RecA mediates MgpB and MgpC phase and antigenic variation in Mycoplasma genitalium, but plays a minor role in DNA repair


E-mail; Tel. (+1) 206 897 5350; Fax (+1) 206 897 5363.


Mycoplasma genitalium, a sexually transmitted human pathogen, encodes MgpB and MgpC adhesins that undergo phase and antigenic variation through recombination with archived ‘MgPar’ donor sequences. The mechanism and molecular factors required for this genetic variation are poorly understood. In this study, we estimate that sequence variation at the mgpB/C locus occurs in vitro at a frequency of > 1.25 × 10−4 events per genome per generation using a quantitative anchored PCR assay. This rate was dramatically reduced in a recA deletion mutant and increased in a complemented strain overexpressing RecA. Similarly, the frequency of haemadsorption-deficient phase variants was reduced in the recA mutant, but restored by complementation. Unlike Escherichia coli, inactivation of recA in M. genitalium had a minimal effect on survival after exposure to mitomycin C or UV irradiation. In contrast, a deletion mutant for the predicted nucleotide excision repair uvrC gene showed growth defects and was exquisitely sensitive to DNA damage. We conclude that M. genitalium RecA has a primary role in mgpB/C–MgPar recombination leading to antigenic and phase variation, yet plays a minor role in DNA repair. Our results also suggest that M. genitalium possesses an active nucleotide excision repair system, possibly representing the main DNA repair pathway in this minimal bacterium.


Mycoplasma genitalium is an emerging sexually transmitted pathogen associated with acute and chronic urethritis in men (Jensen, 2004), and cervicitis, endometritis, pelvic inflammatory disease and tubal factor infertility in women (Cohen et al., 2002; Manhart et al., 2003; Haggerty, 2008). When untreated or inappropriately treated, M. genitalium persists for months or even years (Iverson-Cabral et al., 2006; Cohen et al., 2007), potentially increasing the risk for sexual transmission and serious upper reproductive tract infection. Besides its importance as a pathogen, M. genitalium has the smallest genome (580 kb, encoding 482 predicted proteins) described so far for an organism capable of self-replication (Fraser et al., 1995), and thus has served as a model organism to study the minimal requirements to sustain life (Carvalho et al., 2005; Glass et al., 2006).

The molecular basis of M. genitalium pathogenesis and the mechanisms responsible for persistence are poorly understood, yet are thought to be linked in part to its complex terminal organelle. This distinctive structure mediates adherence, motility and cell division (Burgos et al., 2006; 2007; Lluch-Senar et al., 2010) and is composed of a complex array of unique proteins (Krause and Balish, 2004). Among these proteins are MgpB (also referred to as P140, MgPa or MG_191) and MgpC (also referred to as P110 or MG_192), two surface-exposed proteins essential for cell adhesion and terminal organelle development (Burgos et al., 2006). Although MgpB and MgpC are encoded in a single expression site located at the MgPa operon, certain segments of the mgpB and mgpC genes (referred to herein as mgpB/C) are found in multiple copies throughout the genome (Iverson-Cabral et al., 2006; 2007; Ma et al., 2007). These partial copies are known as mgpB repeat regions B, EF and G, and mgpC repeat region KLM, and are organized in nine distinct chromosomal locations termed MgPa repeats (MgPars 1 to 9). Overall, the MgPars are 78–90% identical to the corresponding sequences within the expression site and represent 4% of the reduced genome (Fraser et al., 1995). These observations prompted the hypothesis that recombination between sequences in the mgpB/C expression site and the MgPars could mediate MgpB and MgpC antigenic variation (Peterson et al., 1995). In addition, the distribution and architecture of the repetitive sequences within the MgPar sites also provides a mechanism to promote deletions in the mgpB/C genes, thus also contributing to MgpB/C phase variation (Burgos et al., 2006). The observation that MgpB and MgpC are among the most immunogenic proteins of M. genitalium (Svenstrup et al., 2006; Iverson-Cabral et al., 2011) sustains the notion that antigenic variation in these proteins may enable this microorganism to evade the immune response and establish persistent infections. In support of this hypothesis, we and others have demonstrated that sequence heterogeneity of mgpB/C is extensive in vitro and in vivo (Iverson-Cabral et al., 2006; Ma et al., 2007) and evolves over time in persistently infected women (Iverson-Cabral et al., 2007).

The isolation and characterization of mgpB/C clonal variants revealed that mgpB/C gene diversity is achieved through segmental and reciprocal recombination with the MgPars (Iverson-Cabral et al., 2007). Of note, this mechanism differs from other known antigenic variation systems based on DNA recombination, which are generally characterized by unidirectional gene conversion events (Vink et al., 2011). Although the specific molecular factors and mechanisms underlying mgpB/C gene variation are not known, the outcome of this diversity suggests that the general recombination apparatus of M. genitalium could be involved. However, genome sequence analysis revealed that many of the genes typically involved in bacterial recombination and DNA repair are apparently absent in this organism (Carvalho et al., 2005). Particularly intriguing is the apparent lack of genes involved in the initiation step of recombination, such as recFOR, recBCD or addAB (Rocha et al., 2005). Thus, how recombination events are initiated remains enigmatic and suggests that additional and novel factors may be required to elicit and possibly regulate the mgpB/C recombination. Nevertheless, the M. genitalium genome contains genes encoding homologues of the basic recombination enzymes required to promote strand exchange, branch migration and resolution of the Holliday junction, all of which are essential steps in recombination (Kowalczykowski et al., 1994). Among these genes are MG_339 which encodes a RecA homologue exhibiting DNA strand exchange activity in vitro (Sluijter et al., 2009), MG_358 and MG_359 which encode homologues for the RuvA and RuvB branch migration proteins (Estevao et al., 2011) and MG_352 which encodes a RecU homologue displaying Holliday junction-resolving activity on synthetic substrates (Sluijter et al., 2010). Although the in vitro activities of these proteins have been recently analysed (Vink et al., 2011), no genetic study elucidating the enzymatic machinery required to promote mgpB/C diversity in M. genitalium cells has been reported.

