Characterization of the molecular chaperone ClpB from the pathogenic spirochaete Leptospira interrogans

Leptospira interrogans is a spirochaete responsible for leptospirosis in mammals. The molecular mechanisms of the Leptospira virulence remain mostly unknown. Recently, it has been demonstrated that an AAA+ chaperone ClpB (a member of the Hsp100 family) from L. interrogans (ClpBLi) is not only essential for survival of Leptospira under the thermal and oxidative stresses, but also during infection of a host. The aim of this study was to provide further insight into the role of ClpB in the pathogenic spirochaetes and explore its biochemical properties. We found that a non-hydrolysable ATP analogue, ATPγS, but not AMP-PNP induces the formation of ClpBLi hexamers and stabilizes the associated form of the chaperone. ADP also induces structural changes in ClpBLi and promotes its self-assembly, but does not produce full association into the hexamers. We also demonstrated that ClpBLi exhibits a weak ATPase activity that is stimulated by κ-casein and poly-lysine, and may mediate protein disaggregation independently from the DnaK chaperone system. Unexpectedly, the presence of E. coli DnaK/DnaJ/GrpE did not significantly affect the disaggregation activity of ClpBLi and ClpBLi did not substitute for the ClpBEc function in the clpB-null E. coli strain. This result underscores the species-specificity of the ClpB cooperation with the co-chaperones and is most likely due to a loss of interactions between the ClpBLi middle domain and the E. coli DnaK. We also found that ClpBLi interacts more efficiently with the aggregated G6PDH in the presence of ATPγS rather than ATP. Our results indicate that ClpB’s importance during infection might be due to its role as a molecular chaperone involved in reactivation of protein aggregates.


Introduction
Bacterial ClpB is a molecular chaperone belonging to the Hsp100 subfamily of AAA + ATPases (ATPases associated with a variety of cellular activities) that cooperates with the DnaK a1111111111 a1111111111 a1111111111 a1111111111 a1111111111

Materials and methods Proteins
ClpB Li was successfully overproduced as an N-terminal hexahistidine fusion protein in E. coli BL21(DE3) strain (Novagen) and then purified by immobilized metal affinity chromatography (IMAC) using Co-NTA agarose (Qiagen) and gel filtration chromatography (Superdex 200, Sigma-Aldrich) as previously reported [24]. After purification, the protein was extensively dialyzed against appropriate buffers (as described below in Materials and methods). The N-terminal histidine tag was removed by proteolytic digestion using the Thrombin Cleavage Capture Kit (Novagen) according to the manufacturer's protocol.

Circular dichroism (CD) spectroscopy
The far-UV CD spectra (200-250 nm) of ClpB Li at a concentration of 0.15 mg/ml were recorded in a 50 mM Tris-HCl pH 7.5, 1 mM EDTA, 1 mM DTT, 20 mM MgCl 2 , 200 mM KCl, 10% glycerol buffer, in 1-mm path-length cells, using a Jasco J-815 spectropolarimeter (Japan) equipped with Jasco Peltier element for temperature control. The mean residue ellipticity was calculated according to [31]. To assess the thermal stability of ClpB Li , far-UV CD signals at 222 nm were recorded between 20 and 75˚C with a scan rate of 0.5˚C/min. A transition mid-point temperature (T m ) was calculated by fitting the sigmoidal Boltzman curve to the ellipticity data using the program OriginPro 9.1 (OriginLab Corp., USA, www.originlab.com).

