Characterization of the Two Mycobacterium tuberculosis recA Promoters

Abstract

The recA gene of Mycobacterium tuberculosis is unusual in that it is expressed from two promoters, one of which, P1, is DNA damage inducible independently of LexA and RecA, while the other, P2, is regulated by LexA in the classical way (E. O. Davis, B. Springer, K. K. Gopaul, K. G. Papavinasasundaram, P. Sander, and E. C. Böttger, Mol. Microbiol. 46:791-800, 2002). In this study we characterized these two promoters in more detail. Firstly, we localized the promoter elements for each of the promoters, and in so doing we identified a mutation in each promoter which eliminates promoter activity. Interestingly, a motif with similarity to Escherichia coli σ⁷⁰ −35 elements but located much closer to the −10 element is important for optimal expression of P1, whereas the sequence at the −35 location is not. Secondly, we found that the sequences flanking the promoters can have a profound effect on the expression level directed by each of the promoters. Finally, we examined the contribution of each of the promoters to recA expression and compared their kinetics of induction following DNA damage.

The protein RecA is very important for a number of processes related to DNA metabolism and has been best characterized in Escherichia coli. It is a central component of recombination, whether by the RecBCD pathway or the RecF pathway (20). In either case, RecA forms a nucleoprotein filament on regions of single-stranded DNA and by its ability to simultaneously bind double-stranded DNA performs the search for homologous sequences. Strand exchange then initiates the actual process of recombination. RecA also has a key regulatory role in the response to DNA damage, owing to the ability of the nucleoprotein filament formed on regions of single-stranded DNA to stimulate the cleavage of the repressor protein LexA (15, 25). Under normal conditions, LexA binds to a specific sequence upstream of the genes it regulates, which include recA and lexA themselves, and inhibits their expression (4, 26); however, following cleavage, the fragments of LexA formed do not effectively bind to this site (3), permitting increased transcription. As well as regulating the expression of other genes following DNA damage, RecA has a direct role in recombinational DNA repair which is particularly important for replication restart following the collapse of a replication fork, for example, when it reaches a damaged segment of DNA (7). Although the majority of studies on RecA function have been conducted with the E. coli protein, RecA is highly conserved across bacterial species (18, 34). Thus, it appears most likely that it performs equivalent functions in all bacteria.

In many species of bacteria, expression of RecA is induced following DNA damage. In the majority of these cases this process is regulated by a homolog of LexA, as in E. coli (12), although the sequence to which the LexA homolog binds varies between species. In both E. coli and the well-studied gram-positive bacterium Bacillus subtilis, recA is expressed from a single promoter (6, 39). It was recently reported that recA of Xanthomonas oryzae pv. oryzae is expressed from more than one promoter, but it appeared that each promoter was regulated by LexA (38). The recA gene of M. tuberculosis is unusual in that it is expressed from two promoters which are both DNA damage inducible but by different mechanisms (10). The promoter further from the coding sequence has a LexA binding site or SOS box located between its putative −10 and −35 regions and is regulated by LexA in a manner analogous to that observed for E. coli. In contrast, the promoter nearer to the coding sequence remains DNA damage inducible both when LexA binding is prevented and when LexA cleavage is blocked. In this study we used lacZ transcriptional fusions to dissect the DNA upstream of the M. tuberculosis recA gene in order to locate regions important for directing recA expression and in particular to identify the promoter elements.

MATERIALS AND METHODS

Bacterial strains and media.

For general cloning, E. coli DH5α was used, while for site-directed mutagenesis E. coli XL1Blue was used (36). The mycobacterial strains used were Mycobacterium smegmatis mc²155 (37) and Mycobacterium tuberculosis H37Rv. E. coli was grown in L broth (36), and M. smegmatis and M. tuberculosis were grown in modified Dubos medium supplemented with albumin (Difco) and 0.2% glycerol. Antibiotics were added as appropriate: kanamycin was used at 50 μg ml⁻¹ for E. coli and 20 μg ml⁻¹ for mycobacteria. E. coli and M. smegmatis were grown at 37°C with shaking in a rotary incubator at 280 rpm, and M. tuberculosis was grown at 37°C in a rolling incubator at 2 rpm. All procedures with M. tuberculosis were carried out under containment level 3 conditions.

Recombinant-DNA techniques.

Plasmids used in this study are listed in Table 1. The oligonucleotide primers used in this study are listed in Table 2. Plasmid DNA was prepared by using SNAP miniprep kits (Invitrogen). Site-directed mutagenesis was performed as described in the QuikChange site-directed mutagenesis kit (Stratagene). For other DNA manipulations, standard DNA protocols were followed (36). For each clone or mutant made, the sequences of the promoter region and the junctions to the vector were determined on an ABI Prism 377 DNA sequencer using the ABI Prism dRhodamine dye terminator cycle sequencing kit (PE Applied Biosystems). Clones were introduced into M. smegmatis or M. tuberculosis by electroporation (17) and verified by PCR and sequencing as described previously (8).

TABLE 1.