In the present study, we examined the role of the MG_339 (recA) gene of M. genitalium on mgpB/C–MgPar recombination and DNA repair. We demonstrate that RecA is required to promote sequence diversity at the mgpB/C locus. Similarly, we show that haemadsorption-deficient (HA-) phase variants arise predominantly by a RecA-dependent mechanism, consistent with MgpB/C phase variation mediated by mgpB/C–MgPar recombination (Burgos et al., 2006). In contrast to other bacterial species, we also show that RecA has a minor role in DNA repair, whereas nucleotide excision repair (NER) could be the main DNA repair pathway in M. genitalium.


Construction of a M. genitalium recA null mutant

The M. genitalium RecA protein is encoded at the MG_339 locus, which is flanked by the MG_338 gene (putatively encoding a lipoprotein) and the MgPa repeat region 9 (Fig. 1A). To investigate the function of RecA in M. genitalium, we constructed a recA null mutant by gene replacement. For this purpose, we produced the suicide plasmid pΔMG_339, which contains the chloramphenicol resistance gene inserted into regions flanking the MG_339 locus. A double crossover event between plasmid pΔMG_339 and the Mycoplasma genome replaces bases 427113 to 430580 by the antibiotic marker (base co-ordinates refer to GenBank Accession Number NC_000908). The resulting 3469 bp deletion includes the full-length recA open reading frame except for the 3′-terminal 23 bp (deletion of the 97.7% coding sequence), and the entire MgPar 9 (Fig. 1A). The intended deletion was confirmed by PCR amplification with selected primers (Fig. 1B) and Southern blot analyses (Fig. S1). To our knowledge, this is the first report of the successful use of the chloramphenicol resistance gene to genetically modify M. genitalium, thus expanding the selective markers available for this microorganism (Pich et al., 2006a; Algire et al., 2009).

Figure 1.

Gene replacement at the MG_339 locus. A. Genome organization at the MG_339 (recA) locus and schematic representation of the construction of a recA deletion mutant by homologous recombination. Small arrows represent the primers used for the mutant screening and dotted lines show the predicted PCR products for the wild-type (WT) and ΔrecA mutant. B. PCR confirmation of the intended deletion at the recA locus. Genomic DNA from wild-type, ΔrecA and ΔrecA+recA strains were used as a template for PCR reactions using primers 5′mg339 and 3′mg340int (PCR 1) and 5′RTMG_339 and 3′mg339 (PCR 2). Primer sequences are shown in Table S1. PCR 1 detects replacement of the recA and MgPar 9 sequences by the chloramphenicol resistance gene (cat). PCR 2 demonstrates the absence of recA gene sequences. PCR products were electrophoresed on a 1% agarose gel. Lane M: kb DNA ladder (Stratagene). C. Immunoblot analysis of wild-type, ΔrecA and ΔrecA+recA strains with polyclonal antibodies raised against M. genitalium RecA recombinant protein. Ten micrograms of whole-cell extract was applied to each lane. Apparent molecular weights in kDa are indicated on the left side. The full-length blot is presented in Fig. S2B.

The absence of M. genitalium RecA protein in the mutant was further demonstrated by immunoblot analyses using polyclonal antibodies raised against M. genitalium RecA (Fig. 1C). Interestingly, we found that these antibodies recognized two main bands on immunoblots of wild-type cell lysates (Fig. 1C), even though the protein is encoded by a single gene. One of these bands migrated slightly below the predicted molecular size of 37.4 kDa, whereas the second band was approximately 34 kDa. A third faint band that migrated slightly higher than the 34 kDa band was also observed. Remarkably, these bands were absent in the ΔrecA mutant and were restored upon complementation, thus confirming their RecA origin (Fig. 1C). To further demonstrate the specificity of our antibodies recognizing M. genitalium RecA species, we analysed M. genitalium and Escherichia coli cell lysates expressing His-tagged and non-tagged RecA variants (Fig. S2). Interestingly, only the slow migrating band of ∼ 37 kDa was detected when M. genitalium RecA protein was expressed in E. coli (Fig. S2D). In contrast, when we analysed cell lysates of a M. genitaliumΔrecA mutant expressing a His-tagged RecA derivative, we detected the three RecA protein species with the expected shift in their electrophoretic mobility (Fig. S2C). Taken together, these results suggest that M. genitalium expresses several isoforms of RecA, although further studies are needed to confirm this hypothesis.

Frequency and rate of mgpB/C gene variation

Several studies have demonstrated that variation within the mgpB/C expression site is a result of recombination with the MgPars (Iverson-Cabral et al., 2006; 2007; Ma et al., 2007), yet the frequency of these events is not known. To address this question, we developed several qPCR-based assays to estimate the frequency of selected mgpB/C variants in a given population. As depicted in Fig. 2A, this assay is based on the use of a pair of primers, one targeting a conserved region in the mgpB/C expression site, the other targeting a unique sequence in one or multiple MgPars, resulting in a PCR product only when the targeted sequence from a MgPar is translocated into the expression site (generating a mgpB/C variant). Thus, when the resulting products are quantified by real-time PCR and normalized to the number of genomes used in the assay, we can estimate the frequency of variants containing the targeted MgPar sequences within the population. Based on this strategy, we designed four different anchored qPCRs, designated B-24, EF-2478, G-135 and KLM-7 (see Experimental procedures). In combination, these qPCR assays detect variants spanning all four of the variable regions (B, EF, G and KLM) and almost all MgPar sites (except for MgPar 9 and the small MgPar 6 site) (Iverson-Cabral et al., 2007).

Figure 2.

Gene variation at the mgpB/C locus. A. Schematic representation of the qPCR-based assay to monitor mgpB/C gene variation. As an example, the KLM-7-anchored PCR is illustrated. This PCR detects new KLM-7 variants using a primer (P1) targeting a conserved region in mgpC gene and a second primer (P2) matching unique sequences (represented by vertical lines) in MgPar 7. P1 and P2 correspond to 5′MgPar7(KLM) and 3′MgPar7(KLM) primers, listed in Table S1. Thus, the KLM-7-anchored PCR detects specific variants generated by homologous recombination between MgPar 7 and the KLM variable region of the MgpC expression site. The different variable regions (B, EF, G and KLM) found in mgpB, mgpC and MgPar 7 are also indicated. B. Comparison of MG_339 gene expression profiles between G37C and ΔrecA+recA strains. Relative MG_339 transcript levels were measured by quantitative RT-PCR normalized to 16S ribosomal RNA. The mean value ± standard deviation is shown from three independent experiments assayed in duplicate. Differences in MG_339 expression were found to be statistically significant (P = 0.0032, two-tailed Student's t-test). C–F. Frequency and accumulation of KLM-7 (C), B-24 (D), G-135 (E) and EF-2478 (F) variants after propagation of G37C, ΔrecA and ΔrecA+recA strains for 2, 4, 6, 8 and 10 passages. Note the different scale for each qPCR assay. The data are presented as the mean ± standard deviation from two independent experiments, with each sample assayed in duplicate.

