Enhanced expression of recX in Mycobacterium tuberculosis owing to a promoter internal to recA.

RecX is a small protein that interacts with, and modulates the activity of, RecA protein. In mycobacteria the recX gene is located immediately downstream of the recA gene, and the coding regions overlap. It has previously been shown that these two genes are co-transcribed in Mycobacterium smegmatis. In this study we examine the expression of recX in Mycobacterium tuberculosis. In addition to being co-transcribed with recA from the DNA-damage inducible recA promoters, we identify a constitutive recX promoter located within the recA coding sequence that is strong enough to make a significant contribution to the expression level of recX in the absence of DNA damage. Intriguingly, this promoter is inactivated in M. smegmatis by a critical base change in the -10 promoter motif, which probably accounts for the lower level of expression of recX relative to recA that we observed in that species. It is possible that this difference in relative expression influences RecA functions including the response to DNA damage of LexA-regulated genes. Crown

Introduction

The RecX protein was first identified as being encoded by a small open reading frame downstream of the recA gene in Pseudomonas aeruginosa and has since been found to be conserved in many bacterial species. The RecA protein is the key protein involved in homologous genetic recombination and recombinational DNA repair. RecA also plays a central role in regulation of gene expression in response to DNA damage by the repressor LexA via its stimulation of LexA cleavage on binding to regions of single-stranded DNA.

A role for RecX as a negative regulator of RecA was initially indicated by data demonstrating that the RecX protein was necessary to overcome the harmful effects of RecA over-expression in bacterial species as diverse as P. aeruginosa, Mycobacterium smegmatis, Streptomyces lividans and Xanthomonas oryzae pv oryzae. Subsequently, the RecX protein was shown to inhibit the ATPase, recombinase and co-protease activities of RecA. Detailed studies in Escherichia coli revealed that RecX acts by binding to the end of the growing RecA filament, thus limiting filament growth and leading to net disassembly. RecX is also able to bind within the major helical groove of the RecA filament at high relative concentrations which would block strand-exchange. Interestingly, in Deinococcus radiodurans, RecX also acts as a negative regulator of recA expression, with deletion of the recX gene resulting in constitutive activation of the recA promoter and over-expression of RecX suppressing recA promoter activity.

In most species studied to date recX is found downstream of the recA gene, and in a number of cases the two genes are co-transcribed, e.g. in S. lividans, M. smegmatis and E. coli. However, in Xanthomonas pathovars, the genes in the conserved lexA-recA-recX loci are each expressed from their own promoter. Exceptions to the typical recA-recX genomic organisation include bacterial species such as Bacillus subtilis and Neisseria gonorrhoeae in which the recA and recX genes are separated on the chromosome by distances as great as 838 kb and 275 kb respectively.

In a number of mycobacterial species, as well as a few other bacteria, the 5′ part of the recX coding sequence overlaps the 3′-region of the recA gene. In M. smegmatis the overlap is 32 bp long, while in Mycobacterium tuberculosis, Mycobacterium leprae Mycobacterium avium and Mycobacterium ulcerans a 35 bp overlap is present. The difference between the M. smegmatis sequence and those of the others is equivalent to the loss of a single codon rather than three separate base deletions. Among mycobacteria M. smegmatis has the simplest recA-recX locus arrangement, with the two genes flanked by genes expressed from the complementary DNA strand. In M. tuberculosis, recA-recX is followed by a single gene in the same orientation, termed Rv2735c, classified as encoding a conserved hypothetical protein with no known function. A further difference between M. tuberculosis and M. smegmatis in this region is the presence of an intein in M. tuberculosis but not M. smegmatis recA; an intein is an intervening sequence that is removed post-translationally in a process termed protein splicing.

The co-transcription of recA and recX has already been described in M. smegmatis. In this study we show that the transcriptional regulation of recX in M. tuberculosis differs from M. smegmatis, with recX being co-transcribed with recA but at the same time possessing its own promoter, the entirety of which is present within the coding region of the preceding recA gene.