Sedimentation velocity analytical ultracentrifugation
Analytical ultracentrifugation was performed at 20˚C with Beckman Optima XL-I analytical ultracentrifuge equipped with a four-or eight-position An-Ti rotor and UV absorption detection at 290 nm in double-sector 1.2 cm cells with charcoal-filled epon centerpieces and sapphire windows. ClpB Li was dialyzed twice against 50 mM Tris-HCl pH 7.5 buffer containing 0.2 M (or 30 mM) KCl, 20 mM MgCl 2 , 1 mM EDTA, 2 mM β-mercaptoethanol, 5% glycerol, and 400 μl of the dialysate was loaded into reference sectors of the cells. Samples (390 μl) contained ClpB Li (at concentrations of 1.2 or 3 mg/ml) alone or with 2 mM nucleotide: ATPγS (adenosine-5'-(γ-thio)-triphosphate); Sigma), AMP-PNP (adenosine 5 0 -(β,γ-imido)-triphosphate; Sigma) or ADP (Sigma). ATP analogues at the same concentrations were also added to the reference sectors. Sedimentation velocity experiments were performed at 50,000 rpm and terminal domain (ND), nucleotide binding domain 1 (NBD1), middle coiled-coil domain (MD), and nucleotide binding domain 2 (NBD2). The functions of the domains are indicated at the top. The amino acid residue numbers are shown for each chaperone and the amino acid sequence identity between ClpB Ec and ClpB Li is indicated for each domain. (B) CD spectra of ClpB Li at 20˚C (folded form) and 75˚C (unfolded form) are shown. The CD signal was expressed as mean molar residue ellipticity (θ). (C) Temperature-induced changes in the CD signal at 222 nm for ClpB Li .
radial absorption scans of protein-concentration profiles were measured at 4.5-or 5-min intervals. The data were analyzed using the SEDFIT program with continuous sedimentation coefficient distribution c(s) model based on Lamm equation [32]. Integration of the c(s) peaks provided the signal-weighted average sedimentation coefficients (s) and the corresponding standard sedimentation coefficients s 20,w (referring to water solvent at 20˚C). Partial specific volume of ClpB Li (from the amino acid composition) as well as density and viscosity of the buffer were calculated using Sednterp program [33].

Proteolytic sensitivity assay
ClpB Li or ClpB Ec (1 μM) was preincubated in 50 mM Tris-HCl pH 7.5 buffer containing 200 mM KCl, 20 mM MgCl 2 , 1 mM EDTA, 1 mM DTT and 10% glycerol for 10 min on ice without or with 5 mM nucleotides: ATP, ATPγS, ADP, AMP-PNP. Trypsin (Sigma) prepared in 1 mM HCl (at a concentration of 1 mg/ml) was then added to the reaction mixtures to a final concentration of 0.2 ng/μl, and the samples were incubated at 37˚C for the indicated periods (from 0 to 60 min). The reactions were quenched by the addition of Laemmli SDS-PAGE buffer and samples were analyzed by 0.1%SDS-12.5%PAGE. The gels were stained with Coomassie blue dye.

ClpB ATPase assay
ClpB Li and ClpB Ec were incubated in assay buffer (100 mM Tris-HCl, pH 8.0, 1 mM DTT, 1 mM EDTA, 10 mM MgCl 2 and 5 mM ATP) at 37˚C for 30 min without or with 0.1 mg/ml κcasein or 0.04 mg/ml poly-lysine, or 2.1 μM aggregated G6PDH. The concentration of ClpB was 0.05 mg/ml for determination of the basal activity and in the presence of κ-casein, and G6PDH or 0.005 mg/ml in the presence of poly-lysine. Inorganic phosphate concentration was determined using the malachite green dye-based colorimetric assay [34] and detection at A 640 .
All absorbance measurements in this study were performed using a model U-1900 Hitachi UV-VIS spectrophotometer.

Aggregate reactivation assays
The purified Fda (2 μM, in buffer A: 100 mM Tris-HCl pH 7.5, 10 mM MgCl 2 and 0.3 mM ZnCl 2 ) was incubated at 55˚C for 10 min. Subsequently, ATP (5 mM) and chaperones: ClpB Li or ClpB Ec (0.65 μM), DnaK Ec (1 μM), DnaJ Ec (0.2 μM) and GrpE Ec (0.1 μM) were added. The total volume of a reaction mixture was 50 μl. The Fda activity was determined as described by Sigma Quality Test Procedure [35], and the decrease in A 340 was measured after 60-and 120-minute incubation at 25˚C using a spectrophotometer. The aggregated Fda in buffer A with 5 mM ATP, but without the chaperones was used as control.
The chaperone-mediated reactivation of aggregated G6PDH and Fda was monitored in the absence of an ATP regenerating system. Therefore, to avoid significant ATP depletion and ADP accumulation, we limited the measurements to an initial stage of the reaction.

