Reduced immunopathology and mortality despite tissue persistence in a Mycobacterium tuberculosis mutant lacking alternative σ factor, SigH

Mutant Construction.

The targeted disruption strategy was based on the vector pCK0686, which contains a hygromycin-resistance cassette flanked by unique multiple cloning sites (MCS). A left flank PCR product containing 1,722 bp proximal to the sigH gene and a right flank PCR product containing 2,059 bp distal to the sigH gene were cloned into the MCSs. The sigH specific flanks were nonoverlapping so that a central region of the sigH gene was deleted from the targeting vector, leaving 82 bp of the 5′ end and 113 bp of the 3′ end of the sigH gene in place. The resulting plasmid pCK0708 was then used to transform M. tuberculosis CDC1551 in a two-step selection process using Southern blotting to confirm allelic exchange. A complemented strain of the sigH mutant (pCK0758) was generated by cloning a 2,862-bp M. tuberculosis CDC1551 genomic fragment encompassing the coding as well as regulatory region of the sigH gene plus the complete coding sequences of three distal genes (MT3319, MT3318, and MT3317) into an L5 phage-based integration-proficient vector for mycobacteria, pMH94 (24), and then transforming the M. tuberculosis ΔsigH mutant with this vector.

Mouse Virulence Assays.

For mouse organ cfu assays, C57BL/6 mice were inoculated by the aerosol route. The liquid inoculum placed in the nebulizer was determined empirically for each bacterial strain to result in the implantation of 100 bacteria per lung at day 1 of infection. Groups of six mice were killed at day 1 and weeks 4, 8, 12, and 16. Lung and spleen tissues were homogenized in their entirely in PBS/0.05% Tween 80, and colonies were enumerated on 7H10 plates grown at 37°C for 3–4 weeks. Intermediate lobes of the right lung of infected mice were saved for histologic analysis (25). For time-to-death assays C3H:HeJ and CB17-severe combined immunodeficient (CB17-SCID) mice were infected intravenously with 10⁶ and 10⁴ bacteria, respectively, and C57BL/6J mice were infected by the aerosol route as described above.

Fluorescence-Activated Cell Sorting (FACS) Analysis and Intracellular Cytokine (ICC) Staining.

Single-cell suspensions of lung and spleen cells were prepared, washed, and plated as described (26–28) and were stimulated with anti-CD3 (0.1 μg/ml) and anti-CD28 (1 μg/ml) antibodies (Ab) in the presence of 3 μM monensin for 5 h. Cells were washed and stained with anti-CD4 (0.2 μg of anti-CD4-CyChrome Ab per 10⁶ cells) and anti-CD8 (0.2 μg of anti-CD8-FITC Ab per 10⁶ cells), washed, and fixed. For ICC, cells were permeabilized with saponin and stained for interferon γ (IFN-γ) (0.4 μg of anti-IFN-γ-PE Ab per 10⁶ cells) or tumor necrosis factor α (TNF-α) (0.4 μg of anti-TNF-α-PE Ab per 10⁶ cells) (PharMingen, San Diego) and analyzed by FACS (Becton Dickinson) as described (26–28). Cells were gated on the lymphocyte population based on cell size. Isotype controls for each Ab were used and uninfected control mice were tested in each experiment.

Microarrays.

Gene-specific PCR primers were designed to amplify internal fragments from a total of 4,016 ORFs from the annotated sequences of M. tuberculosis CDC1551 and H37Rv (29). Individual purified PCR products were spotted in duplicate on high contact angle slides (Corning, Corning, NY) by using a 96-well format IAS arrayer (Intelligent Automation Systems, Cambridge, MA). Bacterial RNA prepared by the Trizol method was reverse transcribed and labeled with Cy3 or Cy5 probes (Amersham Pharmacia) by using the aminoallyl labeling method. The slides were scanned with an Axon Genepix scanner, and the TIGR (The Institute for Genomic Research) Spotfinder and Array Viewer software systems were used to define and quantify spot intensities. For each bacterial growth point, two independent RNA preparations from wild type and mutant were prepared, and for each RNA sample pair reverse labeling was performed (four hybridizations for each growth point). As each amplicon was spotted in duplicate, eight relative hybridization values were available for each gene for each growth point.

Construction of the sigH Mutant and Complemented Strain.

To study the role of M. tuberculosis SigH in the pathogenesis of tuberculosis, we used allelic exchange to delete and replace the sigH gene (MT3320, Rv3223c) (29) with a hygromycin-resistance gene cassette (ΔsigH mutant) in M. tuberculosis strain CDC1551 (wild type) (30). We also complemented the sigH mutation by cloning a complete copy of the sigH gene and its promoter into the single-copy integrating plasmid pMH94. The ΔsigH mutant, the complemented strain, and the wild-type strain showed identical growth rates in Middlebrook 7H9 medium (Difco) in vitro.

Proliferation and Survival in Mouse Tissues.