Methods

Bacterial strains, plasmids and growth conditions

E. coli strains Alpha-select Silver efficiency (Bioline, London, UK) and DH5α (Invitrogen) were used as host strains for plasmid manipulations. E. coli cells were grown at 37 °C on LB agar plates or in LB medium with shaking at 225 rpm; where appropriate antibiotics were added at the following concentrations: ampicillin 100 μg/ml, kanamycin 50 μg/ml. M. tuberculosis strains H37Rv, 1424 (a streptomycin resistant derivative of H37Rv that is the parental strain for the recA mutant) and ΔrecA were grown at 37 °C on Difco Middlebrook 7H11 agar (Beckton Dickinson) plates supplemented with 4% albumin and 0.5% (w/v) glycerol or in modified Dubos medium (Difco) supplemented with 4% albumin and 0.2% (w/v) glycerol rolling at 2 rpm. Where appropriate, antibiotics were added to M. tuberculosis cultures at the following concentrations: kanamycin 25 μg/ml, streptomycin 100 μg/ml, hygromycin 50 μg/ml. All procedures with live M. tuberculosis were carried out under ACDP containment level 3 conditions.

Recombinant DNA techniques

Plasmid DNA was prepared using miniprep kits (QIAGEN) as described by the manufacturer. The plasmids used and their construction are described in Table 1, and the primers used in this study are listed in Table 2. The locations of key primers and other features in the recA-recX sequence are shown in Supplementary figure 1. Reporter constructs were created by cloning PCR fragments generated using the primer pairs indicated in Table 1 into the integrating lacZ transcriptional vector pEJ414. PCR reactions for cloning utilised PfuUltra® Hotstart DNA Polymerase (Stratagene) and buffer; all reactions contained 5% DMSO (Sigma–Aldrich). Site-directed mutagenesis was performed as described in the QuikChange® Site-Directed Mutagenesis Kit (Stratagene) but with two 15-cycle PCR reactions; an extra 1 μl of DNA polymerase was added between reactions. For other DNA manipulations, standard DNA protocols were followed. For each clone or mutant made, the sequences of the promoter region and the junctions to the vector were determined by commercial sequencing using the Illumina Genome Analysis system (Geneservice). Clones were introduced into M. tuberculosis via electroporation as described previously.

Table 1

Plasmids used in this study.

Plasmid	Construction/relevant characteristics
pCR4-Blunt-TOPO	TOPO cloning vector (Invitrogen)
pEJ414	lacZ transcriptional reporter plasmid with mycobacteriophage L5 attachment site and integrase genes. KanR23
pEJ449	recA P1 promoter in pEJ41422
pPR-Hyg	1780 bp fragment encompassing HygR cassette in ΔrecA strain in pEJ414 (PCR fragment from Hyg_PrF and Hyg_PrR)
pPRrecX	580 bp fragment 3′-end M. tuberculosis recA coding sequence upstream of recX in pEJ414 (PCR fragment from recA_F2 and recX_PrR)
pPRrecX-I	314 bp fragment equivalent of first 314 bp of cloned fragment in pPR recX in pEJ414 (PCR fragment from recA_F2 and Intein_R)
pPRrecX-R	266 bp fragment equivalent of last 265 bp of cloned fragment in pPR recX in pEJ414 (PCR fragment from recX_Pr2F and recX_PrR)
pJD01	ATGA to CGTC mutation in −10 region of pPRrecX (site-directed mutagenesis using X2SDMF and X2SDMR)
pJD05	285 bp fragment 3′-end M. smegmatis recA coding sequence upstream of recX in pEJ414 (PCR fragment from Msm_recX_PrF and Msm_recX_PrR)

Table 2

Primers used in this study.

