Title: The GroE chaperonin machine is a major modulator of the CIRCE heat shock regulon of Bacillus subtilis
Abstract: Article1 August 1997free access The GroE chaperonin machine is a major modulator of the CIRCE heat shock regulon of Bacillus subtilis Axel Mogk Axel Mogk Institute of Genetics, University of Bayreuth, D-95440 Bayreuth, Germany Search for more papers by this author Georg Homuth Georg Homuth Institute of Genetics, University of Bayreuth, D-95440 Bayreuth, Germany Search for more papers by this author Christian Scholz Christian Scholz Laboratorium für Biochemie, University of Bayreuth, D-95440 Bayreuth, Germany Search for more papers by this author Lana Kim Lana Kim Institute of Genetics, University of Bayreuth, D-95440 Bayreuth, Germany Search for more papers by this author Franz X. Schmid Franz X. Schmid Laboratorium für Biochemie, University of Bayreuth, D-95440 Bayreuth, Germany Search for more papers by this author Wolfgang Schumann Corresponding Author Wolfgang Schumann Institute of Genetics, University of Bayreuth, D-95440 Bayreuth, Germany Search for more papers by this author Axel Mogk Axel Mogk Institute of Genetics, University of Bayreuth, D-95440 Bayreuth, Germany Search for more papers by this author Georg Homuth Georg Homuth Institute of Genetics, University of Bayreuth, D-95440 Bayreuth, Germany Search for more papers by this author Christian Scholz Christian Scholz Laboratorium für Biochemie, University of Bayreuth, D-95440 Bayreuth, Germany Search for more papers by this author Lana Kim Lana Kim Institute of Genetics, University of Bayreuth, D-95440 Bayreuth, Germany Search for more papers by this author Franz X. Schmid Franz X. Schmid Laboratorium für Biochemie, University of Bayreuth, D-95440 Bayreuth, Germany Search for more papers by this author Wolfgang Schumann Corresponding Author Wolfgang Schumann Institute of Genetics, University of Bayreuth, D-95440 Bayreuth, Germany Search for more papers by this author Author Information Axel Mogk1, Georg Homuth1, Christian Scholz2, Lana Kim1, Franz X. Schmid2 and Wolfgang Schumann 1 1Institute of Genetics, University of Bayreuth, D-95440 Bayreuth, Germany 2Laboratorium für Biochemie, University of Bayreuth, D-95440 Bayreuth, Germany The EMBO Journal (1997)16:4579-4590https://doi.org/10.1093/emboj/16.15.4579 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info Class I heat-inducible genes in Bacillus subtilis consist of the heptacistronic dnaK and the bicistronic groE operon and form the CIRCE regulon. Both operons are negatively regulated at the level of transcription by the HrcA repressor interacting with its operator, the CIRCE element. Here, we demonstrate that the DnaK chaperone machine is not involved in the regulation of HrcA and that the GroE chaperonin exerts a negative effect in the post-transcriptional control of HrcA. When expression of the groE operon was turned off, the dnaK operon was significantly activated and large amounts of apparently inactive HrcA repressor were produced. Overproduction of GroEL, on the other hand, resulted in decreased expression of the dnaK operon. Introduction of the hrcA gene and its operator into Escherichia coli was sufficient to elicit a transient heat shock response, indicating that no additional Bacillus-specific gene(s) was needed. As in B.subtilis, the groEL gene of E.coli negatively influenced the activity of HrcA. HrcA could be overproduced in E.coli, but formed inclusion bodies which could be dissolved in 8 M urea. Upon removal of urea, HrcA had a strong tendency to aggregate, but aggregation could be suppressed significantly by the addition of GroEL. Purified HrcA repressor was able specifically to retard a DNA fragment containing the CIRCE element, and the amount of retarded DNA was increased significantly in the presence of GroEL. These results suggest that the GroE chaperonin machine modulates the activity of the HrcA repressor and therefore point to a novel function of GroE as a modulator of the heat shock response. Introduction Regulation of the heat shock response in bacteria has been studied most extensively in Escherichia coli (for recent reviews, see Bukau, 1993; Yura et al., 1993; Georgopoulos et