Mutations in Mycobacterium tuberculosis Rv0444c, the gene encoding anti-SigK, explain high level expression of MPB70 and MPB83 in Mycobacterium bovis

Differential expression of MPT70 and MPT83 expression across MTC members

Differential in vitro expression of MPB70 has been described in M. tuberculosis and M. bovis as well as other traditional MTC members (Liebana et al., 1996; Cousins et al., 2003). Using the revised MTC phylogeny, we documented MPT70 and MPT83 expression across a panel of MTC members to specifically determine which organisms manifest altered expression. By quantitative reverse transcription-PCR (qRT-PCR) coupled with a molecular beacon to analyse mRNA levels of mpt70, our results demonstrate that this gene is differentially expressed across MTC members in vitro. Specifically, M. tuberculosis H37Rv and H37Ra, M. africanum and M. microti had low expression whereas M. caprae, M. bovis, BCG Russia and the Oryx bacillus were high expressors of mpt70 (Fig. 1A). To show that differential expression of mpt70 (Fig. 1A) also affects protein levels, we performed a Western blot analysis on culture filtrates using a rabbit polyclonal antibody against MPB70. As shown in Fig. 1B, the results correlate perfectly with the qRT-PCR data. We previously noted that the expression of mpb70 by BCG strains is coupled to mpb83 expression in vitro (Charlet et al., 2005). To determine whether this is also the case across natural MTC members, we tested MPT83 expression by Western blot using a polyclonal rabbit antibody; the results parallel those of MPT70 (Fig. 1C). A polyclonal antibody raised against MPB64, a secreted antigenic protein whose expression is independent of SigK, was used as a loading control. As evidenced in Fig. 1D, all strains produced this protein. In addition, for each member of the MTC, we also compared the in vitro transcriptome against that of M. tuberculosis H37Rv, to screen for important transcriptional differences across organisms (data provided in Table S1). Microarray-based interrogation revealed that upregulation of mpb70, mpb83 and neighbouring genes (dipZ, Rv2876) by M. caprae, M. bovis, BCG Russia and the Oryx bacillus figured prominently among their principal differences in expression compared with M. tuberculosis. Based on these results, MTC members analysed could be divided into two groups: M. tuberculosis H37Rv and H37Ra, M. africanum and M. microti had low in vitro production of MPT70 and MPT83, while high production was observed for M. bovis, BCG Russia (early BCG), M. caprae and the Oryx bacillus.

Expression of mpt70 and mpt83 in M. tuberculosis is induced inside THP-1 macrophages

The low levels of MPT70 and MPT83 in M. tuberculosis have been documented in vitro (Hewinson et al., 1996). However, during intracellular infection, the organism modulates its expression, and mpt70/mpt83 figure among the genes reported to be strongly induced during adaptation to the phagosomal environment (Schnappinger et al., 2003). In this report, confirmation of microarray results by qRT-PCR was only performed for mpt83. To confirm that mpt70 expression is also inducible in M. tuberculosis, THP-1 macrophages were infected with M. tuberculosis H37Ra and virulent H37Rv, and bacterial mRNA was isolated to study mpt70 expression by qRT-PCR. As shown in Fig. 2, mpt70 levels are significantly higher in both H37Ra and H37Rv after 24 h inside THP-1 cells pretreated with interferon-γ (IFN-γ) as compared with non-stimulated cells, indicating that mpt70 is induced in stimulated macrophages (Fig. 2). Similar results were obtained for mpb83 (data not shown). These data imply that the wild-type expression phenotype, as exemplified by M. tuberculosis, is low in vitro and inducible during infection.

