SecB-like chaperone controls a toxin–antitoxin stress-responsive system in Mycobacterium tuberculosis

Abstract

A major step in the biogenesis of newly synthesized precursor proteins in bacteria is their targeting to the Sec translocon at the inner membrane. In Gram-negative bacteria, the chaperone SecB binds nonnative forms of precursors and specifically transfers them to the SecA motor component of the translocase, thus facilitating their export. The major human pathogen Mycobacterium tuberculosis is an unusual Gram-positive bacterium with a well-defined outer membrane and outer membrane proteins. Assistance to precursor proteins by chaperones in this bacterium remains largely unexplored. Here we show that the product of the previously uncharacterized Rv1957 gene of M. tuberculosis can substitute for SecB functions in Escherichia coli and prevent preprotein aggregation in vitro. Interestingly, in M. tuberculosis, Rv1957 is clustered with a functional stress-responsive higB-higA toxin–antitoxin (TA) locus of unknown function. Further in vivo experiments in E. coli and in Mycobacterium marinum strains that do not possess the TA-chaperone locus show that the severe toxicity of the toxin was entirely inhibited when the antitoxin and the chaperone were jointly expressed. We found that Rv1957 acts directly on the antitoxin by preventing its aggregation and protecting it from degradation. Taken together, our results show that the SecB-like chaperone Rv1957 specifically controls a stress-responsive TA system relevant for M. tuberculosis adaptive response.

In bacteria, newly synthesized proteins emerging from the ribosome are assisted by essential molecular chaperones and targeting factors, which facilitate folding and/or partition them as cytosolic, inner-membrane, or exported proteins. Generic molecular chaperones such as DnaK-DnaJ (Hsp70-Hsp40), Trigger Factor, GroEL-GroES (Hsp60-Hsp10), and SecB are key players in this process (1, 2). In most Gram-negative bacteria, a substantial number of newly made presecretory proteins are targeted posttranslationally to the Sec translocon located at the inner membrane by the export chaperone SecB (3–5). SecB is a homotetrameric chaperone of 17 kDa monomers assembled as a dimer of dimers (6) that binds nonnative substrates either co- or posttranslationally with high affinity and maintains them in a translocation-competent state. Binding to the chaperone prevents unproductive folding, aggregation, and degradation of the preprotein clients. Substrate selection by SecB is strongly influenced by the folding rate of proteins, where rapidly folding proteins would escape binding to the chaperone (4, 5). This discriminative model called “kinetic partitioning” reflects the preference of SecB for slow-folding presecretory proteins (7–9). SecB specifically interacts with the SecA motor component of the Sec translocon at the cytoplasmic membrane (10). Such a unique and dedicated partnership between substrate-bound SecB and SecA facilitates efficient preprotein targeting and transfer to the Sec translocon as well as the subsequent translocation through the membrane (11). As a chaperone with generic properties, SecB also binds aggregation-sensitive cytosolic protein substrates and prevents their aggregation in the absence of the cytosolic chaperones DnaK and Trigger Factor (TF) (12).

Because of its role in protein export, the chaperone SecB is considered a proteobacterial invention associated with the presence of an outer membrane present in α-, β-, γ-, and some δ-proteobacteria (13, 14). Nonetheless, genes showing homology to secB are infrequently found in Planctobacteria, Spirochaetae, Sphingobacteria, Eurybacteria, and Endobacteria (13). Functions in protein translocation have not been described for any of these putative SecB-like proteins, and no SecB homologs have been found in mycobacteria. Despite their classification as Gram-positive bacteria, recent discoveries have shown that there is a well-defined outer membrane in mycobacteria (15, 16) and that Mycobacterium tuberculosis encodes a significant number of putative outer membrane proteins (17, 18). How these outer-membrane proteins are assisted in the cytoplasmic space and targeted to the inner membrane in these bacteria remains unknown.

In this work, we first show that the major human pathogen M. tuberculosis encodes a SecB-like chaperone, namely Rv1957, which can functionally replace the Escherichia coli SecB chaperone, both in vivo and in vitro. We also demonstrate that Rv1957 mycobacterial chaperone controls the HigB-HigA stress-responsive toxin–antitoxin (TA) system of M. tuberculosis by acting directly on its cytosolic HigA antitoxin substrate. The role of this atypical SecB-like chaperone of M. tuberculosis is discussed.