Plasmids used in this study

Plasmid	Promoter	Insert locationa	Construction (reference)b
pEJ414			lacZ reporter vector (30)
pEJ417	P1 + P2	1-345	345-bp fragment of DNA containing 310 bp from upstream M. tuberculosis recA in pEJ414 (30)
pEJ418	P1	211-386	176-bp PCR product from oligonucleotides RNAP1Xba and RNAP2Spe in SpeI site in pEJ414 (10)
pEJ419	P2	50-250	201-bp PCR product from oligonucleotides RNAP3Xba and RNAP4Spe in SpeI site in pEJ414 (10)
pEJ420	P1	214-269	56-bp ds oligonucleotide PROXPR1 between XbaI and HindIII sites in pEJ414
pEJ421	P2	167-224	58-bp ds oligonucleotide DISTPR1 between XbaI and HindIII sites in pEJ414
pEJ422		1-60	60-bp ds oligonucleotide FARUP into PmlI site in pEJ414
pEJ423	P2	167-224	58-bp ds oligonucleotide DISTPR2 containing mutation in SOS box between XbaI and HindIII sites in pEJ414
pEJ431	P2	50-224	175-bp PCR product from oligonucleotides RNAP3Xba and DISTPR1r between XbaI and PmlI sites in pEJ414
pEJ432		104-161	58-bp ds oligonucleotide UP1 into XbaI site in pEJ414
pEJ435	P2	104-224	59-bp ds oligonucleotide UP1 into XbaI site in pEJ421 in correct orientation; SDM with oligonucleotide XBATOWT to correct sequence of internal XbaI site to wild-type sequence (10)
pEJ445	P2	104-224	59-bp ds oligonucleotide UP1 into XbaI site in pEJ423 in correct orientation; SDM with oligonucleotide XBANCO to change upstream XbaI site to NcoI site; contains mutation in SOS box and internal XbaI site
pEJ447		260-345	86-bp XbaI-HindIII fragment of PCR product from oligonucleotides DOWN and VREV on pFM6 (28) between XbaI and HindIII sites in pEJ414
pEJ449	P1	1-345	SDM of pEJ417 with oligonucleotide P2MUT to inactivate P2 (10)
pEJ463	P1	104-269	166-bp PCR product from oligonucleotides UP1f and PROXPR1r into PmlI site in pEJ414; SDM with oligonucleotides P2MUT to inactivate P2
pEJ465	P1	167-269	103-bp PCR product from oligonucleotides DISTPR1f and PROXPR1r into PmlI site in pEJ414; SDM with oligonucleotide P2MUT to inactivate P2
pEJ481	P2	167-224	61-bp ds oligonucleotide CDSRECA from CDS recA into XbaI site in pEJ421 in sense direction
pEJ482	P2	167-224	61-bp ds oligonucleotide CDSRECA from CDS recA into XbaI site in pEJ421 in reverse orientation
pEJ498	P1	50-269	220-bp PCR product from oligonucleotides RNAP3Xba and PROXPR1r containing P2 mutation between XbaI and HindIII sites in pEJ414
pEJ507	P1	50-313	264-bp PCR product from oligonucleotides RNAP3Xba and BEGNCO containing P2 mutation between XbaI and HindIII sites in pEJ414
pEJ512	P1	104-313	210-bp PCR product from oligonucleotides UP1f and BEGNCO containing P2 mutation between PmlI and HindIII sites in pEJ414
pEJ514	P1	167-313	147-bp PCR product from oligonucleotides DISTPR1f and BEGNCO between PmlI and HindIII sites in pEJ414; SDM with oligonucleotides P2MUT to inactivate P2
pEJ535	P2	124-224	38-bp ds oligonucleotide UP2 into XbaI site in pEJ421 in correct orientation, then SDM with oligonucleotide XBATOWT to correct sequence of XbaI site to wild-type sequence
pEJ626	P2	55-345	Deletion of 54 bp between XbaI and insertion of NheI site of pKKG25
pEJ627	P2	1-224	224-bp PCR product from oligonucleotides FARUPf and DISTPR1r containing P1 mutation into PmlI site in pEJ414
pEJ628		1-161	161-bp PCR product from oligonucleotides FARUPf and UP1r into PmlI site in pEJ414
pKKG5	P2	104-224	pEJ445 with T−10G mutation
pKKG6	P2	104-224	pEJ445 with A−12C mutation
pKKG7	P2	104-224	pEJ445 with C−14A mutation
pKKG8	P2	104-224	pEJ445 with G−16T mutation
pKKG11	P2	104-224	pEJ445 with C−17A mutation
pKKG12	P2	104-224	pEJ445 with A−32G mutation
pKKG13	P2	104-224	pEJ445 with C−36A mutation
pKKG14	P2	104-224	pEJ445 with G−15T mutation
pKKG15	P1	1-345	pEJ449 with TCT→GAG in −10 region
pKKG16	P1	1-345	pEJ449 with T−13G mutation
pKKG17	P1	1-345	pEJ449 with A−12C mutation
pKKG18	P1	1-345	pEJ449 with T−10G mutation
pKKG19	P1	211-386	pEJ418 with TCT→GAG in −10 region
pKKG20	P1	211-386	pEJ418 with T−13G mutation
pKKG21	P1	211-386	pEJ418 with A−12C mutation
pKKG22	P1	211-386	pEJ418 with T−10G mutation
pKKG23	P2	50-250	pEJ419 with A−12C mutation
pKKG24	P1	1-345	pEJ449 with CAGC→ACTA in −35pos region
pKKG25	P2	1-345	pEJ417 with P1 A−12C mutation
pKKG26	P1	1-345	pEJ449 with TTG→GAC in −35hom region
pKKG27	P2	50-224	pEJ431 with A−12C mutation
pKKG28	P2	104-224	pEJ435 with A−12C mutation
pKKG29		50-105	54-bp ds oligonucleotide MEDUP between XbaI and HindIII sites in pEJ414
pKKG30		50-161	112-bp PCR product from oligonucleotides MEDUP and UP1 into PmlI site in pEJ414

^aInsert locations are according to the sequence in Fig. 1, with the first base of that sequence being 1.

