat treatment (to aquire thermotolerance) was given 18 h before, so 78 and 30 h after the two siRNA treatments respectively.
Focused gene arrays
Oligo GEArray Human Toxicology and Drug Resistance microarrays (OHS-401) were purchased from Superarray. RNA was isolated from HEK-293 cells 0, 1, 3 and 6 h after a 30 min 45 °C heat shock as described previously . Subsequently, 3 μg of RNA was used in the first-strand cDNA synthesis reaction as described by the manufacturer. Visualization was performed with ECL (enhanced chemiluminescence) and hyperfilm (Amersham) and the resulting images were scanned and processed.
Relative changes in transcript level were determined on the Icycler (Bio-Rad) using SYBR Green supermix (Bio-Rad) as described previously . Primers were designed with PerlPrimer (http://perlprimer.sourceforge.net/). Calculations were done using the comparative Ct method according to User Bulletin 2 (Applied Biosystems). For each set of primers, the PCR efficiency was between 90 and 105%. Primer sequences are listed in Supplementary Table S2.
Cell extracts and sample preparation
At 24 or 48 h after transfection, cells were recovered by trypsinization, pelleted and resuspended in 1 ml of PBS. The cell suspension was centrifuged at 200 g for 5 min at room temperature (20 °C) and the pellet was resuspended in 75–100μl of RIPA buffer [25 mM Tris/HCl (pH 7.6), 150 mM NaCl, 1% Nonidet P40, 1% sodium deoxycholate and 2% SDS plus protease inhibitors] and sonicated (15 s at 50 W). Protein content was determined with the DC (detergent-compatible) protein assay (Bio-Rad). Western blot samples were prepared at a final concentration of 1μg/μl in SDS/PAGE loading buffer and heated for 5 min at 100 °C. Filter-trap samples were prepared at a final concentration of 100 ng/μl, 20 ng/μl and 4 ng/μl in FTA buffer [10 mM Tris/HCl (pH 8.0), 150 mM NaCl and 50 mM DTT (dithiothreitol)] with 2% SDS, and heated for 5 min at 100 °C. Samples were used immediately or kept frozen at −20 °C.
Western blot analysis
Equal amounts of protein were loaded on to SDS/PAGE (10% or 12.5% polyacrylamide gels). Proteins were transferred on to nitrocellulose membranes and probed with mouse anti-GFP antibody JL-8 (Clontech) at a 1:5000 dilution, mouse anti-V5 antibody (Invitrogen) at a 1:5000 dilution. GAPDH (glyceraldehyde-3-phosphate dehydrogenase) was used a loading control and was detected with a mouse antibody (RDI Research Diagnostics) at 1:10000 dilution. Blots were subsequently incubated with HRP (horseradish peroxidase)-conjugated anti-mouse secondary antibody (Amersham) at a 1:5000 dilution, and visualization was performed with ECL and hyperfilm.
To determine protein aggregates, the filter-trap assay was performed as described previously . Briefly, 10, 2 and 0.4 μg of protein extracts were applied on to a 0.2μm pore cellulose acetate membrane prewashed with FTA+0.1% SDS. Mild suction was applied and the membrane was washed three times. Aggregated proteins trapped in the membrane were probed with mouse anti-GFP antibody JL-8 (Clontech) at a 1:5000 dilution and mouse anti-V5 antibody (Invitrogen) at a 1:5000 dilution followed by HRP-conjugated anti-mouse secondary antibody (Amersham) at a 1:5000 dilution. Visualization was performed using ECL and hyperfilm.
Cell lysis and luciferase activity measurements were performed as described previously . Luciferase activity after treatments were expressed relative to the activity in unheated control cells (=100%). Error bars in plots represent S.D. In the case of a single experiment, the S.D. was calculated according to the rules of error propagation. In the case of multiple experiments, S.D.s were calculated directly from the differences in the percentages of luciferase activity compared with an unheated control.
