Mycobacterium tuberculosis SigF Regulates Genes Encoding Cell Wall-Associated Proteins and Directly Regulates the Transcriptional Regulatory Gene phoY1

Abstract

Mycobacterium tuberculosis SigF is homologous to stress response and sporulation sigma factors in many bacteria. Consistent with a possible role in mycobacterial survival under stress conditions, M. tuberculosis sigF is strongly induced within cultured human macrophages and upon nutrient starvation, and SigF has been implicated in M. tuberculosis entry into stationary phase. On the other hand, SigF appears to contribute to the immune pathology of tuberculosis (TB), and a sigF-deficient mutant has altered cell membrane properties. Using an M. tuberculosis sigF deletion mutant, we show here that sigF is not required for bacillary survival under nutrient starvation conditions and within activated murine macrophages or for extracellular persistence in an in vivo granuloma model of latent TB infection. Using a chemically inducible recombinant strain to conditionally overexpress sigF, we did not observe arrest or retardation of growth in exponentially growing cultures or reduced susceptibility to rifampin and isoniazid. Consistent with our hypothesis that SigF may mediate TB immunopathogenesis by altering cell membrane properties, we found that overexpression of sigF resulted in the differential regulation of many cell wall-associated proteins, including members of the MmpL, PE, and PPE families, several of which have been shown to influence host-pathogen interactions. The most highly upregulated gene by quantitative reverse transcription-PCR at all time points following sigF induction was Rv3301c (phoY1), which encodes a probable transcriptional regulatory protein homologous to PhoU proteins involved in regulation of phosphate uptake. Using in vitro transcription analysis, we show that SigF directly regulates phoY1, whose proposed promoter sequence is GGATTG-N₁₆-GGGTAT.

Alternative RNA polymerase sigma factors are commonly used in many bacterial species to coordinate gene regulation in response to a variety of environmental conditions (30). The Mycobacterium tuberculosis genome encodes 13 sigma factors, including the primary sigma factor SigA, the primary-like sigma factor SigB, and 11 alternative sigma factors comprising 10 extracytoplasmic function subfamily sigma factors and SigF (21). The M. tuberculosis sigF gene (11) is homologous to stress response sigma factors in Staphylococcus aureus, Listeria monocytogenes, and Bacillus subtilis as well as sporulation sigma factors in Streptomyces coelicolor and B. subtilis (10). The activity of M. tuberculosis SigF is regulated posttranslationally by the anti-sigma factor UsfX, encoded by the usfX gene, which is located upstream of the sigF gene, and these two genes are cotranscribed from a SigF-dependent promoter located directly upstream of usfX. In turn, two different anti-anti-sigma factors, RsfA and RsfB, appear to regulate the activity of UsfX by redox potential and phosphorylation, respectively (2). Consistent with a possible role in bacterial survival under stress conditions, M. tuberculosis sigF is strongly induced within cultured human macrophages (13) as well as during the stationary phase of growth and upon exposure to nitrogen depletion, cold shock (11), nutrient starvation (3), and several antituberculous drugs (22).

M. tuberculosis sigF has also been implicated in tuberculosis immunopathogenesis. Specifically, a sigF-deficient mutant of M. tuberculosis was found to have an immunopathology phenotype in mice, characterized by attenuated tissue damage despite persistence at high colony counts in mouse lungs (12) as well as reduced lethality (6) compared to the isogenic wild-type strain. Similarly, a sigF-deficient strain was found to induce ill-defined lung granulomas lacking necrosis relative to the parent strain in guinea pigs (18). Interestingly, the ΔsigF mutation conferred negative neutral red staining to an M. tuberculosis strain (12), suggesting reduced synthesis of cell envelope-associated sulfolipids. In support of this hypothesis, global gene expression analysis of the M. tuberculosis ΔsigF mutant strain during stationary phase revealed relative downregulation of several genes involved in the biosynthesis and structure of the cell envelope as well as genes involved in the biosynthesis and degradation of surface polysaccharides and lipopolysaccharides (12). The cell envelope of M. tuberculosis is known to be important in directing host-pathogen interactions (16), thus providing a potential mechanism by which sigF may modulate immune pathology in the infected host.

In order to test the hypothesis that sigF is a stress response sigma factor involved in M. tuberculosis cell cycle regulation and survival under stress conditions, we studied the survival phenotype of a ΔsigF mutant (6) under nutrient starvation conditions, within activated macrophages in vitro, and in an in vivo granuloma model of M. tuberculosis persistence (17). Using a chemically inducible recombinant strain to conditionally overexpress sigF, we characterized the effect of sigF upregulation on M. tuberculosis growth kinetics and susceptibility to isoniazid and rifampin. We used global gene expression profiling following sigF induction to investigate possible sigF-regulated genes.

MATERIALS AND METHODS

Strains.

