Characterization of Mycobacterium smegmatis sigF mutant and its regulon: overexpression of SigF antagonist (MSMEG_1803) in M. smegmatis mimics sigF mutant phenotype, loss of pigmentation, and sensitivity to oxidative stress

Abstract In Mycobacterium smegmatis, sigF is widely expressed during different growth stages and plays role in adaptation to stationary phase and oxidative stress. Using a sigF deletion mutant of M. smegmatis mc2155, we demonstrate that SigF is not essential for growth of bacterium. Deletion of sigF results in loss of carotenoid pigmentation which rendered increased susceptibility to H2O2 induced oxidative stress in M. smegmatis. SigF modulates the cell surface architecture and lipid biosynthesis extending the repertoire of SigF function in this species. M. smegmatis SigF regulon included variety of genes expressed during exponential and stationary phases of growth and those responsible for oxidative stress, lipid biosynthesis, energy, and central intermediary metabolism. Furthermore, we report the identification of a SigF antagonist, an anti‐sigma factor (RsbW), which upon overexpression in M. smegmatis wild type strain produced a phenotype similar to M. smegmatis mc2155 ΔsigF strain. The SigF‐anti‐SigF interaction is duly validated using bacterial two‐hybrid and pull down assays. In addition, anti‐sigma factor antagonists, RsfA and RsfB were identified and their interactions with anti‐sigma factor were experimentally validated. Identification of these proteins will help decode regulatory circuit of this alternate sigma factor.


Introduction
Mycobacterium smegmatis, a fast-growing saprophytic environmental bacterium, is used as a surrogate to study mycobacterial physiology and gene regulation as it is easy to culture in laboratory conditions. Owing to its habitat, M. smegmatis encounters more diverse conditions than its pathogenic counterparts and consequently its genome (6.98 Mb) has expanded nearly twice to the size of M. tuberculosis (4.4 Mb) to accommodate more genes. There is an unusual expansion of several genes which have acquired many paralogs unlike in other mycobacterial species (Waagmeester et al. 2005). There are 28 sigma factor genes in M. smegmatis in contrast with 13 reported in M. tuberculosis (Cole et al. 1998;Waagmeester et al. 2005;Rodrigue et al. 2006) and there are seven paralogs of sigma factor sigH, which are differentially expressed in M. smegmatis (Waagmeester et al. 2005;Singh and Singh 2009). Sigma factors reversibly associate with RNA polymerase and allow them to specifically direct the expression of specific set of genes. M. smegmatis genome encodes one of each group I, II, and III sigma factors represented by SigA, SigB, and SigF, respectively, and 25 of group IV sigma factors (Kapopoulou et al. 2011). SigA, the primary sigma factor in both M. smegmatis and M. tuberculosis, is essential for bacterial viability (Gomez et al. 1998), while SigB, the primary-like sigma factor is very similar to SigA and is dispensable for growth in M. smegmatis (Fontán et al. 2009). SigF (group III) and extracytoplasmic function (ECF) sigma factors (group IV) constitute alternate sigma factors which enable adaptation to a range of external and internal stimuli. Locus for sigA, sigB, sigD, sigE, sigF, sigG, and sigH are well conserved in M. smegmatis and M. tuberculosis (Sachdeva et al. 2010).
Earlier, the sigF was reported as a late-stage specific sigma factor, present only in the genomes of slow-growing pathogenic mycobacteria (DeMaio et al. 1996(DeMaio et al. , 1997. M. tuberculosis sigF was found strongly induced within cultured human macrophages, during stationary phase of growth, upon exposure to cold shock, nutrient starvation, and several antibiotics (Graham and Clark-Curtiss 1999;Michele et al. 1999;Betts et al. 2002). M. tuberculosis ΔsigF strain grew to a threefold higher density in stationary phase than the wild-type strain (Chen et al. 2000), but showed almost similar sensitivity to heat shock, cold shock, and hypoxia relative to the parental strain (Geiman et al. 2004;Hartkoorn et al. 2010). M. tuberculosis ΔsigF strain was attenuated for virulence in a mouse infection model despite persistence at high bacterial load in lungs compared with the isogenic wild type (Geiman et al. 2004). Overexpression of sigF in M. tuberculosis resulted in the differential regulation of many cell wall-associated proteins and other genes involved in the biosynthesis and degradation of surface polysaccharides and lippolysaccharides, believed to play important roles in host-pathogen interactions (Williams et al. 2007;Hartkoorn et al. 2010). However, we earlier demonstrated that, sigF is conserved in all the mycobacterial species analyzed and proposed that apart from regulating the expression of virulence genes in pathogenic mycobacteria, SigF is likely to play more roles in mycobacterial physiology (Singh and Singh 2008).
In M. smegmatis, sigF is widely expressed during different growth stages (Singh and Singh 2008). M. smegmatis sigF is transcriptionally induced in response to nutrient depletion, cold shock and upon exposure to agents that damage cell wall architecture, like SDS and antibiotics, isoniazid, and ethambutol (Singh and Singh 2008;Gebhard et al. 2008). A sigF mutant of M. smegmatis ATCC 607 strain showed higher transformation efficiency, lack of carotenoid pigmentation, and increased susceptibility to hydrogen peroxide mediated oxidative stress (Provvedi et al. 2008). SigF in M. smegmatis plays role in adaptation to stationary phase, heat, and oxidative stress (Hümpel et al. 2010). While both these studies demonstrate the role of M. smegmatis SigF in oxidative stress, molecular basis of this increased sensitivity to hydrogen peroxide remains unclear. Furthermore, proteins involved in posttranslation regulation of M. smegmatis SigF activity are not characterized, making it difficult to define the regulation circuitry of this alternate sigma factor. Using an insertion deletion mutant of M. smegmatis mc 2 155 sigF, we demonstrate that SigF in M. smegmatis modulates the cell surface architecture and lipid biosynthesis, extending the repertoire of SigF function in this species. We also demonstrate that the increased sensitivity of the sigF mutant to H 2 O 2 mediated oxidative stress is primarily due to loss of the carotenoid pigment. Furthermore, we report the identification of a SigF antagonist, an anti-sigma factor (RsbW), which upon overexpression in M. smegmatis wild type strain produced a phenotype similar to M. smegmatis mc 2 155 ΔsigF strain. The SigF-anti-SigF interaction was duly confirmed using bacterial two-hybrid system and pull down assay. In addition, anti-sigma factor antagonists, RsfA and RsfB were identified and their interactions with anti-sigma factor were verified using two-hybrid system.