Genetic basis behind MPT70/MPT83 differential expression

To elucidate the genetic reason for the variable expression of MPT70 and MPT83 across MTC members, we first focused our attention on M. bovis (high producer) and M. tuberculosis H37Rv (low producer), whose whole genome sequences are publicly available (http://genolist.pasteur.fr/TubercuList, http://genolist.pasteur.fr/BoviList). The sequences of M. bovis mpb70 and mpb83 are identical to their M. tuberculosis counterparts mpt70 and mpt83. As both organisms also have the identical sequence of sigK, the gene encoding the positive regulator of the two antigens (Charlet et al., 2005), we hypothesized that M. bovis may lack a functional anti-SigK. By analysing the sequence of the genes in the vicinity of sigK in M. tuberculosis H37Rv versus M. bovis 2122, we found that only one gene, Rv0444c, located immediately downstream of sigK, varies between the two organisms. In M. bovis, a high producer of MPB70 and MPB83, Rv0444c harbours two non-synonymous single nucleotide polymorphisms (SNPs): C320T and C551T, resulting in an amino acid change of glycine to aspartic acid and glycine to glutamic acid respectively. We then sequenced Rv0444c across different MTC members and found that members with low-level production of MPT70/MPT83 have the wild-type gene (H37Rv sequence) whereas members with high-level expression of these antigens have a mutated Rv0444c (Table 1). Of note, the Oryx bacillus, which produces MPT70/MPT83 at high levels in vitro (Fig. 1), has a different Rv0444c mutation than M. bovis; the stop codon has been substituted with a serine, resulting in a read-through mutation that may affect protein translation and/or stability. The observation that the gene immediately neighbouring sigK had suffered two independent mutations that were associated with the same expression phenotype strongly implicated Rv0444c as the gene encoding anti-SigK.

Complementation of BCG Russia with wild-type Rv0444c results in a significant decrease in mpb70/83 expression

To functionally prove that Rv0444c gene product is the anti-SigK, we complemented BCG Russia, which has a mutated Rv0444c and is a high producer of MPB70/83, with the wild-type Rv0444c amplified from M. tuberculosis H37Rv, expressed under the mycobacterial optimal promoter (MOP) which consists of the BCGhsp70 promoter and the Escherichia coli tac promoter (George et al., 1995). To control for gene copy, we separately complemented BCG Russia with an overexpressed mutant Rv0444c (amplified from BCG Russia itself). Overexpression of wild-type and mutant Rv0444c was verified by qRT-PCR with SYBR green (data not shown). Next, using qRT-PCR, coupled with molecular beacons, we observed that addition of wild-type Rv0444c resulted in a pronounced decrease in mpb70/83 expression as compared with BCG Russia complemented with the empty vector, but complementation with the mutant Rv0444c had no effect in mpb70/83 expression (Fig. 3). This indicated that the decrease in expression we observed was specifically due to the introduction of wild-type Rv0444c.

Lack of Rv0444c expression results in overexpression of mpt70 and mpt83

As a second means of proving that Rv0444c encodes the anti-SigK, we aimed to determine whether the lack of a functional Rv0444c in M. tuberculosis would result in an overexpression of mpt70/mpt83. To this end, we constructed a M. tuberculosis H37Rv mutant that expresses sigK but not Rv0444c in the following manner. A ΔsigK mutant was engineered by allelic exchange, replacing sigK with a kanamycin cassette using the pKO knock-out plasmid as described in Lewis et al. (2003). Deletion of sigK and its replacement by the kanamycin gene was confirmed by Southern blot analysis (Fig. 4A and B) and sequence analysis of the mutant indicated that the kanamycin cassette inserted between nucleotide positions 533 815 and 534 397 of the M. tuberculosis H37Rv genome resulting in the complete removal of sigK (Fig. 4C). As sigK and Rv0444c are in close proximity to each other (only 43 bp between them) and predicted to be co-transcribed, we analysed the expression of both sigK and Rv0444c in the deletion mutant by qRT-PCR. As expected, neither gene was expressed in the mutant, and hence the sigK deletion created a polar mutation affecting Rv0444c expression (Fig. 5A). Next, we separately complemented this strain with sigK alone (single gene complement) and with both sigK and Rv0444c gene (dyad complement). As shown in Fig. 5A, the single gene complement only expressed sigK whereas the dyad complement expressed both sigK and Rv0444c. These data were confirmed by a Western blot analysis of the Rv0444c gene product (Fig. 5B). We then analysed the expression of mpt70 and mpt83 in these clones by qRT-PCR. As shown in Fig. 6A, the single gene complement, which expressed sigK but not Rv0444c, expressed high levels of mpt70/mpt83 in vitro. The expression of mpt70/mpt83 in the ΔsigK::dyad strain was at low levels, similar to H37Rv (Fig. 6A). In perfect agreement with the qRT-PCR data, Western blot analysis of the same strains demonstrated that the sigK complement (ΔsigK::sigK) became a high producer of MPT70 and MPT83 proteins, whereas the dyad complement expressed MPT70 and MPT83 at levels similar to H37Rv (Fig. 6B and C). These results once again showed that Rv0444c encodes a protein with anti-SigK activity.

