Identification of the Acidobacterium capsulatum LexA box reveals a lateral acquisition of the Alphaproteobacteria lexA gene

The genetic material of bacteria is often exposed to damaging agents that can compromise its integrity and the viability of the cell. To overcome this problem, many genes are employed in different pathways to detect and repair DNA damage; alternatively, some lesions in the DNA are tolerated, and thus allow a certain degree of mutation. Most of these genes constitute specific networks that act in a coordinated manner, and one such network, present in most Bacteria clades, is the LexA-regulated SOS system.

In Escherichia coli, the SOS regulon consists of more than 40 genes, whose functions are involved in DNA replication, DNA repair, mutagenesis and control of the cell cycle (Fernández de Henestrosa et al., 2000; Courcelle et al., 2001). Under normal conditions, the SOS network remains repressed by the product of the lexA gene (Walker, 1984). The E. coli LexA protein binds specifically to a 16 bp palindrome (CTGTN₈ACAG) named the LexA box, located in the promoter region of SOS genes, whose expression is normally down-regulated to a basal level (Little et al., 1981). After DNA-damage-mediated stalling of the replication fork, RecA acquires an active conformation (RecA*) by binding to ssDNA (Sassanfar & Roberts, 1990). RecA* mediates the autohydrolysis of the E. coli LexA repressor between residues Ala⁸⁴ and Gly⁸⁵, resulting in the expression of the SOS genes (Little, 1991; Little et al., 1980). Once DNA damage is effectively repaired, RecA* concentration decreases, LexA ceases to be hydrolysed, and the non-cleaved LexA synthesized de novo once again inhibits the transcriptional expression of the SOS network.

The evolutionary history of the SOS system across the domain Bacteria is complex. First of all, and with very few exceptions (such as Cytophaga, Flavobacterium, Bacteroides and Epsilonproteobacteria), the lexA gene seems to be present in all bacterial families in which drastic genetic reduction (such as that observed in Buchnera, Rickettsia, Aquifex and Mycoplasma, among others) has not taken place. Furthermore, a clear relationship seems to exist between the LexA-binding sequence and the branching order of several bacterial phylogenetic groups from their common ancestor (Mazón et al., 2004b). Therefore, and in agreement with the bacterial genome sequences known so far, several groups can be established. The first comprises all the Gram-positive bacteria and closely related phyla, such as the Cyanobacteria and the green non-sulfur bacteria. All these phyla possess the GAACN₄GTTC sequence or a related motif as their specific LexA-binding sequence (Winterling et al., 1998; Fernández de Henestrosa et al., 2002; Mazón et al., 2004a). On the other hand, Alphaproteobacteria have a GTTCN₇GTTC direct repeat as their LexA box (Fernández de Henestrosa et al., 1998; Tapias & Barbé, 1999), whereas Beta- and Gammaproteobacteria (Erill et al., 2003) share the same LexA-binding sequence as that found in E. coli (CTGTN₈ACAG). However, an intermediate phylum (Fibrobacter) between Proteobacteria and Cyanobacteria possesses a LexA box that seems to represent a transition from the Gram-positive one to that seen in E. coli (Mazón et al., 2004b).

Several DNA repair genes, including some belonging to the canonical LexA regulon (recA and uvrA), are usually involved in the resistance of bacteria to acidic pH (Raja et al., 1991; Thompson & Blaser, 1995; Hanna et al., 2001). However, some bacterial species whose environmental growth conditions are very acidic show a constitutive expression of the recA gene that is mediated either by the absence of a LexA box on the recA promoter (e.g. Acidobacillus thioferrooxidans; Ramesar et al., 1989) or by the lack of a lexA regulatory gene (e.g. Helicobacter pylori; Tomb et al., 1997).