Results and Discussion
Construction of Mycobacterium smegmatis sigF knockout mutant and its complementation The sigF deletion (ΔsigF) mutant was created by replacing sigF ORF with the hygromycin (hyg) resistance cassette and molecularly validated (see supplemental material, Fig. S1) as detailed in methods. One of the ΔsigF mutants referred as SFKO1 has been studied and described throughout this manuscript. The SFKO1 was complemented with the sigF gene, cloned downstream of hsp60 promoter, at an ectopic locus in the SFKO1 genome. The sigF complemented strain is designated as SFKO1/sigF.

Role of SigF in stress responses
The effect of sigF deletion on in vitro growth was monitored by comparing the growth of the SFKO1 strain to the wild type M. smegmatis. Both strains were allowed to grow in different media for extended length of time; the sigF mutant strain grew slightly faster than the wild type, attained higher cell density with reduced lag phase, but displayed similar growth characteristics afterwards till extended stationary phase of growth (data not shown). This suggests that the sigF is dispensable for the growth of M. smegmatis under normal physiological conditions. These results are in line with the earlier findings (Provvedi et al. 2008).
SigF has been described as a stress-response sigma factor in slow-growing mycobacteria (DeMaio et al. 1996). Previously, we had shown that sigF is transcriptionally induced in M. smegmatis in response to cold shock, nutrient starvation and after treatment with SDS and antimycobacterial drugs like isoniazid and ethambutol (Singh and Singh 2008). We examined whether SigF is required for survival of M. smegmatis during these stress conditions. No significant differences in survival were noticed between the sigF mutant and the wild type strain under these stress conditions (data not shown). Gebhard et al. (Gebhard et al. 2008) had reported that SigF is required for survival against heat shock and acidic stress in M. smegmatis. We did not test the acidic stress but upon heat shock no appreciable difference in survival of sigF mutant was noticed in comparison to the wild type strain. We checked the sigF deletion mutants of both M. smegmatis mc 2 155 (SFKO1) and M. smegmatis ATCC 607 strains. One of the reasons of this difference could be the temperature as we tested the survival, based on our earlier studies Singh 2008, 2009), at 45°C while they used 50°C for heat stress in their studies.
But, similar to earlier findings (Provvedi et al. 2008), the sigF deletion mutant was found to be more susceptible than the wild type strain upon exposure to hydrogen peroxide mediated oxidative stress (Fig. 1A). Complemented strain (SFKO1/sigF) restored the survival after oxidative stress. Since, sigF was not found to be induced upon oxidative stress in previous studies (Singh and Singh 2008), we examined the sigF expression at RNA and protein level after treatment with hydrogen peroxide. No difference in the sigF expression level was noticed upon oxidative stress using log phase and stationary phase Mycobacterium smegmatis WT (MS), MSΔsigF mutant (SFKO1) and MSΔsigF/sigF complemented (SFKO1/sigF) strains were subjected to oxidative stress (10 mmol L −1 H 2 O 2 ) and their percent survival were calculated as described in methods. Susceptibility of ΔsigF mutant to oxidative stress is completely restored after complementation with sigF. Data were collected from three different experiments. The mean values and standard deviations were plotted for each set of data. **P < 0.01 relative to M. smegmatis wild type (MS) as determined by paired t-test. (B) Quantitative real time RT-PCR analysis of sigF gene expression after oxidative stress (10 mmol L −1 H 2 O 2 ). Relative expression was determined with reference to untreated control (corresponding to 1.0 at Y axis). The expression of genes was normalized with the sigA transcript level. The mean value and standard deviations were calculated from two different experiments and plotted for each set of data. (C) Western blot of SigF protein using protein samples from exponential and stationary phase cultures under treated (10 mmol L −1 H 2 O 2 ) and untreated conditions. Apparently similar levels of SigF proteins were detected in treated and untreated samples. Gel picture shows equal loading of proteins. Loss of carotenoid pigment renders increased H 2 O 2 sensitivity to the sigF mutant Disparate response to oxidative stress was reported in saprophytic and pathogenic mycobacteria (Sherman et al. 1995). Saprophytes like M. aurum and M. smegmatis produce carotenoids, which are known scavengers of free radicals (Levy-Frebault and David 1979) and enhance the strength of the cell wall due to their lipophilic nature and intercalation into the cell membrane (Kubler and Baumeister 1978). M. smegmatis mc 2 155 colonies produce pale yellow pigment (carotenoid isorenieratene) when incubated under light for 5-6 days. Deletion of sigF resulted in loss of pigmentation in SFKO1 ( Fig. 2A) which was mostly restored after complementation with the sigF gene (SFKO1/sigF) ( Fig. 2A), suggesting that the loss of pigmentation is specifically due to deletion of sigF. Because carotenoids are robust antioxidants and fortifiers of cellular barriers they are deemed beneficial for withstanding the stress beard by saprophyte like M. smegmatis. Since, we did not find the appreciable differences in the sigF expression after peroxide mediated oxidative stress despite the marked sensitivity of the ΔsigF mutant to H 2 O 2 , we reasoned that this phenotypic characteristic of the M. smegmatis ΔsigF mutant might be due to absence of carotenoids in the mutant. Moreover, the key detoxifying enzymes of reactive oxygen species in mycobacteria, katG and ahpC were found to be SigF independent (Gebhard et al. 2008;Hümpel et al. 2010). To test our hypothesis, we treated M. smegmatis mc 2 155 cells with diphenylamine (DPA), a known inhibitor of carotenogenesis in mycobacteria (Houssaini-Iraqui et al. 1993), and subjected the DPA-treated bacterial cells to hydrogen peroxide mediated oxidative stress. The DPA-treated bacteria showed pronounced sensitivity to oxidative stress, comparable to M. smegmatis ΔsigF mutant strain (Fig. 2B). This was duly confirmed when SFKO1/crt strain apart from restoring the pigmentation ( Fig. 2A) showed a significant recovery in survival following hydrogen peroxide mediated oxidative stress akin to SFKO1/sigF strain (Fig. 2B).
Carotene isorenieratene is the characteristic pigment of almost all orange-pigmented mycobacteria including M. phlei (Goodwin andJamikorn 1956, 1956), M. aurum (Levy-Frebault and David 1979), M. avium, and M. intracellulare (Tarnok and Tarnok 1970, 1970. The synthesis of isorenieratene requires farnesyl pyrophosphate as a precursor, which leads to isorenieratene in five metabolic steps involving, CrtE, CrtB, CrtI, CrtY, and CrtU (Provvedi et al. 2008). It was postulated that in the absence of SigF, transcription of crt operon is off, hence SFKO1 mutant lacks pigmentation. Evidently, crtI transcript was found to be several-fold downregulated in SFKO1 mutant cells to 80% with respect to untreated control (100%). DPA treated MS cells when exposed to H 2 O 2 showed reduced survival which was relatively similar to H 2 O 2 treated ΔsigF mutant cells and much lower than wild type treated cells. Susceptibility of ΔsigF mutant to oxidative stress is completely restored after complementation with sigF and nearly to a similar extent after complementation with crt locus genes. Data were collected from three different experiments. The mean values and standard deviations were plotted for each set of data. *P < 0.05, **P < 0.01 relative to H 2 O 2 treated M. smegmatis WT (H 2 O 2 /MS) as determined by paired t-test. (C) Expression of crtI gene in SFKO1. In complemented strain SFKO1/sigF expression was restored to almost wild type level. The expression of genes was normalized with the sigA transcript level. The mean value and standard deviations were calculated from two different experiments and plotted for each set of data. in comparison to wild type strain (Fig. 2C) and the expression (Fig. 2C) as well as pigmentation ( Fig. 2A) were restored, almost to the wild type level, in the complemented SFKO1/sigF strain. In M. smegmatis genome, a carotenogenic gene cluster comprises six open reading frames, crtIBYcYdUV, each transcribed in the same direction. The GGPP synthase gene, crtE, was found far away from the crt locus. The upstream regions of crtI gene harbored a canonical SigF promoter signature (Provvedi et al. 2008). When crt locus genes were overexpressed in SFKO1/crt strain, SFKO1/crt akin to SFKO1/sigF, restored the pigmentation ( Fig. 2A) which was lost due to sigF deletion, suggesting that the SigF directly regulates the carotenoid biosynthesis and thereby the pigmentation of bacterial colonies in M. smegmatis. These results established that in M. smegmatis SigF confers resistance to hydrogen peroxide mediated oxidative stress largely through the carotenoid pigments.