Co-transcription of sigK and Rv0444c and physical interaction of their gene products

For other anti-sigma factors, the genes encoding the anti-sigma factor and the sigma factor are co-transcribed, and there is physical interaction between the anti-sigma factors N-terminus and its cognate sigma factor (Yoshimura et al., 2004; Hahn et al., 2005). To determine whether these properties could be observed for Rv0444c and sigK, first RT-PCR was performed using sigK and Rv0444c primers, providing the expected product (data not shown). Next, analysis of the membrane topology of Rv0444c gene product in silico (http://bioinf.cs.ucl.ac.uk/psipred/) predicted that the Rv0444c-encoded protein would be a transmembrane protein with an intracellular N-terminal and an extracellular C-terminal, consistent with other anti-sigma factors of M. tuberculosis (Hahn et al., 2005). We therefore tested for direct interaction between SigK and the N-terminal part of Rv0444c gene product using the yeast two-hybrid system. As shown in Fig. 7A, the yeast β-galactosidase reporter was activated upon coexpression of SigK and the N-terminal region of Rv0444c, indicating that the proteins can physically interact. Additionally, this transformant grew on media lacking uracil (Fig. 7B), indicating activation of the URA3 reporter gene. Consistent with induction of URA3, the same transformant was unable to grow in the presence of 5-fluoroorotic acid (5FOA), a compound converted to the toxic 5-fluorouracil by uracil (Fig. 7C). In sum, our data indicate that sigK and Rv0444c are co-transcribed, that the N-terminal (cytoplasmic) part of the Rv0444c gene product physically interacts with SigK, and that this interaction results in decreased expression of the SigK-regulated genes (mpt70/mpb83). We therefore refer to the protein encoded by Rv0444c as RskA (regulator of SigK).

Defining the transcriptional impact of RskA by microarray

To better define the regulon of the SigK-RskA system, we performed two different whole genome microarray analyses. In the first set of microarray experiments, we compared the transcriptional profile of Russia::pMH416 (complement with empty vector) with Russia::Rv0444c (complement with wild-type Rv0444c under the mop promoter). Genes most significantly downregulated upon overexpression of Rv0444c were Rv2873 (mpb83), Rv2875 (mpb70) and Rv2874 (dipZ) (data provided in Table S2). In a second set of microarray experiments, we analysed the transcriptional profile of the ΔsigK mutant against ΔsigK::sigK (single gene complement). In these microarrays, the addition of sigK significantly upregulated expression of Rv0446c and Rv0449c (both part of the sigK locus), and Rv2873 (mpb83), Rv2875 (mpb70) and Rv2876 (Fig. 8). Of note, the only discordance between these two sets of transcriptome analyses was the apparent downregulation of lipP in the sigK complement arrays. As lipP is next to the attB site (where pMV306 integrates), and complementation of BCG Russia employed an extrachromosomal plasmid, we postulate that this result is an artefact of the complementation process, and not that lipP is a SigK-dependent gene. Taken together, these results revealed that the SigK-RskA regulon is small and restricted to two regions: the sigK and the mpt70/mpt83 regions. These results are in agreement with our previous study in which the addition of sigK in a BCG background affected only the expression of the sigK and mpb70/mpb83 regions (Charlet et al., 2005).