ClpB-aggregate interaction assay
The filtration assay was performed as reported earlier [36]. Aggregated G6PDH (21 μM) was diluted 10-fold by the addition of the refolding buffer B containing ClpB Li (0.65 μM) and 5 mM nucleotide: ADP, ATP, ATPγS, or AMP-PNP. The mixtures were incubated with shaking at 30˚C for 10 min and then applied to the filter devices (Millipore Ultrafree-MC Centrifugal Filter Unit with the membrane pore size 0.1 μm). After 5 min incubation at room temperature, the filter devices were centrifuged at 7,500 g for 5 min to get the flow-through fractions, then washed with the refolding buffer containing an appropriate nucleotide at 30˚C for 5 min and re-centrifuged. Next, SDS-loading buffer (2x) was added to the filter devices and they were incubated at 50˚C for 5 min with shaking. Then, the filter devices were centrifuged to obtain the eluate fractions, which were separated by 0.1%SDS-10%PAGE and stained with Coomassie blue dye. The stained gels were scanned and analyzed with 1Dscan EX, Scananalytics Inc. Sigma program.

Heat-shock survival assay
The clpB Li gene was cloned into a low-copy pGB2 plasmid together with the native E. coli clpB heat-shock promoter (i.e. the σ 32 -dependent promoter). The nucleotide sequence of clpB Li was amplified from genomic DNA of L. interrogans by PCR using AccuTaq LA polymerase MIX (Sigma) with the following PCR primers: CATATGAAATTAATAAA CTTACATCCAAATT with the NdeI restriction site underlined, and AAGCTTTTAAA CTACAACAACTACCTTTCC CTwith the HindIII restriction site underlined. The E. coli σ 32 promoter was amplified from pGB2-ClpB Ec [41] using the following primers: CCCGGGTTCTCGCCTGGTTAGGGC with the XmaI site underlined, and CATATGAACTCCTCCCATAACGGATCwith the NdeI site underlined. First, the PCR products were cloned into pJET1.2 blunt vector (Fermentas), then digested with NdeI, HindIII, and XmaI, and ligated with the linearized pGB2/XmaI-HindIII vector to produce pGB2-ClpB Li . The E. coli MC4100ΔclpB cells were transformed with the empty pGB2, pGB2-ClpB Ec [41], or pGB2-ClpB Li and bacterial survival during heat-shock was determined as described earlier [41]. To detect ClpB in E. coli cultures, Western blotting was performed according to [42] using anti-ClpB Li158-334 serum [18], that recognized both ClpB Li and ClpB Ec , a peroxidase-coupled goat anti-rabbit secondary antibody (Sigma), and visualized with the substrate chromogen, 3,3'-diaminobenzidine tetrachloride (DAB, Sigma) and 30% H 2 O 2 .

Secondary structure and thermal stability of ClpB Li
First, we estimated the secondary structure of the recombinant ClpB Li and its thermal stability by performing CD spectroscopy, which was a prerequisite for further characterization of the chaperone. As shown in Fig 1B, the CD spectrum recorded at 20˚C showed local minima at 208 and 222 nm, which indicates that the recombinant ClpB Li is folded into a structure that is dominated by α-helices. This result is in agreement with the secondary structure of ClpB Ec obtained from spectroscopic measurements [25] and that observed in the crystal structure of ClpB from T. thermophilus [3]. Furthermore, the structure of ClpB Li is thermodynamically stable at the assay temperatures used in this study, as shown by a cooperative unfolding transition ( Fig 1C), with T m of approx. 67˚C. The thermal unfolding of ClpB Li is accompanied by a loss of the α-helical structure (see dotted line in Fig 1B).