al., 1994). The positive regulator of the major heat shock regulon, the sigma-32 regulon, is the σ32 subunit of RNA polymerase (RNAP), which confers to core RNAP the specificity to transcribe heat shock genes (Grossman et al., 1984; Landick et al., 1984; Cowing et al., 1985). Induction of the heat shock response is achieved through a rapid increase in the levels of active σ32 because of enhanced synthesis and stabilization (Grossman et al., 1987; Straus et al., 1987; Tilly et al., 1989). Genetic data show that at least three heat shock proteins, DnaK, DnaJ and GrpE, are involved in negative regulation of the heat shock response (Tilly et al., 1983, 1989; Sell et al., 1990; Straus et al., 1990). Since these genes act synergistically in other reactions (reviewed by Georgopoulos and Welch, 1993), they are often referred to as the DnaK chaperone machine, where the DnaK protein is the major chaperone protein and GrpE and DnaJ act as co-chaperones (Georgopoulos, 1992). Mutants in any of the three genes overproduce heat shock proteins (HSPs) and are deficient in shut off of the heat shock response. At physiological temperatures, the DnaK chaperone system physically interacts with σ32 (Gamer et al., 1992; Liberek et al., 1992; Liberek and Georgopoulos, 1993) and this inhibits the heat shock response (Liberek and Georgopoulos, 1993; Gamer et al., 1996). In Bacillus subtilis, the heat shock response seems to be regulated differently. Three classes of heat shock genes have been described which are all regulated at the level of transcription (for recent reviews, see Hecker et al., 1996; Schumann, 1996). While class I heat shock genes are under negative control by a repressor protein encoded by the hrcA gene (Yuan and Wong, 1995a; Schulz and Schumann, 1996), those of class II are positively regulated by the alternate sigma factor σB (Haldenwang, 1995). Heat shock genes belonging neither to class I or class II have been grouped into class III, including clpP (Völker et al., 1994), clpC (Krüger et al., 1994), lon (Riethdorf et al., 1994), ftsH (Deuerling et al., 1995) and htpG (Schulz et al., 1997). Since the mechanism of regulation of these genes is largely unknown, class III might be heterogeneous. All heat shock genes together constitute the heat shock stimulon, and it is an open question whether there is cross-talk among the three classes (Schumann, 1996). Nine class I heat shock genes have been identified so far and are organized in two operons, the heptacistronic dnaK and the bicistronic groE operon (Schmidt et al., 1992; Homuth et al., 1997). The dnaK operon consists of the genes hrcA–grpE–dnaK–dnaJ–orf35–orf28–orf50 (Homuth et al., 1997), where hrcA encodes a negative regulator of class I heat shock genes (Schulz and Schumann, 1996) which interacts with an operator sequence (Yuan and Wong, 1995a) composed of a perfect inverted repeat of 9 bp separated by a 9 bp spacer designated the CIRCE element (controlling inverted repeat of chaperone expression) by us (Zuber and Schumann, 1994). Transcriptional analysis of the dnaK operon revealed two σA-dependent promoters, one preceding the whole operon and being heat-inducible (Wetzstein et al., 1992), the other not being heat-inducible and located in front of dnaJ (Homuth et al., 1997). The groE operon is bicistronic, as has been described for most bacterial species, and consists of the two genes groES and groEL whose products form the chaperonin machine. There is only one transcript whose synthesis is enhanced transiently after heat shock (Schmidt et al., 1992) at a σA-dependent promoter (Yuan and Wong, 1995b). The repressor protein HrcA and its operator CIRCE are the crucial elements in the regulation of class I heat shock genes. Whereas the hrcA gene has been described in 10 different species so far, the CIRCE element has been found >60 times in >30 bacterial species, suggesting that this regulation mechanism is widespread among eubacteria (Hecker et al., 1996). Here, we address the question of how the HrcA repressor is transiently inactivated to allow enhanced expression of the genes of the CIRCE regulon. Our results unambigously show that the GroE chaperonin machinery is the major modulator of the CIRCE