SigF modulates cell wall architecture by affecting GPL distribution and lipid biosynthesis
Previously, in M. smegmatis, we observed increased sigF expression upon exposure to isoniazid, ethambutol, and SDS (Singh and Singh 2008). Isoniazid and ethambutol specifically target cell wall biosynthesis process in mycobacteria, whereas SDS is an ionic detergent that affects the cell wall architecture. Overexpression of sigF in M. tuberculosis was reported to alter the regulation of many cell wall-associated proteins, suggesting a role for SigF in maintaining cell wall architecture in mycobacteria (Forrellad et al. 2013). To examine the effect of sigF deletion on the cell wall architecture in M. smegmatis, we performed transmission electron microscopy using M. smegmatis WT and ΔsigF mutant cells. In M. smegmatis, GPLs constitute the major cell-surface glycolipids and react with ruthenium red to give the electron-dense appearance to the outermost cell envelope layer (Etienne et al. 2002). We noticed uniform distribution of GPLs on the surface of WT cells (Fig. 3A), while ΔsigF mutant cells displayed patchy GPLs distribution ( Fig 3B). Next, we analyzed the total GPLs in wild type and ΔsigF mutant by TLC and mass analysis (see supplemental material, Fig. S2), but no difference was found in GPLs profile of ΔsigF mutant, suggesting that the uneven distribution of GPLs in the ΔsigF mutant cells is not due to difference in overall content and type of GPLs. Then, we examined the profiles of other cell wall lipids. TLC analysis of polar lipids also did not reveal any differences (data not shown), but nonpolar lipids showed distinct TLC profiles. Lipids spots present in wild type cells ( Fig. 4A and C) were conspicuously missing in ΔsigF mutant cells ( Fig. 4B and D). We also noticed distinct differences in trehalose containing lipids ( Fig. 4E and F), an important component for cell wall integrity, indicating that the SigF alters the cell wall lipid composition by modulating the lipid biosynthesis pathway.