Name	Sequence (5′→3′)	Position relative to translation start site of gene indicated.
Plasmid constructiona
recA_F2	CTGGTCTAGAGATGACCGATGCCGTGCTGAATTATC	−579 to −553 recX
recX_PrR	GACGCGGCCGCTTTGAGGGATCATCGGTCA	−18 to +1 recX
Intein_R	GGGAAGCTTGTTGTGCACGACAACCCCTTCGG	−288 to −266 recX
recX_Pr2F	GTGTCTAGATGTTCGCCCCCCTTCAAGCAGG	−265 to −244 recX
Hyg_PrF	CACTCTAGACGATCTGAGCTTGCATGCCTGC	n/a
Hyg_PrR	GCAAAGCTTGGTCATCTCGATCTGGCTCG	n/a
Msm_recX_PrF	GGTCTAGACGCCGTTCAAGCAG	−255 to −242 recX
Msm_recX_PrR	CGAAGCTTCAGAAGTCAACCGG	+18 to +31 recX
RecA operon co-transcription∗
recX_F1	GCACCCGCGCCGAGTTAGC	+89 to +107 recX
recA_F1	TCGCGGATGCCCTGGATGACAAAT	−426 to −403 recX
recA_F2	CTGGTCTAGAGATGACCGATGCCGTGCTGAATTATC	−579 to −553 recX
Rv2735c_R1	AGTTCCGCGTTCGTGCCCTTCA	+506 to +527 Rv2735c
recA_F3	AAGTCTAGATGCCTCGCAGAGGGCACTCGG	+754 to +774 recA
recA_F4	GCTCAGGCCGCCGGTGGTGTTG	+250 to +271 recA
RecX_R1	CGTTCGCCCGCCTGGACTGA	+216 to +235 recX
5′ RACE
GSP_recX	GCTCGGCGGCCAGCTCGGCGATAAC	+481 to +505 recX
GSP_recX_Nest	GCTTTTCCGCCCGCCCCCGTTCG	+333 to +355 recX
GSP_Hyg	TGTGCGGCGAGTTGCGTGA	n/a
GSP_Hyg_Nest	GTCGTCGTCCCCTCAAACTGG	n/a
qRT-PCR
SigA_F	TCGGTTCGCGCCTACCT	+679 to +695 sigA
SigA_R	TGGCTAGCTCGACCTCTTCCT	+731 to +751 sigA
recA_up_F	ATCGAGAAGAGTTACGGCAAAGG	+52 to +74 recA
recA_up_R	GCCCAGGGCCACGTCTA	+143 to +159 recA
RecA_del_F	ACCGGCGCGCTGAATA	+538 to +553 recA
RecA_del_R	CGCGGAGCTGGTTGATG	+576 to +592 recA
RecX_F	GCACCCGCGCCGAGTTAG	+89 to +106 recX
RecX_R	GGCCAGCCGATCCAATACCC	+152 to +171 recX
2735c_F	TAACCCGCTTGCCTCTGAA	+228 to +246 Rv2735c
2735c_R	ACCTACCGTCACCGGGAAAG	+270 to +289 Rv2735c
Msm_sigA_F	CCCACCGGGAATTCGTAAG	−34 to −16 sigA
Msm_sigA_R	TTGCCGGGCTTGCCTT	+13 to +28 sigA
Msm_recA_F	TGACCGGCGCGTTGA	+539 to +553 recA
Msm_recA_R	CGGAGCTGGTTGATGAAGATC	+573 to +593 recA
Msm_recX_F	TGCGCCGCGAACGT	+412 to +425 recX
Msm_recX_R	CCTGCGGGTGACCTTGAC	+450 to +467 recX
Site-directed mutagenesis†
X2SDM_F	CATTGGTGCCGTGGTGACCGcgtcTCCCTCAAGCGGCCGCCACG	n/a
X2SDM_R	CGTGGCGGCCGCTTGAGGGAgacgCGGTCACCACGGCACCAATG	n/a

Underlined bases indicate restriction sites added to primers to facilitate cloning of PCR products and bases in italics indicate bases that are not homologous to the native sequence.

Bases in lower case indicate bases different from the wild-type sequence.

RNA preparation and reverse transcription

The FastRNA® Pro Blue Kit (Qbiogene) was used for the isolation of total RNA from M. tuberculosis (50–100 ml) grown to OD600 of ∼0.6. Contaminating DNA in the RNA preparations was digested using TURBO RNase-free DNase (Ambion), and the RNA was subsequently cleaned up using an RNeasy MiniKit (QIAGEN). RNA concentrations were determined spectrophotometrically at 260 nm. Removal of DNA was confirmed by performing PCR using an aliquot of the DNase-treated RNA as a template. Reverse transcription for qRT-PCR and shorter transcripts used Superscript II (Invitrogen) according to the product information. Reverse transcription for long transcripts used QuantiTect Reverse Transcriptase (QIAGEN) according to the product information but with an extended (2 h) 42 °C incubation time. PCR reactions following reverse transcription (RT-PCR) were performed using REDTaq ReadyMix PCR Reaction Mix incorporating REDTaq DNA polymerase (Sigma–Aldrich) with an initial denaturation step of 95 °C for 5 min, followed by 10 cycles of 94 °C for 1 min, 68 °C for 1 min, 72 °C for 8 min, then 35 cycles of 94 °C for 1 min, 58 °C for 1 min, 72 °C for 8 min, and a final extension step of 72 °C for 20 min.

