Posttranslational Regulation of Mycobacterium tuberculosis Extracytoplasmic-Function Sigma Factor σ^L and Roles in Virulence and in Global Regulation of Gene Expression

Sigma factors are interchangeable subunits that provide the promoter recognition function to bacterial RNA polymerase (RNAP) holoenzymes (10). Thirteen sigma factor genes were predicted from the genome sequence of Mycobacterium tuberculosis. Several of these were shown to be important for appropriate stress responses and virulence (9, 15-18, 20, 27, 29). Interestingly, the expression of the σ^B-encoding gene (sigB) is induced in response to several stimuli (4, 14, 17, 18), suggesting a potential general role in stress responses (16, 24). Previous studies have shown that transcription of sigB can be initiated at the same nucleotide by RNAP holoenzymes that contain extracytoplasmic-function sigma factor σ^E or σ^H (17, 18, 21). It was also recently reported that the extracytoplasmic-function sigma factor σ^L may also have a role in the expression of sigB (11, 16).

Using an in vitro transcription assay, we have tested each M. tuberculosis sigma factor to determine whether additional sigma factors could elicit transcription of sigB. Reactions were performed as previously described (3) using purified Mycobacterium smegmatis RNAP, 150 fmol of sigB DNA template (encompassing base pairs −361 to +329 relative to the first translated nucleotide), and 3 pmol of recombinant sigma factors (13). As expected, primer extension products were detected for the σ^E- and σ^H-containing holoenzymes (Fig. 1A). Interestingly, the σ^L-dependent transcription signal also originated from the same nucleotide (Fig. 1A). Importantly, the σ^E-, σ^H-, and σ^L-dependent signals matched the sigB transcriptional start site identified in vivo (17, 21). In addition, an upstream σ^F-dependent promoter with striking homology to the proposed σ^F consensus was identified (Fig. 1A) (3, 9, 16). No signal was observed with the nine remaining M. tuberculosis sigma factors (data not shown). Primer extension assays were performed according to standard procedures using the following oligonucleotide: Mtb sigB-EXT (5′-ATCCAGATCGCTGTCAACCCGG-3′). A putative promoter closely resembling the σ^E-σ^H-σ^L-dependent sigB promoter was also found upstream of sigL. In vitro transcription was thus performed as in Fig. 1A using a sigL DNA template (base pairs −345 to +138 relative to the translation start site) and oligonucleotide Mtb sigL-EXT (5′-CTCTTGGACGACGTCTTC-3′). Interestingly, σ^L was the only sigma factor that allowed transcription from this promoter, suggesting that minor divergences in DNA sequences may account for promoter discrimination by σ^E, σ^H, and σ^L (Fig. 1B).

Hahn and colleagues recently proposed that RslA could be a σ^L anti-sigma factor and detected an interaction between these proteins in a bacterial two-hybrid assay (11). Since RslA is a predicted transmembrane protein, we have fused its cytoplasmic moiety (amino acids 1 to 114) to glutathione S-transferase (GST) and performed direct interaction assays with hexahistidine-tagged σ^E, σ^F, σ^H, and σ^L. Assays were performed as previously described (3) and analyzed by immunoblotting using anti-His antibody. Figure 2A shows a direct interaction between σ^L and GST-RslA. No interaction was detected between GST-RslA and σ^E, σ^F, and σ^H or between GST and any sigma factor (Fig. 2A and data not shown). In vitro transcription assays next demonstrated that RslA inhibits σ^L-dependent transcription of the sigB gene in a dose-dependent manner (Fig. 2B). In this latter experiment, increasing concentrations (1.5, 3.0, 6.0, and 12 pmol) of the N-terminally hexahistidine-tagged RslA cytoplasmic moiety were added to a fixed amount of σ^L-containing holoenzyme and sigB promoter template as described above for Fig. 1. Importantly, the σ^L-dependent signal was not affected by the addition of RshA or Rv1222, a σ^H (26) and a σ^E anti-sigma factor, respectively (data not shown). Taken together, these results demonstrate that RslA is a σ^L-specific anti-sigma factor.

The DNA region including sigL and rslA was next replaced in the genome of M. tuberculosis H37Rv by an hygromycin resistance gene using specialized transduction (2). The growth curve and colony morphology of the resulting strain (TB1) was indistinguishable from the wild-type (wt) strain (data not shown). In agreement with the results obtained by Hahn et al. using a similar sigL mutant (11), we have not observed any difference in the sensitivity of the TB1 and wt strains to diamide, cumene hydroperoxide, H₂O₂, and sodium nitroprusside (data not shown) using a previously described disk diffusion assay methodology (18). Moreover, the TB1 and wt strains were not differentially affected by EDTA, vancomycin, tetracycline, heat shock (45°C), acidic pH, hyperosmolarity, and dithiothreitol. However, in contrast to the findings of Hahn and colleagues, we have noticed a small but reproducible sensitivity of TB1 to the superoxide generator plumbagine and to the detergent sodium dodecyl sulfate relative to the wt strain in disk diffusion assays performed as previously described (18); these phenotypes were fully complemented by the introduction of the sigL-rslA locus (or by the sigL gene alone, see TB3 strain described below) at the L5 integration site (data not shown). However, no significant increase in sigL expression was monitored by quantitative reverse transcription-PCR (qRT-PCR) after exposure to plumbagine or sodium dodecyl sulfate (data not shown).