Results and Discussion

Rv1957 from M. tuberculosis Shows Similarities to the SecB Chaperone.

Inspection of the M. tuberculosis genome reveals the unexpected presence of a gene (Rv1957) coding for a protein with about 13% identity to the E. coli SecB chaperone. Remarkably, although Rv1957 is present in the other members of the M. tuberculosis complex (MTBC)—i.e., Mycobacterium canetti, Mycobacterium africanum, and Mycobacterium microti—it is not found in the most closely related species outside the MTBC, M. marinum or Mycobacterium kansasii. A refined alignment based on the known E. coli SecB structures and constructed mostly by increasing gaps in the two adjacent β-sheets, β-1 and β-4, in SecB reveals a sequence that is 19% identical and 31% similar to that of SecB (Fig. 1A). This alignment shows that Rv1957 has several clusters of key residues critical for SecB oligomerization and interaction with substrates or with its specific partner SecA at the Sec translocon (19–25). Previous work has shown that there are several conserved interaction sites for SecB on SecA, i.e., at the N-terminal and at the C-terminal zinc-binding region of SecA (20, 24, 26). Remarkably, the housekeeping SecA1 from M. tuberculosis, which does not have the cysteine residues involved in zinc binding at the extreme C-terminal region, retains all of the conserved neighboring residues involved in contact with SecB (19). Whether Rv1957 is capable of interacting with SecA1 remains to be determined. Finally, both Rv1957 and SecB are highly acidic proteins with comparable isoelectric points (4.49 and 4.26, respectively). Such similarities raise the possibility that Rv1957 encodes a functional SecB-like chaperone in M. tuberculosis.

Fig. 1.

A functional SecB-like chaperone in M. tuberculosis. (A) Sequence alignment of the Rv1957 gene product and the E. coli SecB chaperone using MUSCLE (Multiple Sequence Comparison by Log-Expectation) and further refinement using score-assisted manual alignment with Genedoc. The secondary structure of E. coli SecB based on the crystal structure of SecB (PDB entry 1QYN) is shown. β-Sheets are shown in orange and α-helices in light blue. The colored bars above amino acids indicate residues involved in interaction with SecA (red) and interactions with substrates (green). Asterisks indicate known mutations affecting SecB–preprotein interactions by disrupting the oligomeric state of SecB. (B) Midlog phase cultures of strain W3110 ΔsecB containing plasmids pFr, pFr-SecB, or pFr-Rv1957 were serially diluted and spotted on LB–ampicillin agar plates without or with 50 μM IPTG and incubated at the indicated temperatures. (C) Pulse-chase analysis showing the processing of ³⁵S-labeled preMBP and proOmpA in strain W3110 ΔsecB containing plasmids pFr, pFr-SecB, or pFr-Rv1957. Cell cultures were grown, labeled, and chased at 30 °C for various times following a 1-h incubation at 30 °C in the presence of 500 μM IPTG.

Rv1957 Exhibits SecB-Like Chaperone Functions both in Vivo and in Vitro.

Deletion of secB in E. coli confers a SecB-dependent cold-sensitive growth phenotype at temperatures below 23 °C. This is due to a strong export defect for many proteins (27, 28). To examine whether Rv1957 could replace SecB in vivo, the Rv1957 gene was cloned on a low-copy plasmid under the control of a lac-inducible promoter and tested for its ability to complement the secB mutant phenotype in E. coli (Fig. 1B). Remarkably, plasmid-encoded Rv1957 fully suppressed the cold-sensitive phenotype of the secB mutant, even at the stringent temperature of 16 °C. As expected, bacterial growth at low temperature was also restored by plasmid-encoded SecB and not by the empty plasmid control. Note that SecB was sufficiently expressed without inducer to allow complementation (Fig. 1B). In contrast, full complementation by Rv1957 required induction with IPTG (Isopropyl β-D-1-thiogalactopyranoside), suggesting that Rv1957 may be less efficient than SecB in assisting protein export in E. coli (Fig. 1B). However, the steady-state expression level of Rv1957 was significantly lower than the one observed for SecB (Fig. S1). This could also account, at least in part, for the less efficient complementation by Rv1957. Next, we tested whether Rv1957 could directly assist the export of two well-known SecB substrates, namely OmpA and maltose-binding protein (MBP), in the absence of SecB. Plasmids encoding Rv1957 and SecB were independently transformed into the E. coli secB null strain, and ³⁵S-met pulse-chase experiments followed by immunoprecipitation with anti-OmpA or anti-MBP antibodies were performed. As shown in Fig. 1C, Rv1957 was capable of restoring OmpA and MBP export, even though less efficiently than the E. coli SecB. In this case, rescue by Rv1957 was to some extent more robust for OmpA than for MBP. These results clearly show that Rv1957 from M. tuberculosis exhibits SecB-like chaperone functions in vivo.