We studied the ability of these three M. tuberculosis strains to survive in the tissues of resistant C57BL/6 mice by cfu analysis after aerosol exposure. Resistant C57BL/6 mice were selected for this analysis because they restrict the tissue proliferation of virulent M. tuberculosis and hence would be expected to amplify survival defects in attenuated strains. After aerosol exposure using inocula that implanted 100 bacilli in the lungs (as assessed by cfu counts the day after infection), mice were monitored for 16 weeks. As may be seen in Fig. 1 A and B, the pattern of mouse organ burden of M. tuberculosis seen with infection by the wild-type, ΔsigH mutant, and complemented strains did not differ. All three strains proliferated in lungs and spleen for 4 weeks, and then persisted in both tissues between 4 and 16 weeks with approximately 10⁶ organisms in the lungs and 10⁴ bacilli in the spleen.

Histopathologic Analysis.

Despite the apparent full virulence of the ΔsigH mutant by organ cfu analysis, histopathologic assessments of the organs from C57BL/6 mice demonstrated an attenuated phenotype in mutant-infected lungs. As may be seen in Fig. 2 A–C, whereas the wild-type and complemented-mutant strains produced progressive granulomatous inflammation and loss of functional alveoli attributable to enlarging and coalescing lesions, the mutant infection showed smaller, less abundant granulomas, and these lesions progressed less rapidly. After 16 weeks, mouse lungs infected by the ΔsigH mutant revealed minimal interstitial mononuclear cell infiltrates and peribronchial inflammation without large inflammatory wedges or granulomatous lesions (Fig. 2B) that were clearly observed in the corresponding wild-type and complemented strain infections (Fig. 2 A and C). A similar decrease in intensity of granulomatous inflammation and a delay in its evolution was observed in the lungs of C3H mice infected with the ΔsigH mutant for 8 or 16 weeks (Fig. 2 E and H) in comparison with infections by the wild-type and complemented strains (Fig. 2 D and F, and G and I). In contrast, the lungs of SCID mice infected with the three different M. tuberculosis strains did not show histopathologic differences at 4 weeks (Fig. 2 J–L), indicating that intact cellular immunity is required for the differential effect. Interestingly, analysis of spleens of all three strains of mice (C57BL/6, C3H, and SCID) indicated a similar extent of granuloma formation and tissue damage with mutant and wild-type bacterial strains (data not shown), suggesting that the disease delay phenotype in the ΔsigH mutant may be lung specific.

Cellular Immune Responses in Mouse Infections.

We postulated that the despite the normal bacterial survival in lung tissue, the ΔsigH mutant lacked virulence factors that induce tissue-damaging cell-mediated immune responses by the host (31, 32). Therefore we evaluated T cell responses in C57BL/6 mice after infection by both the wild type and the ΔsigH mutant by using flow cytometry of cellular suspensions obtained from mouse tissues at various times after infection. As may be seen (Fig. 3 A and B), between 3 and 4 weeks, the recruitment of CD3⁺ cells to the lung increased in wild-type-infected lungs, but not in mutant-infected mice. The reduced T cell recruitment to the lungs at 4 weeks by the ΔsigH mutant was observed in both total CD4⁺ and CD8⁺ cells with approximately one-log₁₀ differences in both cell types (Fig. 3 C and D). By 16 weeks, levels of both CD4⁺ and CD8⁺ cells declined in the lungs of wild-type-infected mice but remained constant in mutant-infected mice, resulting in similar numbers of both T cell types. Intracellular cytokine staining for IFN-γ and TNF-α synthesis within pulmonary CD4⁺ and CD8⁺ cells showed that the cell-associated cytokine synthesis paralleled the T cell population curves (reduced IFN-γ and TNF-α positivity in mutant infection at 4 weeks) indicating similar rates of expression of these cytokines per cell in infection with both strains of bacteria (Fig. 3 E–H). By 16 weeks the levels of cytokine-producing CD4⁺ and CD8+ cells in the two types of infection had essentially re-equalized. Thus the recruitment of cytokine-producing CD4⁺ and CD8⁺ cells to the lungs (but not the spleens) of ΔsigH mutant-infected mice was significantly blunted beginning at 4 weeks after infection. Thus the pathologic disease delay phenotype seen in lungs of C57BL/6 mice after ΔsigH mutant infection correlates with diminished T cell recruitment to the lung despite a bacterial burden equivalent to that of wild type.

Time-to-Death Analyses.

We performed time-to-death experiments to determine whether the milder immunopathology observed in ΔsigH mutant infection might manifest as a prolongation of survival. When C57BL/6 mice infected with each of the three strains were held long term for time-to-death analysis, no deaths were observed in any of the three groups for 15 months in keeping with the known innate resistance of this mouse strain to tuberculosis. However, in C3H mice the ΔsigH mutant displayed a pronounced attenuation in time-to-death with the median time-to-death (MTD) for the ΔsigH mutant exceeding 170 days (0/12 mice dead at day 170) as compared with 52 and 51.5 days for infection by the wild-type and complemented strains, respectively (Fig. 4A, P < 0.001 for mutant compared with both other strains). The differences in MTD between wild-type and complemented strains in C3H mice were not statistically significant. SCID mice infected with the ΔsigH mutant, the wild type, and the complemented strain showed no significant difference in mortality with MTD of 32, 30, and 29 days, respectively (Fig. 4B, P > 0.05 for all pairwise comparisons).