^bds, double-stranded; SDM, site-directed mutagenesis; CDS, coding sequence.

TABLE 2.

Oligonucleotides used in this study

Namea	Sequence	Useb
BEGNCO	CGAAGCTTCCATGGTGCCTCTCCTGTGG	PCR
CDSRECA	CTAGAGCACGCGCTGGATCCGGACTATGCCAAGAAGCTCGGTGTCGACACCGATTCGCTGCTGGTT	Cloning
DISTPR1f	CTAGACGCGGCGTGTCACACTTGAATCGAACAGGTGTTCGGCTACTGTGGTGATCATTCGGA	PCR and cloning
DISTPR2f	CTAGACGCGGCGTGTCACACTTGAATCGAACGGGCGTTCGGCTACTGTGGTGATCATTCGGA	Cloning
DOWN	GCTCTAGAGGCCAACCGACCGATACC	PCR
FARUPf	CTGCGACGCCGAAAGGTCAGATCCGGGCCGGTGAGCACGCCGGATCCGGCCAGGCTAGCG	PCR and cloning
MEDUPf	CTAGACCAGGCTAGCGGTGTTGAGCAGATCGTCGGTGATCCGGACCAGCCGCGCACGCA	PCR and cloning
P1MUTf	GTGGCTGTCTCTAGTGTCACGGCCAAC	SDM
P2MUTf	CAGGTGTTCGGCTCCTGTGGTGATCATTC	SDM
PROXPR1f	CTAGATCATTCGGAGCAGCCGACTTGTCAGTGGCTGTCTCTAGTGTCACGGCCAACCGA	PCR and cloning
RNAP1Xba	TATCTAGAGGTGATCATTCGGAGCAG	PCR
RNAP2Spe	GACTAGTAACCTTTGCCGTAACTCTTC	PCR
RNAP3Xba	GATCTAGACCAGGCTAGCGGTGTTGAG	PCR
RNAP4Spe	TACTAGTGAGACAGCCACTGACAAGTCG	PCR
SOSMUTf	CACTTGAATCGAACGGGCGTTCGGCTCCTGTG	SDM
UP1f	CTAGAGTCGGGCCGCACCGCCGCCAGGGCGTTCGACGCGCCGACGAGCGCGGACGCGATGTTT	PCR and cloning
UP2f	CTAGAGGGCGTTCGACGCGCCGACGAGCGCGGACGCGATGTTT	Cloning
VREV	ATTAGGCACCCCAGGCTTTACACT	PCR
XBANCOf	GAGCGGGCTTTTTTTTGTACCATGGGTCGGGCCGCAC	SDM
XBATOWTf	CGGACGCGATGTTGCCACACGCGGCGTGTCAC	SDM

^aFor primers whose names end in “f,” the sequence of the corresponding “r” primer is the complement of the sequence given here.

^bSDM, site-directed mutagenesis.

Induction conditions.

To induce DNA damage in M. smegmatis transformants, mitomycin C (0.2 μg ml⁻¹) was added to a 10-ml aliquot of a culture at an A₆₀₀ of 0.4 to 0.5 and incubated for 5 h; an equal volume of the same culture was incubated for the same period without any addition to provide an uninduced control. For M. tuberculosis, a similar procedure was used except that 30-ml aliquots of a culture at an A₆₀₀ of 0.3 to 0.4 were used and the incubation time following addition of mitomycin C was 24 h, as a longer time is required in this species for full induction (30). Following this, the bacteria were harvested, washed three times in Z buffer (27) without β-mercaptoethanol, and stored as a pellet at −20°C.

Preparation of cell extracts and β-galactosidase assays.

Cell lysates were prepared and protein and β-galactosidase assays were performed as described previously (8). In the case of M. tuberculosis lysates, the supernatants were filtered through a low-binding Durapore 0.22-μm-pore-size membrane filter (Ultrafree-MC; Millipore) to ensure complete removal of bacteria before removal from containment facilities. The specific activity, in units per milligram of protein, was calculated with the formula defined by Miller (27).

RNA isolation.

Commercially available kits were used for the isolation of total RNA (Hybaid Ribolyser Blue kit) from M. tuberculosis cultures treated with mitomycin C (0.2 μg ml⁻¹) for 24 h. Contaminating DNA in the RNA preparations was digested with RNase-free DNase (Roche), and the RNA was subsequently cleaned up by using an RNeasy minikit (Qiagen). RNA concentrations were determined spectrophotometrically at 260 nm.

Primer extension analysis.