Real-time PCR

Real-time PCR analysis was carried out on the 7500 Fast Real-Time PCR System (Applied Biosystems) using the PCR master mix containing SYBR green dye (Applied Biosystems). The 20 μl PCRs consisted of PCR master mix, 900 nM concentrations of each primer, and 5 μl of cDNA template. The sequences of the primers used in the real-time PCR are given in Table 2. In each case, the test gene and the normalising gene (sigA) were assayed along with a set of standard samples (genomic DNA), and the amounts of gene-specific mRNA were normalised to the amount of sigA mRNA.

5′ RACE transcriptional start site mapping

RNA ligase-mediated 5′ RACE using the Generacer kit (Version 2.0; Invitrogen) was used to map the transcriptional start site of recX according to the manufacturer’s guidelines. To facilitate the identification of a transcript start site, which in bacteria carry a 5′-triphosphate, RNA was first treated with tobacco acid pyrophosphatase (TAP), which hydrolyses this group to a 5′-monophosphate to which an RNA oligonulceotide can then be ligated. Following this ligation step and reverse transcription, the resulting cDNA was amplified using the Generacer 5′ forward primer and the appropriate gene-specific primers (Table 2). The products obtained were cloned into pCR4-Blunt-TOPO (Invitrogen) and between five and eight clones were sequenced.

Preparation of cell-free extracts and β-galactosidase assays

Mycobacterial cultures were grown to mid-exponential phase (OD600 0.6) unless otherwise stated; the bacteria were harvested, washed three times in PBS and cell-free extract prepared using a Ribolyser (Hybaid) and glass beads (150-212 microns, Sigma) as described previously. The supernatant was filtered through a low-binding Durapore 0.22 μm membrane filter (Ultrafree-MC; Millipore) to ensure complete removal of bacteria before removal from CL3 facilities. Where mitomycin C induction was required, cultures were grown to early exponential phase (OD600 0.3) and were then split into two; one culture was induced with 0.02 μg/ml mitomycin C and the other was an uninduced control. Both were then incubated at 37 °C for 24 h before harvesting and preparing cell-free extract as above. Total protein was assayed using a BCA kit (Pierce); β-galactosidase activity was determined using ortho-nitrophenyl-β-galactoside (ONPG) as substrate in 500 μl reactions as described and expressed in Miller units per milligram of protein.

Results

Definition of the recA operon

To define the operon structure of recA in M. tuberculosis, we assessed co-transcription with the two downstream genes recX and Rv2735c by RT-PCR. Initially the presence of RNA spanning each of the pairs of genes recA-recX and recX-Rv2735c was demonstrated by the formation of a product following reverse transcription of RNA isolated from the parental wild-type M. tuberculosis strain 1424, but not if the RT step was omitted (Figure 1). To investigate if a single transcript including all three genes was formed, two separate primers within recA were used in conjunction with one in Rv2735c, and in each case a product was formed by RT-PCR that was dependent on the reverse transcription step (Figure 1). Thus, recX and Rv2735c are at least partially co-transcribed with recA.

Figure 1

recX and Rv2735c are co-transcribed with recA in wild-type M. tuberculosis but read-through into recX is not apparent in the recA mutant strain. (A) Schematic showing the relative locations of the three genes recA, recX and Rv2735c, each depicted by a large arrow. The position of the intein in recA is shown by the grey box within the arrow representing recA, and the region deleted in recA is indicated by the double-headed arrow. The locations of the primers used for RT-PCR analysis are indicated by the small arrows and labeled. The expected PCR products, labeled 1 to 6, formed by the various primer combinations listed on the left are shown by the black bars, with the sizes of these products indicated at the right. (B) The RT-PCR products obtained using wild-type genomic DNA (upper left panel) or RNA isolated from wild-type parental 1424 M. tuberculosis (upper middle and right panels) or from the ΔrecA strain (lower panels). The RT− panels in the centre show the results when reverse transcription was omitted, and the RT+ panels on the right show the products obtained from RNA following a reverse transcription step. The lane numbers correspond to the numbering of the PCR products in (A); PCR 3 was not tested for the ΔrecA strain. The positions and sizes of the expected PCR products are indicated by the arrows on the right panels, and the positions of DNA molecular mass markers are indicated by the short lines at the left of the gel images.