Genome-wide gene expression studies of Mycobacterium smegmatis ΔsigF mutant and wild-type strains
A genome-wide gene expression analysis of the M. smegmatis mc 2 155 WT and ΔsigF mutant strains was performed using Agilent microarray platform. SigF-regulated genes during exponential phase and stationary phase were indentified, as described in the methods. Difference in the expression of a gene was calculated as the ΔsigF mutant to WT expression ratio and is expressed as fold-change; only ≥ 2-fold difference in the gene expression (P ≤ 0.05) was considered for analysis. Under these conditions, 142 genes in exponential phase and 158 genes in stationary phase were found to be significantly down-regulated in the ΔsigF mutant. A large number of genes showed reduced expression in both exponential and stationary phase cells, and almost similar numbers of genes were found to be down-regulated exclusively in exponential and stationary phase cells (Table 1). We also identified enhanced expression of 39 genes in exponential phase cells and 49 genes in stationary phase cells in ΔsigF mutant strain. The entire expression data can be found in Data set S1 in the supplemental material. To validate the microarray results, real-time PCR was performed on few randomly selected genes from microarray data. Similar to microarray results, the selected genes showed reduced expressions in real-time PCR experiment (see supplemental material, Fig. S3) as well.
The SigF promoter consensus in M. smegmatis was first identified in silico (Provvedi et al. 2008), and was later improved upon by experimental data (Gebhard et al. 2008;Provvedi et al. 2008;Hümpel et al. 2010). Using an improved SigF promoter consensus from later studies, 1200 bp upstream of the annotated start codon of the down-regulated genes (Table 1) were visually checked for sequence similarities. We searched 1200 bp upstream sequence because several genes were arranged in gene clusters wherein the SigF consensus  was found far upstream of the down-regulated genes or even in the ORFs of the preceding genes. It may be noted that the canonical SigF promoter consensus was located more than 1000 bp upstream of the sigF gene in M. smegmatis genome (Gebhard et al. 2008). We reasoned that the SigFdependent genes are likely to be down-regulated in both stages of growth. Notably, genes that showed reduced expressions commonly in exponential as well as stationary phase Fold-change in expression -ΔsigF strain/wild-type gene expression ratio in log2 scale.SigF consensus (GTTT-N (14-19) -GGGTA) was found in the upstream regions of majority of the down-regulated genes. Locus IDs in bold refer to genes that are clustered as operon in the genome. SigF consensus in such cases was found either in ORFs of preceding genes or in far upstream of the first gene of the cluster, e.g. SigF consensus was present 97 bp upstream of MSMEG_2347, MSMEG_2343-MSMEG_2347 constitute crt locus. a Genes found down-regulated in Hümpel et al. (2010) as well as in this study. cells, most of them showed the presence of the SigF promoter consensus in their upstream regions (Table 1), suggesting that they are SigF-dependent. Majority of genes that showed reduced expressions in this study were also reported to be down-regulated by Humpel et al. (Hümpel et al. 2010). They identified the SigF promoter consensus in the upstream regions of transcriptional regulators, sigH3 (MSMEG_0573), whiB1 (MSMEG_1919), whiB4 (MSMEG_6199), and phoP (MSMEG_5872), but the expressions of these genes were found unaltered in the ΔsigF mutant. In this study, using our selection criteria (≥2-fold, P ≤ 0.05), we identified three transcriptional regulators; MSMEG_5542 (HTH3 family), MSMEG_5731 (GntR family), and MSMEG_6508 (MarR family) which showed reduced expression in exponential phase, and MSMEG_5542, MSMEG_5301 (TetR family) with reduced expression in stationary phase. Of these MSMEG_5542, 5731, 6508 were found to have SigF consensus in their upstream regions. It is likely that the down-regulated genes which did not show SigF foot-prints in their upstream regions are indirectly regulated by SigF-dependent transcriptional regulators. Several of the exclusively down-regulated genes from exponential and stationary phase cells also showed SigF promoter consensus in their upstream regions, while few of them were found lacking the consensus. Based on the SigF promoter sequences, identified from this study, we deduced a profile of the SigF promoter consensus (Table 1), which showed the similar occurrence of the nucleotides at a given position in the earlier reported SigF promoter signature (Hümpel et al. 2010).

Mycobacterium smegmatis ΔsigF mutant phenotype and SigF regulon
The M. smegmatis ΔsigF mutant displayed notable phenotypes likes, loss of pigmentation, pronounced sensitivity to oxidative stress and alteration in the cell wall architecture due to patchy distribution of GPLs in the cell wall. Correlating the loss of pigmentation phenotype the expressions of carotenoid biosynthesis genes (MSMEG_2243-MSMEG_ 2247) were found to be down-regulated during both growth stages (Table 1). The SigF promoter consensus was identified in the upstream of the cluster and the reduced expression of crtI, the first gene of the cluster, was validated by real time PCR (Fig. 2C). Complementation of the ΔsigF mutant restored the original phenotype ( Fig. 2A).
Regarding the sensitivity to oxidative stress the expressions of key enzymes that detoxify reactive oxygen intermediates, katG and ahpC, were found unaltered in the mutant strain, suggesting these genes are not regulated by SigF. We demonstrated that the overexpression of crt locus genes largely restores the susceptibility of ΔsigF strain to oxidative stress. Moreover, several genes which could possibly render resistance to ΔsigF strain against oxidative stress were found to be SigF-dependent and showed reduced expressions in both growth stages of ΔsigF strain. Two potential hydrogen peroxide detoxifying enzymes, exclusively present in M. smegmatis, a manganese containing catalase (MSMEG_6213) and a heme containing catalase KatA (MSMEG_6232), showed reduced expressions in both stages in present study as well as in earlier report (Hümpel et al. 2010). A starvation-induced DNA protecting protein (MSMEG_6467) linked with oxidative stress resistance in bacteria (Gupta et al. 2002) showed reduced expression in both growth stages. M. smegmatis is a saprophyte and dehydrogenase activity is considered to be a good measure of microbial oxidative activity in saprophytes. Many genes (MSMEG_1794, MSMEG_5400, MSMEG_5402, MSMEG_0684) encoding for dehydrogenages and predicted to perform oxidoreductase activity (SmegmaList) were found to be SigF-dependent and down-regulated in both growth stages. These are likely to render susceptibility to the mutant strain toward oxidative stress.
In M. smegmatis, GPL biosynthesis gene cluster maps to a single locus of ~65 kb in the genome, containing nearly 30 ORFs that included genes for the synthesis as well as transport of GPLs (Ripoll et al. 2007). In the genome-wide gene expression study (see supplementary Data set S1) no genes from GPL biosynthesis gene cluster showed altered regulation in the ΔsigF mutant strain. We also did not find the SigF consensus signature in the upstream regions of genes clustered at this locus. This was in line with our earlier observation wherein we did not notice any difference in GPLs profile of ΔsigF mutant. However, a complete analysis of polar and nonpolar lipids from ΔsigF mutants showed distinct differences in 2D-TLC profile of nonpolar lipids in mutant strain. Concomitant with these findings trehalose biosynthesis genes (MSMEG_6514, MSMEG_6515) and mycocerosic acid synthase genes (MSMEG_6765 to MSMEG_6767) were found to be significantly down-regulated in ΔsigF strain (Table 1). MSMEG_6515 encodes for trehalose synthase which enables the conversion of glycogen into trehalose. The SigF promoter consensus was identified in the upstream of these genes, indicating that trehalose and mycocerosic acid synthase (MAS) genes are directly regulated by SigF and affect the cell wall architecture by inhibiting lipid biosynthesis pathway in sigF mutant.