Acidobacterium capsulatum is a Gram-negative, acidophilic, chemo-organotrophic bacterium, containing menaquinone, of which very few isolates have been cultured (Kishimoto et al., 1991; Hiraishi et al., 1995). This micro-organism has been proposed to belong to a new bacterial phylum that branched approximately at the same time as the phylum Fibrobacter, and significantly earlier than the Proteobacteria (Ludwig et al., 1997; Quaiser et al., 2003). Despite its extreme growth conditions, no data exist on the DNA repair system of A. capsulatum. For this reason, and taking advantage of the fact that the A. capsulatum genome is being sequenced (), here we report the cloning of the A. capsulatum lexA gene, the purification of its product, and the characterization of its binding site, in an effort to unravel both the composition of its LexA regulon and the evolution of this gene network across the domain Bacteria.

Bacterial strains, plasmids, oligonucleotides and DNA techniques.
Bacterial strains and plasmids used in this work are listed in Table 1. E. coli was grown in LB medium at 37 °C and antibiotics were added to the cultures at the concentrations reported in Sambrook et al. (1992). A. capsulatum ATCC 51196 was grown in M-269 medium with 0·1 g l¹ yeast extract (pH 3·5) at 30 °C (Hiraishi et al., 1995). E. coli cells were transformed with plasmid DNA as described by Sambrook et al. (1992). Oligonucleotide primers for PCR, restriction enzymes, T4 DNA ligase and polymerase, and the DIG DNA labelling and detection kit, were from Roche. Genomic DNA of A. capsulatum was obtained by standard procedures (Sambrook et al., 1992) from a 10 ml culture grown at 30 °C in M-269. The DNA sequence of all PCR-mutagenized fragments was determined by the dideoxy method (Sanger et al., 1977) on an ALF sequencer (Amersham-Pharmacia).

Table 1. Bacterial strains and plasmids used in this work

Identification, cloning and purification of A. capsulatum LexA.
The A. capsulatum lexA gene sequence was identified by performing a TBLASTN search of its unfinished genome at The Institute for Genomic Research (TIGR) () using Fibrobacter succinogenes LexA protein as a query. The comparison yielded a region with significant homology in part of contig number 83 (Supplementary Table S1), and we obtained this region with 1 kb flanking each side of the hypothetical lexA coding region. Primers Up-Acido-LexA (5'-CTTCTCCTGCACGCAAGC-3') and Dw-Acido-LexA (5'-GGATCCTACTTGCGTCTGAAATCG-3') (italic type shows restriction sites) were designed using this sequence, and were employed to amplify the entire A. capsulatum lexA gene and its promoter region by PCR. The resulting 813 bp fragment was cloned into a pGEM-T vector, resulting in pUA1066. This plasmid was then sequenced to check for the introduction of any changes in the DNA sequence during PCR and to confirm the database DNA sequence. A. capsulatum LexA contains all the conserved residues involved in the repressor autocleavage (Ala, Gly, Ser and Lys), as deduced from a CLUSTALW alignment of different LexA proteins (Fig. 1).

(81K): Fig. 1. CLUSTALW alignment comparison of the LexA proteins from A. capsulatum (Aca), Sinorhizobium meliloti (Sme), Rhodobacter capsulatus (Rca), Bacillus subtilis (Bsu), Fibrobacter succinogenes (Fsu) and Escherichia coli (Eco). Dark shading shows identical conserved amino acids; light shading indicates similar conserved residues. Arrows indicate the Ala, Gly, Ser and Lys residues involved in the LexA autocatalytic cleavage. The residues of the α3 helix, the DNA binding domain, are overlined.

A. capsulatum LexA protein was purified using the TALON Purification kit (Stratagene). Briefly, the coding sequence of the A. capsulatum lexA gene was amplified and cloned using primers NdeI-Acido-LexA (5'-CATATGGCTGTGACCAAGCGTC-3') and Dw-Acido-LexA, which contained NdeI and BamHI restriction sites, respectively. The resulting pUA1067 plasmid was then digested with NdeI/BamHI, and the 625 bp DNA fragment was cloned into pET15b (Novagene). The latter plasmid was transformed into E. coli BL21 (DE) Codon plus cells (Stratagene) to overproduce the LexA protein, which was purified by Co²⁺ affinity chromatography, as described previously (Campoy et al., 2002). Rhodobacter sphaeroides LexA protein, also used in this work, had been previously purified (Tapias et al., 2002).