Using these qPCR assays, we found that the frequency of variants detected per genome in our M. genitalium clone G37C (see Experimental procedures) after two passages (seven generations each) was 2.1 × 10−4, 3.4 × 10−4, 9.1 × 10−4 and 1.6 × 10−3, for KLM-7, B-24, G-135 and EF-2478 respectively (Figs 2C–F and S3). In these experiments, the number of variants detected is well correlated with the number of donor sites targeted by each anchored PCR. Specifically, the frequency of variants for B-24, G-135 and EF-2478 which target, respectively, 2, 3 and 4 MgPars was 1.6, 4.3 and 7.6 times greater than that of KLM-7 which targets a single MgPar. Furthermore, the anchored PCR products showed a variety of sequence variants consistent with the occurrence of different recombination events between mgpB/C and the targeted MgPar sites (data not shown). Importantly, the frequency of variants detected increased over time, demonstrating that new variants arise during in vitro propagation (Figs 2C–F and S3). Consequently, we also estimated the spontaneous recombination rate in vitro, by measuring the change in frequency every two passages for a propagation period of 10 passages. The calculated rates of gene variation for each qPCR assay, after normalizing by the estimated number of generations, are shown in Table 1. These results indicate that all the variable regions constantly undergo sequence diversity. After normalizing all the qPCR assays to one donor site, we estimated that the average recombination rate in a single variable region was 4.48 (± 2.1) × 10−6 events per genome per generation. Since there are four variable regions (B, EF, G and KLM) in mgpB/C, each represented in five to eight copies in the MgPars (Iverson-Cabral et al., 2007), we estimate that there are a minimum of 28 possible recombination events, assuming a single recombination event per variable region and possible donor site. Therefore, if all variable regions recombine at a similar rate and taking this minimum number of possible exchange combinations, we hypothesize that mgpB/C gene variation occurs at rates higher than 1.25 × 10−4 events per genome per generation. These results suggest that the potential to generate gene diversity in M. genitalium is high and could be similar to other well-characterized systems of bacterial antigenic variation (Turner and Barry, 1989; Criss et al., 2005; Helm and Seifert, 2010).

Table 1.  Recombination rates at the mgpB/C locus in G37C, ΔrecA and ΔrecA+recA strains.
StrainRecombination rateaFold differenceb
  • a. 

    Number of new KLM-7, B-24, G-135 or EF-2478 variants arising per genome per generation ± standard deviation. Data were collected from two independent experiments for each strain and represent the mean of the change in frequency every two passages for a total period of culture of 10 passages and divided by the estimated number of generations.

  • b. 

    For each set of data, the recombination rate of the G37C strain has been set at 1.00, and the recombination rate for each of the other strains has been normalized to this value.

  • ND, not detected.

G37C5.79 (± 3.48) × 10−63.26 (± 4.29) × 10−61.22 (± 0.88) × 10−52.58 (± 0.41) × 10−
ΔrecA+recA2.09 (± 1.81) × 10−56.80 (± 5.63) × 10−52.34 (± 1.93) × 10−44.88 (± 3.65) × 10−43.6220.8719.2118.90

Mycoplasma genitalium RecA is required for mgpB/C gene variation

To investigate the role of RecA in mgpB/C gene variation, we determined the occurrence of mgpB/C variants in the ΔrecA mutant. By applying the qPCR-based assay we found that loss of RecA drastically reduced the frequency of mgpB/C variants, although a few recombinants were still detected (ranging from 2.6 × 10−5 KLM-7 variants to 1.1 × 10−4 EF-2478 variants; Fig. 2C–F). Moreover, in the absence of RecA we were unable to see accumulation of variants during the 10 passages of propagation (Table 1). To examine in more detail whether these variants detected in the ΔrecA mutant were genuine or PCR artefacts generated in the absence of recombinant template (Paabo et al., 1989), we adapted our anchored PCR assay to a nested PCR format as follows. We amplified the KLM variable region using primers targeting flanking conserved sequences in the mgpC gene, and used this PCR product as a template for the KLM-7-anchored PCR assay. Using this strategy, we still detected variants in the ΔrecA mutant (data not shown), reinforcing the idea that these recombinants were genuine.

Growth defects upon recA inactivation have been described for several bacteria (Sciochetti et al., 2001). To ensure that this potential phenotype did not impact the recombination rates, we also compared the growth curves of the wild-type and ΔrecA mutant. As shown in Fig. 3, the growth characteristics of the ΔrecA mutant were comparable to that of the parental strain, with doubling times of approximately 17 h.

Figure 3.

Growth curves of wild-type (WT) G37 and its derivatives strains, ΔrecA, ΔrecA+recA, ΔuvrC and ΔuvrC+uvrC. Growth was monitored over time by measuring the number of genomes by qPCR. Data are shown in a semi-logarithmic plot and represent the mean ± standard deviation of three separate experiments.

Reintroduction of a wild-type copy of the recA gene into the ΔrecA mutant restored the frequency of mgpB/C variants, and even increased the recombination rates for all the variable regions analysed (Fig. 2C–F and Table 1). To examine if variations in expression of RecA could be the cause of this recombination enhancement, we compared the quantity of recA transcripts in the G37C and ΔrecA+recA strains by qRT-PCR. As shown in Fig. 2B, the complemented mutant exhibited a ∼ 2.5-fold increase in the expression levels of recA compared to the G37C strain. Southern blot analyses demonstrated that a single transposon was inserted in the genome of the ΔrecA+recA strain, ruling out the possibility that the increase in expression could be due to the presence of multiple copies of recA (Fig. S4). Alternatively, the genomic environment of the insertion site may explain the increase in the expression of the delivered recA gene. The transposon insertion was located at repeat region G of the MgPar 4. Importantly, this insertion does not disrupt coding regions, and does not affect any of the variable regions in MgPar 4 targeted by the anchored PCR assays (Fig. S4). Similar results were also obtained when we analysed different ΔrecA+recA transformants also expressing increased levels of recA transcripts (data not shown). Taken together, these results demonstrate that RecA mediates mgpB/C gene diversity and suggest that variations in the RecA levels influence the rates of mgpB/C gene variation.