A chemically inducible acetamidase promoter-sigF fusion gene and a kanamycin resistance gene were integrated into the chromosome of M. tuberculosis CDC1551 by site-specific recombination between attP on plasmid pSCW35 and attB in the mycobacterial chromosome (sigF-inducible strain), as previously described (1). The control strain of M. tuberculosis CDC1551 was generated by the same method with plasmid pSCW38 (which is identical to pSCW35 but lacks the sigF gene fused to the acetamidase promoter). A ΔsigF M. tuberculosis CDC1551 mutant was previously constructed using allelic exchange, and its complemented strain was generated using a single-copy, integrative plasmid (pPC51) harboring the entire sigF operon and upstream elements (6).

Antibiotic exposure.

Control and sigF-inducible strains grown to mid-exponential phase in supplemented Middlebrook 7H9 broth were exposed to 0.2% acetamide beginning 24 h prior to the addition of antibiotics and throughout the entire duration of antibiotic exposure. For determination of isoniazid MIC, 10⁵ organisms from acetamide-induced control and sigF-inducible strains were grown for 14 days at 37°C in the presence of the following final concentrations of isoniazid: 0, 0.015, 0.03, 0.06, 0.12, 0.25, 0.5, 1, 2, and 4 μg/ml. For determination of rifampin MIC, 10⁵ organisms from acetamide-induced control and sigF-inducible strains were grown for 14 days at 37°C in the presence of the following final concentrations of rifampin: 0, 0.0625, 0.125, 0.25, 0.5, 1, 2, 4, 8, and 16 μg/ml. Samples were incubated in standing tubes, and the MIC was defined as the lowest antibiotic concentration in which a bacterial pellet was absent.

Macrophage experiments.

Alveolar macrophage J774A.1 cells were cultured in RPMI media supplemented with 0.2 mM l-glutamine (Invitrogen) and 10% heat-inactivated fetal bovine serum (Sigma) in a humidified 37°C, 5% CO₂ incubator. After growth to a confluent monolayer, the cells were harvested using a sterile rubber scraper and a single cell suspension was prepared and cultured in 24-well plates (Costar). When the density of cells reached 10⁶ per well, the cells were activated by addition of 500 U/ml gamma interferon (Roche) overnight and 200 ng/ml lipopolysaccharide (Sigma) for 3 h prior to infection. After removal of the media, the cells were infected for 2 h with wild-type M. tuberculosis, the ΔsigF mutant, or the sigF complemented strain at a multiplicity of infection of 1:1. The macrophages were then washed five times with RPMI and cultured in media as described above. For each time point following infection, intracellular bacteria were recovered by removing media and lysing cells for 5 min in 0.1% Triton X-100 (Sigma). Bacterial CFU were counted 21 days after being plated on 7H10 plates.

Nutrient starvation.

Wild-type M. tuberculosis, the ΔsigF mutant, and the sigF complemented strain were grown in supplemented 7H9 broth to exponential phase (optical density at 600 nm [OD₆₀₀] = 0.4). Each culture was pelleted, washed twice with phosphate-buffered saline, and resuspended in phosphate-buffered saline with 0.05% Tween 80 prior to being transferred to standing flasks and incubated at 37°C, as previously described (3). The viability of each strain during starvation was assessed by plating triplicate cultures on 7H10 plates at the following time points: 24 h, 48 h, 72 h, 7 days, 14 days, and 28 days.

Hollow-fiber infection.

The CDC1551 wild-type, ΔsigF mutant, and complement strains were passaged in the mouse and frozen. Each strain was grown up to an OD of ∼1.0 in supplemented Middlebrook 7H9 broth prior to infection. Sixty-nine 6-week-old female SKH1 mice (Charles River) each underwent subcutaneous placement of two hollow fibers containing one of the three M. tuberculosis strains (23 mice per group) on day 0 (∼10⁴ organisms/hollow fiber). On the day after infection (day 1), three mice from each group were sacrificed for intrafiber CFU counts to determine the numbers of CFU implanted. At subsequent intervals (days 7, 14, and 28), three mice from each group were sacrificed for intrafiber CFU counts. Log-transformed CFU values were used to calculate averages and standard errors for graphing purposes.

Acetamide induction.

Early-exponential-phase cultures (A₆₀₀, ∼0.3) of control and sigF-inducible strains grown in supplemented Middlebrook 7H9 broth (10% oleic acid-albumin-dextrose-catalase, 0.05% Tween, and 0.1% glycerol) were exposed to 0.2% acetamide for the following time intervals prior to RNA extraction for microarray analysis: 0 h, 6 h, 12 h, and 24 h. For growth kinetics experiments, each strain was grown in supplemented Middlebrook 7H9 broth in the presence of 0.2% acetamide for 6 h, 12 h, 24 h, 48 h, 72 h, and 7 days.

Microarray preparation and data analysis.