Mobility shift assays.
LexADNA complexes were detected by electrophoretic mobility shift assays (EMSAs), using purified LexA proteins. Typical 20 µl reactions containing 10 ng DIG-DNA labelled probe and 20 nM final concentration of either pure A. capsulatum or pure R. sphaeroides LexA were incubated in binding buffer (5 mM Tris/HCl, 5 mM MgCl₂, 25 mM NaCl, 5 %, v/v, glycerol, 1 µg bulk carrier DNA and 50 µg BSA ml¹). After 30 min incubation, the reaction mixture was loaded onto a 6 % non-denaturing Tris/glycine/polyacrylamide gel. DNAprotein complexes were separated at 100 V for 1 h, and then transferred to a Biodine B membrane (Pall Gelman Laboratory), followed by the detection of DIG-labelled DNAprotein complexes, following the manufacturer's protocol (Roche).

RT-PCR analysis.
Mitomycin C-mediated induction of several genes was analysed by real-time quantitative RT-PCR of total A. capsulatum RNA with LightCycler (Roche), as described previously (Campoy et al., 2002; Bustin, 2002), and using the appropriate primers. The Titan One Tube RT-PCR system (Roche) was used, as described by Campoy et al. (2002), to perform RT-PCR experiments and to determine the transcriptional organization of the imuA-imuB-dnaE gene cassette.

Phylogenetic analysis.
Sequence data of F. succinogenes, Myxococcus xanthus and A. capsulatum were obtained from TIGR through their website at . Protein sequences for all other organisms were acquired from the Microbial Genome Database for Comparative Analysis website () and the US Department of Energy Joint Genome Institute website (). All proteins used for phylogenetic analyses are listed in Supplementary Table S1, where their locus, accession numbers and micro-organism abbreviations are also given. Alignments of protein sequences were carried out using a combined procedure to improve alignment quality. Protein sequences were first aligned through CLUSTALW version 1.83 (Thompson et al., 1994), using default (10), 25 and 5 gap-opening penalties (GOPs) for the multiple alignment stage, and generating three different alignments. These three alignments, together with a local alignment generated by the T-COFFEE lalign method, were integrated as libraries into T-COFFEE version 1.37 (Notredame et al., 2000) for optimization. The optimized alignment was then visually inspected with BioEdit version 5.0.9 (Hall, 1999) and submitted to Gblocks version 0.91b (Castresana, 2000) with the half-gaps setting and, otherwise, default parameters, in order to select conserved positions and discard poorly aligned and phylogenetically unreliable information.

Phylogenetic analyses of the refined alignments were carried out using MrBayes version 3.1.1 (Ronquist & Huelsenbeck, 2003), applying a mixed four-category Gamma distributed rate model plus proportion of invariable sites model (invgamma). MrBayes Metropolis-Coupled Markov Chain Monte Carlo runs were carried out with four independent chains for 10⁶ generations. The resulting phylogenetic trees, which are the product of four independent MrBayes runs, were plotted with TREEVIEW version 1.6.6 (Page, 1996). In all cases, only branch support values over 0·85 are shown.

Identification of the A. capsulatum LexA-binding site
EMSAs with the purified A. capsulatum LexA were used to determine the ability of this protein to bind to its own promoter. The addition of increasing concentrations of LexA to a fragment extending from 149 to +112 of the A. capsulatum lexA gene promoter (with respect to its predicted translational starting point) produced one retardation band whose intensity was directly related to the amount of protein used (Fig. 2b). To locate more precisely the LexA-binding region, the A. capsulatum lexA promoter was divided into two fragments (designated LexA1 and LexA2), which were obtained by PCR amplification with suitable oligonucleotides. The two fragments were DIG end-labelled and then used as probes in an EMSA (Fig. 2a). A stable DNAprotein complex was observed when both fragments were incubated in the presence of purified A. capsulatum LexA (Fig. 2c). These data indicate that the region to which the LexA protein binds was located downstream of position 35 of the lexA promoter.