regulon. Results Genes grpE, dnaK and dnaJ are essential for growth and survival only at high temperatures Recently, we reported on the isolation of a deletion/insertion mutation within the chromosomal dnaK gene (Schulz et al., 1995). This mutant turned out to be viable and to form colonies in the temperature range 16–51°C, but was unable to grow at temperatures >51°C and exhibited a filamentous phenotype. Knockouts in genes grpE and dnaJ obtained by insertion of a cat cassette near their 5′ ends exhibited a comparable phenotype. Both genes could be inactivated, demonstrating that they are non-essential. To prove that insertion of the cat cassette led to inactivation of the three genes, the presence of their proteins was analysed by immunoblotting before and after exposure to heat. Low amounts of DnaK, DnaJ and GrpE were present in the wild-type strain 1012, which increased after thermal upshift (Figure 1). In the three knockouts, not only were the proteins encoded by the inactivated genes absent, but also those of the downstream genes, due to a polar effect exerted by the cat cassette (Figure 1). These results indicate that the genes encoding the DnaK chaperone machine are dispensable at temperatures below 51°C, which is in contrast to data published for E.coli, where dnaK and dnaJ null mutants grow extremely poorly at all temperatures (Bukau et al., 1988) and where the grpE gene could not be deleted in a wild-type background (Ang and Georgopoulos, 1989). Figure 1.Synthesis of class I heat shock proteins in wild-type and mutant strains. Total protein cell extracts were prepared from wild-type and three mutant strains LK06 (grpE::cat), BT02 (dnaK::cat) and GH05 (dnaJ::cat) grown at 37°C or subjected to a 15 or 30 min thermal upshift at 50°C. Cell concentrations were equivalent prior to extract preparations. Each set of samples was first loaded on an SDS–PAGE gel and stained with Coomassie brilliant blue to ensure that sample concentrations were equivalent. Samples run on an SDS–polyacrylamide gel were identified by immunoblot analysis. Western blots of the same membrane were performed sequentially using GrpE-, DnaK-, DnaJ- or GroEL-specific antisera. Download figure Download PowerPoint The DnaK chaperone machine does not modulate regulation of class I heat shock genes In E.coli, the DnaK chaperone system formed by the HSPs DnaK, DnaJ and GrpE negatively modulates transcription of the heat shock response by affecting the synthesis, activity and stability of σ32 (reviewed by Bukau, 1993; Yura et al., 1993; Missiakas et al., 1996). Here, we wished to determine whether these genes also regulate expression of the CIRCE regulon. To assess expression of the dnaK and groE operons quantitatively in vivo, we measured the β-galactosidase activity of hrcA–bgaB and groE–bgaB operon fusions in the three knockout strains described above. These two transcriptional fusions have been integrated at the amyE locus of the B.subtilis chromosome to ensure unimpaired expression from the heat shock operons. As controls, the β-galactosidase activity was also determined in the wild-type strain 1012 and in two different isogenic hrcA null mutants. While the activity of the reporter enzyme increased ∼10-fold in the wild-type strain after temperature upshift, high constitutive levels were observed in the hrcA deletion and in the hrcA::cat strains, which were comparable with the value observed 15 min after heat induction in the wild-type strain, thereby confirming the role of HrcA as a repressor of both operons (Table I). In the three mutant strains involving grpE, dnaK or dnaJ, both the basal and the induction levels were comparable with those measured in the wild-type strain (Table I). We conclude from these experiments that the DnaK chaperone machine in B.subtilis influences neither the basal nor the heat-induced level for the two operons of the CIRCE regulon. We also tested for a putative influence of the DnaK chaperone machinery on expression of class II (ctc–bgaB) and III (clpC–bgaB) heat shock genes, and could not find any (data not shown). Table 1. The grpE, dnaK and dnaJ genes do not influence induction of the heat shock response of class I genes Strain