TF and DnaK chaperones assist the folding of newly synthesized cytosolic proteins in E. coli. As a consequence, an E. coli strain lacking both of these major chaperones exhibits a strong temperature-sensitive phenotype and accumulates high levels of aggregated cytosolic proteins (12, 29, 30). Overexpression of SecB partially rescues such defects, thus indicating that SecB has a general chaperone function in the absence of these chaperones (12). To further investigate the chaperone function of Rv1957 in a more stringent in vivo system, we asked whether Rv1957 could rescue protein folding and bacterial growth in the absence of TF and DnaK (12). As shown in Fig. 2A, plasmid-encoded Rv1957 partially suppressed protein aggregation in the absence of DnaK and TF. Nevertheless, suppression was considerably less efficient than that exhibited by SecB possibly due to the fact that a fraction of Rv1957 copurified with the aggregates. In this case, aggregation of Rv1957 would decrease its functional concentration. However, it may be that the copurification reflects a bona fide chaperone interaction. In contrast to SecB, suppression of the temperature-sensitive growth phenotype of the Δtig ΔdnaKdnaJ strain by Rv1957 was very weak and only visible after prolonged incubation (2 d) at 32 °C and required expression from the high-copy-number plasmid pSE380 (Fig. 2B). This indicates that Rv1957 is more efficient in assuming the role of SecB in protein export than in rescuing generic chaperones in E. coli.

Fig. 2.

Rv1957 chaperone functions. Rv1957 partially prevents protein aggregation and rescues bacterial growth of the Δtig ΔdnaKdnaJ triple chaperone mutant. (A) Aggregates were extracted from strain MC4100 Δtig ΔdnaKdnaJ containing plasmids pSE380-ΔNcoI vector, pSE-SecB, or pSE-Rv1957 grown at 22 °C following a 2-h expression of SecB and Rv1957 with 500 μM IPTG and a temperature switch at 33 °C for 1 h. Whole-cell extracts (wce) and their corresponding aggregates (agg) are shown. The arrow indicates the fraction of Rv1957 copurifying with the aggregates. (B) Fresh transformants of MC4100 Δtig ΔdnaKdnaJ containing plasmids pSE380ΔNcoI, pSE-SecB, or pSE-Rv1957 were grown to midlog phase at 22 °C. IPTG inducer (0, 0.05, or 0.1 mM) was added, and cultures were incubated for 1 h at the same temperature. Cultures were then serially diluted and spotted on LB–ampicillin agar plates for 48 h at the indicated temperature. (C) Rv1957 prevents purified proOmpC aggregation in vitro. Aggregation kinetics of denatured proOmpC (2 μM) were followed at 25 °C by measuring light scattering at 320 nm in the presence of SecB (dark gray line) or Rv1957 (light gray line) or in the absence of chaperone (black line). The SecB₄ and Rv1957₄ concentrations (1, 2, or 4 μM) are indicated above each graph. The percentage of aggregation was normalized to proOmpC aggregation obtained when no chaperone was added.