Expression in the ΔrecA strain

The mutation in the ΔrecA strain consists of deletion of 1259 bp between the two PstI sites within the coding sequence for RecA and insertion of a 1770 bp DNA fragment encoding hygromycin resistance. This mutation would be expected to have a polar effect on the expression of downstream genes that are part of an operon with recA owing to the presence of a transcriptional terminator within the fragment inserted. Therefore, we evaluated the expression of recA, recX and Rv2735c in the parental wild-type strain 1424 and the recA mutant by quantitative RT-PCR with normalisation to the housekeeping gene sigA.

When a primer set located in the part of the recA coding sequence remaining upstream of the deletion was used, there was no significant difference between the two strains in the expression level observed, while, as expected, no expression was detectable in the ΔrecA strain when a primer set located in the deleted region was used. The apparent expression level in the wild-type strain appeared to differ between the two primer sets, but this was not statistically significant (p > 0.05, Students t-test). The observed expression level for Rv2735c was at a lower level than recA or recX. It has been observed previously that downstream genes in an operon are expressed at a lower level than promoter proximal genes, including in actinomycetes. The most significant finding was that expression of recX and Rv2735c was still detectable, albeit at a reduced level, in the ΔrecA strain (Figure 2). This raised the possibility that this residual expression might arise from a promoter other than those already characterised upstream of recA.

Figure 2

recX and Rv2735c are still expressed in the ΔrecA strain, although at a lower level than in the wild-type. qRT-PCR was performed as described in the Methods using RNA isolated from wild-type M. tuberculosis (black bars) or the recA mutant (white bars). Expression levels were normalised to that of the housekeeping gene sigA. Primer set RecA-up was located in recA upstream of the deleted region, while primer set RecA-del was located within the deleted region. Primer sets RecX and Rv2735c were within the coding sequences of the respective genes. Values are the means from three independent cultures, each of which was assayed in triplicate; error bars show standard deviations.

An alternative explanation could be that the residual expression of recX in the ΔrecA strain might originate from transcriptional read-through from the recA promoters owing to incomplete termination in the hygromycin resistance cassette. To assess this possibility, we compared the ability of RNA isolated from the ΔrecA strain with that from the wild-type parental strain to act as template in RT-PCR reactions when one primer was located in recA and the other was in recX or Rv2735c. The precise locations of the primers within the recA-recX sequence are shown in Supplementary figure 1 along with other key features. When wild-type RNA was used, we obtained a product of the expected size for each primer pair combination used following reverse transcription, but not if the reverse transcription step was omitted (Figure 1). However, none of the primer combinations that included a primer located in recA yielded an RT-PCR product when RNA isolated from the ΔrecA strain was used (Figure 1). In contrast, a product was formed when a primer in recX was used in combination with one in Rv2735c (Figure 1), demonstrating that the RNA isolated from the mutant strain was capable of being reverse-transcribed into cDNA that could act as a substrate for PCR. Although the binding site for one of the primers used in recA would have been removed in the deletion strain, for one of the other primers in recA the binding site was upstream of the deletion, while another would bind immediately downstream of the deletion. Therefore, the lack of RT-PCR product formation for these two primer pairs with RNA from the ΔrecA strain argues against both transcriptional read-through from the recA promoters and expression originating from a promoter within the hygromycin resistance fragment.

Localisation of promoter activity

The results described above suggested that the residual expression of recX observed in the ΔrecA strain most likely arose from a promoter located in the remaining part of the recA gene downstream of the deletion/insertion, although the possibility of a promoter in the hygromycin resistance fragment driving expression was not completely excluded. To distinguish between these possibilities, we made two transcriptional fusions to the reporter gene lacZ. The first, pPR-Hyg, contained the DNA fragment conferring resistance to hygromycin present in the ΔrecA strain and the second, pPRrecX, contained the 3′-end of recA from the end-point of the deletion in the ΔrecA strain to the beginning of the RecX coding sequence. Following transformation into wild-type H37Rv bacteria, we were surprised to detect promoter activity from both fragments (Figure 3), although the activity residing within the 3′-end of recA was substantially greater than that from the hygromycin resistance fragment.