Hahn and colleagues have shown that a sigL mutant of M. tuberculosis is attenuated in BALB/c mice in a high-dose intravenous infection model. We have performed a similar analysis using a low-dose aerosol infection model for two different mouse strains with different natural resistances to infection. C57BL/6 and DBA/2 mice were challenged, with ∼100 and 200 CFU/mouse of the TB1 or H37Rv strain (6). No major differences were observed in the abilities of the two strains to colonize lung and spleen tissues or in the size and extent of lesions as assessed by histopathology analysis at 28 and 60 days postinfection (Table 1 and data not shown). DBA/2 mice were also held for observation and determination of the time to death. A significant extension of the median survival time was observed for mice infected with the TB1 mutant compared to the parental strain, thus indicating that sigL is important for virulence (Fig. 3).

Using qRT-PCR and primer extension assays, Hahn and colleagues have concluded that sigL is constitutively expressed through mid-exponential and stationary phase by a σ^L-independent promoter located 130 bases upstream of sigL (11). However, by using a plasmid containing a fusion of the first 564 nucleotides upstream of sigL to lacZ, β-galactosidase activity in the TB1 and wt strains (measured by the method of Miller [19]) tended to increase with culture density, especially in the transition from lag to exponential phase (Fig. 4).

In order to study the role of σ^L by mimicking naturally inducing conditions (release of σ^L by RslA), we have complemented the TB1 mutant strain by the introduction of sigL and its upstream region at the L5 integration site (12). The resulting “partially complemented” strain (TB3) thus contains sigL but not rslA, which should allow a constitutive activity of σ^L. A 25-fold induction of sigL was indeed measured by qRT-PCR in this strain with respect to the H37Rv strain (data not shown). The TB1 and TB3 strains were grown to mid-exponential phase, and the RNA expression profiles were compared using oligonucleotide microarrays as previously described (18). The resulting data were analyzed using the significance analysis of microarray method (28). A total of 27 genes from 12 putative transcription units were up-regulated in TB3 relative to TB1 (Table 2). No repressed genes were identified. Six induced genes were chosen, and their induction levels were confirmed by qRT-PCR (Table 2). The mRNA levels of the selected genes in H37Rv and TB1 mutant strains were also compared. All genes were expressed at similar levels in both strains, suggesting that inhibition of σ^L activity by RslA is virtually complete under these conditions (Table 2).

Hahn and colleagues have recently identified 19 genes induced by the overexpression of σ^L from an acetamide-inducible promoter (11). Some genes (pks10, pks7, Rv1138c, mpt53, and Rv2877c) were identified by both approaches. However, the remaining four genes from the pks10 operon, three genes from the putative mmpL13 operon, four genes from the putative operon starting with Rv3166c, mce2F and its upstream gene, sigB, and others, were identified only in this work. Putative promoter boxes similar to those of in vitro-identified σ^L-dependent promoters (Fig. 1) and compatible with the proposed σ^L consensus sequence (11) were found upstream of eight of these genes (Table 2). Rapid amplification of cDNA end (5′-RACE) experiments, performed as previously described (17), confirmed that the transcriptional start sites of sigL, mmpL13A, and mpt53 were suitably located to allow σ^L to recognize these promoter elements (data not shown). Interestingly, although σ^E, σ^H, and σ^L all recognize the same promoter upstream of sigB, no other genes seem to be regulated by any combination of these sigma factors (17, 18). This suggests that although the consensus sequences recognized by these sigma factors are similar, few differences are sufficient for promoter discrimination.

Many of the genes regulated by σ^L could be involved in processes related to the cell envelope (Table 2). For example, the pks10 operon and ppsA gene products are likely to be involved in the biosynthesis of dimycocerosyl phthiocerol, an important component of the mycobacterial cell wall (1, 5, 8, 22, 23). Moreover, the mmpL13a and mmpL13b gene products belong to a family of proteins involved in lipid transport (5), sulfolipid biosynthesis (7), and peptidoglycolipid biogenesis (25). From this perspective, it is noteworthy that RslA is a predicted transmembrane protein that could possibly sense the condition of the cell envelope or an external signal. The absence of σ^L could thus result in a weakening the cell surface or in an inappropriate modulation of the host immune response by the bacterium, which could at least partially explain the attenuation of the sigL mutant reported in this paper and by others (11).

• Open in new tab
• Download powerpoint

FIG. 1.

Identification of M. tuberculosis (Mtb) sigma factors involved in the transcription of selected genes. (A) High-resolution mapping of transcription signals observed using an in vitro transcription screening of the sigB upstream region. σ^E, σ^H, and σ^L initiate transcription at the same nucleotide, while σ^F use an upstream promoter. (B) sigL is expressed from an autoregulated promoter. In vitro transcription with the corresponding sigma factors was followed by a primer extension procedure. Proposed promoter boxes are underlined. Arrows indicate transcriptional start sites.

• Open in new tab
• Download powerpoint

FIG. 2.

RslA is a σ^L-specific anti-sigma factor. (A) GST-RslA interacts with σ^L. Results of GST pull-down assays between GST-RslA and hexahistidine-tagged σ^H or σ^L revealed by immunoblotting using an anti-His antibody are shown. L, loading control; S, supernatant; P, pellet. (B) RslA inhibits σ^L activity in a dose-dependent manner. In vitro transcription inhibition assays at the sigB promoter were carried out in the absence (−) or presence (+) of σ^L and increasing amounts (triangle) of hexahistidine-tagged RslA cytoplasmic moiety.

• Open in new tab
• Download powerpoint

FIG. 3.

Time-to-death analysis in DBA/2 mice after aerosol infection with M. tuberculosis H37Rv (squares) and with the sigL mutant strain (circles).

• Open in new tab
• Download powerpoint

FIG. 4.

Measurement of sigL expression. M. tuberculosis H37Rv harboring a reporter plasmid carrying the sigL upstream region fused to lacZ was grown in Middlebrook 7H9 medium, and β-galactosidase activity was measured at different points of the growth curve. The bars show β-galactosidase activity (in Miller units), and the line shows the optical density at 540 nm (OD 540).

• American Society for Microbiology