Leaving on the lights: host-specific derepression of Mycobacterium tuberculosis gene expression by anti-sigma factor gene mutations

A fundamental requirement for microbial growth and survival is the ability to adapt to the environment. These adaptations are needed to allow efficient nutrient utilization, survival in nutrient poor conditions and the ability to tolerate a range of damaging stresses. In the case of pathogenic bacteria, specific adaptations are geared towards replication or survival in the host, and often are critical determinants of virulence.

At the population level, adaptation often results from selection of advantageous mutations with outgrowth of new, fitter populations. At the level of the individual microbial cell, however, adaptation requires the ability to alter physiology, often very rapidly. The major means by which such alterations can be achieved on a time scale rapid enough to allow survival is through changes in the patterns of transcription. Bacteria have thus incorporated a remarkable array of mechanisms to regulate transcription in response to specific stimuli. Among these, the alternative sigma factors, which recognize promoter sequences distinct from those recognized by the primary sigma factor, play an important role in many bacterial species. As recently reviewed, a range of recent evidence supports the long-standing concept that the competitive association of sigma factors with core RNA polymerase (RNAP) allows rapid reprogramming of bacterial transcription (Mooney et al., 2005).

The number of alternative sigma factors varies widely among bacterial species, and correlates roughly with the variability of the environments encountered by a given bacterial species, ranging from two alternative sigma factors in Streptococcus pyogenes, whose primary niche is limited to the human oropharynx, to more than 60 encoded by the soil actinomycete Streptomyces coelicolor, whose natural habitat is highly variable in terms of nutrients, stresses and competing microbial flora (Ferretti et al., 2001; Bentley et al., 2002). Mycobacterium tuberculosis has 12 alternative sigma factors, a relatively large number for an organism whose natural environment is limited to the mammalian host.

Perhaps the most intensively studied alternative sigma factor, rpoS of Escherichia coli, has an extraordinary range of direct and indirect effects on transcription, and an equally remarkable range of mechanisms that control RpoS activity (Hengge-Aronis, 2002). Among the many ways that sigma factor activity can be controlled is through the action of regulatory proteins that directly bind the sigma factor, i.e. anti-sigma factors, and which prevent the association of the sigma with core RNAP. The mechanisms leading to release of sigma from its cognate anti-sigma are thus the means by which the environmental signal is transduced to alter transcription, and reflect the function of the specific sigma factor. These mechanisms may incorporate multiple proteins, allowing for the sensing and integration of multiple inputs and complex signals. E. coli RpoE, for example is released from the membrane spanning anti-sigma factor RseA through a proteolytic cascade that involves at least three other proteins that are activated in response to accumulation of misfolded outer membrane proteins in the cell envelope (Grigorova et al., 2004). In M. tuberculosis, the oxidative stress response sigma factor SigH is regulated by its anti-sigma factor RshA (Song et al., 2003). This protein, like its S. coelicolor homologue RsrA, functions as a redox switch where sigma–anti-sigma binding is disrupted by the oxidation of critical intramolecular disulphide bonds in the anti-sigma factor, as the result of oxidative stress (Li et al., 2003; Bae et al., 2004).

In this issue of Molecular Microbiology, Said-Salim et al. provide evidence that mutations in a putative anti-sigma factor gene of Mycobacterium bovis disrupt negative regulation of the alternative sigma factor SigK. In undertaking this work, these investigators followed up on long-standing observations that M. tuberculosis isolates from humans produce low levels of two secreted antigens, Mpt70 and Mpt83, whereas isolates from cattle (M. bovis) secrete large amounts of these proteins (Harboe and Nagai, 1984; Hewinson et al., 1996). This group had recently shown that mutations in sigK accounted for the low-level secretion of these proteins in some variants of the vaccine strain M. bovis BCG, suggesting that SigK is required for expression of these genes (Charlet et al., 2005). The absence of any differences in the nucleotide sequence of sigK in M. tuberculosis and M. bovis led them to examine adjacent genes for an explanation for these differences in protein secretion, based on the fact that alternative sigma factors are often regulated by anti-sigma factors encoded by genes cotranscribed with the sigma factor gene. These investigations identified two single nucleotide polymorphisms in Rv0444c, the gene immediately 3′ of sigK, in tubercle bacilli isolated from cattle (M. bovis) and goat (Mycobacterium caprae). They also found that a strain from an African gazelle (Oryx) contained a distinct mutation in this gene. Through a series of genetic, transcriptional and protein interaction studies, the authors provide strong evidence that Rv0444c is an anti-sigma factor of SigK and that the mutations in this gene interfere with its negative regulation of SigK.