Transcriptional fusion U/mg of protein 0 min 15 min 30 min after heat induction 1012 PhrcA–bgaB 24 210 220 ΔhrcA PhrcA–bgaB 170 260 300 hrcA::cat PhrcA–bgaB 155 330 360 grpE::cat PhrcA–bgaB 27 224 184 dnaK::cat PhrcA–bgaB 22 180 190 dnaJ::cat PhrcA–bgaB 24 194 196 1012 PgroE–bgaB 80 280 300 ΔhrcA PgroE–bgaB 280 580 550 hrcA::cat PgroE–bgaB 200 440 520 grpE::cat PgroE–bgaB 70 250 280 dnaK::cat PgroE–bgaB 75 240 250 dnaJ::cat PgroE–bgaB 85 240 260 Wild-type strain 1012 and several isogenic mutant strains carrying transcriptional fusions to the two class I operons were grown in LB medium to mid-exponential phase at 37°C and then heat-induced at 50°C. Samples were withdrawn immediately before (0 min) and at 15 and 30 min after temperature upshift, and β-galactosidase activities were determined as described (Mogk et al., 1996). To verify these results by an independent experiment, an immunoblot analysis was performed to determine the relative level of GroEL protein. Heat induction resulted in comparable increases in the amount of GroEL protein in the wild-type strain and in the three mutants (Figure 1). These results unequivocally demonstrate that induction of groEL occurred unimpaired in all three mutants, and confirm our previous conclusion that the DnaK chaperone machine influences neither the basal level nor the heat induction of class I genes. Next, we tested whether the DnaK system is involved in turning off the CIRCE regulon by quantifying the amount of groEL mRNA by slot-blotting. In the wild-type strain, the level of groEL-specific transcript dramatically increased within the first 5 min after heat induction, peaked at ∼10 min and then declined (Figure 2). The same transcription profiles were observed in grpE, dnaK and dnaJ null mutants. In summary, all these results clearly demonstrate that the DnaK chaperone system does not modulate expression of class I shock genes. Figure 2.Concentration of groEL mRNA in different B.subtilis strains. Slot-blot analysis of total RNA isolated before (0 min) and after a heat shock from 37 to 50°C (5, 10, 15, 30 and 60 min). Bacillus subtilis strains 1012 (WT), grpE::cat (E::cat), dnaK::cat (K::cat) and dnaJ::cat (J::cat) were analysed. Digoxigenin-labelled riboprobe RNA complementary to groEL mRNA was used as probe. Aliquots of 1 μg of total RNA were applied per slot. Download figure Download PowerPoint The amount of GroES/GroEL influences expression of the dnaK and groE operons Because the DnaK chaperone system is not involved in regulating expression of class I heat shock genes, we searched for a possible role for the groE operon, the second operon of the CIRCE regulon. Since isolation of null mutations in either groES or groEL was not possible (Li and Wong, 1992), we decided to fuse the chromosomal copy of the groE operon to a controllable promoter, thereby allowing depletion of the two proteins (strain AMX1). We have chosen a xylose-inducible promoter system where expression of the desired operon is negatively controlled at the level of transcription by the xylose repressor protein (Rygus et al., 1991). The advantage of this promoter system is that it is tightly controlled in the absence of inducer and that its expression can be modulated by different xylose concentrations added to the medium (Kim et al., 1996). When plated on LB agar in the absence of xylose, strain AMX1 failed to form colonies, confirming that expression of the groE operon is absolutely required for growth. When growth of AMX1 was followed in liquid LB medium, the generation time was strictly dependent on the concentration of added inducer at between 0.1 and 2.0% of xylose (data not shown). Does the amount of GroEL protein within the cells correlate with the xylose concentration added to the medium? GroEL was not detectable in cells incubated for ∼4 h in the absence of xylose, whereas addition of increasing amounts of inducer up to 2% resulted in a linear increase in the amount of GroEL (Figure 3), and the 2% value was slightly above the basal level found in wild-type cells and significantly less than the amount detected in heat-shocked cells (data not shown). These data clearly reveal that a certain amount of GroEL and, most probably, GroES