Ni-NTA (Ni2+-nitrilotriacetate) precipitation of His6-tagged proteins
Cells were trypsinized and washed with PBS. Subsequently, the cross-linker DSP [dithiobis(succinimidyl propionate)] was added to a final concentration of 1 mM and incubated for 30 min on ice. The cross-linking was stopped by the addition of glycine to a final concentration of 2 mM and incubated for 15 min on ice. Cells were washed with PBS and the cell pellet was harvested in 0.5 ml of lysis buffer [150 mM NaCl, 50 mM NaH2PO4, 10 mM imidazole, 0.5% Nonidet P40 (Igepal), 1.5 mM MgCl2, 3% glycerol, 0.9 mM DTT and protease inhibitors (pH 8.0)]. Cells were lysed by passage through a 26 gauge needle five times and subsequently centrifuged twice at 200 g for 15 min. Then, 30μl of supernatant was taken and mixed with 30μl of 2×Laemmli sample buffer. The rest of the supernatant was transferred to a new tube and 10μl of Ni-NTA agarose beads (Qiagen) was added and incubated at 4 °C for 1 h with slow agitation. Next, the beads were washed four times with wash buffer [300 mM NaCl, 50 mM NaH2PO4, 20 mM imidazole, 0.5% Nonidet P40 (Igepal), 1.5 mM MgCl2 and 3% glycerol (pH 8.0)]. The pellet was eluted in 30μl of lysis buffer and 30μl of 2×Laemmli sample buffer and boiled. Then, 10μl was subsequently loaded for Western blot analysis.
Immunolabelling and confocal microscopy
At 16–24 h after transfection, indirect immunofluorescence of the V5 tag was performed to detect the chaperones. Cells were fixed with 3.7% formaldehyde for 15 min, washed three times with PBS, permeabilized with 0.2% Triton-X100 and blocked for 30 min with 0.5% BSA and 0.1% glycine in PBS. Incubation with mouse anti-V5 monoclonal antibody (Invitrogen) at a 1:100 dilution was performed overnight at 4 °C followed by a 1 h incubation with Cy5 (indodicarbocyanine)-conjugated anti-mouse secondary antibody (Jackson Immunoresearch) at 1:200 dilution. Aggregates were visualized using the eGFP (enhanced GFP) tag. To visualize nuclei, cells were stained for 10 min with 0.2 μg/ml DAPI (4′,6-diamidino-2-phenylindole). Coverslips were mounted in Citifluor. Images of Cy5 and DAPI fluorescence were obtained using the Leica confocal laser-scanning microscope (Leica TCS SP2, DM RXE) with a 63×/1.32 oil lens. The captured images were processed using Leica Confocal Software and Adobe Photoshop.
In vitro ATPase, luciferase refolding and citrate synthase aggregation protection assay
HSPA1A and HSPA6 were cloned in the pET151 prokaryotic expression system (Invitrogen) and isolated according to the manufacturer's procedures. DNAJB1 was purified as described previously . ATPase experiments were performed as described previously . The luciferase refolding assay was a modified version of one previously described . Luciferase (Promega, 12.66 mg/ml) was diluted 100-fold in the mixture containing 10μM HSP90, 25 mM Tris/HCl (pH 7.8), 8 mM MgSO4, 1% BSA, 10% glycerol and 0.25% Triton X-100, and incubated for 20 min at 40 °C. Then, the renaturation process was carried out at room temperature in the buffer containing the following components at final concentrations: 40 mM Tris/HCl (pH 7.4), 50 mM KCl, 5 mM DTT, 15 mM MgSO4, 100 μg/ml creatinine kinase, 20 mM creatinine phosphate, 0.015% BSA, 5 mM ATP, 80 nM denatured luciferase, 0.4μM HSP90, 1μM DNAJB1 and either 2.4μM HSPA1A or HSPA6. At time points from 0 to 80 min, 5μl aliquots were taken, and the activity of renatured luciferase was measured in a luminometer (BMG Labtechnologies) after addition of the Bright-Glo substrate (Promega).
The citrate synthase aggregation protection assay was performed as described previously .
Cellular survival assays
Cells stably expressing HSPA1A and HSPA6 under the tet-inducible promoter were treated with tetracycline (1 μg/ml) to induce expression and incubated for 24 h. Cells were treated with siRNA as described above and incubated for 72 h. Subsequently, cells were heat-shocked for 30 min at 45 °C or left at 37 °C and incubated at 37 °C for 18 h to acquire maximal thermotolerance as described previously . Subsequently, cells were treated with a second heat shock of 45 °C for 15, 30, 45, 60 and 120 min and plated in triplicate (200 cells) on 10 cm Petri dishes in 10 ml of culture medium containing tetracycline (1 μg/ml), and allowed to grow for 14 days. The colonies were fixed and stained, and colonies containing 50 cells were scored.
Primary amino acid alignments were performed in ClustalX2 using the neighbour-joining algorithm and Blosum matrixes at the default settings . Bootstrap analysis was performed using 1000 random number generator seeds and 1000 bootstrap trials. Phylograms were made by importing the homology tree output of ClustalX in TreeView .