To firmly demonstrate that Rv1957 has a bona fide SecB-like chaperone function, we purified Rv1957 and tested its ability to prevent protein aggregation in vitro. We noted that purified Rv1957 migrated in SDS/PAGE at a significantly higher molecular weight than predicted, i.e., 27 kDa versus 20 kDa (Fig. S2A), in agreement with our in vivo experiments shown in Fig. S1. Gel filtration analysis revealed that native Rv1957 has a molecular weight of about 115 kDa, corresponding to that of a homotetrameric protein, as observed for SecB (Fig. S2 A and B). The homotetrameric nature of Rv1957 reinforces the fact both SecB and Rv1957 might be evolutionarily related.

Purified Rv1957 was then compared with SecB for its ability to prevent aggregation of denatured proOmpC, a model outer membrane protein known as a SecB substrate in E. coli (28). Rv1957 efficiently prevented proOmpC aggregation in a manner comparable to that exhibited by the native E. coli SecB (Fig. 2C). These in vitro results are in complete agreement with the previous in vivo experiments and clearly demonstrate that M. tuberculosis possesses a bona fide SecB-like chaperone.

Rv1957 Controls a Toxin–Antitoxin Stress-Responsive System.

The observation that Rv1957 is not found in all Mycobacteria suggests that it may not fulfill a general SecB-like chaperone function in protein export under normal growth conditions. Remarkably, Rv1957 is clustered together with genes that are part of a functional type II TA locus of unknown function, which is related to the higBA family (host inhibition of growth) (Fig. 3A) (31, 32). TA loci encode two-component systems composed of a toxin and an antitoxin, which are essential elements of the prokaryotic response to environmental insults (33–35). Under specific stress conditions such as nutritional stress, the antitoxin, which normally forms an inactive complex with its specific toxin partner, is degraded by activated proteases. The resulting free active toxin subsequently degrades or inhibits its cellular targets (RNAs or proteins, respectively), thus modulating the global level of translation or DNA replication (33–35). Such a TA-dependent response significantly reduces cell growth and favors adaptation to stress and long-term survival. In this respect, it is interesting to note that TA systems are abundant in some pathogenic bacteria including Vibrio cholerae and M. tuberculosis (31, 32, 34) and thus represent a promising class of antimicrobial targets (34).

Fig. 3.

SecB-like chaperone Rv1957 specifically controls the HigB-HigA toxin–antitoxin system in M. tuberculosis. (A) On-scale schematic of the genomic organization of Rv1957 (546 bp) together with higB toxin (378 bp), higA antitoxin (450 bp), and the Rv1954a (303 bp) gene of unknown function. The two known promoters P1 and P2 are also shown. (B) Suppression of HigB toxicity by HigA and Rv1957 in E. coli. Strain W3110 containing the plasmid pSE380ΔNcoI or pSE-Rv1957 was transformed with pMPMK6 vector, pK6-HigB, pK6-HigA, or pK6-HigB-HigA on LB–ampicillin–kanamycin agar plates containing 0.2% glucose at 37 °C. Double transformants were grown in LB–ampicillin–kanamycin 0.2% glucose to midlog phase, serially diluted, and spotted on LB–ampicillin–kanamycin agar plates with or without IPTG inducer as indicated. Plates were incubated at 37 °C overnight. (C) Overexpression of the E. coli chaperone SecB can replace Rv1957 in modulating the HigB-HigA toxin–antitoxin system. Experiments were carried out as described in B except that pSE-Rv1957 was replaced by pSE-SecB. (D) Suppression of HigB toxicity by HigA and Rv1957 in mycobacteria. Plasmids encoding the toxin HigB, HigB-HigA, or HigB-HigA plus the Rv1957 chaperone under the control of the hsp60 promoter were electroporated into the M. marinum M strain, and plates were incubated for 2 wk at 30 °C.