protein is absolutely required for growth, thereby confirming the data published by Li and Wong (1992), and further demonstrate that the amount of GroEL in strain AMX1 can be manipulated by the xylose concentration in the medium. Figure 3.Cellular levels of class I heat shock proteins depend on the amount of GroEL present in the cells. The amount of three different heat shock proteins was measured in B.subtilis strain AMX1 grown in the presence of different xylose concentrations. Other details are as for Figure 1. Download figure Download PowerPoint The next step was to assess the basal level of expression of the dnaK and the groE operons in the presence of varying amounts of GroE proteins. Expression of the lepA gene (lep stands for leader peptidase) was included as a control since the lepA gene does not respond to a heat shock (Homuth et al., 1996). This was achieved by transforming the hrcA−, groE− and lepA–bgaB fusions (all inserted at the amyE locus) separately into AMX1, resulting in strains AMX2, AMX3 and AMX4, respectively, and measuring the BgaB activity. In the absence of inducer, the β-galactosidase activity of hrcA–bgaB was 10.7-fold higher at 37°C as compared with its level in the uninduced wild-type strain, and reached a level comparable with that found either in the absence of repressor or 30 min after thermal upshift in the wild-type strain (Figure 4A). Growth in the presence of 0.1% xylose still produced a 3.8-fold increase in the basal level, and growth in the presence of 0.5 or 2.0% xylose resulted in β-galactosidase activities comparable with those found in the uninduced wild-type strain. A thermal upshift in the absence of inducer produced only a slight induction of the operon fusion, most probably due to its already high expression, whereas heat induction in the presence of 0.5 or 2% xylose resulted in a normal heat shock response (data not shown). For groE–bgaB, a comparable expression pattern was obtained, resulting in an ∼6-fold higher expression in the complete absence of xylose (Figure 4B). In contrast, the β-galactosidase activity in cells carrying the lepA–bgaB fusion remained about constant, independently of the xylose concentrations within the medium (Figure 4C). Figure 4.The groE operon modulates expression of the dnaK operon. Expression of the three transcriptional fusions hrcA–bgaB (A), groE–bgaB (B) and lepA–bgaB (C) was measured in strain AMX1 in the absence (first column) or in the presence of increasing amounts of xylose (next three columns). Wild-type 0 and 30 mean strain 1012 before and 30 min after thermal upshift to 50°C. The data in the last column refer to 1012 carrying a deletion of hrcA. Download figure Download PowerPoint In parallel, we measured the amount of DnaK and HrcA by immunoblotting. Whereas the amount of DnaK was high in the absence of GroEL (Figure 3), it decreased when the xylose concentration, and thereby that of GroEL, was enhanced. In accordance with these findings, the amount of HrcA repressor was also high when cells were grown in the absence of xylose (Figure 3). This result was astonishing since high amounts of repressor should result in strong repression of the dnaK operon. Since this was not the case, the repressor protein seems to be unable to interact with its operator. We conclude from these experiments that the groE operon modulates expression of the dnaK operon and of itself by acting as a negative regulator. If reduced levels of GroE proteins enhance the basal level of expression of class I genes, overexpression of GroEL on the other hand should result in reduced expression of these genes. To test this assumption, groEL was fused to an isopropyl-β-D-thiogalactopyranoside (IPTG)-inducible promoter on the high copy number plasmid pREP9, and the new plasmid pREP9-groEL was transformed into three different reporter strains. Growth in the presence of 2 mM IPTG resulted in an ∼25-fold overexpression of GroEL as compared with the amount present in wild-type cells before heat induction (data not shown). This high amount of GroEL reduced the basal level of hrcA–bgaB expression by a factor of ∼4 and, in addition, greatly diminished the heat shock response by a factor of 