Bacterial strains and culture conditions

Mycobacterium africanum 60914 (previously characterized in Mostowy et al., 2004b), the Oryx bacillus 51145, M. caprae 60312 (previously characterized in Mostowy et al., 2005) and M. bovis 62389 are generous gifts from Louise Thibert. M. microti and M. tuberculosis H37Rv are gifts from Dr David Sherman. Mycobacteria and BCG strains were grown at 37°C in Middlebrook 7H9 medium (Difco Laboratories, Detroit, MI) containing 0.05% Tween 80 (Sigma-Aldrich, St Louis, MO) and 10% albumin-dextrose-catalase (Becton Dickinson and Co., Sparks, MD) supplement on a rotating platform (Wheaton). For solid media, Middlebrook 7H10 (without Tween) supplemented with OADC was used. E. coli DH5-α used for cloning purposes was cultured at 37°C in Luria broth (Difco). Antibiotics were added as needed at the following concentrations: kanamycin: 50 μg ml⁻¹ for E. coli and 25 μg ml⁻¹ for mycobacteria; hygromycin: 100 μg ml⁻¹ for E. coli and 50 μg ml⁻¹ for mycobacteria; apramycin: 30 μg ml⁻¹ for both E. coli and mycobacteria.

DNA extraction and sequence analysis

DNA from MTC members was extracted using a protocol based upon lysozyme and proteinase K (Van Soolingen et al., 1991), where samples were used as template for targeted sequence analysis. Primers were designed to amplify Rv0444c and flanking DNA (Rv0444cL: 5′-GGCGCTCATGACTGAACATA-3′ and Rv0444cR: 5′-CTAGTGGTACCCGGCGTGTT-3′) where amplicons were subject to direct dideoxy terminal sequencing at the McGill University and Genome Quebec Innovation Center (http://genomequebec.mcgill.ca/). Sequenced products were analysed by alignment searches against published genome sequences, namely M. tuberculosis H37Rv using Tuberculist (http://genolist.pasteur.fr/TubercuList/), M. tuberculosis 210 and CDC1551 using the sequences provided at NCBI (http://www.ncbi.nlm.nih.gov/sutils/genom_table.cgi), M. bovis AF2122/97 using Bovilist ( http://genolist.pasteur.fr/BoviList/), and the assembly sequence of BCG Pasteur (http://www.sanger.ac.uk/cgi-bin/blast/submitblast/m_bovis).

Preparation and infection of THP-1 cells

To prepare the human macrophage-derived THP-1 cell line for infection as previously described (Lewis et al., 2003), cells were pelleted by centrifugation, resuspended in RPMI 1640 medium plus 10% fetal calf serum (FCS) with 100 nM phorbol 12-myristate 13-acetate (PMA; Sigma), and delivered (1–10 × 10⁷ cells in 40 ml) into 175 cm² tissue flasks (Falcon). Stimulated THP-1 cells were additionally provided with human IFN-γ from BioDesign International at this time (100 U ml⁻¹). Cells were incubated at 37°C with 5% CO₂ for 24 h. PMA (with or without IFN-γ)-containing medium then was removed from the flasks, cells were washed with warm RPMI 1640 medium, given fresh RPMI 1640 medium plus FCS, and re-incubated for 24 h during the infection process. For each infection, bacteria stocks (M. tuberculosis H37Rv and H37Ra) were grown to mid-log phase in 7H9 media, diluted in warm RPMI 1640 medium plus 10% FCS (THP-1 complete media), and added to each flask containing 1–10 × 10⁷ THP-1 cells at a multiplicity of infection of 5–10. Cells were incubated at 37°C in 5% CO₂. After 24 h, extracellular bacteria were removed, and RNA was isolated from internalized bacteria (see below).

Ex vivo RNA isolation

Based on published RNA extraction protocols (Mangan et al., 2002; Schnappinger et al., 2003), RNA was isolated from bacteria within infected macrophages, yielding approximately 1 μg from 10⁷ host cells. Specifically, 24 h after infection, macrophages were lysed in low detergent concentrations in presence of guanidinium isothiocyanate, followed by phagosome isolation and bacterial harvest by ultracentrifugation. Bacterial RNA was subsequently extracted and analysed by qRT-PCR as described below.