Expansion of DNAK-like and DNAJ-like chaperones in the three domains of life
The reason for having high chaperone gene numbers in human cells is currently unclear. Possible evolutionary forces to drive chaperone gene duplication are compartmentalization, emergence of multi-domain proteins, multi-cellularity, cellular specialization and development (Figure 1A). The HSPA/DNAK/DNAJ gene families are the most abundant HSP-encoding gene families found in the human genome. DNAK and DNAJ genes were lacking in the archaea Thermococcus kodakaraensis and Methanococcus jannaschii (Figures 1B and 1C), but were found in other archaea such as Thermoplasma acidophilum and Methanosarcina mazei. Both DNAK and DNAJ sequences were found in all eubacteria analysed. The number of DNAK/HSPH/HSPA dramatically increased in eukaryotes (Figure 1B), which was partially due to an increase in the number of HSPA genes (Figure 1B). However, the number of HSPA genes did not change markedly within eukaryotes, whereas the DNAJ gene numbers differed from approx. 20 genes in yeast, up to over 49 genes in humans and over 80 genes in plants (Figure 1C). These results suggest that a higher number of DNAJ proteins seem to specialize the HSPA chaperone machines .
Humans contain a highly homologous subfamily of cytosolic HSPA family members
One reason for increases in the gene numbers of DNAK/HSPA/HSPH between prokaryotes and eukaryotes could be cellular compartmentalization. Neighbour-joining phylogeny analysis and computational subcellular localization prediction methods show that all of the HSPH/HSP110-like members cluster together in one subfamily and are cytosolic/nuclear proteins (Figure 2A), except HSPH4/Grp170, which is expressed in the ER (endoplasmic reticulum). HSPA13/HSPA14, HSPA12A and HSPA12B are distinct members of the HSPA family and are poorly studied. A very homologous subfamily includes HSPA1A/B, HSPA1L, HSPA2, HSPA6 and HSPA8, which are all predicted to be cytosolic/nuclear. The slightly more distantly related members HSPA5 and HSPA9 are the ER and the mitochondrial equivalents of HSPA8 respectively. Ectopic expression of the V5-tagged members was in line with the predicted localization (Figure 2A). Thus, although part of the HSPA family did expand due to cellular compartmentalization, most of the expansion was independent of this event, and these members may exert their specialized functions within the cytosolic/nuclear compartment.
Many DNAJ members are predicted to localize to the cytosolic and nuclear compartment
A previous study in yeast showed that most DNAJ members are cytosolic proteins . Bioinformatics predicts that three of the four human DNAJA members localize in the cytosol (Figure 2B). Neighbour-joining phylogeny analysis pointed to the existence of two subfamilies within the DNAJB family. The first family shows homology with DNAJB1, DNAJB4 and DNAJB5, whereas the second consists of the members DNAJB2, DNAJB6, DNAJB7 and DNAJB8 (Figure 2C). For both subfamilies, most members were predicted to be in the cytosol or nucleus (Figure 2C). For the members that we use in the present study this could be confirmed for the ectopically expressed V5-tagged proteins (Figure 2C). These results indicate that also in cultured human cells most DNAJ members are cytosolic/nuclear proteins.
Expression profiles of HSPH, HSPA, DNAJA and DNAJB chaperones
As most of the HSP proteins are located in the cytosolic/nuclear compartment, we used QPCR to precisely analyse the constitutively expressed levels and the heat inducibility of HSPH, HSPA, DNAJA and DNAJB transcripts in HEK-293 cells (Figures 3A–3C). These results showed that HSPH1 was the major heat-inducible HSPH gene (Figure 3A), and also the most abundant member under non-heat conditions (Figure 3B). HSPA1A/B, HSPA1L and HSPA6 were the major heat-inducible HSPA genes within the HSPA family (Figure 3A). Interestingly, mRNAs for both HSPA1A and HSPA6 are not detected under non-stress conditions, but constitute a significant part of the HSPA transcript levels after heat stress (Figure 3B). Although HSPA1L shows high heat-inducibility, the transcript levels after heat shock remain relatively low (Figure 3B). HSPA1B is expressed at high levels and is heat-inducible, whereas HSPA8 and HSPA9 are also expressed at high basal levels, but are hardly heat-inducible (Figures 3A and 3B).