At the above-defined time points, 50-ml samples were removed from incubating cultures and centrifuged at 3,500 rpm for 10 min at 4°C. The supernatants were removed, and the pellets were resuspended in 1 ml TRIzol reagent (GIBCO/BRL). Mycobacterial membranes were disrupted using 0.1 mm zirconia/silica beads in a Bio-Spec bead beater (8 cycles at 30 s/cycle). Mycobacterial RNA was recovered by centrifugation, chloroform extraction, and isopropyl alcohol precipitation, as previously described (3, 29). Fluorescently labeled cDNA was generated using Powerscript (Clontech) with fluorescent dyes Cy3 and Cy5 (Amersham). Labeled cDNA from the sigF-inducible strain was competitively hybridized to arrays with labeled cDNA from the control strain for each time point after acetamide exposure. The cDNAs were simultaneously hybridized on microarray slides containing commercially available (Operon) M. tuberculosis 70-mer oligonucleotides representing all open reading frames (ORFs) annotated in the H37Rv genome sequencing project, and fluorescence intensity data were collected with a GenePix 4000B scanner (Axon Instruments) with GenePix Pro 4.0 software. GenePix result files were loaded into the R statistical language using the limma package from Bioconductor (www.bioconductor.org) for normalization and statistical analysis. Median spot intensities were calculated using the read.maimages function with weighted spots set to zero and normalized using the normalizeWithinArrays function with the nonlinear printtiploess normalizing algorithm and the normexp background correction algorithm with a correction factor of 50. A linear model was then applied to the normalized data using the lmFit function, and a Bayesian method for determining differential expression was applied using the eBayes function.

Quantitative RT-PCR assay.

Prior to reverse transcription (RT), control and experimental RNA (10 ng) isolated from acetamide-exposed cultures was treated with RNase-free DNase (Invitrogen) and subjected to 36 cycles of PCR to ensure that all DNA had been removed, as assessed by ethidium bromide-stained agarose gel analysis. After RT, cDNA corresponding to each transcript was subjected to 34 cycles of RT-PCR for quantification. The cycle threshold value (C_T) obtained for each gene of interest (GOI) was normalized with that of sigA, a housekeeping gene (HKG) with constant expression under different experimental conditions (20), in order to obtain the normalized C_T (nC_T; calculated as nC_T = GOI C_T − HKG C_T). Levels of sigA were constant by microarray analysis in both control and sigF-inducible strains throughout the time course. The change in C_T (ΔC_T) was calculated using the following formula: ΔC_T = C(nC_T) − S(nC_T), where C represents the control strain and S represents the sigF-inducible strain.

Recombinant M. tuberculosis SigF protein purification and in vitro transcription assay.

The M. tuberculosis sigF gene fused to a His tag was cloned into pET22b(+) vector (Novagen), which was used to infect Escherichia coli BL21 (DE3) cells. The fusion protein was expressed by inducing the bacterial cells with 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG) for 2 h at room temperature. These cells were harvested; resuspended in 10 mM Tris-Cl (pH 8.0), 0.5 M NaCL, 10 mM imidazole; broken by silicon bead beating; and centrifuged. The resulting supernatant was collected, and expression of the fusion protein was tested using anti-His tag antibody. The fusion protein was isolated by Ni-nitrilotriacetic acid metal affinity chromatography (Novagen), following the manufacturer's instructions, and its purity was tested by loading the eluted material on a sodium dodecyl sulfate-polyacrylamide gel electrophoresis gel. Single runoff in vitro transcription assays were performed as previously described (23). Briefly, RNA polymerase holoenzyme was formed by incubating the purified SigF protein (20 pmol) with E. coli core RNA polymerase (4 pmol) at 37°C for 30 min in transcription assay buffer (10 mM Tris, 50 mM KCl, 10 mM MgCl₂, 0.1 mM EDTA, 0.25 μg/μl bovine serum albumin). Then, 0.09 μg of template DNA was added and incubated with 0.25 mM of ATP, GTP, UTP, and [³²P]CTP. The transcript was analyzed by 5% denaturing polyacrylamide gel, which was developed on film (Kodak BioMax MR). Negative-control assays were performed as described above, except purified SigF was omitted from the reaction mixtures. DNA templates corresponding to usfX (Rv3287c, MT3386) and phoY1 (Rv3301c, MT3400) were PCR amplified using the primers listed in Table 2. PCR products were purified using a QIAGEN PCR product purification kit prior to use in the in vitro transcription assays.

TABLE 2.

Primers used in this study

Analysis and ORF	Gene name	Sense primer	Antisense primer
Quantitative RT-PCR
Rv3286c/MT3385	sigF	CTGCATCTGCGGCTAGGTA	TCGATGGACAAGGTGTGGTA
Rv3287c/MT3386	usfX	ACCTTCGAGGACCTGGATTT	CCTCCACCACAACTTCGTCT
Rv3301c/MT3400	phoY1	GCAAGACAAGTGCTGGTGAA	GAAGCGCTCAAAGAAACGAC
Rv2884/MT2952		GTGGACGGCGATTGGATATT	AGCTGGCTGGGTGAAGGA
Rv1823/MT1871		GGTAGCCAGCAAATCATCCT	GCCAAGATGGTGTAGGGACT
Rv1824/MT1872		AGGTCATCCAGCCGTATCTG	TTGGTCACCGACATAGACGA
Rv0202c/MT0212	mmpL11	GGTGTTTGTGTTGTCGATGC	TAGCGCATCAGGATGAACAG
Rv3136/MT3221	PPE51	GGATGACCTGCATAGGCATT	GAAGCGCTCAAAGAAACGAC
Rv1441c/MT1486	PE_PGRS26	CCAAATGGCCAGATTTCATC	CGAGCACCTGTTGTTCGAG
Rv0152c/MT0161	PE2	CTACCACCGCCGTCTATGTT	CCCTCTTCCACTGACGTGTT
Rv1195/MT1233	PE13	GCGATGTATCAGTCCGTGAG	GCATTGGCGACTTCAGTG
Rv1738/MT1780		GATCGACGAACACGAAGGAT	GGGTCGACACCTTCAACATT
Rv2919c/MT2987	glnB	ATCACTGCGATCGTGAAGC	GGAATCGTCAACAACGACCT
Rv1853/MT1901	ureD	ATGCTCACTGGGAGATCGAC	CGATGACGACCAAAATCCTT
Rv2703/MT2777	sigA	CTCGACGCTGAACCAGACCT	AGGTCTTCGTGGTCTTCGTC
In vitro transcription
Rv3287c/MT3386	usfX	GCTGGTATACCGA ACATCCC	AGCAACGTTCAGCTCGACGG
Rv3301c/MT3400	phoY1	ACATGTTGCGGG CCAGCGCA	GCCTGCGTTGCGCGTTTCAT