Figure 3

Identification of promoter activity upstream of recX. (A) Schematic showing the arrangement of the recA locus in the wild-type and in the ΔrecA strain at the top, with the various fragments cloned in the lacZ reporter vector aligned with the corresponding part of the locus below. The RecA coding sequence is shown in black, the intein sequence in dark grey, and the recX sequence in light grey, with the hygromycin resistance fragment indicated by a white box and the lacZ transcriptional reporter gene by a white arrow. (B) Promoter activity conferred on a lacZ reporter gene by transcriptional fusion with the DNA fragments indicated in (A), or with the recA P1 promoter in the positive control pEJ449. pPR-Hyg contains the fragment encoding resistance to hygromycin in the the recA mutant, pPRrecX contains the 3′end of recA from the end-point of the deletion in the recA mutant to the beginning of the RecX coding sequence, pPRrecX-I contains the intein-encoding part of the 3′end of recA and pPRrecX-R contains the RecA-encoding part of the 3′end of recA. pEJ414 is the vector control. β-galactosidase activity was determined using untreated cultures of M. tuberculosis (white bars) or cultures exposed to 0.02 μg/ml mitomycin C for 24 h (black bars). Values are the means from three independent cultures, each of which was assayed in duplicate; error bars show standard deviations.

As part of the DNA remaining at the 3′-end of recA in the deletion strain corresponds to the intein, we wanted to know if the promoter in this region was located in the intein-encoding DNA or in the RecA-encoding sequence. We, therefore, subdivided this region and made two further lacZ reporter constructs, one pPRrecX-I containing just the intein sequence and the other pPRrecX-R just the RecA sequence. Analysis of these clones in M. tuberculosis H37Rv clearly demonstrated that the promoter activity resided in the RecA-encoding DNA only (Figure 3).

To assess if the promoter identified at the 3′-end of recA was DNA damage-inducible, we compared the expression levels obtained from the relevant reporter clones following exposure to the DNA damaging agent mitomycin C (0.02 μg/ml for 24 h) with an unexposed control. The presence or absence of mitomycin C did not alter the expression levels obtained (Figure 3). In contrast, the recA P1 promoter contained within pEJ449 used as a positive control was clearly inducible, as previously demonstrated (Figure 3). Thus, the promoter within the 3′-end of recA is constitutive.

Identification of promoter elements

In order to confirm the presence of promoters, we sought to identify the promoter elements and then assess the effect of introducing mutations in the −10 regions. To facilitate the identification of promoter elements, we first mapped the transcript start sites by 5′ RACE.

This revealed that in the reporter clone containing the hygromycin resistance fragment the transcript start site was actually located within vector sequence between the insert and the lacZ gene and that the −10 region was located at the junction of the insert and vector. It appears that in constructing this clone, combining vector sequence with very limited homology to the SigA −10 consensus (normally incapable of driving expression of the reporter gene as demonstrated by assays on the empty vector in Figure 3), with insert sequence with a reasonable match to the SigA −35 consensus created a weak artificial promoter. Thus, the promoter activity apparently originating from the hygromycin resistance fragment is an artefact of cloning, and was not investigated further.

The transcript start site for recX was mapped to the A corresponding to the first base of the ATG initiation codon. Upstream of this, motifs (TTGGCA-N17-GATGAT) resembling both the −35 and −10 consensus sequences for SigA (TTGACW-N17-TATAMT) were identifiable at the appropriate locations (Figure 4). To confirm that the promoter had been correctly identified, we introduced a mutation in the −10 motif, changing the central ATGA to CGTC in pJD01. When assayed in M. tuberculosis H37Rv the corresponding promoter activity was reduced to levels comparable with the vector control (Figure 4). Thus, the presence of an active promoter within the coding sequence of recA was verified.

Figure 4

The location of the recX promoter at the 3′-end of recA and the effect of mutation of the −10 region. (A) The sequence of the 3′-region of recA and the 5′-region of recX showing the location of the recX promoter at the 3′-end of recA in relation to the intein and the two coding regions. The 3′-end of the intein is underlined with a dotted line. The recX sequence is in italics. The region of overlap of recX with recA is underlined, with the start codon of recX and the stop codon of recA double underlined. The −35 and −10 promoter elements are boxed and the transcription start site for this promoter is circled. The SigA consensus sequence is shown in lower case above the promoter elements. The bases replacing the native −10 motif in the mutated reporter clone pJD01 are indicated in italic lower case below the −10 motif of the wild-type sequence. (B) Effect of introducing the sequence change shown in (A) on recX promoter activity, compared with the vector pEJ414. β-galactosidase activity was determined from cultures of M. tuberculosis grown to OD600 of 0.6; values are the means from three independent cultures, each of which was assayed in duplicate and the error bars show standard deviations.