For the DNAJ family, we found DNAJB1 and DNAJB4 to be the major heat-inducible DNAJ genes; DNAJB6, DNAJB9 and DNAJB11 were only mildly inducible, and the rest of the members investigated were not up-regulated after heat stress at all (Figure 3A). Under normal growth conditions, DNAJA1 was found to be the most abundant DNAJ transcript (Figure 3C). DNAJA3, DNAJB1, DNAJB6 and DNAJB11 were also found to be expressed under non-stress conditions, but only at moderate levels (Figure 3C). Interestingly, DNAJB1 was found to be by far the most abundant DNAJ transcript after heat stress (Figure 3C). These results are consistent with previous studies that show that DNAJB1 is the major partner of the heat-inducible HSPA1A/B protein. Whether it also serves as a cofactor for HSPA6 is currently unknown. So surprisingly, only a minority of the HSPA and DNAJ family is expressed in HEK-293 cells and only a small number of these genes are heat-inducible.
To test the generality of these findings in HEK-293 cells, we measured the global constitutive expression pattern of various HSPH-, HSPA- and DNAJ-encoding genes in various cell lines and human tissues. For this purpose, dedicated mini-arrays were used which contain a relative large number of probe sets against the various chaperones. Although expression levels for some members such as HSPA1A, DNAJC8 and HSPA9 varied widely between cell lines, we found that the profile of chaperone expression was surprisingly similar among the cell types investigated, indicative that our findings in HEK-293 cells are representative and do not largely deviate from the other cell lines tested (Figure 3D). Also the heat-inducibility pattern of the investigated HSPs in HeLa cells  was comparable with that observed in HEK-293 cells (Figures 3A–3C; Supplementary Figure S1 at http://www.BiochemJ.org/bj/435/bj4350127add.htm). In fact, data from the public databases on HEp2, a human epithelial cell line, and U397, a human lymphoma of myelomonocytic origin, also revealed the same pattern of (non) heat-inducibility of the various HSP members used in the present study (Supplementary Figure S1), strongly supporting the idea that the heat-inducibility of the selected HSP genes is widely conserved.
Stimulation of protein refolding and aggregation suppression is facilitated by distinct chaperone members
To explore putative differences in function, we expressed each of the selected chaperones in HEK-293 cells and performed two functional tests. First, firefly luciferase was co-expressed with the chaperones and cells were heated to measure the refolding of luciferase after the heat stress. Expression of chaperones such as HSPA1A have been found to enhance the refolding of heat-denatured luciferase [15,26]. Secondly, Huntingtin fragments with expanded polyQ were co-expressed with the selected chaperones. These polyQ-expanded proteins are known to aggregate and the expression of chaperones, such as DNAJB1, have been found to suppress its aggregation [27,28]. We used versions of luciferase and the expanded polyQ proteins that are expressed in both the cytosol as well as the nucleus simultaneously  because all cloned chaperones except DNAJA3 and DNAJB9 were predicted or found to be expressed in the cytosol and/or nucleus (Figures 2B, 2C and results not shown).
As shown in Figure 4(A), HSPA1A showed the highest activity on luciferase refolding, whereas various HSPA paralogues, such as HSPA1L, HSPA2 and HSPA8, showed some, albeit lower, activity. HSPA14 did not show an effect on refolding, but this was not unexpected as this member does not contain the canonical substrate-binding domain and is known to have a specialized function in translation . Unexpectedly, however, the HSPA6 paralogue also showed no activity on luciferase refolding (see below). In addition, the Escherichia coli DNAK or Saccharomyces cerevisiae SSA-1, both well known to support luciferase refolding in the host organisms [29,30], did not support luciferase refolding in HEK-293 cells (Figure 4A). Although SSA1 was not expressed to comparable levels as the other chaperones, DNAK was expressed at similar levels as HSPA1A. DNAK's lack of refolding stimulating activity may suggest that cofactors of the mammalian HSPA machine may be unable to co-operate with non-mammalian chaperones. In addition to HSPA members, overexpression of the DNAJA2 and DNAJA4 also resulted in stimulation of luciferase refolding. All other DNAJ members tested were not effective or even slightly suppressed refolding, although for some members (DNAJA3, DNAJB4 and DNAJB5) the results were inconclusive as they were only expressed at low or undetectable levels.