Primers were end labeled with [γ-³²P]ATP at their 5′ termini by means of T4 polynucleotide kinase as described in the primer extension kit (Promega). Labeled LACR (8) primer (1 pmol) was annealed to 50 μg of total RNA at 58°C for 1 h. Extension was carried out with avian myeloblastosis virus reverse transcriptase (Promega) according to the manufacturer's instructions at 42°C for 1 h. The products were ethanol precipitated, resuspended in 5 μl of Tris-EDTA (pH 7.5) plus 5 μl of loading dye (98% [vol/vol] formamide, 10 mM EDTA, 0.1% xylene cyanol, 0.1% bromophenol blue), and denatured at 90°C for 10 min before being separated alongside sequencing reactions performed on plasmid pEJ417 with the same primer using a Sequenase kit (USB/Amersham) on an 8% (wt/vol) polyacrylamide-urea gel.

RESULTS

Previous studies on the expression of the M. tuberculosis recA gene have shown that it is expressed from two promoters which are regulated by different mechanisms (10). The promoter nearest to the recA coding sequence was termed P1, and the other was termed P2. The putative promoter elements for each of the promoters based on homology to known E. coli promoters are indicated in Fig. 1, along with the transcriptional and translational initiation sites. We wished to determine experimentally whether these sequences are indeed important for promoter function in mycobacteria.

FIG. 1.

Sequence of the DNA upstream of M. tuberculosis recA. The sequence of the original fragment demonstrating recA promoter activity contains two promoters. The fragment begins 310 bp upstream of recA and extends 35 bp into the coding region. The transcription start sites mapped previously (28) by primer extension are shown by the arrows. The putative promoter elements are underlined and labeled (P1 and P2), and the SOS box is boxed. The translation initiation codon is indicated by double underlining. The numbering system indicating relative positions within the fragment and used to describe the inserts contained in various reporter constructs in Table 1 is shown on the right.

The initial experiments were conducted in M. smegmatis, as the same primer extension products had previously been identified when the M. tuberculosis recA gene was cloned in M. smegmatis as were found in M. tuberculosis itself (28), suggesting that the same promoters were used in both species. The majority of the constructs were subsequently analyzed in M. tuberculosis, as we found some differences in expression in the two species.

Analysis of the P2 promoter sequence.

Upstream of the P2 transcriptional start site there are sequence motifs resembling those of various E. coli promoters to different extents. Sequences similar to the σ³² heat shock promoter consensus had been identified (9, 28) for which the −35 sequence matched the consensus quite well but the −10 sequence matched only poorly (Fig. 2a). These sequences are also similar to those of gearbox promoters (1), in this case with a better match in the −10 region than in the −35 region (Fig. 2a). Furthermore, there is a reasonable agreement with the σ⁷⁰ promoter consensus in the same region (Fig. 2a). To determine whether these sequences are important for expression from P2 in mycobacteria, we chose to examine a set of mutant sequences with changes at various positions within these motifs for alterations to the expression level.

FIG. 2.

Effect of specific mutations in the predicted promoter elements of P2. (a) Comparison of the sequence of the P2 promoter region with consensus E. coli promoters. (b) Expression of constructs bearing individual base changes in the predicted −10 and −35 regions of the P2 promoter in pEJ445 containing a mutation in the SOS box (shown in lowercase) to prevent LexA binding in M. smegmatis. The base change in each construct is indicated, with dashes representing bases identical to the control; the graph shows β-galactosidase activities. Values are means from duplicate assays of at least three independent inductions, with error bars indicating standard deviations.

The binding site for LexA is located between the putative −10 and −35 sequences for P2, so to dissociate the effects of induction and of promoter activity we initially examined the P2 promoter mutations in the context of a mutation in the SOS box which prevents LexA binding and thus renders expression constitutive (8). The control plasmid for this set of experiments was pEJ445, which contains 120 bp of DNA encompassing P2, the constitutive SOS box mutation, and an introduced XbaI site in the center of the fragment to allow cloning of different promoter-containing oligonucleotides.

Individual base changes at six positions in the −10 region and two positions in the −35 region were assessed for their effects on promoter function in M. smegmatis (Fig. 2b). Three of the mutations (P2A−12C, P2C−14A, and P2C−36A) yielded reduced expression, with the most dramatic effect being due to the change at position −12. This mutation eliminated promoter activity with only background levels of β-galactosidase activity being detected. Two of the mutations resulted in increased expression (P2G−16T and P2C−17A), with the first of these resulting in approximately fivefold-increased β-galactosidase activity. The other three changes had no significant effect on the expression level. Overall, these observations support a role for the identified motifs in promoter function.

To confirm that the P2A−12C mutation also eliminated the activity of the wild-type inducible promoter in both M. smegmatis and M. tuberculosis, this change was introduced into plasmid pEJ435 (10), which contains the same 120-bp fragment as pEJ445 described above but with all wild-type sequence. In both species of mycobacteria, this mutation again reduced the expression to background levels (10) (see Fig. 6).

FIG. 6.

Indication that there may be a third, weak promoter. Introducing the change P2A−12C in the longer constructs pEJ419 and pEJ431 resulted in a low-level residual promoter activity in both M. smegmatis and M. tuberculosis, in contrast to the effect of the same mutation in the smaller clone pEJ435, where expression is eliminated. The inverted black triangle represents the mutation A−12C in P2. The β-galactosidase activity determined with (grey bars) or without (black bars) exposure to the DNA-damaging agent mitomycin C (0.2 μg/ml) is shown. Values are means from duplicate assays of at least three independent inductions, with error bars indicating standard deviations.

Analysis of the P1 promoter sequence.