While the results presented in this work provide a clear picture of this regulatory mechanism, the origin of these mutations and the consequences of the resulting constitutive high-level expression of SigK-regulated genes bear further investigation. Anti-sigma factor inactivation as a means to increase specific gene expression and enhance pathogen virulence was first described in the early 1990s in Pseudomonas aeruginosa. Expression of the gene encoding the alternative sigma factor AlgU (AlgT), which is required for expression of genes involved in alginate synthesis, is negatively regulated by two adjacent 3′ genes, mucA and mucB. P. aeruginosa strains infecting the lungs of cystic fibrosis (CF) patients had long been observed to change over time to a highly mucoid phenotype, caused by high-level production of the exopolysaccharide alginate. Mutations in the mucA gene, which encodes an anti-sigma factor of AlgU, were identified in several independent mucoid strains and shown to be the genetic basis of this alginate overproduction [reviewed in the study by Deretic et al. (1994)]. The findings of Said-Salim et al. thus potentially fit into this paradigm of derepression of virulence factor gene expression through anti-sigma factor inactivation.

Differences in these systems, however, raise several interesting questions. First, in the case of P. aeruginosa, the mucA mutations clearly become established in bacterial populations in response to selection in a specific environment, the CF lung. The rapid reversion of mucoid strains to non-mucoidy upon in vitro growth, i.e. in the absence of this selection, supports this interpretation. In the case of M. tuberculosis Rv0444c, the origin of the observed mutations is not known. While selection in the environment of specific animal hosts, as posited by the authors, is plausible, the limited number of strains analysed and the inability to observe the acquisition of mutations de novo preclude drawing a definitive conclusion.

Alternatively, given the remarkable stability of the M. tuberculosis genome across human isolates and among isolates from other animal hosts, it is possible that the observed mutations are the result of a ‘founder effect’ (Mayr, 1963). In this scenario an isolate containing these mutations, which might not be advantageous, became established early during the spread of M. tuberculosis from humans to cattle. Though this alternative is possible, the presence of independent mutations in the same gene in the strain infecting the Oryx, which, like the cow and the goat is a member of the family Bovidae, supports the authors' interpretation that these Rv0444c inactivating mutations arose and became established in these bacterial populations because they provide a selective advantage in these animal hosts. Further, the inactivation of sigK in M. bovis BCG with prolonged passage in vitro suggests that these mutations leading to elevated protein antigen production are not neutral and would not persist in the absence of a selective advantage in vivo. Taken together, these data support the concept that these mutations are present in these Bovidae-infecting strains as the result of positive selection in vivo. Characterization of the Rv0444c gene in a range of isolates from diverse geographic regions and from a range of animal species would provide additional insight into this question.

A second interesting question results from the nature of the observed mutations. In the case of mucA, the inactivating mutations were point mutations causing premature stop codons or frameshifts that dramatically altered the MucA protein, presumably resulting in its complete inactivation. The mutations observed in M. bovis and in the Oryx bacillus are all non-synonymous single nucleotide polymorphisms. In M. bovis, both mutations result in substitution of an acidic residue for glycine, changes that would be expected to have major effects on protein structure. The first of these substitutions occurs in a predicted transmembrane helix, whereas the other substitution occurs in the carboxy terminal region of the protein that is predicted to be extracellular. In the case of the Oryx bacillus, the mutation occurs in the native stop codon, resulting in a gene that would encode a protein that contains an additional 28 residues. None of these mutations map to the region encoding the amino terminal domain (residues 1–92), shown by the authors to be required for the interaction of Rv0444c with SigK. These observations, and the striking difference in these mutations from the types of mutations observed in mucA, raise the possibility that these altered Rv0444c proteins might retain some functional activity.

Additional questions raised by the studies of Said-Salim et al. include what is the function of the SigK regulon, and how and why it is regulated in human-infecting M. tuberculosis but is constitutively expressed in Bovidae-infecting strains. The SigK regulon appears to be very small, including just two loci: (i) mpt83 (Rv2873), dipZ (Rv2874) and mpt70 (Rv2875); and (ii) the sigK locus. As discussed by the authors, the secreted antigens Mpt70 and Mpt83 might have a role in virulence in addition to their established role in immunity. The host-specific differences in their regulation may provide leads to new insights into host–pathogen interactions. The function of dipZ, as well as of the other genes in the sigK-encoding operon (Rv0449c-Rv0444c) are not known; whether they play a role in virulence or immunity and whether their differential regulation is important in pathogenesis remain to be determined. The nature of the host signal to which Rv0444c responds in the human host, and the mechanism by which this signal is detected and presumably leads to release of SigK from Rv0444c are also unknown. Indeed, whether Rv0444c is a transmembrane protein capable of binding extracellular ligands has not yet been investigated. The answers to these questions will provide insight into both specific mechanisms of SigK regulation and the role of this regulon in tuberculosis pathogenesis.