(20K): Fig. 2. (a) Fragments amplified from the A. capsulatum lexA promoter region and used in electrophoretic mobility shift assays (EMSAs). In all cases, positions are indicated with reference to the translational start point of the lexA gene. (b) Electrophoretic mobility of the LexA1 probe in the presence of decreasing concentrations of LexA protein. (c) Effect of a 500-fold molar excess of unlabelled A. capsulatum lexA promoter (PlexA) and pBSK(+) plasmid DNA (pBSK) in the migration of the LexA1 probe in the presence of 20 nM A. capsulatum LexA protein. Migration of LexA2 probe in the presence (+) or absence () of A. capsulatum LexA is also shown. (d) Effect of a tetranucleotide substitution (LexA2-Mut) in the migration of the LexA2 probe (Mut) in the presence of 20 nM pure A. capsulatum LexA protein. The mobility of the wild-type LexA2 probe (Wt) in the presence (+) and absence () of LexA protein is also shown.

A close inspection of this region revealed the presence of an Alphaproteobacteria-like LexA-binding site (GTTCN₇GTTC), extending from positions 29 bp to 15 bp with respect to the lexA translational start codon (Fig. 2a). To test whether this sequence was actually the A. capsulatum LexA box, an EMSA experiment with a LexA2-derivative fragment, in which the first GTTC motif was substituted by an AGGT tetranucleotide, was performed. Fig. 2(d) shows that the GTTC motif was needed by A. capsulatum LexA to successfully bind its promoter. Furthermore, R. sphaeroides LexA protein was also able to bind the wild-type LexA2 fragment, but not its mutant derivative (Fig. 3a). Likewise, the A. capsulatum LexA protein was able to bind the wild-type R. sphaeroides recA promoter but not a mutant derivative with a modified LexA Box (Fig. 3b). These experiments clearly demonstrate that R. sphaeroides and A. capsulatum recognize the same LexA-binding sequence. In this respect, it is worth noting that the same LexA box (GTTCN₇GTTC) is present in the lexA gene promoter region of Solibacter usitatus sp. Ellin6076, another member of the phylum Acidobacteria whose sequence is being completed by the Joint Genome Institute ().

(20K): Fig. 3. (a) Migration of both A. capsulatum LexA2 (Wt) and LexA2-Mut (Mut) probes in the presence of 20 nM LexA protein of either A. capsulatum (LexA Aca) or R. sphaeroides (LexA Rsp). (b) Migration of both R. sphaeroides recA wild-type (Wt) promoter and a mutant derivative (Fernández de Henestrosa et al., 1998) lacking its LexA box (Mut) in the presence of 20 nM LexA protein of either A. capsulatum (LexA Aca) or R. sphaeroides (LexA Rsp).

Analyses of the LexA-regulon gene core
Comparative analyses of the SOS system in different subclasses of the Proteobacteria (Alpha, Beta and Gamma) indicate that a common set of genes (lexA, recA, ssb, uvrA and ruvA) is directly repressed by LexA in all these subclasses, and therefore constitutes the canonical gene composition of the SOS regulon in this phylum (Erill et al., 2003, 2004). Some of these genes (lexA, recA and ruvA) are also regulated by LexA in Bacillus subtilis (Dubnau & Lovett, 2002). Furthermore, it has recently been reported that a gene (dnaE2) that encodes a second alpha subunit of DNA polymerase III, and which is found either isolated in the chromosome (Mycobacterium tuberculosis) or belonging to a multiple gene cassette (many Beta- and Gammaproteobacteria species), is also under LexA control (Boshoff et al., 2003; Abella et al., 2004; Galhardo et al., 2005). To analyse the constitution of the A. capsulatum LexA regulon, TBLASTN was employed to search the A. capsulatum genomic database for recA, dinP, uvrA, ruvAB, ssb and dnaE2, using their respective homologues in the Alphaproteobacteria. It is worth noting that, besides all the searched proteobacterial or B. subtilis lexA-regulated genes (recA, dinP, uvrA, ruvAB and ssb), two independent copies (iid1 and iid2) of the multiple gene cassette constituted by the imuA-imuB-dnaE2 genes were also detected. Furthermore, the iid1 and iid2 cassettes both constitute single transcriptional units when analysed by RT-PCR (Fig. 4).