Nucleotide-induced oligomerization of ClpB Li
It has been shown that the self-association of ClpB from E. coli (ClpB Ec ) into hexameric ringshaped structures is tightly regulated by protein concentration and enhanced by the presence of nucleotides [9,43]. It has been also shown that hexamerization of ClpB Ec is necessary for its ATPase activity and the biological function [25]. Therefore, we decided to study the selfassembly of the recombinant ClpB Li and answer the question whether it forms nucleotideinduced oligomers. For this purpose, we carried out sedimentation velocity experiments (Figs 2 and 3).
As shown in Fig 3A, ClpB Li in the absence of nucleotides sedimented as a single predominant species with the standard sedimentation coefficient s 20,w = 4.5 S, which agreed with the previously determined sedimentation coefficient of the monomeric ClpB Ec [25]. The addition of a weakly hydrolyzed ATP analogue, ATPγS (Fig 3B), but not AMP-PNP ( Fig 3D) induced efficient self-association of ClpB Li , manifested by the presence of a major peak at 14.1 S in the sedimentation coefficient distribution. The maxima of the distribution are broad, indicating that several types of oligomers are in rapid equilibrium in the presence of ATPγS. In such cases, the peaks appear at intermediate positions, which do not correspond to the s-values of the sedimenting species [44]. The value of 14.1 S is lower but close to the sedimentation coefficient of the hexameric ClpB Ec [9]. Thus, we conclude, that in the presence of ATPγS, the hexamer of ClpB Li is most likely the largest oligomeric species. As shown in Fig 3C, in the presence of ADP, two major components are observed with sedimentation coefficients~4.6 S and 12.1 S. However, the position of the fastest sedimenting peak is about 2 S lower than for the distribution in the presence of ATPγS. This result suggests that ADP does not support full association of ClpB Li . Since a high salt concentration is known to promote dissociation of subunits in oligomeric proteins, we tested whether a low salt concentration may stabilize ClpB hexamers without nucleotides present. Fig 3E shows the sedimentation velocity result obtained for ClpB Li (at a concentration of 3 mg/ml) in a buffer with 30 mM KCl (vs. 200 mM KCl in panels A-D). We found that low salt concentration and an increased ClpB Li concentration did not result in efficient self-association of the chaperone, because ClpB Li sedimented primarily at 5.6 S.
Taken together, our observations (Fig 3 and Table 1) indicate that the ATP analogue, ATPγS, but not ADP, induces an efficient self-association of ClpB Li into hexamers. ADP does induce oligomerization of ClpB Li , but less efficiently than ATPγS.
We also investigated the effect of nucleotides on the structure of ClpB Li by monitoring changes in its proteolytic degradation (Fig 4A). We compared the proteolytic pattern obtained for ClpB Li with that found for ClpB Ec (Fig 4B). The ClpB proteins were digested with trypsin in the absence of nucleotides and in the presence of ATP, ATPγS, ADP and AMP-PNP. Subsequently, the degradation products were separated by SDS-polyacrylamide gel electrophoresis and stained with Coomassie blue dye (Fig 4A and 4B). We found that in the absence of nucleotides, ClpB Li was digested into several fragments in the~25-to~60-kDa range with a complete loss of the intact~100-kDa ClpB Li after 20 min of incubation with trypsin. In the presence of  Characteristics of ClpB from L. interrogans nucleotides, ClpB Li showed varying degrees of protection against trypsin digestion, which indicates that ClpB Li undergoes structural changes upon binding of all tested nucleotides. Interestingly, ClpB Li in the presence of either ATPγS or ADP was more resistant to proteolysis than ClpB Ec , (compare Fig 4A and 4B), as shown by a lack of prominent digestion fragments in ClpB Li with ATPγS and ADP and their presence in ClpB Ec .