We next studied the capacity of the same set of chaperones to suppress the aggregation of a 74 glutamine-containing Huntingtin exon-1 fragment. As shown in Figure 4(C), the most active aggregation suppressors were DNAJB1, DNAJB2, DNAJB6 and DNAJB8. Interestingly, these best preventers of polyQ aggregation did not support refolding (Figure 4A). Inversely, none of the members that enhanced luciferase refolding (Figure 4A) could suppress polyQ aggregation (Figure 4C). One possibility to explain these results is substrate specificity. To test whether this could explain the differential effects of the chaperones on refolding luciferase or on polyQ aggregation suppression, we tested the effect of the various chaperones on luciferase aggregation. For this, cells were heated and lysed and the lysates were run on SDS/PAGE. Heat-denatured luciferase can be detected as SDS-insoluble aggregates in the stacking gel (Figure 4D). These aggregates physically differ from polyQ aggregates, as they are not retained in a filter-trap assay (results not shown). Yet, the subset of chaperones that were most potent in suppressing polyQ aggregation (DNAJB1, DNAJB2a, DNAJB6 and DNAJB8) also substantially suppressed luciferase aggregation (Figure 4D) without supporting its refolding (Figure 4A). Inversely, however, the two strongest supporters of refolding (HSPA1A and DNAJA4) also suppressed luciferase aggregation, yet were rather ineffective in preventing polyQ aggregation. It must be noted that their reduction of luciferase aggregation might merely be a reflection of their stimulating effect on luciferase refolding. So, the difference of the various members to prevent aggregation or stimulate refolding is not due to a difference in substrate specificity; rather, the results suggest that the different HSPs have different client affinity and/or differently affect client processing.
To exclude the possibility of cell-type-specific effects, we next tested the effects of the subset of chaperones on refolding and polyQ aggregation suppression in hamster lung fibroblast cells (O23) (Figures 5A and 5B). In the O23 hamster fibroblasts, refolding of luciferase (Figure 5A) and suppression of polyQ aggregation (Figure 5B) were supported by the same set of different chaperones as in HEK-293 cells, with the exception of DNAJB5, which was active for the suppression of polyQ aggregation in O23, but not HEK-293, cells (Figure 5B). Taken together, these results indicate that higher eukaryotic cells indeed seem equipped with at least two molecular distinct chaperone activities, one set to suppress protein aggregation and one set able to support refolding of denatured substrates (Figure 5C).
HSPA6 is a heat-inducible HSP70 member distinct from HSPA1A and HSPA1B
One very striking observation was that the strongly heat-inducible HSPA6 (Figure 3), one of the least investigated HSPA members and probably the most recently evolved member that exists only in large mammals , did not support luciferase folding like the strongly heat-inducible HSPA1A. Yet, HSPA6 shows high sequence homology with HSPA1A (81% identity) and has been linked to cellular heat responses, especially when grown at low cell densities . Compared with HSPA1A, HSPA6 shows the highest degree of divergence in the C-terminal lid domain (Figure 6A). To investigate whether this C-terminus may have evolved to serve different functions of HSPA6, aside from those tested here, we constructed chimaeras of HSPA1A and HSPA6 using two shared restriction sites (Figure 6B). One site (Van91I) is positioned in the middle of the ATPase domain and the other (PstI) is located near the outer boundary of the ATPase domain (Figure 6B). Surprisingly, the chimaeric proteins containing the C-terminus of HSPA6 fused to the N-terminal ATPase fragment of HSPA1A were found to be fully active in supporting luciferase refolding. In contrast, both chimaeric proteins containing the ATPase domain from HSPA6 were ineffective (Figure 6C). We next repeated this experiment in HEK-293 cells depleted for HSPA1A and HSPA1B using shRNA (short hairpin RNA) and found essentially the same results (Figure 6D). In addition, experiments in hamster lung fibroblast O23 lacking basal expression of HSPA1A/B showed similar results (results not shown), implying that the effect of the chimaera was not due to formation of functional interactions with the wild-type endogenous HSPA1 members. Thus these results show that the complete C-terminus including the spacer domain, the substrate-binding domain, as well as the lid domain from HSPA6 are functional and can recognize firefly luciferase as a client like HSPA1A. Yet, efficient refolding is not supported, probably due to a difference in the more homologous ATPase domain. One possibility is that the HEK-293 cells do not contain the required (heat-inducible) cofactors needed for adequate regulation of the HSPA6 ATPase that leads to refolding activity. To test whether HSPA6, which is exclusively expressed under heat-shock conditions, depends on other heat co-inducible factors, we primed cells with a short heat shock in the presence or absence of ectopically expressed HSPA6 or HSPA1A. Such priming leads to a resistance to second heat treatments, referred to as thermotolerance which, among others effects, improves luciferase refolding [18,26] (Figure 6E). Ectopically expressed HSPA1A led to an even further improvement of refolding in thermotolerant cells, but this effect was not seen after overexpression of HSPA6 (Figure 6E). Thus, also after the induction of ‘all’ heat-inducible factors, HSPA6 remained dysfunctional in refolding luciferase. Inversely, we blocked the heat-shock-induced expression of HSPA6 or HSPA1 by RNAi (RNA interference) during thermotolerance development. As shown in Supplementary Figure S2 (at http://www.BiochemJ.org/bj/435/bj4350127add.htm), the expression of both HSPA1A and HSPA6 could be blocked selectively and efficiently. Using these selective RNAi molecules, we found that RNAi against HSPA1A significantly reduced luciferase refolding, not only in control or HSPA1A transfected cells, but also in thermotolerant cells (Figure 6F). In contrast, the knockdown of exogenous HSPA6 expression did not show a reduction in refolding under all these conditions (Figure 6F). Thus HSPA1A, but not HSPA6, contributes to enhanced luciferase refolding during thermotolerance, and the absence of HSPA6 activity for this endpoint seems not to be due to a lack of heat-inducible cofactors.