Post-translational regulation of SigF in
Mycobacterium smegmatis: overexpression of rsbW mimics the M. smegmatis ΔsigF mutant phenotype Sigma factors activity is post-translationaly regulated by their cognate anti-sigma factors, which sequester them and make them unavailable for RNAP. In M. tuberculosis, SigF is post-translationally regulated by its cognate antisigma factor RsbW, which is, in turn, regulated by two anti-anti-sigma factors, RsfA and RsfB (Beaucher et al. 2002). Both are able to disrupt the RsbW-SigF complex, releasing SigF to allow its association with RNA polymerase. In M. smegmatis rsbW (MSMEG_1803) is colocalized (Fig. S1) and cotranscribed with sigF (MSMEG_1804) (Gebhard et al. 2008). But, barring the sequence similarity with M. tuberculosis RsbW (Rv3287c), there has been no experimental evidence till date which demonstrates that MSRsbW binds to SigF and regulates it negatively. We argued that if MSMEG_1803 is indeed the anti-SigF, RsbW, negatively regulating the SigF in M. smegmatis, overexpression of MSMEG_1803 in M. smegmatis wild type cells should sequester the prevailing pool of SigF and thereby making them unavailable for binding to RNA polymerase. This will impede the expression of SigF regulon and the MSMEG_1803 overexpressing M. smegmatis cells will produce a phenotype akin to M. smegmatis ΔsigF mutant. As shown in Fig. 5(A) and (B), we observed loss of pigmentation and increased susceptibility to oxidative stress in strain MS:MSrsbW nearly similar to SFKO1, the ΔsigF mutant strain. This proved that MSMEG_1803 indeed encodes for the cognate anti-SigF protein which binds to SigF in M. smegmatis and regulates it negatively. Similar observations were made with M. smegmatis wild type cells overexpressing M. tuberculosis rsbW (MS:MtbrsbW) ( Fig. 5A and B), which further established that MSMEG_1803 is true ortholog of MtbrsbW, as both strains produced similar phenotypes akin to SFKO1. To establish that the observed phenotypes of MS:MSrsbW and MS:MtbrsbW strains are indeed due to overexpression of rsbW and sequestering of SigF proteins we performed real time semiquantitative RT-PCR of these genes in M. smegmatis wild type, SFKO1 and overexpressing recombinant strains. We also examined the expression levels of two putative anti-anti-sigF proteins RsfA (MSMEG_1786) and RsfB (MSMEG_6127) from M. smegmatis, which were identified based on their homology to M. tuberculosis RsfA and RsfB. As observed in Fig. 5(C) the expression levels of rsbW, rsfA, and rsfB were found to be similar to wild type, while the sigF was nearly absent, owing to its deletion, in SFKO1 strain. However, the expressions of these genes were found to be similar in MS:MSrsbW and MS:MtbrsbW strains, suggesting that MSrsbW (MSMEG_1803) is indeed similar to MtbrsbW. A negligible expression of sigF gene was noticed in both strains, which implies that enhanced cellular level of RsbW protein, owing to its overexpression (Fig. 5C), completely sequestered the SigF protein, and, in turn shut down the expression of sigF gene, which is transcriptionally autoregulated. Since the sigF is cotranscribed with rsbW the increased rsbW level in MS:MSrsbW and MS:MtbrsbW strains amounts to the ectopically expressed rsbW under the control of hsp60 pr in these strains. Interestingly, the expressions of rsfA and rsfB were also found to be induced, similar to rsbW, in both recombinant strains. RsfA and RsfB are known to antagonize RsbW, therefore, it is possible that some feedback machinery in the bacterial cell would have sensed the increased cellular level of RsbW and invoked an ensuing response by transcriptionally upregulating the expression of both anti-sigF antagonists. It may be noted that the expression levels of RsfA (MSMEG_1786) and RsfB (MSMEG_6127) were not significantly altered in ΔsigF mutant strain in genome wide gene expression analysis performed in this study and by Hümpel et al. 2010. Also both these genes lacked SigF footprints in their upstream regulatory regions.
Furthermore, using bacterial two-hybrid experiment we analyzed the interactions of M. smegmatis anti-SigF RsbW with SigF and its two antagonists RsfA and RsfB. M. smegmatis RsbW showed very strong interactions with SigF and RsfA while a comparatively weak interaction was noticed with RsfB (Table 2). Similar results were obtained when we allowed M. tuberculosis RsbW to interact with M. smegmatis SigF, RsfA, and RsfB (Table 2). On the other hand, we did not notice any interaction when M. smegmatis RsbW was allowed to interact with M. smegmatis SigA, which confirmed the specificity of MSRsbW to its cognate sigma factor SigF. To further confirm these interactions we performed GST pull down assay. M. smegmatis RsbW was overexpressed as GST tagged protein (GST-MSRsbW) using pET41a+ vector in Escherichia coli, purified and immobilized on GST beads. A column was prepared with GST-MsRsbW immobilized beads and whole cell lysates of recombinant E. coli strains overexpressing M. smegmatis SigF, RsfA, and RsfB proteins were applied and allowed to bind to GST-MsRsbW. Subsequently, interacting proteins were eluted using reduced glutathione and electrophoresed on SDS-PAGE (Sodiumdodecyl sulfate polyacrylamide gel electrophoresis) (Fig. 6). Individual bands were excised and sequenced using MALDI/MS (data not shown). We noticed similar level of interactions between RsbW, SigF, RsfA, and RsfB proteins as it was observed in bacterial two-hybrid assay. Thus, combined together, bacterial two-hybrid and GST pull down results clearly established that MSMEG_1803 encodes for anti-SigF RsbW protein in M. smegmatis which specifically and strongly interacts with its cognate sigma factor SigF and its antagonists RsfA and RsfB. The fact that these proteins showed similar level of interactions with M. tuberculosis RsbW suggests that most likely, similar to M. tuberculosis, in M. smegmatis SigF is post-translationally regulated by its anti-sigma factor RsbW, which is in turn regulated by its antagonists RsfA and RsfB. However, further experiments are required to elucidate the regulation of these interactions with respect to different physiological states of mycobacterial cells. It would be of interest to examine whether some more SigF antagonists are present in M. smegmatis genome as predicted by Hümpel et al. (2010) in their studies.