Microarray Identification of SigH-Regulated Genes.

We studied the global expression patterns of the M. tuberculosis ΔsigH mutant by using complete genomic microarrays (33–36). Total RNA from the M. tuberculosis ΔsigH mutant and wild-type CDC1551 was prepared from cultures grown in 7H9 medium to OD₆₀₀ values of 0.3, 0.6, 1.2, and >2.0. In late exponential phase of in vitro growth (OD₆₀₀ = 1.2) approximately 120 genes were underexpressed and 60 genes were overexpressed in the M. tuberculosis ΔsigH mutant relative to the wild type (Table 1 and Table 2, which is published as supporting information on the PNAS web site, www.pnas.org). Genes encoding hypothetical unknown proteins accounted for 30% of differentially expressed transcripts.

Several components of the thioredoxin–thioredoxin reductase system demonstrated a relative decrease in expression in the ΔsigH mutant by microarray analysis; these include MT4032–4033 (thioredoxin reductase and thioredoxin, repressed 3.9-fold), MT1516–1517 (thioredoxins, repressed 3.8-fold), and MT0838 (a putative thioredoxin, repressed 3.2-fold). In addition, related oxidative stress response proteins MT2541 (glutaredoxin-like, repressed 3.4-fold) and MT2063 (ferredoxin, repressed 3.1-fold) showed significantly decreased relative expression in the ΔsigH mutant. Several other types of stress-response genes, including MT0265 (hsp20 family member, 3.5-fold repressed), the heat-shock gene dnaK (MT0365, hsp70 family members, 2.9-fold repressed), MT2052, MT2061, MT2698, and MT2699 (each universal stress protein family members, repressed 2.1-, 2.2-, 4.5-, and 2.5-fold, respectively), were relatively underexpressed in the ΔsigH mutant. Since thioredoxins and heat shock proteins reduce and refold proteins damaged by oxidation, their σ factor dependent expression may represent a stress-management strategy by the pathogen aimed at maintaining activity of essential proteins (37). Microarray analysis of N,N,N′,N′-tetramethylazodicarboxamide (diamide)-treated strains for the most part intensified gene expression differences between mutant and wild type (data not shown), as would be expected because diamide is a known thiol-oxidizing agent.

Because there is a 68-bp overlap between the sigH gene remnant in the M. tuberculosis ΔsigH mutant and the sigH amplicon spotted on our microarray, it is possible to detect the level of ineffective sigH transcription in the mutant by microarray analysis. Rather than producing a compensatory increase of ineffective sigH transcription, lack of functional SigH protein resulted in a relative decrease in sigH (MT3320) transcription in the mutant by 4.9-fold. This finding suggests that the M. tuberculosis sigH gene is self-regulated, a hypothesis strengthened by the presence of a consensus sigH-binding sequence that was found in the upstream untranslated region of the sigH gene (Fig. 5).

The relative expression of gene MT3318 (Rv3222Ac), which is located distal to sigH in the same transcriptional unit and exhibits homology to the anti-σ factor gene class, was also diminished in the mutant strain (4.2-fold). The arrangement of genes at the M. tuberculosis sigH locus resembles that of the Streptomyces coelicolor A3(2) sigR locus (38, 39), and it has been shown that the gene adjacent to sigR (rsrA) inhibits SigR activity by functioning as SigR-sequestering anti-σ factor (40). Because MT3318 transcriptional levels parallel those of sigH, it appears likely that the SigH and the MT3318 gene product are produced in equimolar quantities. The expression of σ factor genes sigE (MT1259, 4.0-fold repressed) and sigB (MT2783, 3.4-fold repressed) was relatively repressed by ΔsigH mutation, suggesting that sigH may occupy a proximal position in the hierarchy of mycobacterial regulatory mechanisms.

SigH Promoter Recognition Consensus Sequence.

Analysis of the untranslated regions of genes underexpressed in the sigH mutant revealed a consensus sequence (Fig. 5) among at least 31 of these genes. These genes harboring a consensus element are likely to be under direct SigH control, whereas those genes that did not exhibit a consensus sequence may be indirectly controlled by SigH. The consensus sequence comprises two modules, CGGGRAC at the −35 region and CGTTR at the −10 region (R = A or G), both modules being separated by about 14–18 bp. This SigH consensus is in strong agreement with the consensus derived for six mycobacterial genes in a recent study by Raman et al. (41) and is similar to that observed for the sequence recognized by the S. coelicolor SigR protein (40).