It still could be argued that the HEK-293 cells may lack the correct DNAJ partner, even after priming the cells with heat shock, required to support HSPA6. Therefore we tested whether there were differences in the spectrum of DNAJ proteins that could physically interact with either HSPA1A or HSPA6. We used Ni-NTA precipitation followed by Western blot analysis to investigate this. Like HSPA1A, HSPA6 was also found to physically interact with all DNAJ members tested (Figure 7) in a specific manner, since none of the V5-tagged DNAJ proteins bound to the Ni-NTA beads (Figure 7). HSPA1A and HSPA6 did not interact with either Troponin I or Troponin T, two proteins unrelated to molecular chaperones, further demonstrating that the interaction is specific. These results not only indicate that the DNAJ domain is the main determining factor for binding of DNAJ to HSPA proteins, but also reveal that differences in specificity for DNAJ proteins do not underlie the observed difference between HSPA1A and HSPA6 in stimulating luciferase refolding.
We next set up a series of in vitro cell-free experiments to directly analyse the functionality and ATPase activity of HSPA6. HSPA1A and HSPA6 were purified and incubated alone or together with purified DNAJB1 to measure their in vitro steady-state ATP hydrolysis activities. Surprisingly, the ATPase of HSPA6 was found not to be defective, but its activity was even higher than that of HSPA1A (Figure 8A). DNAJB1 stimulated the ATPase of HSPA1A, but even then, the activity was lower than for HSPA6 without DNAJB1. Moreover, the ATPase activity of HSPA6 was not (further) enhanced by DNAJB1 (Figure 8A). Differences in the ATPase activity of HSP70 members were previously shown to exist for the yeast Hsp70s Ssa and Ssb, and these differences were suggested to be potentially relevant for the functional uniqueness of these Hsp70s . To test whether this may indeed be true, we compared the activity of purified HSPA1A and HSPA6 on three different substrates. First, we heat-denatured purified firefly luciferase in the absence or presence of HSPA1A and HSPA6 with and without DNAJB1. Consistent with our cellular data, HSPA6 either alone or together with DNAJB1 showed no detectable activity on luciferase refolding after thermal denaturation, whereas HSPA1A, together with DNAJB1, clearly supported refolding (Figure 8B). We next measured the ability of HSPA6 to suppress aggregation of the thermosensitive enzyme citrate synthase using a citrate synthase aggregation protection assay. As shown in Figure 8(C), the presence of recombinant HSPA1A suppressed the heat-induced aggregation of citrate synthase, whereas recombinant HSPA6 did not (Figure 8C). Finally, we used the tumour suppressor p53, for which it was previously shown that HSPA1A can restore its activity after heat shock [22,33]. This chaperone activity involves a remodelling of the p53 tetrameric complex, which might depend on mechanistically distinct chaperone characteristics than assisting in the renaturation of heat-unfolded monomeric proteins. Interestingly, now HSPA6, like HSPA1A, did show a chaperone-like activity as it was found to rescue the heat-induced impairment of human recombinant wild-type p53 to bind to the p21/WAF1 promoter (Figure 8D). In fact, HSPA6 was even more efficient in rescuing p53 activity and, unlike HSPA1A, did not rely on the presence of J-proteins.