Previous work has shown that transcription of the higB-higA-Rv1957 operon is induced by DNA damage (36, 37), heat shock (38), nutrient starvation (39), and hypoxia (32, 40), suggesting a role in an adaptive response of M. tuberculosis to stress. Furthermore, deletion of the antitoxin gene higA alone inhibited bacterial growth whereas deletion of the entire higB-higA-Rv1957 operon had no apparent effect (41). Remarkably, disruption of Rv1957 resulted in a slow growth phenotype (42). Because molecular chaperones are often key players of the stress response, we hypothesized that the SecB-like chaperone Rv1957 could be involved in the modulation of such a TA system in M. tuberculosis. It was previously shown that, as a functional antitoxin, overexpression of HigA, separately from HigB, in a heterologous E. coli protease-deficient strain neutralized the HigB-dependent inhibition of growth (31). To investigate such a putative TA-chaperone (TAC) system under more stringent conditions, and thus avoid nonnative unbalanced HigA/HigB levels, we decided to keep the native higB-higA gene organization intact. We cloned higB (toxin), higA (antitoxin), or both higB-higA on a low-copy plasmid under the control of a tightly regulated pBAD promoter to minimize background toxicity of the toxin. All three constructs were expressed in an E. coli wild-type strain in the presence or in the absence of the Rv1957 chaperone cloned on a compatible plasmid under the control of a lac-inducible promoter. As shown in Fig. 3B, expression of the toxin HigB alone or in the presence of either the HigA antitoxin or the Rv1957 chaperone efficiently blocked bacterial growth. This indicates that, under the conditions tested, neither the antitoxin nor the chaperone was capable of inactivating the toxin in vivo. In sharp contrast, toxicity of the toxin was totally abolished when the antitoxin and the chaperone were jointly expressed. To further demonstrate that inhibition of the toxin by Rv1957 is associated with its SecB-like chaperone function, we replaced Rv1957 by the E. coli SecB in the same strain expressing the toxin and the antitoxin and monitored antitoxin rescue in vivo as performed in Fig. 3B. In this case, we found that overexpression of E. coli SecB also resulted in toxin inactivation (Fig. 3C). However, inactivation by SecB required at least a 10-fold increase in inducer concentration and was thus significantly less efficient compared with Rv1957. Taken together, these results strongly suggest that Rv1957 controls the HigB-HigA TA system in M. tuberculosis.

To ensure that toxin inactivation by both the antitoxin HigA and the chaperone Rv1957 was not due to the heterologous E. coli system, different fragments of the higB-higA-Rv1957 operon (i.e., higB, higB-higA, or higB-higA-Rv1957) were cloned in a mycobacterial shuttle vector under control of the hsp60 promoter and introduced into wild-type M. marinum, which is closely related to M. tuberculosis but lacking a chromosomal ortholog of the higB-higA-Rv1957 operon. As observed in E. coli, only the construct containing the entire operon could be efficiently introduced in M. marinum. Attempts to introduce constructs containing only the toxin or the toxin–antitoxin-encoding genes resulted in only a very few viable colonies (Fig. 3D). Plasmids from four “survivors” were sequenced together with plasmids from four colonies of the construct containing the entire operon. Although all plasmids in the transformants with the entire operon were intact, all plasmids obtained from the survivors without Rv1957 carried mutations (deletions of variable size covering the toxin-encoding gene and/or the hsp60 promoter region or a mutation within the hsp60 promoter region) (Fig. S3A). Together, these experiments show that both the chaperone and the antitoxin are required to mitigate the toxic effects of HigB toxin in mycobacteria as well as in the E. coli heterologous host. Remarkably, all three constructs could be introduced efficiently into Mycobacterium bovis bacillus Calmette–Guérin, which possesses orthologs of higA, higB, and Rv1957 (Fig. S3B). This suggests that this strain is able to counteract the toxic effects of the toxin by chromosomally expressed antitoxin and a SecB-like chaperone. Taken together, these in vivo results therefore demonstrate that the mycobacterial chaperone specifically controls a functional stress-responsive cytoplasmic TA system.

Rv1957 Chaperone Specifically Assists the HigA Antitoxin.

Antitoxins are generally unstable protease-sensitive proteins (33, 34). This suggests that Rv1957 could act directly on the antitoxin HigA to prevent its degradation and/or assist its correct folding. To shed light on such a mechanism, we tested whether Rv1957 coexpression affected HigA expression and/or solubilization in vivo in E. coli. As shown in Fig. 4A, the antitoxin was efficiently detected in whole-cell fractions only when the Rv1957 chaperone was coexpressed, indicating that Rv1957 protects HigA from degradation by a yet unknown protease(s) (28, 34). A direct in vivo interaction between Rv1957 and HigA was confirmed by pull-down experiments performed in E. coli (Fig. S4). In addition, we found that, in the presence of the chaperone, the antitoxin was found mainly in the soluble fraction (Fig. 4A). Conversely, in the absence of Rv1957, the small fraction of HigA antitoxin that escaped proteolysis was found in the insoluble fraction (Fig. 4A), suggesting that, in the absence of the chaperone, HigA is either rapidly degraded or aggregates. These results were further confirmed using a functional strep-tagged variant of HigA and subsequent visualization with a more sensitive anti-strep antibody (Fig. S5).