Complementation assays

To complement BCG Russia, a 1068 bp DNA fragment containing the Rv0444c gene plus 292 bp upstream was PCR amplified from M. tuberculosis H37Rv or from BCG Russia using the following primers: Rv0444F-XbaI: 5′-ATAAAATCTAGAGGTGCGGCCAACGTCGATC-3′; Rv0444R-HindIII: 5′-ATAAAAGCTTCCGGCGTGTTCGTCGCGATGC-3′. The PCR amplicon was first cloned into pCR 4-TOPO cloning vector (Invitrogen, Carlsbad, CA). The fragment was then removed from the TOPO vector with the restriction enzymes EcoRI (found in the TOPO vector) and HindIII and cloned into pMH416 digested with the same enzymes. pMH416 (a kind gift of D.R. Sherman and M.J. Hickey) is a derivative of pMH29 (George et al., 1995) that has the apramycin-resistance gene (Cangelosi et al., 2006). The resulting plasmid pMH416::Rv0444c as well as the pMH416 empty vector were electroporated into BCG Russia. Complementation was PCR-confirmed and amplicons were sequence-confirmed for all transformants.

Construction of sigK mutant and complemented strains

Mycobacterium tuberculosis H37Rv sigK mutant was created by allelic exchange in a two-step selection process, replacing sigK with a kanamycin cassette. For this purpose, we used the plasmid pKO, which contains the sacB (conferring sucrose sensitivity) and a hygromycin cassette in its backbone as described in Lewis et al. (2003). pKO multiple cloning sites are also flanked by a kanamycin-resistance gene. DNA fragments of 1464 bp and 1666 bp spanning the regions proximal and distal of sigK, respectively, were PCR amplified from H37Rv genomic DNA using the following primers: sigKproxF: 5′-ATAAGCATGCAGCGATGCGTTGGGAGAG-3′; sigKproxR: 5′-ATAAAAGCTTGGATGCAGCTGAGGGTCTG-3′ and sigKdistF: 5′-ATAAGGTACCTAGCCGTGCACTATGACCTG-3′; sigKdistR: 5′-ATAAGGTACCCGATGGGTAGCCTATCGCCA-3′. The PCR amplicons were independently cloned into pCR 4-TOPO cloning vector (Invitrogen, Carlsbad, CA). The distal fragment was digested from TOPO with KpnI and ligated to the pKO vector digested with the same enzyme, making the plasmid pKO-distal. PCR and restrictions digests were used to confirm the presence and the correct orientation of the insert. The proximal region was digested from TOPO with SpHI and HindIII and cloned into pKO-distal plasmid digested with the same enzymes. The final construct pKOSigK contains the regions proximal and distal to sigK on each side of the kanamycin-resistance gene. The construct was confirmed by amplifying and sequencing the two regions (proximal and distal). We then transformed the pKOSigK plasmid into M. tuberculosis H37Rv using the method previously described (Belley et al., 2004). Hygromycin-resistant colonies obtained from the first selection were selected and analysed for site-specific integration using the Expand Long Template PCR System (Roche). The clones in which the plasmids integrated at the right place were plated on 7H10 plus kanamycin and 2% sucrose. Kanamycin- and sucrose-resistant colonies would have lost the plasmid backbone containing sacB (Pelicic et al., 1996). These colonies were tested on hygromycin and kanamycin independently. Those colonies that grew on kanamycin but not on hygromycin were selected for further analysis. sigK deletion was first confirmed by PCR with primers flanking the sigK gene and sequencing, and also confirmed by Southern blot analysis.