Since HSPA6 is exclusively expressed under conditions of severe stress [34,35], we wondered how its rescue of p53 activity and maybe other critical clients or structural elements in cells would affect cell survival after a severe heat shock. For this, HEK-293 cell lines were generated that stably expressed either HSPA1A or HSPA6 under the control of tet-regulated promoters (Figure 9A). Whereas expression of HSPA1A reduced heat-induced mortality, measured using the colony-formation assay, expression of HSPA6 did not show such cytoprotective effects (Figure 9B). To test whether HSPA6 might require cofactors that are only expressed in heat-stressed cells to protect cells, we also used thermotolerant HSPA1A- and HSPA6-expressing cell lines. Thermotolerant cells overexpressing ‘all’ heat-inducible HSPs showed a high level of resistance towards heat-induced cell killing (Figure 9B) which was much higher than the resistance seen for HSPA1A overexpression, as described previously . The overexpression of HSPA1A (or HSPA6) in thermotolerant cells did not show a supertolerant phenotype, indicating that a maximal survival is reached by the general induction of heat-stress proteins under these thermotolerant conditions (Figure 9B). Therefore the effects of HSPA6 expression in thermotolerant cells could not be assessed using overexpression. We therefore selectively depleted either HSPA1A or HSPA6, or blocked their induction during thermotolerance development by the selective RNAi described above (Figure 9C). Although siRNA against HSPA1A strongly enhanced heat-induced cell killing in both control and thermotolerant cells (Figure 9D), the knockdown of HSPA6 showed no effect for both conditions. Taken together, our results show that the strongly heat-inducible HSPA6 appears to play no significant role in protecting HEK-293 cells against heat-induced cell death.
Human cells contain many members of the HSP70/HSPA and DNAJ/HSP40 family, and most members of the HSPA and DNAJ family, that form the core of HSP70 machines, were found to reside in the cytosolic and nuclear compartment of cells. Except for those that could not be studied due to poor expression, we found that their ectopic overexpression resulted in different phenotypes with regard to the handling of two fundamentally different clients: refoldable heat-denatured luciferase compared with non-refoldable rapidly aggregating polyQ proteins. One subset showed a high activity to stimulate luciferase refolding, whereas the second subset was very effective in suppressing polyQ aggregation. This was not related to client specificity itself, as the polyQ aggregation inhibitors also suppressed heat-induced aggregation of luciferase. Probably the largest functional differentiation was seen for HSPA6 that, despite its high sequence homology with HSPA1A, could not handle a typical HSP70 client, such as heat-unfolded luciferase or citrate synthase, but that did rescue p53 activity after heat denaturation.
Although our comparison involved only two different clients, they prompt us to propose a (admittedly simplified) model of how different HSP members may primarily act on two fundamentally different classes of substrate, i.e. those that are potentially (re)foldable (generated by acute stresses) and those that are non-(re)foldable proteins (due to chronic stresses). Many acute stresses (such as heat shock) will cause unfolding of normally native proteins that, upon stress relief, can ultimately regain their original native state. Strikingly, most of the HSPs that were effective in refolding luciferase (Figures 4 and 5) were heat-inducible (Figure 3 and Supplementary Figure S1). In contrast, aggregation of non-refoldable proteins, such as polyQ proteins, is prevented by other members, most of which are not (strongly) heat-inducible (Figures 3–6 and Supplementary Figure S1). In fact, we previously found a similar paradigm for members within the HSPB family of small HSPs in which the strongest heat-inducible HSPB1 supported luciferase refolding, without showing a major effect on polyQ aggregation, whereas members such as HSPB7, HSPB8 and HSPB8 did not support luciferase refolding, but strongly suppressed polyQ aggregation . Intriguingly, all of these three polyQ-aggregation-inhibiting HSPB members were directly involved in protein degradation pathways [36,37]. So, it appears that a specialized subset of cytosolic chaperones may have evolved to dispose of (rather than refold) non-foldable proteins, such as those linked to inherited genetic mutations [e.g. polyQ-expanded Huntingtin, CFTR (cystic fibrosis transmembrane conductance regulator) or α-synuclein]. What the molecular basis for this differentiation in client recognition and processing is, remains to be elucidated. However, it is striking to note that for stimulating luciferase refolding HSP70 is often limited . Also, including for effects of HSPB1, refolding is always entirely dependent on an adequately regulated nucleotide cycle of the HSP70 machine because accelerating this cycle with the HSP70 nucleotide-exchange factor BAG-1 inhibits refolding [39,40]. In contrast, for polyQ the effects of the best inhibitors of aggregation in both DNAJ families and HSPB families are largely independent of HSP70 and not inhibited by BAG-1 [36,41].