4–5, but the induction factor remained the same in both strains (7- to 8-fold; Figure 5A). A comparable result was found for the groE–bgaB fusion, where the basal level of expression was also reduced ∼4-fold, and the induction factor remained the same under both conditions (Figure 5B). When the lepA–bgaB fusion was analysed under these conditions, overproduction of GroEL did not influence its expression either before or after a heat shock (Figure 5B). These results are consistent with the conclusion that groEL acts as a negative modulator of class I heat shock genes. Taking all these results together, the GroE chaperonin rather than the DnaK chaperone machine modulates expression of the CIRCE regulon in B.subtilis. Figure 5.Overproduction of GroEL leads to a reduced class I heat shock response in B.subtilis. Strain 1012 carrying transcriptional fusions between two different promoter regions and bgaB and pREP9-groEL were grown in the absence or presence of 2 mM IPTG to induce synthesis of GroEL. Strains were grown to mid-logarithmic phase, the BgaB activities were measured and the amount found in the absence of IPTG prior to heat shock was set as one. (A) Strain AM03 (hrcA–bgaB), (B) strain AM02 (groE–bgaB) and (C) strain AM04 (lepA–bgaB) carrying pREP9-groEL, respectively. Empty bars, no IPTG; closed bars, 2 mM IPTG added. Download figure Download PowerPoint A complete heat shock response can be obtained in E.coli in the presence of the HrcA repressor and its operator, the CIRCE element After having identified the groE operon as a negative modulator of the CIRCE regulon, we asked whether additional gene(s) will influence expression of this regulon. To identify this gene(s), we decided to transfer the known components into E.coli, thereby creating a reporter strain and allowing screening for additional functions by transforming this strain with B.subtilis gene banks. Yuan and Wong (1995a) have already shown that a transcriptional fusion between the groE promoter and the bgaB reporter gene was expressed at a high level in E.coli; addition of hrcA on a second plasmid resulted in a drastic decrease in the β-galactosidase activity. They further reported that treatment of these cells with heat resulted in an increase in enzymatic activity, and speculated that the repressor protein itself might be temperature sensitive (Yuan and Wong, 1995a). First, we performed similar experiments by constructing two different plasmids. One carried a transcriptional fusion between the hrcA promoter including the CIRCE regulatory element and bgaB (pAM100), the other contained, in addition, the hrcA gene expressed from its own promoter but devoid of the CIRCE element, resulting in constitutive expression of hrcA (pAM101). Both plasmids were transformed separately into E.coli strain Ω394, and β-galactosidase activities were measured in both strains before and after temperature upshift. The enzymatic activity found with pAM101 prior to heat induction was set at one relative unit. In the absence of hrcA, ∼30 relative units of β-galactosidase were measured, which did not increase after thermal upshift, as expected (Table II). Upon addition of hrcA, the enzymatic activity dropped to one relative unit in E.coli cells grown at 30°C and increased ∼5-fold after a heat shock to 42°C, thereby confirming the results of Yuan and Wong (1995a). The failure to obtain full induction of the operon fusion after a heat shock can be explained by the fact that the level of HrcA protein is strongly enhanced due to the absence of its operator on pAM101, as visualized by immunoblotting (data not shown). Table 2. Heat regulation of class I genes occurs in E.coli Strain Plasmid hrcA present Relative β-galactosidase activities at 0 15 30 min after heat shock Wild-type pAM100 − 30 32 32 Wild-type pAM101 + 1.0 4.3 5.1 dnaK756 pAM101 + 0.5 1.7 2.9 groEL140 pAM101 + 8.7 11.4 21.3 groES30 pAM101 + 19.2 27.9 32.4 Wild-type strain Ω394 and its isogenic derivatives were transformed with either pAM100 carrying only the hrcA–bgaB transcriptional fusion (no hrcA gene present) or pAM101 containing the fusion and the hrcA gene. The amount of β-galactosidase obtained in the wild-type strain in the presence