(27K): Fig. 4. Structural arrangement of both copies (iid1 and iid2) of the A. capsulatum imuA-imuB-dnaE transcriptional unit. Positions are indicated with reference to the translational start point of the imuA gene of each operon, and mark the 5' end of the primers designed to amplify the imuA-imuB and imuB-dnaE fragments by RT-PCR. (a) RT-PCR analysis of iid1 with imuA-1up and imuB-1dw primers (lane 2) and imuB-1up and dnaE-1dw primers (lane 3); as a control, the same reactions were done using DNA (lanes 5 and 6) or RNA in a PCR reaction with imuA-1up and imuB-1dw primers (lane 1). (b) RT-PCR analysis of iid2 with imuA-2up and imuB-2dw primers (lane 1) and imuB-2up and dnaE-2dw primers (lane 2); as a control the same reactions were done using DNA (lanes 5 and 6) or RNA in a PCR reaction with imuA-2up and imuB-2dw primers (lane 3). The HinfI-digested DNA of φX174 used as a molecular weight (MW) marker is also shown.

After the identification of these genes and operons (iid1, iid2, ruvA, dinP, recA and uvrA) in the A. capsulatum database, DNA fragments containing their upstream regions were isolated by PCR using suitable oligonucleotides, and the ability of these fragments to bind the A. capsulatum LexA protein was tested. Accordingly, competitive EMSAs were carried out with the LexA2 DIG-labelled probe and a 500-fold excess of non-labelled promoter fragments of all these genes. Fig. 5 shows that only iid1, iid2 and dinP upstream regions were able to bind the A. capsulatum LexA protein. It must be noted that a sequence search revealed that there are no ORFs immediately upstream of these genes, indicating that each gene is the first one of a transcriptional unit (data not shown). In accordance with these results, A. capsulatum LexA-binding sites were only found in iid1, iid2 and dinP promoter regions (data not shown).

(15K): Fig. 5. EMSA experiments using the LexA2 probe incubated with purified LexA protein of A. capsulatum in the presence of different DNA competitors. The LexA2 probe was incubated in the presence (+) or absence () of 20 nM purified A. capsulatum LexA protein and, at the same time, in the presence of 20 nM LexA and a 500-fold molar excess of unlabelled fragments containing promoter regions from recA, uvrA, ruvA and dinP genes, and the iid1 and iid2 transcriptional units.

To further characterize the behaviour of the DNA-repair-related A. capsulatum genes identified, the effect of mitomycin C on their expression was analysed. Table 2 shows that, as expected, lexA, iid1 and iid2 transcriptional units are induced by DNA damage. It is worth noting also that, among the canonical proteobacterial LexA-regulated genes present in A. capsulatum, only recA is DNA-damage inducible in this organism, despite the fact LexA does not bind to its promoter region (Fig. 5).

Table 2. Behaviour of several DNA-repair-related A. capsulatum genes towards mitomycin exposure

Phylogenetic comparison of the A. capsulatum lexA gene
It has been shown that the phylum Acidobacteria branched long before the Proteobacteria on the evolutionary tree (Ludwig et al., 1997), and the same holds true for the phylum Fibrobacter (Griffiths & Gupta, 2001). Nevertheless, F. succinogenes possesses a LexA-binding sequence that seems to represent a transition between that of Gram-positive bacteria and that of Beta- and Gammaproteobacteria (Mazón et al., 2004b). In contrast, the A. capsulatum LexA box seems clearly related to that of the Alphaproteobacteria, which branched later than Deltaproteobacteria, a subclass in which F. succinogenes-related LexA recognition motifs can be observed (Campoy et al., 2003; Mazón et al., 2004b). This fact strongly suggested the possibility of a lateral gene transfer (LGT) event concerning the lexA gene between Acidobacteria and Alphaproteobacteria.