Characteristics of ClpB from L. interrogans
A protective effect of nucleotides on ClpB Li was weaker in the presence of ATP or AMP-PNP than with ATPγS or ADP, as shown by low-molecular weight fragments appearing after 10-20 min of incubation with trypsin. A partial loss of protection against trypsin for ClpB Li with ATP, as compared to the ATPγS-bound state, can be due to a fraction of nucleotide-free ClpB Li populated during the ATP turnover. ADP does induce oligomerization of ClpB Li , but not as efficiently as ATPγS (Fig 3B and 3C). Apparently, an incomplete oligomerization induced by ADP provides ClpB Li with a significant protection against trypsin, comparable to that of ATPγS. As for AMP-PNP, a lower extent of the ClpB Li protection can be explained by a low population of the ClpB Li oligomers induced by that ATP analogue (Fig 3D). Overall, the extent of the ClpB Li protection against trypsin in the presence of different nucleotides (Fig 4) correlates with the capability of those nucleotides to stabilize the ClpB Li oligomers (Fig 3).

ATPase activity of ClpB Li
We determined the ATPase activity of ClpB Li under the same conditions as the previously tested ClpB Ec [25,36] using a malachite green phosphate-detection assay. As shown in Fig 5, ClpB Li exhibits a slightly lower basal ATPase activity than ClpB Ec . The presence of unstructured polypeptides: κ-casein or poly-lysine, which are known to enhance the ATPase activity of ClpB Ec up to 5-and 25-fold, respectively [45], caused an increase in the ATPase activity of ClpB Li. Thus, the ATPase of ClpB Li resembles in this respect the ATPase of ClpB Ec . Aggregated glucose-6-phosphate dehydrogenase (G6PDH) did not significantly affect the ATPases of ClpB Li and ClpB Ec .

Chaperone activity of ClpB Li
It is known that ClpB Ec efficiently reactivates aggregated proteins in cooperation with DnaK/ DnaJ/GrpE [1,25]. Therefore, we tested the ability of ClpB Li to reactivate aggregated protein substrates in the presence of the E. coli DnaK chaperone system. In the reactivation assays, we used chemically denatured G6PDH that was previously tested in vitro as a ClpB substrate [4,36] and two new model substrates: thermally aggregated FBP aldolase (Fda) and inclusion bodies (IBs) of VP1-β-galactosidase (VP1LAC). Using the pull-down strategy coupled with the mass spectrometry (MS) analysis, we identified Fda as a putative substrate for ClpB Li (unpublished data). As we reported earlier, ClpB Ec significantly increased the reactivation yield of βgalactosidase aggregated in the form of IBs [38,46].
The reactivation yield for the aggregated G6PDH (Fig 6A) in the presence of ClpB Li was significantly higher than for ClpB Ec or the E. coli DnaK/DnaJ/GrpE system. In the case of reactivation of the aggregated Fda (Fig 6B), ClpB Li was again more effective than ClpB Ec and equally potent as DnaK/DnaJ/GrpE. These results demonstrate an intrinsic disaggregase activity of ClpB Li , which, for the selected substrates, exceeds that of ClpB Ec . However, the aggregate reactivation yields in Fig 6A and 6B obtained with ClpB Li in the presence of the E. coli DnaK/ DnaJ/GrpE were similar to those obtained with ClpB Li alone. This result suggests that ClpB Li does not interact with the E. coli DnaK chaperone system during aggregate reactivation in vitro. The reactivation of β-galactosidase from IBs produced in E. coli (Fig 6C) occurred more efficiently with ClpB Li than in the absence of ClpB, but did not reach the efficiency of ClpB Ec . This observation, again, suggests that ClpB Li is capable of reactivating protein aggregates, but does not cooperate with the E. coli co-chaperones. Altogether, these properties of ClpB Li are similar to the previously investigated ClpB from a zoonotic bacterium Ehrlichia chaffeensis [36].