of hrcA was set as one relative unit. What happens to the repressor protein after a heat shock in E.coli? Will it be inactivated irreversibly due to an intrinsic temperature sensitivity, as suggested by Yuan and Wong (1995a)? If this should turn out to be the case, a prolonged incubation of E.coli cells carrying pAM101 at 42°C should not result in a shut off of the bgaB reporter gene. To answer this question, we measured the amount of bgaB mRNA after a heat shock from 30 to 42°C. A rapid increase after thermal upshift was followed by a decline (Figure 6), as has already been reported for the dnaK operon in B.subtilis (Wetzstein et al., 1992). Since hrcA on pAM101 is not subject to down-regulation (its operator is not present), we additionally infer from this result that turning off of the heat shock response is independent of regulated synthesis of HrcA. Measuring the amount of HrcA protein present before and after temperature upshift by immunoblotting in E.coli wild-type cells revealed that it remained constant (Figure 7). Most importantly, these data clearly argue against an intrinsic temperature-sensitivity of HrcA. They further show that a heat shock response comparable with that found in B.subtilis can be obtained in E.coli in the presence of just the HrcA protein and its operator. We conclude that additional B.subtilis-specific factors are not involved in regulation of the HrcA activity. Figure 6.The dnaK operon of B.subtilis is transiently heat induced in E.coli. Slot-blot analysis with RNA isolated from E.coli strain Ω394 carrying pAM101 grown in LB to mid-logarithmic phase at 30°C and at different times after a shift to 42°C. The bgaB-specific mRNA was detected using digoxigenin-labelled complementary riboprobe RNA, and the amount present before heat induction was set as one. Download figure Download PowerPoint Figure 7.The amount of HrcA protein remains constant in E.coli wild-type and in different isogenic mutant strains. Other details are as for Figure 1. Download figure Download PowerPoint Will the amount of β-galactosidase be influenced by the dnaK or groE allele in E.coli cells? To answer this question, the plasmid pAM101 carrying hrcA and the transcriptional hrcA–bgaB fusion was transformed into different mutant strains. The dnaK756 allele reduced expression of the bgaB gene 2-fold before and after heat shock, but the induction factor remained identical. In contrast, two different groE alleles dramatically increased expression of bgaB 10- to 20-fold (Table II). In all three mutants, the amount of HrcA protein remained constant (Figure 7). We conclude that, though the repressor is present, it is inactive. These data are in agreement with those found with B.subtilis and further underline the role of the GroE chaperonin machine in modulating expression of class I heat shock genes of B.subtilis. Addition of GroEL partially suppressed aggregation of HrcA All these in vivo data point to a physical interaction between the HrcA repressor and GroEL. To examine this assumption, the influence of GroEL on HrcA protein was studied in vitro using purified components. To facilitate purification of HrcA, the B.subtilis hrcA gene was cloned into an expression vector allowing its overexpression in E.coli. Upon addition of the inducer IPTG, large amounts of HrcA protein were synthesized and formed aggregates. We tried unsuccessfully to prevent formation of inclusion bodies by varying the temperature and IPTG concentration or by fusing HrcA to thioredoxin (R.Emmerich, unpublished results). The inclusion bodies could be dissolved in denaturing solvents containing 6 M guanidinium hydrochloride or 8 M urea. Upon removal of the chaotropic agent either by dialysis or by dilution into non-denaturing conditions (several buffers were tried), the HrcA protein aggregated spontaneously. Similar observations have been made by S.-L.Wong also for HrcA of B.subtilis, by T.Ohta for HrcA of Staphylococcus aureus, and by H.Bahl for HrcA of Clostridium acetobutylicum (personal communications). As an alternative, we cloned the hrcA homologue from Bacillus stearothermophilus because this protein from a thermophilic organism should be more stable. When this gene wa