RESULTS

The role of sigF in M. tuberculosis cell cycle regulation and survival under stress conditions in vitro.

We directly tested the role of sigF as a molecular switch for entry into stationary phase by inappropriately expressing the gene during exponential-phase growth. An M. tuberculosis strain containing a chromosomally integrated, chemically inducible acetamidase promoter-sigF fusion gene (sigF-inducible strain) and a control strain containing the remaining construct genes but lacking sigF were each grown to OD₆₀₀s of 0.1, 0.2, and 0.4 and exposed to 0.2% acetamide. As shown in Fig. 1, the growth curves of the sigF-inducible and control strains are identical following acetamide exposure, regardless of starting OD₆₀₀.

FIG. 1.

Growth kinetics of the sigF-inducible (shaded squares) and control (open squares) strains in supplemented 7H9 broth at 37°C as a function of number of hours after exposure to 0.2% acetamide. Replicate experiments produced similar results.

We hypothesized that if sigF is involved in regulating M. tuberculosis replication, overexpression of sigF may result in increased resistance of M. tuberculosis to the rifamycins and to isoniazid. However, in both the sigF-inducible and the control strains, the MICs were equivalent to those for rifampin (0.25 μg/ml) and isoniazid (0.06 μg/ml), suggesting that SigF and its positively regulated gene products do not represent direct molecular targets of these drugs and that sigF expression does not retard bacillary growth sufficiently to induce phenotypic drug tolerance.

We directly tested the requirement for sigF expression in M. tuberculosis survival during nutrient starvation using a ΔsigF M. tuberculosis mutant (6) and compared the survival of this strain with that of the isogenic wild-type and complement strains. As demonstrated previously (3), we observed no loss of viability of the wild-type strain under prolonged nutrient starvation conditions. Similarly, the viability of the sigF-deficient mutant was equivalent to that of the wild-type and complement strains as late as 28 days after depletion of nutrients, suggesting that sigF is not essential for M. tuberculosis survival under nutrient starvation conditions (data not shown).

Previous studies showed no survival defect for a ΔsigF M. tuberculosis mutant in an in vitro model using human monocytes (6). However, this study did not address adequately the survival of the ΔsigF mutant in activated macrophages. Cytokine activation of macrophages has been shown to enhance phagosome maturation and macrophage-mediated mycobacterial killing (34). Prior to infection with M. tuberculosis, alveolar macrophage J774A.1 cells were activated with the addition of interferon-gamma and lipopolysaccharide, as previously described (33). As shown in Fig. 2, the sigF gene is not required for short-term survival of M. tuberculosis within activated murine macrophages.

FIG. 2.

Intramacrophage survival of the ΔsigF mutant (open triangle) compared to that of the isogenic wild-type (closed triangle) and complement (closed square) strains as a function of number of days after infection of activated alveolar macrophage J774A.1 cells.

The role of sigF in M. tuberculosis survival in an in vivo granuloma model of persistence.

Recently, we have developed an accelerated in vivo granuloma model of M. tuberculosis extracellular persistence (17). In this model, organisms are encapsulated within semidiffusible hollow fibers, which are implanted subcutaneously in mice. Progressive granulomatous inflammation surrounds the fibers, and within this microenvironment, the encapsulated organisms rapidly display a dormancy phenotype characterized by stationary-state CFU counts, decreased metabolic activity, and increased susceptibility to rifampin compared to isoniazid.

As shown in Fig. 3, sigF is not required for extracellular M. tuberculosis survival within artificial granulomas in vivo, as the numbers of intrafiber bacilli were virtually identical for the ΔsigF, wild-type, and complement strains 28 days after hollow-fiber implantation. Gross peri-fiber inflammation and normalized mouse spleen weights were equivalent in all three groups (data not shown).

FIG. 3.

Extracellular persistence of the ΔsigF mutant (open triangle) compared to that of the isogenic wild-type (closed triangle) and complement (closed square) strains within semidiffusible hollow fibers implanted subcutaneously in mice.