The most striking functional diversity found in the present study concerns HSPA6. This HSP70 member that is only found in large mammals is not expressed at 37 °C, but shows a strong induction after heat stress ( and the present study) and after proteasome inhibition , especially when cells are grown at low density . Yet, HSPA6 was unable to assist refolding of heat-unfolded luciferase (in cells and in vitro) or prevent aggregation of citrate synthase (in vitro), activities that were seen for all other cytosolic HSPA members (except HSPA14 that lacks the full peptide-binding domain). Domain-swapping experiments showed that HSPA6 has a functional peptide-binding domain, but an abnormal N-terminal ATPase domain that was related to its inability to chaperone luciferase. In vitro experiments next revealed that the (intrinsic) basal ATPase activity of HSPA6 is much higher than that of HSPA1A, again pointing to the possibility that the nucleotide cycle is critical for the general chaperone-like activity of HSP70 proteins (see above). Indeed, also for DNAK, stimulation of luciferase refolding was reduced when the ATPase activity was too high . Similarly, deleting the C-terminal EEVD motif from HSPA1A resulted in an increased ATPase activity and loss of chaperone activity in vitro  and reduced activity in cells . Intriguingly, the ATPase activity of the EEVD mutant and its (residual) chaperone activity could no longer be boosted by DNAJ proteins [38,44], resembling the effects we show in the present study for the ATPase activity of HSPA6 (Figure 8A). Strikingly, however, HSPA6 was able to rescue p53 activity after thermal denaturation even more effectively than HSPA1A. So, higher rates of intrinsic ATPase activity of HSP70 may be disadvantageous for certain substrates (luciferase, citrate synthase), while being advantageous for other substrates (p53). Also in yeast, the ATPase activities among different HSP70s were found to vary significantly . Whereas these differences were mainly attributed to differences in the C-terminal domains, our domain-swapping experiments suggest that for HSPA1A and HSPA6 the functional differences we found are intrinsic to the ATPase domain itself. Yet, irrespective of the mechanism responsible for the different ATPase activities, these differences in the ATPase activity may be relevant to the functional uniqueness of HSP70s. Another striking feature of the purified HSPA6 was that, unlike HSPA1A, it did not require additional J-proteins to rescue heat-denatured p53 activity. Although we do not completely rule out that our purified HSPA6 contains traces of prokaryotic DNAJ proteins, this would suggest that, given its higher basal ATPase activity than HSPA1A, HSPA6 might be able to chaperone a subset of critical, special clients under conditions where the availability of J-proteins is limited.
Whatever the physiological function of HSPA6 and its chaperone action on p53 (and maybe other) clients could be, HSPA6 up-regulation or its depletion did not affect the clonogenic ability of naïve or primed (thermotolerant) HEK-293 cells after heat shock. As such, it clearly differs from HSPA1A, implying that, after heat shock, canonical protein damage-repair type activities, as stimulated by HSPA1A, are more critical compared with the more specialized activities of HSPA6 for the reproductive ability of cells like HEK-293.
In summary, in the present study we have shown that cells seem to be equipped with many functionally distinct subsets of chaperones, some of which seem to be dedicated to (re)folding and some that may have evolved to dispose of non-foldable proteins. Even more so, within the highly sequence conserved HSPA family, functional differentiation is larger than expected, which is consistent with recent findings in yeast . HSPA6 herein represents an extreme example: it lacks the generic chaperone-like properties of other HSP70s and may have evolved to maintain specific critical functions under conditions of severe stress.
Jurre Hageman designed and conducted most of the experiments and wrote the manuscript. Maria van Waarde participated in most experiments. Alicja Zylicz and Dawid Walerych conducted in vitro chaperoning experiments of which the data are presented in Figure 8. Harm Kampinga was the grant holder, designed the experiments, and supervised the study and writing of the manuscript.
This work was supported by Innovatiegerichte Onderzoeksprogramma Genomics [grant number IGE03018], and by the European Commission FP7 grant “Proteins in Health and Disease”, HEALTH-PROT, GA [grant number 229676].
Abbreviations: CLIPS, chaperones linked to protein synthesis; Cy5, indodicarbocyanine; DAPI, 4′,6-diamidino-2-phenylindole; DTT, dithiothreitol; ECL, enhanced chemiluminescence; eGFP, enhanced green fluorescent protein; ER, endoplasmic reticulum; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GFP, green fluorescent protein; HEK, human embryonic kidney; HIP, Hsc (heat-shock cognate) 70-interacting protein; HRP, horseradish peroxidase; HSP, heat-shock protein; Ni-NTA, Ni2+-nitrilotriacetate; polyQ, polyglutamine; QPCR, quantitative PCR; RNAi, RNA interference; siRNA, small interfering RNA; tet, tetracycline
- © The Authors Journal compilation © 2011 Biochemical Society