To complement the ΔsigK mutant with sigK only, we used the plasmid pH37Rv, which is a derivative of pMV306 containing the sigK region (including the complete gene and 288 bp upstream) as previously described (Charlet et al., 2005). To complement the ΔsigK with the dyad (sigK and Rv0444c), we cloned a 1703 bp PCR fragment containing sigK, Rv0444c plus 288 bp upstream of sigK into pMV306, using the following primers: SigKF-NotI: 5′-ATAAGCGGCCGCACGCGTCACCCCAACTACT-3′; Rv0444R-EcoRV: 5′-ATAAGATATCCCGGCGTGTTCGTCGCGATGC-3′. NotI and EcoRV sites were added to facilitate the cloning of this fragment into pMV306. Because pMV306 originally has a kanamycin-resistance gene and the ΔsigK mutant generated was kanamycin-resistant, we genetically replaced the kanamycin cassette with a hygromycin-resistance marker. The presence of the correct insert was confirmed by restriction digest and PCR followed by sequencing. The resulting plasmids pMV306-hyg::sigK and pMV306-hyg::dyad were electroporated into the ΔsigK mutant. Complementation was PCR-confirmed by amplifying the inserts with primers specific for the regions of pMV306 (Charlet et al., 2005) flanking the insert and amplicons were sequence-confirmed for all transformants.

RNA extraction and RT-PCR

Natural and genetically engineered strains were grown in vitro to an OD₆₀₀ of 0.2–0.6, where they were pelleted by centrifugation, resuspended in 1 ml of wash buffer (0.5% Tween 80, 0.8% sodium chloride) and transferred to 2.0 ml screw-cap tubes. RNA was isolated by a modified phenol-chloroform extraction protocol as previously described (Charlet et al., 2005). Genomic DNA contamination was removed by RNAeasy on-column digestion, following the manufacturer's protocol (Qiagen, Mississauga, ON). Once the quality of RNA was confirmed by denaturing gel electrophoresis (formaldehyde), it was converted to cDNA using Fermentas cDNA kit.

To determine mRNA levels for mpb70, mpb83 and sigA, we performed qRT-PCR with molecular beacons, applying sigA levels as a normalization control (Manganelli et al., 1999). The qRT-PCR method as well as primers and beacons for sigA, mpb70 and mpb83 used in this study are previously described in Charlet et al. (2005). To determine levels of sigK and Rv0444c as function of our deletion and complementation mutants, we modified this protocol to test expression by qRT-PCR with SYBR green. To test for co-transcription of sigK and Rv0444c, primers within each gene (SigK-CTXL: 5′-GGGCTGACGTATGTCGAAGT-3′ and Rv0444-CTXR: 5′-CTCGTTCATCGTCGGACAC-3′) were employed in PCR analysis of mRNA with and without reverse transcriptase, to demonstrate message bridging the genes.

Protein preparation and immunoblot analysis

For secreted proteins, culture supernatants were filtered with a 0.22 μm membrane filter and concentrated with an Millipore Ultra-10 Centrifugal Filter Unit, 10 000 MW cut-off. For membrane-associated proteins, culture pellet was resuspended in 1 ml PBS and samples were subjected to the Fast Prep (BIO 101 Savant) for 15 s at 6.0 rpm. Samples were incubated on ice for 10 min. Cell debris were removed by spinning down the samples and the supernatant was boiled for 20 min. SDS-PAGE was performed under reducing conditions using the Mini-PROTEAN 3 electrophoresis system (Bio-Rad) with 12% polyacrylamide gels. Proteins were transferred to a polyvinylidene difluoride membrane. Rabbit polyclonal antibodies against MPB70 and MPB83 (gifts of Harald Wiker) were blotted onto the membrane and the ECL Western kit (Millipore, Billerica, MA) was used for protein detection. As internal loading control, the same samples were transferred to a membrane and blotted with a mouse polyclonal antibody against MPT64 (gift of Mark Doherty), a secreted antigen whose expression is not thought to be affected by SigK. To detect levels of Rv0444c-encoded protein, we used rabbit serum raised against recombinant Rv0444c (Chemicon International, Temecula, CA) obtained as follows. Rv0444c was amplified from H37Rv and cloned into the pET21 vector. After sequence confirmation, the plasmid was transformed into E. coli BL21 (DE3) pLysS (Novagen) and the protein expressed through overnight induction with 0.5 mM IPTG. Cells were harvested by centrifugation (5000 g, 10 min and 4°C) and disrupted by sonication at 4°C. Cell debris was removed by centrifugation, and the supernatant was loaded onto a 1 ml HiTrap chelating column (Amersham Biosciences). The column was extensively washed with loading buffer and eluted using a 0–600 mM imidazole gradient in loading buffer to reveal a single band of the expected length on SDS-PAGE gel. This product was then sent for preparation of serum.