The role of sigF in regulating expression of mycobacterial cell wall components.

In order to gain insight into M. tuberculosis genes that encode cell wall components and are regulated by sigF, we studied the genome-wide transcriptional profile of M. tuberculosis strains at various time points following overexpression of sigF. The sigF-inducible and control strains were grown in vitro to early exponential phase, when sigF is normally not expressed in wild-type organisms (11). Cultures of each strain were exposed to 0.2% acetamide for 6 h, 12 h, or 24 h. These time points were chosen based on the observation that sigF upregulation was readily detectable by RT-PCR at 6 h but not at 2 and 4 h following acetamide exposure of the knock-in strain (data not shown).

We reasoned that genes which were differentially regulated at all three time points studied were more likely to be directly regulated by sigF than those which were regulated at two or fewer time points. Induction of sigF resulted in significant (P ≤ 0.05) differential regulation of 347 different genes at all time points studied, including 145 upregulated genes and 202 downregulated genes (see the supplemental material). As expected, sigF induction resulted in a significant increase in sigF and usfX transcripts at all time points studied. Consistent with a regulatory role for SigF in the biosynthesis and structure of the cell envelope of M. tuberculosis, we observed significant differential regulation of 73 genes encoding cell wall-associated proteins, representing over 20% of all differentially regulated genes. Furthermore, sigF overexpression resulted in significant differential regulation of 20 genes encoding proteins of the PE and PPE families, several members of which have been shown to localize to the extracellular surface of M. tuberculosis and to have antigenic potential (4, 9, 24).

Recognizing the role of M. tuberculosis SigF as an alternative RNA polymerase sigma factor that positively regulates transcription by directing core RNA polymerase to its cognate promoter, we reasoned that genes found to be downregulated following induction of sigF were likely to be indirectly regulated. Therefore, we focused our attention on sigF positively regulated genes (Table 1). Microarray analysis revealed significant (P ≤ 0.05) upregulation of 36 genes annotated as encoding cell wall-associated proteins. Among these genes were members of the mycobacterial membrane large (mmpL) family, including mmpL2, mmpL5, and mmpL11, which encode proteins involved in transmembrane transport, and which are virulence factors promoting in vivo survival of M. tuberculosis (5, 19). In addition, we detected significant upregulation of Rv2884, which was shown biochemically to have a sigF-dependent promoter (15). Induction of sigF also resulted in significantly increased expression of a putative operon containing Rv1823, Rv1824, Rv1825, and Rv1826 that has been shown to be under the control of the SigF-dependent Rv1823 promoter (15). Finally, the sigF-inducible strain showed increased expression of phoY1 (Rv3301c) and Rv3300c, which appear to be members of an operon.

TABLE 1.

M. tuberculosis genes significantly upregulated (P < 0.05) at 6, 12, and 24 hours following sigF induction according to functional category

Category	ORF	Gene name	ORF	Gene name
Regulatory proteins	Rv1267c/MT1305	embR	Rv0890c/MT0914
	Rv2884/MT2952	Rv0043c/MT0049
	Rv0158/MT0167	Rv3291c/MT3390
		Rv3164c/MT3253
Intermediary metabolism	Rv0296c/MT0310	atsG	Rv0186/MT0195	bglS
	Rv1916/MT1967	aceAb	Rv1652/MT1690	argC
	Rv0815c/MT0837	cysA2	Rv0118c/MT0126	oxcA
	Rv2496c/MT2571	pdhB	Rv0866/MT0889	moaE2
	Rv1310/MT1350	atpD	Rv2202c/MT2258	cbhK
	Rv0322/MT0337	udgA	Rv1511/MT1561	gmdA
	Rv3290c/MT3389	lat	Rv1309/MT1349	atpG
	Rv2200c/MT2256	ctaC	Rv2178c/MT2234	aroG
		Rv0542c/MT0567	menE	Rv1470/MT1516	trxA
		Rv1296/MT1335	thrB
Cell wall and cell processes	Rv1469/MT1515	ctpD	Rv0676c/MT0705	mmpL5
	Rv3476c/MT3580	kgtP	Rv1315/MT1355	murA
	Rv0419/MT0433	lpqM	Rv0924c/MT0951	nramp
	Rv1270c/MT1308	lprA	Rv1280c/MT1317	oppA
	Rv1252c/MT1291	lprE	Rv1281c/MT1318	oppD
	Rv0173/MT0182	lprK	Rv3301c/MT3400	phoY1
	Rv0202c/MT0212	mmpL11	Rv1342c/MT1383	pks14
	Rv0507/MT0528	mmpL2	Rv1302/MT1341	rfe
		Rv2400c/MT2471	subI	Rv1824/MT1872
Information pathways	Rv3287c/MT3386	usfX	Rv1299/MT1338	prfA
	Rv0001/MT0001	dnaA	Rv0652/MT0681	rplL
	Rv1298/MT1337	rpmE	Rv1080c/MT1111	greA
	Rv0003/MT0003	recF	Rv3286c/MT3385	sigF
	Rv0182c/MT0191	sigG	Rv1007c/MT1036	metS
	Rv1536/MT1587	ileS
Lipid metabolism	Rv2930/MT2999	fadD26	Rv1193/MT1230	fadD36
	Rv1467c/MT1514	fadE15	Rv0437c/MT0453	psd
	Rv1185c/MT1222	fadD21	Rv1925/MT1976	fadD31
Conserved hypotheticals	Rv1825/MT1873		Rv3300c/MT3399
PE/PPE	Rv3136/MT3221		Rv0152c/MT0161
	Rv1441c/MT1486	Rv1646/MT1684
	Rv1195/MT1233	Rv1172c/MT1209
	Rv1089/MT1120