Putative promoter elements similar to the −10 and −35 sequences of E. coli σ⁷⁰ promoters had been identified for P1 (28), but these two motifs were much closer together (9 bp) than in E. coli (16 to 19 bp) (35). It was not known whether the mycobacterial RNA polymerase recognized this sequence with its unusual positioning or a different sequence in the normal location.

We initially sought to confirm that the −10 motif had been correctly identified. Individual base changes at three positions in the −10 region, along with a triple change, were made in each of the P1 reporter clones pEJ418 and pEJ449 (10) and assessed for their effects on promoter function in M. smegmatis. pEJ418 contains 100 bp of upstream DNA which does not include P2, while pEJ449 contains 310 bp of DNA upstream of recA with P2 inactivated by the P2A−12C mutation described above and yields higher levels of expression (a phenomenon which is addressed further below). While the changes at positions −10 (P1T−10G) and −13 (P1T−13G) had little effect in either plasmid, both the triple change and the single change at position −12 (P1A−12C) eliminated promoter activity in pEJ418 (pKKG19 and pKKG21) (Fig. 3a), resulting in background levels of β-galactosidase activity. However, when the same P1 mutations were introduced into pEJ449, which already carries the P2 null mutation, a low residual expression of β-galactosidase was observed in induced cultures (pKKG15 and pKKG17, respectively) (Fig. 3b). A similar difference in the behavior of the P1A−12C mutation in pEJ418 and pEJ449 was observed in M. tuberculosis (Fig. 3). This observation suggested either that one of the promoter mutations was not a complete null mutation or that a third weak promoter might exist; these possibilities are investigated below.

FIG. 3.

Effect of specific mutations in the predicted promoter elements of P1. The expression of constructs bearing the indicated base changes in pEJ418 (a) and pEJ449 (b) in M. smegmatis and M. tuberculosis is shown. The base changes in each construct are indicated, with dashes representing bases identical to the control. The β-galactosidase activity determined with (grey bars) or without (black bars) exposure to the DNA-damaging agent mitomycin C (0.2 μg/ml) is shown. Values are means from duplicate assays of at least three independent inductions, with error bars indicating standard deviations. nt, not tested.

Having confirmed that the proposed −10 motif did function as part of the promoter, we addressed the question of whether the motif with homology to E. coli −35 sequences (−35hom) or the motif actually located at the −35 position (−35pos) was more important for promoter function in mycobacteria. To ensure that the mutation we examined would disrupt the relevant motif maximally, we changed three bases in −35hom and four bases in −35pos in pEJ449. In both M. smegmatis and M. tuberculosis, the disruption of −35pos (pKKG24) had a relatively small effect on the induced levels of expression, which remained 72 and 70%, respectively, of that seen with pEJ449. In contrast, mutation of −35hom (pKKG26) caused much greater decreases in the expression level, to 26% in M. smegmatis and to 12% in M. tuberculosis (Fig. 3b). Thus, it appears that in M. tuberculosis a sequence motif located between −25 and −30, and not one around −35, is important for the promoter function of the recA P1 promoter.

Expression from the P1 promoter is enhanced in the presence of flanking DNA.

As mentioned above, the two P1 reporter constructs pEJ418, which begins at position −53 with respect to the P1 transcription start site, and pEJ449, which runs from position −264, yielded substantially different expression levels when assayed in both M. smegmatis and M. tuberculosis (Fig. 3). In an attempt to identify the regions of DNA required for optimal expression, the length of the DNA fragment contained in the reporter vector pEJ414 was progressively shortened relative to that in pEJ449 first at one end and then at the other, while maintaining the mutation inactivating P2, and the expression directed by each construct was assessed with and without DNA damage. It should be noted that the level of expression of all the P1 constructs was significantly higher in M. tuberculosis than the corresponding expression in M. smegmatis.

Initial limited deletion at both ends of the insert, yielding pEJ507, resulted in expression similar to that directed by pEJ449 (Fig. 4). In this construct the 5′ end of the cloned fragment begins at position −214 relative to the P1 transcription start site and the 3′ end is coincident with the translation start site of the recA gene at position +50. Further reduction in the amount of cloned DNA at the 5′ end, to position −160 in pEJ512 or to position −97 in pEJ514, while maintaining the 3′ end point at the translation start site of the recA gene, resulted in a progressive decrease in the expression level (Fig. 4) such that the expression directed by the smallest construct was markedly less than that by the largest of the three clones with this 3′ end point. Reducing the length of the cloned DNA compared with that in pEJ507 at the 3′ end to position +6, while maintaining the 5′ end at −214, in pEJ498 also caused a decrease in the expression level (Fig. 4). In this case this may be due to altered stability of the RNA transcript, as these two constructs produce different lengths of transcribed recA RNA. Again, progressive deletion at the 5′ end to positions −160 in pEJ463, −97 in pEJ465, or −50 in pEJ420, while keeping the 3′ end fixed at position +6, resulted in further decreases in expression (Fig. 4) with more marked effects seen in M. tuberculosis than in M. smegmatis. This analysis revealed that maximal expression from P1 required the presence of DNA between positions −214 and +50 relative to the P1 transcription start site.

FIG. 4.

Effect of varying the amount of flanking DNA on P1 promoter activity. The extent of the DNA fragments cloned in the lacZ reporter plasmid is indicated schematically on the left. The inverted black triangle represents the mutation (A−12C) in P2 which inactivates the P2 promoter. The β-galactosidase activity determined with (grey bars) or without (black bars) exposure to the DNA-damaging agent mitomycin C (0.2 μg/ml) is shown for both M. smegmatis and M. tuberculosis. Values are means from duplicate assays of at least three independent inductions, with error bars indicating standard deviations.