Conclusions
In this study, we report that in M. smegmatis the SigF is not essential for growth of bacterium. Deletion of sigF results in loss of carotenoid pigmentation which rendered increased susceptibility to H 2 O 2 induced oxidative stress as complementation of ΔsigF mutant with carotenoid genes largely restores the phenotype. In M. smegmatis, sigF deletion altered the outer most layer of the cell envelope and the cell wall lipid composition by modulating the lipid biosynthesis pathway. M. smegmatis SigF regulon Table 2. Interactions of anti-SigF (RsbW) with its antagonists (RsfA and RsfB) and SigF.

DNA manipulation, construction of sigF mutant, and its complementation
Recombinant DNA techniques were performed as per standard procedures (Sambrook et al., 2001) using E. coli DH5α as the initial host. Restriction and DNA modifying enzymes were obtained from Fermentas. Primers used in this study are described in Table 4. Preparation of electrocompetent cells and electroporation were done as previously described (Singh and Singh 2008). M. smegmatis mutant lacking sigF was constructed using allelic exchange method. For this, a hygromycin resistance cassette flanked by nearly 1 kb flanking regions of each side of the sigF gene was cloned into pDrive plasmid vector generating pDΔsigF. The final allele exchange cassette contained 5′flank/Hyg r /3′flank in pDΔsigF. 5′ and 3′ flanking regions contained a few nucleotide sequences of sigF gene which was later used for PCR amplification of sigF ORF from wild type and ΔsigF mutant. pDrive contains only E. coli origin of replication and, therefore, fails to multiply in mycobacteria and serves as suicide vector in mycobacteria. pDΔsigF was electroporated into M. smegmatis mc 2 155 and transformants were selected on hygromycin (50 μg mL −1 ) plates. The expected double crossover event would exchange sigF gene with hygromycin resistance marker in mutant strain. Selected colonies were first screened by PCR using MSSF1 and MSSF2 primers followed by sequencing and finally validated using Southern blotting. Southern blot was carried out using SmaI digested genomic DNA of M. smegmatis wild type and putative sigF deletion mutants using two probes, one specific for sigF-rsbW (Probe 1) and another for hyg (Probe 2) (Fig. S1). The probe was labeled using Dig High Prime DNA labeling

Susceptibility of Mycobacterium smegmatis strains to oxidative stress
For stress experiments, different M. smegmatis strains were grown to 0.6-0.8 OD 600 (exponential phase) and 2.6-2.8 OD 600 (stationary phase) and then cultures were split into aliquots. For oxidative stress, cultures were treated with H 2 O 2 (10 mmol L −1 ), allowed to grow for 4 h at 37°C and plated thereafter in duplicates following 10-fold serial dilution for CFU analysis. Untreated cultures were taken as control for stress experiments. The total number of colonies that appeared in the untreated control was considered 100%. Data were collected from three different experiments. The mean values and standard deviations were plotted for each set of data. For inhibition of carotenoid biosynthesis, initially the dose of diphenylamine (DPA) was set so that ≥ 80% of M. smegmatis mc 2 155 wild type cells survive after DPA treatment. 0.1 mmol L −1 DPA treatment for 4-6 h ensured the survival of 80% wild type cells. Further experiments with different M. smegmatis strains (Fig. 2) were performed with exponentially grown culture at similar OD values (0.6-0.8). Cultures were incubated with 0.1 mmol L −1 DPA for 2 h before H 2 O 2 treatment and stress susceptibility was analysed as described above.