Quantitative RT-PCR analysis was used to confirm significant regulation for a subset of upregulated genes in the sigF-inducible strain compared to what was found for the control strain. The primers used in RT of each gene are listed in Table 2. Gene expression levels in each strain were normalized to those of sigA, whose expression was not altered in the mutant strain at any time point studied (data not shown). As expected, the sigF-inducible strain showed significantly increased expression of sigF and usfX at 6, 12, and 24 h after acetamide induction (Table 3). The most highly upregulated gene at all time points after acetamide induction in the sigF-inducible strain was phoY1 (Rv3301c). A subset of genes found to be significantly downregulated by microarray analysis in the sigF-inducible strain after acetamide exposure were confirmed to be downregulated by RT-PCR (Table 3).

TABLE 3.

Quantitative RT-PCR of a subset of genes significantly regulated after sigF induction

ORF	Gene name	Normalized ΔCT at indicated time point
		6 h	12 h	24 h
Rv3286c/MT3385	sigF	6.64	7.37	5.27
Rv3287c/MT3386	usfX	2.07	3.74	4.74
Rv3301c/MT3400	phoY1	7.71	8.34	9.57
Rv2884/MT2952		2.71	2.90	1.41
Rv1823/MT1871		2.60	4.40	5.90
Rv1824/MT1872		3.10	3.00	1.21
Rv0202c/MT0212	mmpL11	1.05	3.91	0.80
Rv3136/MT3221	PPE51	4.91	4.85	1.16
Rv1441c/MT1486	PE_PGRS26	1.05	2.11	4.05
Rv0152c/MT0161	PE2	−0.95	1.16	2.90
Rv1195/MT1233	PE13	2.66	1.71	2.41
Rv1738/MT1780		−3.31	−1.30	−4.00
Rv2919c/MT2987	glnB	−1.30	−1.10	−2.20
Rv1853/MT1901	ureD	−2.90	−1.30	−1.71

In vitro transcription analysis.

Although many genes were found to be upregulated at various time points after sigF induction, we reasoned that some of these genes could be directly regulated by SigF, while others may be indirectly regulated. Using the Search Pattern function in Tuberculist (http://genolist.pasteur.fr/TubercuList/), we screened the M. tuberculosis genome for genes containing the consensus recognition sequence NGTTTN-N₁₅-GGGTAT, which was shown by biochemical approaches to constitute the SigF-dependent promoter sequence of usfX (Rv3287c), the gene located upstream of sigF and encoding the anti-sigma factor UsfX or RsbW (2). The search was limited to intergenic regions up to 250 bp from the putative translation start site, and search criteria allowed no more than one mismatch total in either of the −35 NGTTTN and −10 GGGTAT sequences, with a spacer size ranging from 14 to 18 nucleotides (nt). A list of 51 such genes was identified and then compared to the list of genes found to be significantly regulated at least at two consecutive time points by microarray analysis following sigF induction. This comparison yielded a group of nine genes, including Rv1823, which previously was shown to have a SigF-dependent promoter (15), as well as mmpL11 and phoY1. Because expression of phoY1 was markedly increased following sigF induction, in vitro transcriptional analysis was used to determine if SigF directly regulates phoY1. Single runoff in vitro transcription assays were performed as described in Materials and Methods, using E. coli core RNA polymerase in the presence or absence of purified SigF. Similar assays were performed using a usfX template containing the upstream sigF-dependent promoter sequence as a control. The phoY and usfX templates were generated using primers listed in Table 2. His-tagged M. tuberculosis CDC1551 SigF was used in these assays, since C-terminal fusions previously have been shown not to interfere with sigma factor activity (2). Although some usfX transcript was detected with core RNA polymerase alone, transcript levels were greatly enhanced by the addition of SigF (Fig. 4B). Based on the location of the usfX promoter consensus sequence in relation to the phoY1 gene, the expected transcript size is >98 nt but <185 nt (Fig. 4A). As shown in Fig. 4B, in vitro transcription of the phoY1 template resulting in a transcript of the expected size was observed only in the presence of SigF, suggesting that SigF directly and specifically regulates phoY1. Therefore, the SigF-dependent promoter sequence of phoY1 appears to be GGATTG-N₁₆-GGGTAT.

FIG. 4.