Fig. 4.

Rv1957 chaperone specifically assists the HigA antitoxin. (A) Rv1957 prevents both the aggregation and degradation of the antitoxin HigA in vivo. Steady-state levels of HigA in E. coli strain W3110 cotransformed either with the pSE-vector and pK6-HigA or with pSE-Rv1957 and pK6-HigA, grown in the presence of 0.1% arabinose and 50 μM IPTG inducers at 37 °C. Whole-cell extracts (WCE) and soluble and insoluble fractions are shown. HigA antitoxin was detected using rabbit anti-HigA antibodies. (B) Rv1957 efficiently protects the antitoxin HigA from aggregation in vitro. Aggregation kinetics of urea-denatured HigA (3 μM) were followed at 37 °C by measuring light scattering at 320 nm in the presence of Rv1957 (light gray line; 1 μM) or SecB (dark gray line; 1 μM) or in the absence of chaperone (black line). (C) Proposed model for Rv1957 functions based on known SecB properties. The Rv1957 chaperone (C) co- and/or posttranslationally binds nonnative HigA antitoxin (A*), thus protecting it from aggregation and degradation by proteases and maintaining it in a form competent for subsequent interaction with the toxin HigB (T). Binding of HigA to the toxin may trigger chaperone release. The possible existence of a transient tripartite complex between the Rv1957 chaperone and HigB-HigA is not known yet. Alternatively, the chaperone could release the antitoxin before its interaction with the toxin.

To firmly demonstrate that Rv1957 efficiently stimulates HigA folding and activity by preventing its aggregation, we next performed aggregation prevention assays in vitro using purified urea-denatured HigA protein in the presence or absence of the chaperone (Fig. 4B). Because of its tendency to aggregate (Fig. S5), the HigA antitoxin was purified in high amounts from the aggregated fraction (SI Materials and Methods). As expected from our in vivo results, HigA rapidly aggregated upon dilution from the denaturant. However, in sharp contrast to proOmpC, aggregation of the antitoxin was very efficiently prevented by Rv1957 (at a 1:3 chaperone:antitoxin ratio), thus emphasizing the specific nature of the interaction between Rv1957 and HigA (Fig. 4B). Remarkably, a similar chaperone:substrate ratio was sufficient for SecB to prevent aggregation of its native substrate proOmpA (43). In contrast to Rv1957, SecB did not significantly rescue HigA aggregation under the same conditions (Fig. 4B). These results strongly support the idea that Rv1957 activates the antitoxin function in vivo (Fig. 3 B and D) and further demonstrate that the chaperone specifically controls the HigB-HigA TA system in M. tuberculosis. A model presenting such atypical control of a TA system by a SecB-like chaperone is depicted in Fig. 4C.

Conclusion

Our results demonstrate that the major human pathogen M. tuberculosis encodes a functional SecB-like chaperone. This is unexpected in view of the known distribution of SecB among bacterial species (13,14). However, we have shown that the mycobacterial chaperone has a unique and specific function involving the fine tuning of a stress-responsive TA system from the higBA family. Why does M. tuberculosis possess such a unique and sophisticated “ménage à trois”? The presence of a distinct outer membrane and a remarkably large number of putative outer membrane proteins in M. tuberculosis (18) suggests that this bacterium could make use of a functional export chaperone under certain circumstances. The fact that Rv1957 possesses key predicted structural elements of the E. coli SecB chaperone that are necessary for its interaction with preproteins and with its SecA partner at the Sec translocon (Fig. 1A) suggests that Rv1957, as part of the TAC system, could indeed be responsive to export stress. In this case, the nonspecific cytoplasmic accumulation of outer membrane precursor proteins or the synthesis of stress-specific outer membrane proteins could result in recruitment of the mycobacterial SecB-like chaperone to assist their efficient targeting to the Sec translocon. Such increasing presecretory clients of Rv1957 would be expected to compete with the antitoxin for binding to the chaperone, thus facilitating degradation of the free antitoxin and the subsequent activation of the toxin until stress conditions subside.