The recX promoter is not conserved in all mycobacteria

In order to gain an indication as to how widespread the recX promoter internal to recA is, we aligned the sequences of the 3′-end of recA and the 5′ end of recX from a variety of mycobacterial species (Figure 5). Not surprisingly, the sequence was completely conserved in Mycobacterium bovis and M. bovis BCG. Highly conserved motifs were present in M. avium and M. avium subsp. paratuberculosis, differing from the M. tuberculosis sequence by only a single base in each of the −10 and −35 motifs. In contrast, the −10 region was completely absent in Mycobacterium marinum, M. ulcerans and M. leprae. In M. smegmatis although the corresponding motifs were recognisable, one of the base changes in the −10 region (A to C at −12) would be likely to inactivate the promoter.

Figure 5

Extent of conservation of the recX promoter amongst mycobacterial species. Alignment of the 3′-end of recA and the 5′-end of recX from M. tuberculosis H37Rv (MtRv), M. bovis AF2122/97 (Mbov), M. bovis BCG str. Pasteur 1173P2 (MBCG), M. avium 104 (Mav), M. avium subsp. paratuberculosis K-10 (Mpar), M. smegmatis mc2155 (Msm), M. marinum ATCC BAA-535 (Mmar), M. ulcerans Agy99 (Mulc), and M. leprae TN (Mlep). The locations of the −10 and −35 motifs are boxed with solid lines and labeled above the sequences, the RecX start codon (position 51–53 in MtRv) is boxed with a dashed line and labeled, and the RecA stop codon is boxed with a dotted line and labeled. Bases that differ from the M. tuberculosis sequence are shaded. The −10 motif is absent in M. marinum, M. ulcerans and M. leprae, and has a potentially inactivating base change in M. smegmatis.

To test if this is the case, we cloned the corresponding region from M. smegmatis into the lacZ transcriptional reporter vector, creating pJD05. When assayed in M. smegmatis, the expression driven by this sequence (3.0 ± 0.4 units/mg protein) did not differ significantly (p > 0.05, Students t-test) from the background level seen with the vector control (2.2 ± 0.4 units/mg protein). Thus, recX does not have an additional promoter other than those upstream of recA in M. smegmatis.

What is the significance of the recX promoter in M. tuberculosis?

The presence of an additional promoter for recX in M. tuberculosis compared with M. smegmatis would be expected to result in a higher recX expression level under non-DNA damaging conditions. To assess if this is the case we compared the expression of recX with that of recA in wild-type H37Rv M. tuberculosis and M. smegmatis by qRT-PCR, normalising to sigA. In M. tuberculosis, there was no significant difference (p > 0.05, Students t-test) in the expression levels of the two genes (Figure 6). In contrast, in M. smegmatis recX was expressed at a significantly lower level (p < 0.01, Students t-test) than recA (Figure 6), although the level of recA was similar in the two species. Thus, the ratio of recX to recA in M. smegmatis was a quarter of the corresponding value for M. tuberculosis (0.21 compared with 0.84).

Figure 6

Comparison of the expression level of recX relative to recA in M. tuberculosis and M. smegmatis. qRT-PCR was performed as described in the Methods using RNA isolated from wild-type M. tuberculosis (Mtb) or M. smegmatis (Msm). Expression levels of recA (black bars), using primer set RecA_up for M. tuberculosis, and recX (white bars) were normalised to that of the housekeeping gene sigA for each species. Values are the means from three independent cultures, each of which was assayed in triplicate; error bars show standard deviations.

How common are promoters internal to genes in M. tuberculosis?

We wondered if promoters were also present within the 3′-ends of other M. tuberculosis coding sequences. To gain some insight into this question, we took a bioinformatic approach. First we searched the M. tuberculosis genome sequence for sequences matching the SigA consensus that were located no more than 200 bp upstream of genes, permitting a maximum of 3 mismatches in total, of which no more than 2 could be in a single motif (either the −10 or −35 motif). We identified 231 regions matching these criteria. We then looked to see how many of these were located upstream of the stop codon of the preceding gene. We found that 84 were within 200 bp (Supplementary table 1), and 44 were within 100 bp, of such a stop codon, representing 36% and 20% of predicted promoters respectively. For comparison, we performed the equivalent searches on the E. coli genome but using the σ70 consensus sequence. This revealed the presence of 457 predicted promoters in the 200 bp regions upstream of genes. However, only 49 of these were within 200 bp of the stop codon of the preceding gene (Supplementary table 2), and only 22 within 100 bp, representing 11% and 5% of the predicted promoters respectively. Thus, although almost twice as many promoters were predicted within 200 bp upstream of genes in E. coli than in M. tuberculosis, a much smaller proportion of these were located upstream of the stop codon of the preceding gene, despite the gene density in the two organisms being very similar. This suggests that promoters internal to coding regions may be more common in M. tuberculosis than in some other bacteria.