In vitro transcription analysis of the usfX (control) and phoY1 templates by use of reconstituted SigF holoenzyme containing E. coli core RNA polymerase. (A) Position of usfX promoter consensus sequence relative to the known translation start site of gene usfX (3) and relative to the putative translation start site of gene phoY1. (B) Lane 1, molecular mass markers; lane 2, E. coli core RNA polymerase and usfX template; lane 3, E. coli core RNA polymerase, usfX template, and purified SigF; lane 4, E. coli core RNA polymerase and phoY1 template; lane 5, E. coli core RNA polymerase, phoY1 template, and purified SigF. Arrows indicate the positions of the transcripts.

We identified a total of four genes that have been shown to be transcribed by SigF in vitro, including usfX (2), Rv1823, Rv2884 (15), and phoY1 (this study), which were also found to be more highly expressed in vivo following overexpression of sigF (this study). Based on knowledge of the usfX promoter sequence (2), we examined the 5′ untranslated regions of the other three genes for sequence similarity in order to discern a sigF promoter consensus sequence. Using WebLogo software (http://weblogo.berkeley.edu/logo.cgi), we determined that two nonconsecutive nucleotides are the most highly conserved within the −35 region, while the sequence GGGTA is highly conserved in the −10 region (Fig. 5). The sigF promoter consensus sequence appears to be NGNTTG-N_14-18-GGGTAT.

FIG. 5.

A group of four genes that have been shown to be SigF dependent both in vitro and in vivo, yielding a sigF promoter consensus sequence.

DISCUSSION

We examined the possibility that sigF expression may mediate tuberculosis immunopathogenesis by regulating expression of cell wall-associated proteins. Using an M. tuberculosis strain containing a chromosomally integrated, chemically inducible acetamidase promoter-sigF fusion gene, we studied the global gene expression profile of M. tuberculosis following sigF overexpression. The most highly upregulated gene by quantitative RT-PCR at all time points following sigF induction was Rv3301c (phoY1). Since the 5′ region of this gene contains a sequence highly similar to that described for the SigF-dependent usfX promoter (2), we hypothesized that sigF may directly regulate phoY1. In vitro transcriptional analysis confirmed that transcription of phoY1 is indeed SigF dependent, with a proposed promoter sequence of GGATTG-N₁₆-GGGTAT, containing a mismatch at the third and sixth base pair positions in the −35 region as well as an extra base pair in the spacer region relative to the SigF-dependent usfX promoter sequence, GGTTTC-N₁₅-GGGTAT (2). Consistent with these findings, a recent study using chromatin immunoprecipitation assays and a single-array model predicted a SigF binding site in the intergenic region between phoY1 (Rv3301c) and glpD2 (Rv3302c) in a Mycobacterium bovis BCG-Russia strain (28).

M. tuberculosis phoY1 appears to encode a 221-amino-acid transcriptional regulatory protein homologous to members of the PhoU family of proteins in other bacteria (32). In E. coli, the phoU gene is the last cistron in the pstSCAB operon, which encodes an ATP binding cassette transporter involved in the high-affinity phosphate-specific transport system induced by P_i starvation (27). The pstSCAB operon is part of the phosphate (PHO) regulon, which is coregulated by the PhoR-PhoB two-component regulatory system. Although phoU is located in the pstSCAB operon, its encoded protein appears to act as a repressor of the PHO regulon and may be involved in intracellular P_i metabolism, likely related to ATP synthesis (32). In M. tuberculosis, the phoY1 gene is likely the first member of an operon containing Rv3300c, which encodes a protein of an unknown function; Rv3299c (atsB), which encodes a probable arylsulfatase; and Rv3298c (lpqC), which encodes a possible esterase lipoprotein. Consistent with the hypothesis that these genes are cotranscribed, we detected significantly increased expression of Rv3300c at all time points following sigF overexpression. Unlike that in E. coli, the homologous two-component signal transduction system in M. tuberculosis, PhoPR, does not seem to regulate phoY1 (35). The regulation of phoY1 and the function of its product in M. tuberculosis remain to be elucidated.

Consistent with our hypothesis, over 20% of all genes significantly (P ≤ 0.05) regulated by sigF in our studies encode putative cell wall-associated proteins. Induction of sigF resulted in significant upregulation of several genes encoding members of the MmpL family of transmembrane proteins, including mmpL2, mmpL5, and mmpL11. Of note, the mmpL11 gene contains a sequence in the 5′ upstream region that is nearly identical to that of the sigF-dependent usfX promoter (2), suggesting that sigF may directly regulate mmpL11. Although these mmpL genes are not essential for in vitro growth, in vivo growth of an mmpL2 mutant was observed to be compromised when signature-tagged transposon mutants of M. tuberculosis Mt103 were screened (5). Similarly, M. tuberculosis transposon mutants with disruptions in mmpL5 and mmpL11 were shown to be significantly attenuated for growth in mouse lungs (19). Overexpression of sigF also resulted in differential regulation of many genes encoding PE and PPE proteins, several members of which have been localized to the cell membrane (4, 9, 24) and shown to influence mycobacterial-host cell interactions (4, 24, 31). Differential regulation of PE and PPE proteins has been described previously for other M. tuberculosis alternative RNA polymerase sigma factors, including sigC (33), sigD (25), sigL (14), and sigM (26).