Interaction of ClpB Li with aggregated protein substrates
Previous studies demonstrated that nucleotides regulate interactions of bacterial ClpBs with the aggregated substrates. It has been shown that only ATPγS promotes significant binding of ClpB Ec to the aggregates [13,36], while ClpB from E. chaffeensis interacts more efficiently with aggregates in the presence of the hydrolysable ATP [36]. Other nucleotides, such as ADP and AMP-PNP did not induce significant binding of ClpB Ec to protein aggregates. We tested whether binding of ClpB Li to the aggregates is more efficiently stimulated by ATPγS (which induces a "frozen" ATP-like ClpB state) or by the hydrolysable ATP. We incubated ClpB Li with the aggregated G6PDH in the absence and presence of the tested nucleotides and then determined the amount of ClpB Li bound to aggregates (Fig 7) by performing a filtration assay (see Materials and methods). Only background amounts of ClpB Li were detected in the Characteristics of ClpB from L. interrogans absence of the aggregates. We found that interactions of ClpB Li with the aggregated G6PDH were more effective in the presence of ATPγS than in the presence of ATP, similar to ClpB Ec . Under the conditions of ATP turnover, ClpB Li appears to lose the capability of binding stably to protein aggregates.

Heat-shock survival of the clpB-null E. coli in the presence of ClpB Li
In E. coli, ClpB is necessary for bacterial survival under heat shock [12]. It was demonstrated that the lack of a functional ClpB decreased the growth rate of E. coli at 45˚C and inhibited cell survival at 50˚C. We investigated whether ClpB Li can function in the E. coli cells and substitute for ClpB Ec . We cloned the clpB Li gene into a low-copy pGB2 plasmid together with the native E. coli σ 32 -dependent clpB heat shock promoter. Next, the resulting plasmid pGB2-ClpB Li and the control plasmids, pGB2 and pGB2-ClpB Ec were introduced into the E. coli ΔclpB mutant and a heat-shock survival assay was performed (see Materials and methods). As shown in Fig  8A, the heat-inducible expression of pGB2-ClpB Li produced a in the~100-kD protein detectable by Western blotting using anti-ClpB Li158-334 serum [18]. We observed, however, that ClpB Li was unable to functionally substitute for ClpB Ec and consequently, it did not rescue E. coli ΔclpB mutant under heat shock conditions (Fig 8B and 8C). In contrast, the expression of the clpB Ec gene from pGB2 complemented the effect of the ΔclpB mutation. With ClpB Ec , 80% of the bacteria survived a severe heat shock at 50˚C for 90 minutes, which is consistent with the previous results [12].