Generation of anti-SigF antibody and immunodetection of SigF
The M. smegmatis sigF ORF was amplified using genespecific primers and cloned into PCR cloning vector pTZ57R/T. The clone was verified by DNA sequencing following which the ORF was relocated to the pET28a+ expression vector generating pETSigF. SigF was overexpressed as N-terminal His 6 -tagged recombinant in E. coli C41 cells, purified using Ni-NTA affinity chromatography and the purified His 6 -SigF was used to raise anti-SigF antibody in female New Zealand white rabbit, as described previously (Biswas et al. 2013). Immunodetection was performed with the primary antibody (polyclonal sera at 1:2000), followed by washing and incubation with the secondary antibody (anti-rabbit IgG horseradish peroxidase conjugate at 1:40,000). The blots were developed using the chemiluminescent substrate (Pierce) and the signals were captured on the Bio-Rad Chemidoc system.

Transmission electron microscopy
Electron microscopy samples were prepared as described previously (Paul and Beveridge 1992). Briefly, fully grown cultures of M. smegmatis strains were diluted (1:100) in fresh LBGT broth and allowed to grow till 0.5 OD 600 . Cultures were centrifuged at 400 × g for 2 min to separate homogenous cell suspension from cell aggregates. Homogenous suspensions were transferred to new tubes and cells were harvested by centrifugation at 2600 × g for 5 min. Cells were washed five times with 0.1 mol L −1 cacodylate buffer (pH 6.8) and pellets (~50 mg wet weight) were fixed in 2.5% (w/v) glutaraldehyde, 0.05% ruthenium red in 0.1 mmol L −1 cacodylate buffer in dark at 4°C overnight. Cells were collected by centrifugation, washed thrice in 0.1 mol L −1 cacodylate buffer before fixing for 2 h in dark in 1% (w/v) osmium tetroxide, 0.05% ruthenium red in 0.1 mol L −1 cacodylate buffer. After this cells were washed thrice in 0.1 mol L −1 cacodylate buffer for 5 min each and embedded in 2% agarose gel. Blocks were dehydrated through a graded ethanol series of 20, 40, 60, 80, and 95% for 5 min each followed by two 10 min washes in absolute ethanol. Samples were embedded in EPON 812 resin at 60°C for 48 h. Ultra thin sections (50-70 nm) were obtained using Ultracut Ultra Microtome (Leica) and picked upon 200 mesh copper grids. Sections were poststained with uranyl acetate and Reynold's lead citrate. Microscopy was performed on a Philips FEI Technai-12 Twin Transmission Electron Microscope and images were recorded using a SIS mega View II CCD camera attached with the microscope.

Extraction and analysis of GPLs and total lipids from Mycobacterium smegmatis
GPLs extraction and analysis were performed as described earlier (Vats et al. 2012). The M. smegmatis wild type and mutant strains were grown in Middlebrook 7H9 medium supplemented with 10% ADC till late stationary phase (2.8-3.0 OD 600 ). GPLs were extracted with CHCl 3 / CH 3 OH (2:1) at room temperature for 24 h. The supernatant was dried using rotatory evaporator till dryness. The lipid extract was deacetylated by 0.2 mmol L −1 NaOH in methanol at 37°C for 1 h followed by neutralization with glacial acetic acid. After drying, lipids were dissolved in CHCl 3 /CH 3 OH (2:1), spotted onto the TLC plate (Aluminium baked silica gel 60 F254) (Merck) and developed in CHCl 3 /CH 3 OH/H 2 O (90:10:1) solvent. GPLs were visualized by spraying with 5% αnaphthol/sulfuric acid in ethanol followed by charring at 120°C for 10 min. The four de-O-acetylated GPLs (dGPLs) were named dGPL I, II, III, and IV, starting from the solvent front. For mass analysis GPLs were analysed and identified by ESI-Q-TOF-MS (Absciex).
Extractions and analysis of lipids were performed as described earlier (Slayden and Barry 2001). Lipids were extracted from freeze dried stationary phase grown M. smegmatis cells. Bacterial cells were resuspended in equal volume of methanolic saline and petroleum ether, mixture was stirred for 12-16 h and then allowed to separate following which nonaqueous phase containing the nonpolar lipids were removed and stored. An equal volume of petroleum ether was added to lower aqueous phase, mixture was stirred for 2 to 4 h, nonaqueous layer was removed and pooled with the first one. Nonpolar lipids were dried using a rotatory evaporator and resuspended in dichloromethane. Extraction of polar lipids was performed by adding chloroform (CHCl 3 ), CH 3 OH, and 0.3% aqueous NaCl (9:10:3) to the extract. The entire mixture was stirred for 4 h and the solvent extract was separated from the biomass. Furthermore, the residues were extracted with CHCl 3 , CH 3 OH, and 0.3% aqueous NaCl (3:10:4) for 4 h. The polar lipid extracts were mixed with CHCl 3 and 0.3% aqueous NaCl in equal ratio and the lower organic layer was separated discarding the upper aqueous layer. Polar lipids were dried using rotatory evaporator and resuspended in CHCl 3 and CH 3 OH (2:1). 100 μg of lipid extracts were spotted onto the TLC plate (aluminium baked silica gel 60 F254) (Merck) and developed using solvent systems described below. Lipids were detected by charring with 5% phosphomolybdic acid (MPA, Sigma-Aldrich) in ethanol.