Expression from the P2 promoter is stimulated by flanking sequences.

The original construct demonstrating P2 promoter activity, pEJ419 (10), contained a 201-bp PCR fragment of DNA beginning at position −168 and extending to +33 relative to the P2 transcription start site cloned in the lacZ reporter plasmid pEJ414 (30). Shortening the 3′ end of the insert so that it ended at position +7 while maintaining the 5′ end (pEJ431) resulted in an equivalent (in M. smegmatis) or even higher (in M. tuberculosis) expression level following induction. A series of constructs bearing progressive deletions of the 5′ end of the cloned DNA was then constructed, and the expression directed by them was compared. Truncation of the 5′ end of the insert to position −114 in pEJ435 (10) resulted in similar expression levels. However, deletion of a further 20 bp to yield an insert beginning at position −94 (pEJ535) resulted in a decrease in expression by a factor of about 2, while deletion to position −51 (pEJ421) caused an even greater reduction in expression (Fig. 5). Inserting a random sequence (which lacked intrinsic promoter activity; data not shown) taken from within the coding sequence of M. tuberculosis recA upstream of the promoter in pEJ421 in either the sense (pEJ481) or the antisense (pEJ482) direction caused a small stimulation in expression which may be a consequence of moving the promoter further from the transcriptional terminators present in the reporter vector to prevent read-through from vector promoters. The expression level directed by pEJ535 was similar to that seen with these control constructs. Thus, these results suggest that the 20-bp sequence located between positions −114 and −95 plays a role in stimulating expression from P2.

FIG. 5.

Effect of varying the amount of flanking DNA on P2 promoter activity. The extent of the DNA fragments cloned in the lacZ reporter plasmid is indicated schematically. The inverted black triangle represents the mutation (A−12C) in P1 which inactivates the P1 promoter. The dotted arrow indicates the inclusion of a random control sequence taken from the coding region of the recA gene. The β-galactosidase activity determined with (grey bars) or without (black bars) exposure to the DNA-damaging agent mitomycin C (0.2 μg/ml) is shown for both M. smegmatis and M. tuberculosis. Values are means from duplicate assays of at least three independent inductions, with error bars indicating standard deviations.

As we wished to use equivalent clones differing only in which promoter was active to assess the contribution of each promoter to expression (see below), we analyzed expression from a clone in which the 5′ end was extended to the end of the sequence under investigation, i.e., to position −217, while maintaining the 3′ end point at +7 (pEJ627). Surprisingly, this construct directed a higher level of expression than any of the P2 constructs previously tested (Fig. 5), suggesting that there may also be sequences enhancing expression between positions −217 and −168. We also tested the effect of extending pEJ431 at the other end by using a construct in which the 5′ end was fixed at −168 and the 3′ end was extended to the end of the sequence under investigation (pEJ626) in the presence of the P1 A−12C mutation, which appeared to eliminate P1 activity. Again, an increase in the expression level was observed (Fig. 5), but in this case it might represent an effect on the stability of the RNA transcript produced, as the amount of transcribed recA RNA differs between these two constructs. Finally, we combined the extension at the 3′ end with that at the 5′ end in pKKG25; this clone directed the highest level of expression of all the P2 constructs tested (Fig. 5), with the individual effects seen with pEJ626 and pEJ627 appearing to be additive.

Is there a third promoter?

One possible explanation for the increased expression from reporter clones containing longer inserts is that these fragments include an additional promoter. Further, the incomplete blockage of expression when both the P1A−12C and the P2A−12C mutations were present together in the full-length fragment studied (pKKG17) (Fig. 3), despite the total elimination of promoter activity by P1A−12C in pEJ418 and by P2A−12C in pEJ435, raised the possibility that a third, although weak, promoter for recA might exist in the extra DNA present in the full 310 bp of upstream sequence compared with the smaller clones.

To try to locate such a promoter, the P2A−12C mutation was introduced into the larger P2 clones pEJ419 (which has extra sequence upstream and downstream compared with pEJ435) and pEJ431 (which has extra sequence upstream compared with pEJ435). In both cases, the residual expression was apparent (Fig. 6) in both M. smegmatis and M. tuberculosis, indicating that any third promoter must lie upstream of P2. However, when various fragments of DNA from this region were examined directly for the presence of a promoter by inserting them in the reporter vector pEJ414, only background levels of β-galactosidase activity were obtained (Table 3). Furthermore, only background expression was found with a clone (pEJ447) containing the final 51 bp of the sequence under study (Fig. 1 and data not shown), demonstrating that no further promoter was located downstream of P1.

TABLE 3.