Generation of spontaneous non-adherent phase variants is dependent on RecA expression

It has been previously shown that M. genitalium HA- variants arise spontaneously at high rates (Mernaugh et al., 1993). Further characterization of these variants revealed that they originated by deletions in mgpB and mgpC genes as a consequence of recombination events with the MgPar sites, suggesting a possible mechanism for MgpB/C phase variation (Burgos et al., 2006). These observations suggest that, in addition to promoting gene variation, RecA might also be required for the generation of non-adherent phase variants. To test this hypothesis, we examined the presence of HA- variants in the G37C and ΔrecA strains. We found that the frequency of phase variants in G37C was 0.6% at passage 1 and increased to 5% after 10 passages (Fig. 4). In contrast, the ΔrecA mutant exhibited a frequency of 0.04% throughout the experiment. Complementation of the ΔrecA mutant restored its capacity to generate HA- variants at a frequency that was threefold higher than G37C at passage 1, consistent with the previous finding of enhancement of mgpB/C gene variation (Fig. 2 and Table 1). After 10 passages, the ΔrecA+recA strain showed a similar frequency of HA- variants as the G37C strain, likely indicating the achievement of equilibrium. These results demonstrate that HA- phase variants are predominantly derived by a RecA-dependent mechanism.

Figure 4.

Role of RecA in generating haemadsorption-deficient (HA-) phase variants. A. Frequency and accumulation of HA- phase variants after propagation of G37C strain, ΔrecA and ΔrecA complemented mutant for 10 passages. The frequency of phase variants was calculated by dividing the number of colonies that failed to adsorb erythrocytes by the total number of colonies analysed. The data are presented as the mean ± standard deviation from two independent experiments assayed in duplicate. The asterisk indicates a statistically significant difference when compared to G37C strain (P < 0.001, two-tailed Student's t-test). B. Representative image showing M. genitalium colonies after the HA assay. Arrows indicate phase variant colonies, which fail to adsorb erythrocytes.

Mycoplasma genitalium RecA protein plays a minor role in DNA repair

Given the central role that RecA plays in DNA repair in other organisms, we also examined the DNA repair capabilities of the M. genitaliumΔrecA mutant. First, we compared the sensitivity of wild-type and ΔrecA strains to DNA damage induced by UV light. Surprisingly, no differences were observed even at the highest dose (Fig. 5A). Subsequently, we tested the sensitivity to increasing levels of mitomycin C (MMC). This DNA-damaging agent causes DNA strand cross-links, resulting in single- and double-strand breaks, the latter requiring homologous recombination for repair. After 24 h of treatment with increasing concentrations of MMC, we observed a progressive decrease in cell viability for both strains (Fig. 5B). Although the sensitivity exhibited by the ΔrecA mutant was greater than the parental strain at higher dosages, these differences were not statistically significant. Taken together, these results suggest that RecA plays a minor role in DNA repair in M. genitalium.

Figure 5.

Survival of wild-type (WT) G37, ΔrecA, ΔuvrC and ΔuvrC complemented strains after (A) UV irradiation and (B) mitomycin C (MMC) treatment. Survivals are shown in semi-logarithmic plots and represent the mean ± standard deviation of three independent experiments. Survival of wild-type and ΔuvrC mutant strains differed significantly after UV and MMC treatment with P-values of P < 0.05 and P < 0.1 respectively (two-tailed Student's t-test). Differences between wild-type and ΔrecA mutant did not achieve statistical significance for any of the matched UV or MMC treatments.

Nucleotide excision repair may be the major DNA repair pathway in M. genitalium

Nucleotide excision repair is an important mechanism that can repair UV-induced DNA damage and other bulky lesions. In E. coli NER is executed by the cooperative action of the encoded products of the uvrA, uvrB, uvrC and uvrD genes (Van Houten et al., 2005). Homologues of these genes have also been identified in the M. genitalium genome, representing the only complete DNA repair pathway identified in this genomically limited pathogen (Carvalho et al., 2005). The minor role of RecA in DNA repair in M. genitalium led us to further characterize the DNA repair capabilities of this microorganism. To accomplish this, we constructed a null mutant of the MG_206 gene, which is predicted to encode a protein with 27% amino acid identity to E. coli UvrC. To do so, we constructed the suicide plasmid pΔMG_206, engineered to promote the replacement of 88.1% of the coding sequence of the MG_206 gene with the tetM438 antibiotic marker by homologous recombination (Fig. 6A). The intended allelic exchange was confirmed by PCR amplification with selected primers (Fig. 6B) and Southern blot analyses (Fig. S1).

Figure 6.

Gene replacement at the MG_206 locus. A. Genome organization at the MG_206 (uvrC) locus and schematic representation of the construction of a uvrC deletion mutant by homologous recombination. Small arrows represent the primers used for the mutant screening and dotted lines show the predicted PCR products for the wild-type (WT) and ΔuvrC mutant. B. PCR confirmation of the intended deletion at the uvrC locus. Genomic DNA from wild-type, ΔuvrC and ΔuvrC+uvrC strains were used as a template for PCR reactions using primers 5′mg206 and 3′mg206 (PCR 1) and 5′mg206int and 3′mg206int (PCR 2). Primer sequences are shown in Table S1. PCR 1 detects replacement of the uvrC gene by the tetM438 resistance gene. Note that the primers used in PCR 1 are the same as those employed for the complementation test, and thus two bands are detected in the ΔuvrC+uvrC strain. PCR 2 demonstrates the absence of uvrC gene sequences. PCR products were electrophoresed on a 1% agarose gel. Lane M: kb DNA ladder (Stratagene).

When propagating the ΔuvrC mutant we noticed that colonies were smaller compared to the wild-type strain (data not shown) and that the mutant also grew slower in liquid media. Comparison of the growth curves of the ΔuvrC mutant with the wild-type strain revealed a similar growth rate but a delay in the onset of the logarithmic phase (Fig. 3). This growth defect was partially complemented by the reintroduction of a wild-type uvrC copy (Fig. 3).