Materials and Methods

Bacterial Strains.

Genetic experiments were performed in E. coli strains W3110, W3110 ΔsecB::Cm^R (27), and MC4100 Δtig::Cm^R ΔdnaKdnaJ::Kan^R (30) as well as in M. marinum wild-type strain M (44) or M. bovis bacillus Calmette–Guérin strain Copenhagen (45). E. coli strains BL21(λDE3) and BL21(λDE3) ΔdnaKdnaJ::Kan^R (28) were used for protein purification with the pET expression system (Novagen).

In Vivo Chaperone Assays.

Complementation of SecB chaperone activity in vivo at a low temperature of growth was carried out as follows. Fresh transformants of W3110 ΔsecB::Cm^R containing pFr, pFr-SecB, or pFr-Rv1957 were grown at 37 °C to midlog phase in LB ampicillin, serially diluted, and spotted on LB ampicillin (50 μg/mL) agar plates in the absence or presence of IPTG inducer. Plates were incubated for 1 d at 37 °C, for 2 d at 20 °C, or for 5 d at 16 °C. Complementation in the absence of DnaK and TF chaperones was tested in the MC4100 Δtig::Cm^R ΔdnaKdnaJ::Kan^R temperature-sensitive strain. Midlog phase cultures of fresh transformants of MC4100 Δtig::Cm^R ΔdnaKdnaJ::Kan^R containing plasmids pSE380ΔNcoI, pSE-SecB, or pSE-Rv1957 grown at 22 °C in LB ampicillin (50 μg/mL) were first diluted to an OD₆₀₀ of ∼0.4 at 22 °C; IPTG inducer (0, 0.05, or 0.1 mM) was then added, and cultures were incubated for an extra hour at the same temperature. Cultures were serially diluted and spotted on LB ampicillin (50 μg/mL) agar plates at 22 °C and 32 °C for 2 d, with or without IPTG inducer.

In Vivo Protection from HigB Toxin.

Suppression of HigB toxicity by HigA and Rv1957 was analyzed as follows. E. coli strain W3110 was transformed with plasmids pSE380ΔNcoI or pSE-Rv1957 on LB ampicillin agar plates at 37 °C. Transformants were then grown in LB ampicillin at the same temperature and cotransformed with pMPMK6 vector, pK6-HigB, pK6-HigA, or pK6-HigB-HigA on LB ampicillin kanamycin agar plates containing 0.2% glucose at 37 °C. Fresh transformants were grown in LB ampicillin kanamycin 0.2% glucose to midlog phase, serially diluted, and spotted on LB ampicillin kanamycin agar plates with or without IPTG inducer as indicated in the figure legends. Plates were incubated at 37 °C overnight.

In Vitro Protein Aggregation Assay.

Unfolded proOmpC [200 μM in 20 mM Tris (pH 7.5), 200 mM NaCl, 1 mM DTT, 20% glycerol, 8 M urea] or HigA [300 μM in 20 mM Tris (pH 7.5), 150 mM NaCl, 1 mM DTT, 20% glycerol, 8 M urea] prewarmed for 10 min at 60 °C was diluted 100-fold in reaction buffer [30 mM Hepes (pH 7.5), 40 mM KCl, 50 mM NaCl, 7 mM magnesium acetate, and 1 mM DTT] preincubated at 25 °C or at 37 °C, respectively, with or without chaperones. Immediately after addition of denatured proOmpC or HigA, aggregation was monitored continuously at 320 nm for 50 min at 25 °C or at 37 °C, respectively, on a Cary Scan UV-visible spectrophotometer from Varian. All solutions used in this assay were filtered through a 0.22-μm filter.

Details on culture conditions, pulse-chase analysis, plasmid constructs, E. coli cell fractionation, pull-down assays, Western blot analysis, and protein purifications are given in SI Materials and Methods.