Discussion
To date, little is known about the structure and biological role of molecular chaperones from Leptospira spp., including ClpB. It has been demonstrated that the L. interrogans ClpB is essential for the pathogen survival under stress conditions and also during infection of the host [18]. Moreover, a recently reported immunoreactivity of ClpB Li with the sera collected from Leptospira-infected animals [24] and the fact that clpB Li is up-regulated in leptospiral cells [18] suggest that the ClpB activity may be important for pathogenicity of Leptospira. In this work, we described the biochemical and structural properties of ClpB from L. interrogans.
As demonstrated by the sedimentation velocity experiments (Figs 2 and 3), ClpB Li forms hexamers in the presence of the ATP analogue, ATPγS (Fig 3B), while it exists as a monomeric protein in the absence of nucleotides (Fig 3A). In contrast to ATPγS, ADP induces partial selfassociation of ClpB Li (Fig 3C). Insofar as ATPγS-induced hexamerization is rather typical for bacterial ClpB homologues, the effect of ADP on their oligomerization in vitro appears species-dependent. Specifically, ADP stabilized the hexameric forms of the yeast ClpB orthologue, Hsp104 [47,48] or ClpB from the halophilic lactic acid bacterium Tetragenococcus halophilus [49], whereas ClpB Ec [43] and the T. thermophilus ClpB [50] did not fully assemble into hexamers in the presence of ADP, similar to our results for ClpB Li . Unexpectedly, another ATP analogue, AMP-PNP, did not induce an effective assembly of oligomeric ClpB Li (Fig 3D). This result indicates that AMP-PNP binding inhibits self-association of ClpB Li or that the affinity of AMP-PNP towards ClpB Li is low. A similar effect, where AMP-PNP did not induce oligomerization was observed for ClpB from Tetragenococcus halophilus [49]. We also tested the effect of a low salt concentration on the ClpB Li oligomerization in the absence of nucleotide and at an increased concentration of the chaperone (Fig 3E). In contrast to ClpB Ec [51], a decreased salt concentration did not stabilize the ClpB Li oligomers. The results presented in Fig 4A show a protective effect of nucleotides on ClpB Li during trypsin digestion with the strongest effect for ATPγS and ADP and a weaker one with AMP-PNP, which correlates with the extent of oligomerization induced by these nucleotides (see Fig 3).
Like other previously characterized ClpB proteins, ClpB from L. interrogans catalyzes the hydrolysis of ATP (Fig 5), stably binds to aggregated substrates in the presence of an ATP analogue (Fig 7), and shows a disaggregase activity towards aggregated proteins: G6PDH, Fda and β-galactosidase trapped in IBs (Fig 6). It is worth noting that the reactivation yield of Fda (see Fig 6B) in the presence of ClpB Li alone was similar to that obtained with the E. coli DnaK chaperone system, but ClpB Li was significantly more efficient in mediating the aggregate reactivation than ClpB Ec . It has been previously shown that the DnaK chaperone system can disaggregate some substrates, specifically small-size and lowcomplexity aggregates [28,52,53]. Thus, the intrinsic disaggregase activity of ClpB Li in Fig  6A and 6B manifesting in the absence of the co-chaperones and exceeding that of ClpB Ec suggests that the range of potential aggregated substrates of ClpB Li may be broader than for ClpB Ec and may overlap with that of DnaK.
Furthermore, the DnaK chaperone system from E. coli increased the efficiency of aggregate reactivation mediated by ClpB Ec , but not ClpB Li (see Fig 6A and 6B). A similar effect was observed before for ClpB from a parasite Plasmodium falciparum [8]. The apparent lack of cooperation between ClpB Li and the E. coli co-chaperones in vitro (Fig 6) is consistent with the result of an in vivo assay presented in Fig 8 (see panels B and C), which shows that ClpB Li is unable to restore the viability of E. coli ΔclpB cells after heat shock. This property of ClpB Li resembles ClpB from E. chaffeensis, for which, however, the E. coli DnaK system potentiated the chaperone activity during reactivation of aggregated proteins in vitro [36]. The apparent lack of an efficient cooperation of ClpB Li with E. coli co-chaperones during protein disaggregation both in vivo and in vitro is likely due to the species-specificity of multi-chaperone networks, as reported before [5,6,8]. As has been previously shown, the middle domain of ClpB is responsible for the species-specific cooperation among the chaperones [6]. The sequence identity between ClpB Li and ClpB Ec within the middle domain is only~45% (Fig 1A), which is apparently insufficient to support a cooperation between ClpB Li and the E. coli DnaK system. The lack of cooperation between ClpB Li and the E. coli co-chaperones also possibly explains an inability of ClpB Li to rescue the survival of E. coli under heat shock (Fig 8, panels B and C). However, the results in Fig 8 could be also explained by a lower potency of ClpB Li towards proteins aggregated in heat-shocked E. coli vs. those accumulating in the chaperone's native environment of Leptospira during infection of the host.

Conclusions
Our studies revealed several crucial structural and biochemical properties of the molecular chaperone ClpB from Leptospira, which may support its role in pathogenicity of spirochaetes. We showed that ClpB Li forms hexameric assemblies that are stabilized and interact with protein aggregates in the ATP-bound state. Moreover, ClpB Li exhibits a protein disaggregase activity that may contribute to the survival of L. interrogans under the host-induced proteotoxic stress. Our studies make a novel contribution to the largely uncharacterized biology of Leptospira and suggest that the L. interrogans and E. coli chaperones evolved differently to respond to the different nature of stress that the proteomes of either bacteria can be exposed to.