To test a possible role of UvrC in gene variation, we analysed the frequency of mgpB/C variants in the ΔuvrC mutant, but no significant differences were observed when compared to G37C (data not shown). The ΔuvrC mutant was then tested for survival following DNA damage induced by UV irradiation and MMC. As expected, this mutant was extremely sensitive to UV light, exhibiting a ∼ 104-fold decrease in survival relative to the wild-type strain at a UV dose of 10 J m−2 (Fig. 5A). Similarly, the ΔuvrC mutant also exhibited a greater sensitivity to MMC (Fig. 5B). For example, after treatment with 80 ng ml−1 MMC, the viability of the wild-type strain and the ΔrecA mutant was, respectively, 150- and 46-fold greater than that of the ΔuvrC mutant. Complementation of the ΔuvrC mutant with a wild-type copy of uvrC restored the UV and MMC survival to wild-type levels (Fig. 5). Indeed, an enhanced UV survival was observed in the complemented mutant strain (Fig. 5A), reinforcing the important role of NER pathway in repairing UV-induced damage. We conclude that M. genitalium has the capacity to repair its DNA and that the uvrC gene is critical to M. genitalium survival after DNA damage. These observations support the hypothesis that M. genitalium possess an active NER system, which could represent the main DNA repair pathway in this minimal bacterium.


Phase and antigenic variation are common mechanisms employed by pathogenic bacteria to overcome immune pressure or to respond to different niches and environments between or within hosts (Henderson et al., 1999; van der Woude and Baumler, 2004). Mycoplasmas possess a number of sophisticated systems to diversify their surfaces, including switching of gene expression by DNA slippage or site-specific recombination events, or promoting antigenic variation by unidirectional gene conversion events as occurs with the VlhA protein of Mycoplasma synoviae (Citti et al., 2010). Despite its reduced genome, M. genitalium also undergoes phase and antigenic variation in two of its surface proteins (MgpB and MgpC), but accomplishes this by a unique mechanism based on reciprocal recombination events (Iverson-Cabral et al., 2007). To understand the mechanisms governing MgpB/C phase and antigenic variation, we defined the role of the M. genitalium RecA protein, a pivotal enzyme required for recombination and DNA repair in other organisms. We found that RecA is required for mgpB/C sequence diversity as well as MgpB/C phase variation, yet plays a minor role in DNA repair in M. genitalium.

To undertake these studies, we developed an anchored qPCR-based assay that allows the straightforward determination of mgpB/C diversity within the population. Using this assay, we estimated that variation within mgpB/C genes occurs at rates of at least 1.25 × 10−4 events per genome per generation. This value is expected to be much higher, since our assay and rate estimates only consider a subset of the possible recombination events between the expression locus and donor sites. Therefore, to measure the full potential of mgpB/C gene variation, other unbiased approaches should be considered, such as those previously described based on large-scale DNA sequencing (Criss et al., 2005; Helm and Seifert, 2010). Regardless, our results suggest that M. genitalium may undergo gene variation at rates similar to that exhibited by other species having recombination-based antigenic variation systems. For example, Neisseria meningitidis and Neisseria gonorrhoeae exhibit rates of 1.6 × 10−3 and 4 × 10−3 events per cell per generation respectively (Criss et al., 2005; Helm and Seifert, 2010). Trypanosoma brucei, a eukaryotic parasite transmitted through insect vectors, possesses a similar degree of sequence diversity (Turner and Barry, 1989), although switching rates are drastically reduced in laboratory-adapted strains (Turner, 1997).

Our anchored qPCR-based assay has proven to be particularly useful to compare mgpB/C gene variation between wild-type and mutant strains. Using this assay we have found that mgpB/C gene variation is drastically decreased in a ΔrecA mutant, demonstrating the involvement of RecA in mgpB/C diversity. Given the recombinatorial nature of the MgPar system, this result is not surprising, since RecA is the central player in homologous recombination, promoting genetic exchange by pairing homologous DNA substrates in an ATP-dependent fashion (Kuzminov, 1999; Sluijter et al., 2009). Similarly, antigenic variation systems from N. gonorrhoeae and T. brucei are also dependent, respectively, on the action of RecA and Rad 51, the eukaryotic homologue of RecA (Koomey et al., 1987; McCulloch and Barry, 1999). However, a different scenario is found in Borrelia burgdorferi, in which the RuvAB helicase, but not RecA, is required for vlsE antigenic variation (Liveris et al., 2008; Dresser et al., 2009; Lin et al., 2009). An intriguing observation of this study was that, despite the significant reduction in recombination rate, a small but detectable number of mgpB/C variants were still detected in the ΔrecA mutant. Therefore, although a RecA-dependent mechanism appears to be the main pathway, we cannot exclude the existence of an alternative mechanism to generate mgpB/C gene variation in M. genitalium. Of note, evidence for RecA-independent recombination events has been previously reported in E. coli (Dutra et al., 2007). Similarly, we also determined that spontaneous HA- phase variants arise principally by a RecA-dependent mechanism, yet a limited number of HA- variants were detected in the absence of RecA. These results are consistent with our previous work showing that HA- variants (designated class I and class II mutants) were the result of large deletions affecting the MgpB/C locus, driven by recombination between the expression site and the MgPar sequences (Burgos et al., 2006). The origin of the few HA- variants found in the recA background could be due to spontaneous point mutations in the mgpB/C locus or other cytadherence-related genes. However, we also cannot rule out the possibility that these variants were originated by a RecA-independent recombination mechanism involving the MgPar sequences. Further analysis of these HA- variants is warranted to more clearly elucidate the mechanisms involved.

Our observation that phase and antigenic variants accumulate in vitro suggests that the mechanism triggering mgpB/C gene variation is stochastic and independent of selective pressure. However, we cannot rule out the possibility that gene variation could be enhanced by specific conditions. In this regard, we found that differences in the expression levels of recA have a direct impact on gene variation rates, indicating that gene diversity can potentially be enhanced. It is tempting to speculate that M. genitalium may be able to increase mgpB/C recombination in vivo by adjusting the RecA levels in response to the host environment. For instance, B. burgdorferi gene variation at the vlsE locus occurs exclusively in the mammalian host, suggesting the involvement of host factors in inducing or selecting variants in this microorganism (Zhang and Norris, 1998).