Discussion

In this study we have demonstrated that M. tuberculosis recX is expressed from a constitutive promoter located within the coding sequence of recA, the gene immediately upstream of recX, in addition to being co-transcribed with recA from its two DNA-damage inducible promoters. The occurrence of promoters internal to coding sequences responsible for transcription of downstream genes in bacteria has only recently been recognised. Our bioinformatic analysis suggests that promoters so located may be more common in M. tuberculosis than in some other bacteria. The development of genomic RNA sequencing approaches is now facilitating the identification of such promoters, such that over 400 internal transcription start sites have been identified in the pathogen Helicobacter pylori. It will be interesting to see what this approach reveals in the case of M. tuberculosis.

Transcription of recX from the promoter internal to recA initiated at the A of the recX ATG codon. Although relatively uncommon, a few examples of such leaderless mRNAs have been described previously from M. tuberculosis: purC, oxyR, eis and lexA. Leaderless mRNAs interact preferentially with 70S ribosomes in bacteria and similar findings in eukaryotes have led to the suggestion that this type of mRNA might be evolutionary ancient. Cross-linking studies have revealed that leaderless mRNAs are recognised by the ribosome initially via the AUG codon; indeed, the addition of an AUG triplet to a random RNA sequence was found to confer ribosome binding. The presence of initiator tRNA conferred greater stability on the interaction, as revealed by gel shift and toeprinting assays. Translation of leaderless mRNAs has also been shown to be preferentially stimulated by initiation factor 2 (IF2), and the ratio of IF2 to IF3 can modify the efficiency of translation initiation on such RNAs, potentially providing a means of modulating translation of leaderless mRNAs in response to conditions causing a variation in the availability of these components of the translational machinery.

The expression of the reporter gene lacZ driven by the recX promoter (ca. 200 units) was at a comparable level to that conferred by either of the recA promoters under non-DNA damaging conditions (P1 ca. 100 units and P2 ca. 230 units) when present on fragments of similar size. Thus, it can be expected to make a significant contribution to the expression of recX in the absence of DNA damage. Intriguingly, this recX promoter is missing in some other mycobacterial species (M. marinum, M. ulcerans and M. leprae) and is not active in M. smegmatis owing to base changes within the promoter elements. At present it is not clear whether M. avium and M. avium subsp. paratuberculosis possess an active promoter in this region. As would be expected from the observations described above, the expression of recX relative to that of recA was lower in M. smegmatis than in M. tuberculosis.

RecX protein interacts with RecA to modulate its activities. Thus, the elevated ratio of RecX to RecA in M. tuberculosis compared with M. smegmatis may influence RecA function. RecX inhibits the ATPase and strand-exchange activities of RecA in vitro, and so may affect DNA repair processes in the cell. Indeed, the lack of recX has been reported to have a small effect on survival following UV irradiation in E. coli. RecX also inhibits the ability of RecA to stimulate cleavage of the repressor protein LexA in vitro and, when overexpressed, in E. coli cells. This latter property results in a reduced rate of induction of LexA-regulated genes in response to DNA damage under the tested conditions. Thus, it is possible that higher basal expression of RecX relative to RecA in M. tuberculosis compared with M. smegmatis may contribute to the previously observed slower kinetics of induction of the SOS response in M. tuberculosis. However, it should be remembered that once induction has occurred, the expression of recX will be dominated by the elevated activity of the two recA promoters which are both induced by DNA damage. To definitively assess the biological significance of the recX promoter located within the recA coding sequence it would be necessary to introduce point mutations that inactivate this promoter and compare the resulting phenotype with a strain having the wild-type sequence. However, this is problematical as the point mutation most likely to inactivate the promoter (A to C at −12) would also result in an amino acid change in RecA (aspartic acid to alanine in the above example) that could affect its activity. The only silent point mutation in terms of coding sequence (T to C at −11) seems less likely to inactivate the promoter, although this could be tested by introducing the base change into a transcriptional fusion plasmid such as pPRrecX.

Funding:

This work was funded by the UK Medical Research Council (programme number U1175 32056).

Competing interests

None declared.

Ethical approval

Not required.