Accordingly, phylogenetic analyses using the LexA protein reveal a strong relationship between Acidobacteria and Alphaproteobacteria. As can be seen in Fig. 6, both types of LexA protein cluster together with a high support value. Conversely, when other proteins belonging to the Alphaproteobacteria SOS regulon (such as RecA, UvrA, Ssb and UvrD) are used for phylogenetic analysis, the placement of the Alphaproteobacteria is in concordance with the established phylogenetic location of that phylum (Ludwig et al., 1997), and is also supported by high support values (Fig. 7).

(14K): Fig. 6. LexA protein sequence tree. The light-shaded area correspondsto the phylum Acidobacteria, while dark shading designates the Alphaproteobacteria subclass. Only branch support values over 0·85 are shown. The relevant branch support value is encircled. The scale bar indicates 0·1 expected changes per site. Name abbreviations are as follows: Aca, Acidobacterium capsulatum; Ban, Bacillus anthracis; Bbr, Bordetella bronchiseptica; Bja, Bradyrhizobium japonicum; Bme, Brucella melitensis; Bsu, Bacillus subtilis; Cpe, Clostridium perfringens; Eco, Escherichia coli; Fsu, Fibrobacter succinogenes; Mtu, Mycobacterium tuberculosis; Mxa, Myxococcus xanthus; Ppu, Pseudomonas putida; Rpa, Rhodopseudomonas palustris; Rso, Ralstonia solanacearum; Sau, Staphylococcus aureus; Sco, Streptomyces coelicolor; Sme, Sinorhizobium meliloti; Son, Shewanella oneidensis; Sus, Solibacter usitatus sp. Ellin6076. Additional information for each LexA is provided in Supplementary Table S1.

(24K): Fig. 7. RecA (a), UvrD (b), UvrA (c) and Ssb (d) protein sequence trees. The light-shaded area corresponds to the phylum Acidobacteria, while dark shading designates the Alphaproteobacteria subclass. Only branch support values over 0·85 are shown. Relevant branch support values are encircled. The scale bars indicate 0·1 expected changes per site. Name abbreviations are as in Fig. 6. Additional information for each protein is provided in Supplementary Table S1.

In the last few years, different LexA-binding motifs have been described for a number of bacterial clades and, as is to be expected, most of these motifs share a substantial number of features. In particular, almost all the LexA boxes described so far present a palindromic dyadspacerdyad structure, and are monophyletic to the clade in which they have been described. Thus, a GAACN₄GTTC palindrome is the consensus sequence for Gram-positive bacteria (Winterling et al., 1998), and this motif, with slight variations, extends also to green non-sulfur bacteria (Fernández de Henestrosa et al., 2002) and Cyanobacteria (Mazón et al., 2004a). The LexA-binding motif is overspecified in terms of information content (LexA boxes are apparently too long for the number of LexA sites in the genome), and this may be the reason behind the vast diversification observed in LexA-binding motifs, which has led to the proliferation of seemingly dissimilar LexA boxes, such as the aforementioned Gram-positive one and that observed in E. coli (CTGTN₈ACAG). In spite of this, several major features can be identified among these divergent motifs, and we have previously provided experimental support for a basic evolutionary history of LexA-binding motifs based on a vertical inheritance model (Mazón et al., 2004b).

Among the different LexA motifs described so far, the Alphaproteobacteria LexA box is perhaps the most divergent and the most intriguing. First described in R. sphaeroides (Fernández de Henestrosa et al., 1998), the Alphaproteobacteria LexA box is monophyletic for this proteobacterial subclass (Erill et al., 2004), and presents a direct-repeat structure that, up to now, had not been described for any other bacterial clade. This fundamental departure from the standard palindromic structure of all other known LexA-binding motifs, together with a highly divergent protein sequence, have made it difficult to reconstruct the precise evolutionary history of the Alphaproteobacteria LexA, but we have previously shown that its most probable origin lies in an evolutionary pathway different from the one that gave rise to the Gamma- and Betaproteobacteria LexA proteins (Mazón et al., 2004b).