Another interesting observation from our study was the detection of three different RecA protein species in M. genitalium cell lysates. The possibility that these additional bands are the result of artefacts during protein electrophoresis or cell lysate preparation seems unlikely, because only one M. genitalium RecA species is detected when expressed in E. coli. Thus, the generation of these RecA species may be dependent on the M. genitalium cell metabolism. Although further work is needed to confirm this view, the expression of these potential RecA isoforms could be explained by specific posttranslational cleavage events on the full-length RecA protein. Alternatively, they may originate from additional translational start codons present in the recA coding sequence, which are recognized by the M. genitalium translational machinery. Supporting this hypothesis is the presence of two in-frame triplets downstream from the proposed ATG start codon. These triplets are ATT and TTG, encoding residues I19 and L24, and may serve as alternative initiation codons (Golderer et al., 1995), resulting in predicted products with the observed molecular sizes. These hypotheses are currently being tested and work is in progress to determine the functional relevance of the expression of these putative isoforms.

Our studies also revealed notable differences between M. genitalium and the E. coli paradigm in DNA repair. We found that inactivation of recA in M. genitalium has no effect on cell survival after UV irradiation. Similarly, only a moderate but not significant reduction in cell viability was observed after DNA damage induced by MMC to the ΔrecA mutant. Such findings are in contrast to results obtained with E. coli as well as other species, in which recA mutants exhibit an exquisite sensitivity to these DNA-damaging treatments (Kuzminov, 1999). In E. coli, RecA performs two major functions in DNA repair: (i) it mediates homologous recombination, a process particularly important for restoring stalled replication forks and for repairing double-strand breaks (Lusetti and Cox, 2002) and (ii) it plays a key regulatory role inducing the SOS response (Kuzminov, 1999). In the presence of DNA damage, the co-protease activity of RecA becomes activated and triggers the autocatalytic cleavage of the LexA repressor, leading to the co-ordinated induction of different SOS genes, including recA and genes from the NER pathway. It is unknown whether M. genitalium RecA protein conserves this co-protease activity, but genome analysis does not reveal orthologues of the lexA gene. Despite the apparent lack of a SOS system, we found that M. genitalium is proficient in repairing photoproducts induced by UV irradiation, probably through the action of the NER pathway, as suggested by the high UV sensitivity exhibited by the ΔuvrC mutant. Although we cannot exclude the possibility that NER genes are induced through a RecA-independent mechanism, our results suggest that the basal expression level of the uvr genes is sufficient to guarantee the genome stability in M. genitalium. This hypothesis is favoured by the extremely slow growth rate of this microorganism (∼ 17 h). As previously suggested (Hanawalt, 1966), a reduced growth rate may allow the NER pathway sufficient time to repair DNA damage and prevent the formation of stalled replication forks without the need to induce a rapid SOS response. This scenario would minimize the requirement of the recombinational DNA repair pathway mediated by RecA. Consistent with this idea, it has been shown that in Mycoplasma pulmonis, which has a faster doubling time (∼ 2 h) (Dybvig et al., 1989), a recA mutant is sensitive to UV irradiation (French et al., 2008).

Recombinational DNA repair is a complex pathway that requires, in addition to RecA, the action of multiple proteins (Kuzminov, 1999). Since M. genitalium apparently lacks many of these genes (Carvalho et al., 2005), it is also conceivable that recombinational DNA repair is not efficient in this organism. How M. genitalium deals with interstrand cross-links (ICLs) induced by MMC exposure is also unclear. The commonly accepted pathway to repair this severe damage is the combined action of NER and recombinational DNA repair, although an alternative mechanism involving NER and translation synthesis polymerases has also been suggested (Dronkert and Kanaar, 2001). The observation that the ΔuvrC mutant is sensitive to MMC reinforces the important role of NER in repairing MMC-induced DNA damage. However, the finding is different than that observed in E. coli, where a recombination-deficient, SOS-proficient recA142 mutant shows a similar sensitivity to a uvrB mutant and greater than a uvrC mutant (Vidal et al., 2006). One explanation for these differences could be that ICLs' repair in M. genitalium is biased to the polymerase translation repair pathway. Alternatively, M. genitalium may be unable to efficiently repair ICLs, and the higher sensitivity observed in the ΔuvrC mutant is due to the action of NER repairing other MMC-induced DNA lesions different than ICLs (Dronkert and Kanaar, 2001).

Remarkably, in a study of saturating whole-genome transposon mutagenesis, no transposon insertions were identified in uvrA, uvrB and uvrC, suggesting that the core components of the NER pathway are essential for growth in M. genitalium (Glass et al., 2006). Our successful deletion of the uvrC gene suggests that is not the case, at least for the uvrC endonuclease. Nevertheless, the growth defects observed in the ΔuvrC mutant and its severe DNA repair-deficient phenotype support the hypothesis that NER is the main DNA repair pathway to guarantee the genomic stability in M. genitalium.

In conclusion, to our knowledge, this study is the first report of the molecular requirements for mgpB/C phase and antigenic variation and provides fundamental insights on DNA repair in the context of a minimal genome. Studies are underway to investigate the role of other putative recombination enzymes in these processes, possible regulatory mechanisms for M. genitalium RecA expression, and the derivation and function of putative RecA isoforms.

Experimental procedures

Bacterial strains and growth conditions

Wild-type M. genitalium strain G37 (ATCC 33530) and its derivatives were grown in SP-4 broth (Tully et al., 1979) at 37°C under 5% CO2 in tissue culture flasks (Corning), unless otherwise indicated. SP-4 medium was supplemented with 0.8% agar (Difco) for colony development, and with chloramphenicol (15 µg ml−1), tetracycline (2 µg ml−1) or gentamicin (100 µg ml−1) during the selection of transformants. For the analysis of mgpB/C recombination we used a G37 single colony-filtered clone designated G37C, previously described elsewhere (Iverson-Cabral et al., 2007). The G37C strain is homogenous and contains mgpB/C sequences identical to those published for the G37 strain. E. coli strains TOP10 and BL21(DE3) (Invitrogen), used for cloning and protein overexpression, respectively, were grown at 37°C in LB broth or LB agar plates containing ampicillin (100 µg ml−1) and X-Gal (40 µg ml−1) as needed.