This study provides the first evidence that the Rv2884 gene and the Rv1823 operon, which have sigF-dependent promoters (15), are, in fact, positively regulated by SigF in vivo. The Rv2884 gene encodes a probable transcriptional regulator with significant sequence similarity to many response regulators of bacterial two-component signal transduction systems. Unlike other genes encoding response regulators, which are transcriptionally coupled to genes encoding a corresponding sensor kinase, Rv2884 appears to be independently transcribed and not associated with a histidine protein kinase gene. The Rv1823 gene appears to be the first gene in an operon also containing Rv1824, Rv1825, and Rv1826, whose gene products serve unknown functions.

Our results are largely discordant with those of Geiman et al. (12), in which microarray analysis of a sigF-deletion mutant of M. tuberculosis CDC1551 and its isogenic wild-type strain revealed a significantly different set of genes downregulated in the mutant. Of the 14 genes reported to be likely directly regulated by SigF in that study, based on usfX promoter sequence similarity in their upstream promoter regions, our microarray results revealed upregulation of only two genes, Rv0893c and Rv0012. However, 3 of the 14 genes from the sigF deletion mutant study could not be verified biochemically using an E. coli two-plasmid system (15), suggesting that these genes are not directly regulated by SigF, or that their regulation requires an additional transcriptional activator absent in the in vitro transcription system. Conversely, Geiman et al. did not find downregulation of the Rv1823 operon, Rv2884, and phoY1 (Rv3301c) in the sigF deletion mutant, despite significant upregulation of these genes in the sigF-inducible strain and demonstration of the sigF dependence of their promoters in vitro by our group and others (15). The discrepancy between our results and those of Geiman et al. may be potentially explained by differences in methodology. When sigF is overexpressed during exponential growth, transcriptional activators present only in stationary phase may be absent, potentially leading to our missing sigF-regulated genes. On the other hand, neither sigF nor usfX was significantly downregulated during late stationary phase in the sigF-deletion mutant (12), consistent with other data suggesting that, unlike in M. bovis BCG (11), sigF is not upregulated during late stationary phase in M. tuberculosis (20). Although recent in vitro transcription experiments identified a sigF-dependent promoter in the upstream region of the sigB gene (8), we did not observe upregulation of sigB following overexpression of sigF by microarray analysis or by quantitative RT-PCR (data not shown), and the upstream region of the sigB gene was not identified as a SigF binding site by chromatin immunoprecipitation assays (28). The majority of genes found to be positively regulated in this study are likely to be indirectly regulated by SigF, since the usfX promoter sequence could not be identified in their 5′ untranslated regions. We believe that the genes of the sigF regulon described here are likely valid, whether directly or indirectly regulated by sigF, since quantitatively confirmed overexpression of sigF was necessary and sufficient for their differential regulation. The conditional overexpression/“knock-in” approach provides information complementary to the earlier deletion mutant data and helps further elucidate the M. tuberculosis sigF regulon.

M. bovis BCG harboring the M. tuberculosis sigF gene showed increased expression of sigF upon exposure to several antituberculous agents, anaerobic conditions, cold shock, oxidative stress, and nutrient depletion and after entry into stationary phase (22). However, in M. tuberculosis, increased expression of sigF was detected only upon nutrient depletion of M. tuberculosis cultures (3, 20) and during infection of cultured human macrophages (13), suggesting differences in the regulation of this gene in M. tuberculosis and M. bovis BCG. We show here that in fact sigF is not essential for M. tuberculosis survival under nutrient starvation conditions or within activated macrophages. These conflicting results potentially may be explained by differences in the experimental systems used or by the observation that gene function may not be ascertained on the basis of transcriptional data alone. We also demonstrated that sigF is not required for extracellular persistence of M. tuberculosis within mouse granuloma-like lesions, conditions which may simulate those experienced by at least some dormant bacillary populations during in vivo infection (17). In previous studies, a sigF-deficient M. tuberculosis mutant was shown to grow to a higher density in stationary-phase cultures than the isogenic wild-type strain (6), suggesting a possible role for sigF in cell cycle regulation under stress conditions. However, we found that overexpression of sigF during early-exponential-phase growth, when sigF mRNA is normally not detectable (11), did not result in detectable growth arrest. One potential explanation for these discordant findings is that although sigF is required for entry into stationary phase, expression of this gene alone may not be sufficient for bacillary growth retardation. Consistent with this hypothesis, we did not detect reduced susceptibility of the sigF-inducible strain to either rifampin or isoniazid, as was observed for nonreplicating bacilli following deprivation of nutrients (3) or exposure to progressive hypoxia in vitro (36).

In conclusion, our data challenge the classification of M. tuberculosis SigF as a stress response alternative RNA polymerase sigma factor involved in stress adaptation and cell cycle regulation and are consistent with a regulatory role for SigF in the structure and function of the mycobacterial cell wall. M. tuberculosis SigF appears to directly regulate Rv1823, Rv2884, and phoY1. Based on genes which have been shown to be transcribed by SigF in vitro and positively regulated by SigF in vivo, the sigF promoter consensus sequence appears to be NGNTTG-N_14-18-GGGTAT.