Assays to test for a third promoter

Clone	Insert location	β-Galactosidase activity (U/mg of protein)a
		M. smegmatis		M. tuberculosis
		U	I	U	I
pEJ414	Vector	2.1 ± 0.23	2.6 ± 0.26	4.0 ± 0.1	5.0 ± 0.4
pEJ422	1-60	2.2 ± 0.13	2.5 ± 0.22	2.1 ± 0.9	2.9 ± 0.5
pKKG29	50-105	1.9 ± 0.20	2.9 ± 0.65	1.6 ± 0.3	1.7 ± 0.2
pEJ432	104-161	2.2 ± 0.12	2.5 ± 0.12	3.2 ± 0.5	4.5 ± 0.3
pKKG30	50-161	2.2 ± 0.17	2.1 ± 0.38	3.7 ± 0.8	4.2 ± 1.7
pEJ628	1-161	2.1 ± 0.15	3.4 ± 0.21	3.9 ± 2.2	5.7 ± 2.1

^aβ-Galactosidase activity was determined with (I) or without (U) exposure to the DNA-damaging agent mitomycin C (0.2 μg/ml). Values are means ± standard deviations from duplicate assays of at least three independent inductions.

Thus, no further promoters could be identified, demonstrating that the enhanced expression from the longer P1 and P2 reporter clones was not due to the presence of additional promoters. It may be that some residual low-level expression results from the P2A−12C mutation in the context of additional upstream DNA, but this is only on the order of 7% of that of the wild-type P2 sequence. Thus, the available evidence is consistent with there being only two promoters for M. tuberculosis recA.

Contribution of each of the promoters to recA expression.

To assess the contribution of the individual promoters to recA expression, we wished to use the reporter clones pEJ449 for P1 and pKKG25 for P2, as these constructs gave maximal activity for each promoter and had sequences flanking the promoters in common. In fact, each of these clones contained the same fragment of DNA as pEJ417 apart from a single base change introduced to inactivate the promoter not under study. Before using these clones to compare the expression level directed by each promoter, we established that in the context of this longer fragment expression was being driven primarily by a single promoter in each case. This was determined by primer extension analysis of RNA from induced cultures of M. tuberculosis transformed with pEJ417, pEJ449, and pKKG25 individually by using a primer located within the reporter gene lacZ to eliminate background from the native recA gene. While two primer extension products were obtained from M. tuberculosis carrying pEJ417 as expected, M. tuberculosis carrying pEJ449 yielded only the smaller product corresponding to P1, and M. tuberculosis carrying pKKG25 gave only the longer product corresponding to P2 (Fig. 7). Thus, within the limits of detection by this assay, pEJ449 and pKKG25 each report expression directed by a single promoter.

FIG. 7.

Confirmation that only a single promoter is active in pEJ449 and pKKG25. Primer extension analysis used RNA isolated from induced cultures of M. tuberculosis carrying pEJ417 (lane 1), pEJ449 (lane 2), or pKKG25 (lane 3) with primer LACR located in the lacZ reporter gene. Lanes G, A, T, and C show sequencing reactions performed on plasmid pEJ417 with the same primer. P1 and P2 indicate the transcription start sites from P1 and P2, respectively. Both P1 and P2 are active in pEJ417, whereas only P1 is active in pEJ449 and only P2 is active in pKKG25.

Comparison of the expression of β-galactosidase activity directed by each of these reporter plasmids in M. smegmatis clearly shows that in this heterologous organism P2 is the dominant promoter (Fig. 8), being responsible for approximately 5 times the level of expression directed by P1. In contrast, in M. tuberculosis the levels of expression following induction are similar for pEJ449 and pKKG25 (Fig. 8), indicating that both promoters play equally important roles in the expression of recA following DNA damage in the native organism.

FIG. 8.

Contribution of P1 and P2 to recA expression. Expression from pEJ449 (P1) and pKKG25 (P2) was compared in M. smegmatis and M. tuberculosis. The β-galactosidase activity determined with (grey bars) or without (black bars) exposure to the DNA-damaging agent mitomycin C (0.2 μg/ml) is shown. Values are means from duplicate assays of at least three independent inductions, with error bars indicating standard deviations.

As the two recA promoters are regulated by different mechanisms (10), they may respond to DNA damage differently. This possibility was investigated with pEJ449 and pKKG25 by following the kinetics of induction by mitomycin C (0.2 μg/ml) in M. tuberculosis, where both promoters are important for recA expression following DNA damage. However, the increase in expression of β-galactosidase with time followed the same pattern for each promoter (Fig. 9), with only a slight increase up to 6 h followed by a more rapid increase at the same rate for each promoter.

FIG. 9.

Kinetics of induction of P1 and P2 in M. tuberculosis. Expression from the P1 promoter clone pEJ449 and the P2 promoter clone pKKG25 was assessed following various times of exposure to mitomycin C (0.2 μg/ml). Values are means from duplicate assays of at least three independent inductions, with the error bars indicating standard deviations.

DISCUSSION

In this study, we localized the promoter elements of each of the two M. tuberculosis recA promoters by a mutational analysis. We found that at both promoters changing the native A at position −12 to C severely impaired promoter function. This A occurs in a region of similarity between the two otherwise dissimilar promoters at positions −14 to −8: CTAGTGT at P1 and CTACTGT at P2. In E. coli, a TA dinucleotide in a similar position is conserved in around 80% of promoters recognized by σ⁷⁰ (24) and has been shown to be important for recognition of the promoter by RNA polymerase (14, 33). In addition, an A at position −11 in the promoters of the gal operon is vital for the formation of a transcriptionally active open complex (23). We speculate that the A at −12 in the M. tuberculosis recA promoters may perform similar functions.