Construction of M. genitalium mutants and complementation experiments

General DNA manipulations were performed following standard procedures (Sambrook and Russell, 2001). All PCR products were obtained from genomic DNA of strain G37 using the primers summarized in Table S1.

To generate a recA null mutant, the suicide plasmid pΔMG_339 was engineered as follows. A 1100 bp fragment spanning the 5′ flanking sequence of MG_339 gene was amplified with primers 5′RAKOmg339 and 3′RAKOmg339. Similarly, a 1035 bp fragment from the 3′ flanking region was amplified with primers 5′LAKOmg339 and 3′LAKOmg339. We then amplified the chloramphenicol resistance gene (cat) under control of the Spiroplasma citri spiralin promoter (SpPr) (Lartigue et al., 2002) from a plasmid containing that antibiotic resistance gene expression cassette (sequence available from GenBank, Accession Number JX02666) using primers 5′cat-(SpPr) and 3′cat-(SpPr). The three PCR fragments generated overlapped each other by 50 bp and were joined by polymerase chain assembly (PCA) (Stemmer et al., 1995). The PCA product was amplified again using the terminal primers 5′LAKOmg339 and 3′RAKOmg339, and cloned into the pCR-Blunt II-Topo vector (Invitrogen).

To generate a null uvrC mutant the suicide plasmid pΔMG_206 was constructed as follows. A 981 bp PCR fragment spanning the 5′ flanking sequence of the MG_206 gene was amplified using primers 5′LAKOmg206 and 3′LAKOmg206, which incorporate Acc65I and EcoRI restriction sites at their 5′ ends respectively. A second 993 bp fragment containing the 3′ flanking sequence of the MG_206 gene was amplified using primers 5′RAKOmg206 and 3′RAKOmg206. These primers contained BamHI and XbaI restriction sites at their 5′ ends respectively. Both PCR fragments were cloned into an EcoRV-digested pBE (Pich et al., 2006a), excised with the corresponding restriction enzymes, and ligated together with a 2 kb fragment containing the tetM438 selectable marker and an Acc65I/XbaI-digested pBSKII+ (Invitrogen). The tetM438 selectable marker was released from the pMTnTetM438 plasmid (Pich et al., 2006a) by digestion with EcoRI and BamHI.

For the complementation of the ΔrecA mutant, the mini-transposon plasmid pMTnTetM438–MG_339 containing a wild-type recA gene was constructed as follows. A fragment encompassing the first 219 bp upstream of the MG_339 gene and its coding sequence was amplified with primers 3′mg339 and 5′mg339, which contain XhoI and EcoRI restriction sites at their 5′ ends respectively. The PCR product was cloned into an EcoRV-digested pBE (Pich et al., 2006a), excised with the corresponding restriction enzymes, and cloned into a XhoI/EcoRI-digested pMTnTetM438 vector (Pich et al., 2006a).

For the complementation of the ΔuvrC mutant, the mini-transposon plasmid pMTnGm–MG_206 containing a wild-type uvrC gene was constructed as follows. A fragment encompassing the first 503 bp upstream of the MG_206 gene, and its coding sequence was amplified with primers 5′mg206 and 3′mg206, which contain XhoI and EcoRI restriction sites at their 5′ ends respectively. The PCR product was cloned into an EcoRV-digested pBE (Pich et al., 2006a), excised with the corresponding restriction enzymes, and cloned into a XhoI/EcoRI-digested pMTnGm vector (Pich et al., 2006a).

Transformation of strain G37 was accomplished by electroporation as previously described (Pich et al., 2006a) with few modifications. Briefly, 30 µg of knockout construct DNA (pΔMG_339 or pΔMG_206) dissolved in 30 µl of electroporation buffer (8 mM HEPES pH 7.2, 272 mM sucrose) was mixed with 90 µl of cells (approximately 109 cells per millilitre) also resuspended in electroporation buffer. The mixture was transferred to a 2 mm gapped electroporation cuvette (Bio-Rad), kept on ice for 15 min and electroporated using the Gene Pulser® II electroporation system (Bio-Rad) with the Pulse Controller set at 2.5 kV and 100 Ω. The cuvette was placed on ice for 15 min and then 900 µl of SP-4 medium was added. After 3 h of incubation at 37°C, transformants were selected onto SP-4 plates containing the appropriate antibiotic. After 2 weeks of incubation at 37°C and 5% CO2, single colonies were picked, propagated in 5 ml of SP-4 containing the appropriate antibiotic and stored at −80°C. The intended genetic modifications were then confirmed by PCR using selected primers (Figs 1 and 6) and Southern blot analyses (Fig. S1). Complementation of the resulting null mutants was achieved via transposition, by electroporating 5 µg of pMTnTetM438–MG_339 or pMTnGm–MG_206 plasmids into the ΔrecA or ΔuvrC mutants, respectively, using similar techniques as described above. The insertion of a single transposon in the genome of the ΔrecA and ΔuvrC complemented strains was confirmed by Southern blot analysis (Fig. S4). The specific insertion sites were further determined by sequencing the flanking genomic DNA with 5′ Tc_seq and 3′ Tn_seq primers respectively (Table S1, Fig. S4).

Expression and purification of M. genitalium RecA and antibody production

The MG_339 coding sequence was amplified from genomic DNA of strain G37 using primers 5′ExpRecA and 3′ExpRecA, which contain XhoI and NcoI restriction sites at their 5′ ends respectively. The resulting PCR product was cloned into an EcoRV-digested pBE (Pich et al., 2006a), excised with the corresponding restriction enzymes, cloned into a XhoI/NcoI-digested pET21-d expression vector (Novagen) and transformed into E. coli BL21(DE3). Transformant cells were grown in 400 ml of LB broth at 37°C to an optical density at 600 nm of 0.6. After 3 h of induction with 1 mM IPTG, cells were harvested, resuspended in 20 ml of binding buffer (5 mM imidazole, 0.5 M NaCl, 20 mM Tris-HCl, pH 7.9), lysed by sonication in the presence of 1 mM PMSF and clarified by centrifugation. The supernatant was filtered through a 0.45 µm filter and applied to a column containing Ni-NTA-Agarose (Qiagen) pre-equilibrated with binding buffer. The column was washed with 60 mM imidazole, 0.5 M NaCl, 20 mM Tris-HCl (pH 7.9) and proteins were eluted in 1 M imidazole, 0.5 M NaCl, 20 mM Tris-HCl (pH 7.9). The eluted fraction was dialysed against 2 l of PBS and purity was estimated by SDS-PAGE and Coomassie blue staining to be > 95% (Fig. S2A).