The identification here of an Alphaproteobacteria-like LexA box in the Acidobacteria, a lineage that is supposed to branch deeply from either the Chlamydia/Planctomyces or the Gram-positive lines (Hiraishi et al., 1995), thus represents a key piece of evidence in this respect and gives rise to several questions. On the one hand, the identification of the A. capsulatum LexA box demonstrates that the GTTCN₇GTTC LexA-binding motif existed well before the Alphaproteobacteria emerged. Based on the tendency towards diversification observed in the LexA-binding sequence, it seems very unlikely that the same motif, and precisely such a divergent motif, should have arisen independently in two different phyla. Moreover, the ability reported here of Alphaproteobacteria and Acidobacterium LexA proteins to effectively bind one another's promoters is further strong evidence of a common origin for both proteins.

Taking into account that no motif similar to that of the Alphaproteobacteria has been identified so far in any of the studied phylogenetic groups that are intermediate between the Alphaproteobacteria and Acidobacteria, the most plausible explanation for the experimental evidence reported here seems to be an LGT event. In agreement with this line of reasoning, the incongruent placement of the Alphaproteobacteria in the LexA tree, clustering precisely with the Acidobacteria at the natural branching point of the latter, gives further credence to the hypothesis that an LGT event took place between the two clades. Furthermore, the fact that studies using RecA, UvrA, Ssb and UvrD proteins locate both Acidobacterium and the Alphaproteobacteria at their accepted branching points, in accordance with observations using well-established phylogenetic markers such as the 16s RNA, suggests that the transfer proposed here occurred from Acidobacteria, or an immediate ancestor presenting the GTTCN₇GTTC motif as its LexA box, to the Alphaproteobacteria. In accordance with this line of reasoning, an alignment of several LexA proteins that are representative of different bacterial groups (Fig. 1) reveals that the Alphaproteobacteria LexA protein possesses a 33 amino acid insertion downstream of the predicted third helix (α3) of its binding domain that is not seen in any other bacterial group. This gives further credence to the direction of the LGT deduced from the LexA tree, and also supports the hypothesis that this insert does not affect the DNA binding domain (DBD) (Knegtel et al., 1995).

The data reported in this work also demonstrate that the A. capsulatum recA gene is DNA-damage inducible (Table 2), although the A. capsulatum LexA protein is unable to bind to the recA gene promoter (Fig. 5). A similar result has also been described for other bacterial species, such as M. tuberculosis, Xylella fastidiosa and Myx. xanthus, in which several DNA-repair-related genes (such as uvrA and ssb) are DNA-damage inducible, even though their expression is not directly under LexA protein control (Brooks et al., 2001; Campoy et al., 2002, 2003). However, to date, the mechanism controlling the LexA-independent DNA-damage-mediated induction of these genes is not known. Likewise, it is also not known if these bacterial species (M. tuberculosis, X. fastidiosa and A. capsulatum) possess a similar control mechanism to regulate the expression of DNA-damage-inducible genes that lack a LexA box in their promoters. Further work is necessary to understand how this lexA-independent gene induction takes place, as well as to determine its relevance in the bacterial response systems against DNA damage.

This work was funded by grant BFM2004-02768/BMC from the Ministerio de Educación y Ciencia (MEC) de España, and by the Consejo Superior de Investigaciones Científicas (CSIC). S. C. is a recipient of a postdoctoral contract from Instituto Nacional de Investigaciones y Tecnología Agraria y Alimentaria (INIA) Institut de Recerca i Tecnologia Agroalimentàries (IRTA). We are deeply grateful to Joan Ruiz and Dr Pilar Cortés for their excellent technical assistance.