Protein-protein interaction analyses using bacterial two-hybrid
BacterioMatch II two-hybrid system (Agilent Technologies) was used for analyses of protein-protein interactions. The system utilizes a double HIS3-aadA reporter cassette which identifies interacting partners with plausibly reduced background. Detection of protein-protein interactions is based on transcriptional activation of the HIS3 reporter gene, which allows growth in the presence of 3-amino-1, 2, 4-triazole (3-AT), a competitive inhibitor of His3 enzyme. Positives are reconfirmed by using the aadA gene, which confers streptomycin resistance, as a secondary reporter.
Mycobacterium smegmatis sigF, sigA, anti-sigF rsbW (MSMEG_1803), and anti-sigF antagonists, rsfA (MSMEG_1786) and rsfB (MSMEG_6127) were amplified using gene specific primers (Table 4) and cloned into bait vector pBT at given enzyme sites (Table 3). Similarly, anti-sigma factors from M. smegmatis (MSrsbW) and M. tuberculosis (MtbrsbW) were amplified using gene specific primers (Table 4) and cloned into target vector pTRG at given enzyme sites (Table 3). All cloning steps were performed in E. coli XL1Blue strain, and the clones were verified by restriction digestion and DNA sequencing. To analyze interactions between two proteins, plasmid pairs carrying ORFs in pBT and pTRG vectors were cotransformed in XL1Blue derived reporter strain, provided with two-hybrid system. Cotransformants were selected on M9 and M9-3AT plates. The cotransformant containing pBT-LGF2 and pTRG-GaL11 P (Agilent) was used as a positive control for expected growth on the selective screening medium (M9 with 5 mmol L −1 3-AT). A cotransformant containing the empty vectors pBT and pTRG was used as a negative control. Further positives were verified using second reporter gene (aadA), conferring streptomycin resistance. The interaction between the bait and target proteins was revalidated by patching cells from a putative positive colony from a selective screening medium (M9-3AT) plate onto a dual selective screening medium (M9-3AT + streptomycin 15 μg mL −1 ) plate. CFU obtained on the nonselective screening medium (M9 without 3AT) and selective medium (M9-3AT) plates were counted, and values were used to determine the percent interaction. The average and standard deviations were determined from data generated from two different experiments.
Cloning, expression, purification of RsbW, SigF, RsfA and RsfB and GST pull down assay Mycobacterium smegmatis rsbW ORF was amplified using gene specific primers and cloned into pET41a+ at SpeI and XhoI sites to generate pET41a-MSrsbW. This allowed MSrsbW to be cloned in fusion with GST at its N-terminal.
India, official service partner of Agilent Technologies (USA). Array was spotted using 60 mer oligo probes (features) in 8 x15K format (Ref No: AMADID: 016421). Average number of probes per gene in each array is 3. Probes were designed in such a way that multiple probes for a given gene specifically hybridize to different parts of the transcript. Each array carried Agilent proprietary probes for quality control purpose. M. smegmatis microarray slides were hybridized with the labeled cRNA. Before hybridization 0.6 μg of each Cy3 labeled cRNAs were fragmented to uniform size of 200 bp to avoid folding up of long transcripts and also remove any steric hindrance which may arise due to secondary structure in long RNA molecules during hybridization. Fragmentation and hybridization were carried out using the Gene Expression Hybridization kit (Part # 5188-5242, Agilent Technologies). Hybridization was carried out in Agilent's Surehyb Chambers at 65°C for 16 h. After hybridization slides were washed using Agilent Gene Expression wash buffers, first at RT and then twice at 37°C. Slides were quickly dried and scanned using the Agilent Microarray Scanner G Model G2565BA at 5 micron resolution. The images were manually verified and found to be devoid of uneven hybridization, streaks, blobs, and other artifacts.

Feature extraction and data analysis
Data extraction from images was done using Feature Extraction software v 10.5.1.1 (Agilent). Feature extracted data were analyzed using GeneSpring GX v 7.3.1 software (Agilent). Normalization of the data was done in GeneSpring GX using the recommended one color Per Chip and Per Gene Data Transformation: Set measurements <0.01 to 0.01 per Chip: Normalize to 50th percentile per Gene: Normalize to Specific Samples. The gene expression ratio (ΔsigF/WT) of ≤ 0.5 or ≥2.0 (P ≤ 0.05) was considered differentially regulated and filtered from the data. Fold-chage refers to expression ratio of ΔsigF strain to wild-type and is expressed in log2. Ratios were tested for significance using student T-test from Agilent's Gene Spring GX version 7.3 software.

Real-time reverse transcription-PCR (RT-PCR) analyses
RNA was extracted from exponential and stationary phase cultures of M. smegmatis wild type and derivative strains (SFKO1, SFKO1/sigF, MS:MSrsbW and MS:MtbrsbW) as described earlier (Singh and Singh 2009). DNase treatement was carried out to remove any DNA contamination, and post-treatment RNA was reverse transcribed using random primers and Transcriptor reverse transcriptase (Roche). qRT-PCR was performed in triplicates using SYBR Green master mix on a Roche 480 LightCycler, as described previously (Singh and Singh 2009). Expression of target genes was normalized with the sigA transcript level. RNA samples that had not been reverse transcribed were included as controls in all the experiments. The mean relative expression levels and SD were determined from the data generated from two different experiments. Each experiment was set up in triplicates.

Microarray data accession number
All experimental details and data have been deposited at the Gene Expression Omnibus (GEO, NCBI) under accession number GSE19774.

Statistical analysis
Significant differences between experimental groups were determined using Student's t-test (GRAPHPAD PRISM 5, GraphPad Software, Inc., La Jolla, CA). For all analyses, a P-value of <0.05 was considered statistically significant.

Supporting Information
Additional supporting information may be found in the online version of this article: Data S1. Log phase and stationary base. Figure S1. Schematic of sigF locus and construction of sigF mutant. Figure S2. TLC profile of the de-O-acetylated GPLs, extracted from the Mycobacterium smegmatis WT (MS) and mutant strain (SFKO1), as described in methods. dGPL I, II, III, and IV, starting from the solvent front. Mass spectra profile of GPLs (I, II, III, and IV) extracted from M. smegmatis wild type (A) and ΔsigF mutant (B). Figure S3. Real time RT-PCR analysis of select genes from microarray data that were found to be down-regulated in ΔsigF mutant.