Yeast two-hybrid analysis

Protein interaction was assayed using the ProQuest^TM two-hybrid system (Invitrogen, Carlsbad, CA). By recombination cloning, we engineered a bait plasmid designed to overexpress recombinant SigK linked at its N-terminus to the Gal4 DNA-binding domain (pDEST32). Similarly, the prey plasmid expressed the N-terminal part of Rv0444c (amino acid residues 1–92) fused to GAL4 activation domain (pDEST22). pDEST32 and pDEST22 contain TRP1 and LEU2, respectively, permitting selection on media lacking tryptophan or leucine (synthetic complete media leu–trp–). Bait and prey plasmids were transformed into the mating-type yeast strain MAV203 using the method essentially described by Gietz et al. (1992). Briefly, yeast strains were grown in YPAD medium to an OD₆₀₀ of ∼1. Competent cells were prepared with 1× TE/LiAc solution and transformation was carried out by mixing yeast cell suspension with 50 μg of single stranded salmon sperm DNA, plasmid DNA (and 0.3–1 μg) and 40% PEG. Cells were incubated at 30°C with agitation (roller drum) for 30 min, followed by a heat shock at 42°C for 15 min. The MAV203 yeast strain contains integrated reporter genes such as ura3 and lacZ and positive interactions are detected when the GAL4 binding and activation domains are bridged via a direct protein–protein interaction, and activation of the reporter genes. Transformants were plated on SC-L-T-Ura and 5FOA plates to test for growth. For the β-galactosidase assay, transformants were grown overnight on nitrocellulose filter paper placed on YPAD agar plate as described in the invitrogen ProQuest Two-hybrid system manual. The filter was soaked in Z-buffer containing β-ME and Xgal after rapid exposure to liquid nitrogen and incubated overnight on Whatman paper.

Microarray analysis

Microarray hybridization and analysis were performed as previously described (Charlet et al., 2005). In brief, mRNA from BCG strains and complemented strains was extracted during log-phase in vitro growth and labelled with Cy3 or Cy5 dUTP by reverse transcriptase (Amersham Biosciences). Labelled cDNA was hybridized to microarrays composed of oligonucleotide probes from the TB Array-Ready Oligo Set^TM (Operon) printed onto Sigmascreen^TM microarray slides (Sigma). Initial comparisons were M. tuberculosis H37Rv versus select MTC members (Table S1). After complementing BCG Russia with wild-type Rv0444c, comparisons of Russia::pMV306 versus Russia::Rv0444c were also performed (Table S2). Hybridized arrays were scanned with ScanArray 5000XL and hybridization results were quantified with ScanArray software (PerkinElmer, Freemont, CA). Array analysis was performed as previously described (Mostowy et al., 2004c; Charlet et al., 2005) and can be summarized as such. Subtracting total spot intensity minus the surrounding background produced a corrected spot intensity. Negative corrected spot intensities were set to +1. All spots flagged as misrepresentative (array artefacts) or unrecognizable (quiescent in both channels) by ScanArray were analytically ignored. Intensity ratios (Cy3/Cy5 or Cy5/Cy3) were determined using corrected spot intensities and log10 transformed. Values for each gene were obtained in duplicate for each array (inherent to array design) and averaged. For each array, a representative Z-score, indicative of how many standard deviations a data point lies from the population mean, was calculated for each gene. Z-scores for each gene were determined across replicates within each experiment to minimize the probability of observing such variation by chance alone. Genes with Z-scores of two or greater across arrays are presented.