The analysis of the P1 promoter has shed light on the importance of elements in addition to the −10 motif in mycobacteria. It has previously been reported that mycobacterial promoters can accommodate a large variety of sequences at the −35 region (2, 19). Our results are in agreement with this, as changing four bases at the −35 position of P1 had only a small effect on expression. However, a major reduction in promoter activity resulted from altering three bases in a motif closer to the −10 motif. Although it is possible that this could reflect inhibition of binding of a regulatory protein, if this were the case we might expect to see an effect only in induced cultures, whereas the expression of both uninduced and induced samples was affected. Thus, it seems more likely that at the recA P1 promoter the mycobacterial RNA polymerase makes important contacts in the region around −28, rather than −35 as is the case for most bacterial RNA polymerases.

It was shown recently in E. coli that the sigma factor domains which interact with the −10 and −35 sites (regions 2.4 and 4.2, respectively) are located too close to each other for this interaction to occur in the free sigma factor and that these domains move apart on binding core RNA polymerase (5). Furthermore, this movement, and hence the ability to recognize the −35 motif, is dependent on the “flexible flap” domain of the β subunit, so that in its absence very little transcription occurs from standard −10/−35 promoters (21). Those authors hypothesized that factors which affect the interaction of the β-flap with sigma factors might permit recognition of promoters with suboptimal spacers. Thus, this may be how the unusual spacing between the motifs at the recA P1 promoter is recognized, as the sequence of the rpoB gene of M. tuberculosis is highly conserved compared with that of E. coli, particularly in the β-flap region.

The P2 promoter elements bear some similarity to different classes of E. coli promoters: heat shock, gearbox, and housekeeping promoters (Fig. 2). Of the mutations introduced into the −10 region of the P2 promoter the one with the most severe effect (A−12C) was in a base conserved in gearbox and housekeeping promoters but not in heat shock promoters, perhaps suggesting that the P2 promoter is not a heat shock promoter. Of two other mutations which matched the gearbox consensus only, one (G−15T) had no effect and one (G−16T) resulted in increased promoter activity. Thus, the P2 promoter did not appear to be of the gearbox type. In addition, there was no significant change in promoter activity with growth phase (data not shown), whereas gearbox promoters exhibit increased activity in stationary phase (1). Thus, although the mutations studied do not allow an unequivocal conclusion to be reached, it may be that the P2 promoter is of the housekeeping type or alternatively that the similarity of parts of the P2 promoter to the E. coli promoters is fortuitous.

It was particularly noticeable that the level of expression driven by each of the M. tuberculosis recA promoters was highly dependent on the amount of flanking DNA present in the reporter construct. Many of the constructs studied here excluded a segment of DNA from upstream of recA previously thought to be required for recA expression (28); this discrepancy was found to be due to an error in the clone contained in M. smegmatis in the previous study. In this study we had access to automated sequencing and every clone was isolated from each mycobacterial species and resequenced to verify it. The requirement for flanking regions of DNA for optimum expression has also been reported for some other mycobacterial promoters. One example is the M. tuberculosis katG gene, which required a 155-bp region located 300 bp upstream of the coding region for maximal expression (29). Another is the Mycobacterium leprae 18-kDa-protein gene, where a reporter clone containing 256 bp of DNA from upstream of the coding sequence exhibited 3 to 4 times the level of expression seen with an equivalent clone containing only 136 bp of DNA in both M. smegmatis and Mycobacterium bovis BCG, although only a single transcriptional start site was found (11).

In the case of the recA promoters, the fact that altering the length of the DNA being tested at either end had a similar effect on the level of expression suggests that the effect is not sequence specific. It seems most likely that the primary effect is simply due to the length of the native DNA influencing its topology in some way, e.g., its degree of supercoiling. It is well established that supercoiling can alter promoter activity (32). Furthermore, it has been shown that the ability of RNA polymerase from M. smegmatis to transcribe mycobacterial promoters in vitro is strongly dependent on the template being supercoiled. A relaxed plasmid substrate resulted in 100-fold-lower yields than a supercoiled plasmid, whereas only a 2-fold difference was observed when E. coli RNA polymerase was used (22).

It appears that in addition to this general effect of sequence length, there may be a more specific role for a 20-bp sequence located between positions −114 and −95 upstream of P2 in stimulating expression. Thus, one possibility is that an activating protein might bind in this region. Other potential mechanisms for the stimulation of expression by this upstream DNA come from studies with other bacterial species. In one case in E. coli, integration host factor (IHF) has been shown to bend the DNA such that the upstream sequence makes contacts with the back of RNA polymerase (13). In another example, IHF is proposed to bend the DNA to allow a distal UP element to come into closer proximity with RNAP (16). Unfortunately, the mycobacterial UP sequence has not been defined, and mycobacterial IHF binding is thought to not be sequence specific (31), so we cannot tell whether either of these factors is involved in the stimulation of expression seen with the M. tuberculosis recA promoters.

The parallel study described here of the various promoter fragments in M. smegmatis and M. tuberculosis revealed that both the M. tuberculosis recA promoters were clearly expressed at higher levels in the native species, M. tuberculosis. This suggests that there are significant differences in the efficiency of promoter recognition or of transcript initiation and/or elongation in the two species when presented with a promoter derived from M. tuberculosis. The most marked difference in expression between the two species was observed for the P1 promoter on induction. It may be that the alternative mechanism for DNA damage induction in M. tuberculosis is not fully conserved in M. smegmatis. These observations reinforce the importance of examining gene expression in the native organism.