Polyclonal antisera against M. genitalium RecA was commercially produced by immunization of rabbits with the purified recombinant RecA protein (Pacific Immunology, CA, USA). Antibody specificity was further assessed by immunoblot using E. coli and M. genitalium cell extracts expressing His-tagged and non-tagged M. genitalium RecA variants (Fig. S2C and D).

Immunoblot analyses

Cell extracts were subjected to electrophoresis through a 12% SDS-polyacrylamide gel, transferred electrophoretically to a nitrocellulose membrane (Protran BA 85) and probed for 2 h at room temperature with a 1:1000 dilution of rabbit anti-M. genitalium RecA antibodies, following standard procedures (Sambrook and Russell, 2001). Goat anti-rabbit IgG conjugated to alkaline phosphatase (Sigma) was used as secondary antibody. Antigen–antibody complexes were visualized using nitroblue tetrazolium (Bio-Rad) and 5-bromo-4-chloro-3-indolyl-phosphate (Roche) reagents according to the supplier's instructions.

Assessment of growth rates

Growth curves were obtained by monitoring the number of genomes over time by real-time quantitative PCR (qPCR). To do so, the M. genitalium G37 and mutant strains were grown in 20 ml of SP-4 broth in 50 ml conical tubes (Corning) for 10 days at 37°C under 5% CO2. The screw caps of the tubes were loosened to permit gas exchange. Cultures grown to a similar density of approximately 1 × 108 cfu ml−1 were used as inoculum for each strain. The qPCR confirmed that a similar number of bacterial cell equivalents were inoculated. Aliquots of 0.5 ml were taken every 48 h from each culture, centrifuged at 20 000 g for 15 min and the resulting pellets were maintained at −20°C until needed. DNA extraction from each sample was performed using the MasterPureTM DNA purification kit (Epicenter) according to the manufacture's instructions. The number of genomes in each sample was quantified by real-time PCR as described below for the mgpB/C gene variation assay.

mgpB/C gene variation assay

To monitor gene variation over time, G37C, ΔrecA and ΔrecA+recA strains were propagated as follows. Five millilitres of SP-4 were inoculated and cultivated for 5 days (seven generations). Cells were scraped off from the flasks in the same culture media, thereby harvesting both adherent and non-adherent cells. One millilitre from this 5 ml cell suspension was taken and centrifuged, and the pellet was stored at −20°C for further DNA extraction. Another 1 ml was kept as a frozen stock at −80°C for the haemadsorption assay (see below). Twenty-five microlitres of culture was then used to inoculate the next culture and the process was repeated for 10 passages. Each passage was performed in duplicate.

A real-time qPCR-based assay was used to determine the frequency of mgpB/C variants in each passage. Based on the genome sequence of G37 strain (GenBank Accession Number NC_000908), we designed four different anchored PCR assays designated B-24, EF-2478, G-135 and KLM-7, with letters and numbers indicating, respectively, the variable region within mgpB or mgpC genes and the MgPar sites targeted (e.g. B-24 targets recombinants between region B and MgPar sites 2 or 4). These anchored PCR assays detect specific variants in which the MgPar sequences targeted by the primers have translocated into the MgpB/C expression site. The set of primers used for each PCR assay are listed in Table S1. To normalize the number of variants to the number of genomes in each sample, we used a second qPCR targeting the 16S ribosomal RNA gene which is present in a single copy by using primers 5′RT16SrRNA and 3′RT16SrRNA (Table S1). Real-time qPCR was performed in a LightCycler instrument (Roche), using the LightCycler-FastStart DNA Master SYBR Green I Kit (Roche). PCR reactions were performed in a final volume of 20 µl containing 2 µl of DNA (approximately 1 × 107 genomes), 0.5 µM of each primer and 2 µl of 10× SYBR Green mix. The MgCl2 concentration was optimized for all reactions at 3 mM, except for KLM-7 and G-135 qPCRs, which required 4.5 and 4 mM respectively. After an initial 10 min incubation at 95°C, each qPCR assay was performed using the following amplification conditions: (i) KLM-7: 40 cycles of 95°C for 10 s, 52°C for 10 s and 72°C for 18 s, (ii) B-24 and G-135: 40 cycles of 95°C for 10 s, 56°C for 10 s and 72°C for 16 s and (iii) 16SrRNA and EF-2478: 40 cycles of 95°C for 10 s, 54°C for 10 s and 72°C for 10 s. Fluorescence readings were acquired at 79°C for KLM-7 and 77°C for the other qPCR assays. Each set of primers generated amplification reactions with efficiencies ranging from 90% to 100%. Specificity of PCR products was verified by melting curve analysis and agarose gel electrophoresis. PCR products were also cloned into an EcoRV-digested pBE (Pich et al., 2006a) and the expected sequence from five clones for each anchored PCR was confirmed by sequencing. Absolute quantification was performed by creating standard curves using ScaI-linearized pBE (Pich et al., 2006a) plasmids harbouring the PCR product. The plasmid copy number was determined spectrophotometrically using an ND-1000 spectrophotometer (NanoDrop Technologies). The qPCR reactions were performed in duplicate and output data were analysed by using the LightCycler 3.5 software (Roche).

RNA isolation and quantitative reverse transcriptase PCR (qRT-PCR)

Total RNA was extracted from 20 ml of mid-logarithmic phase cultures using the RNAaqueous Kit (Ambion) following the manufacturer's instructions. One microgram of DNAse-treated RNA (TURBO DNAse, Ambion) was reverse-transcribed by using the SuperScript III First-Strand synthesis kit (Invitrogen) with random hexamers according to the manufacturer's instructions. Parallel reactions were performed with no reverse transcriptase to control for contaminating DNA. Real-time qPCR analysis of the resulting cDNAs was performed in duplicate as described for the mgpB/C gene variation assay using 2 µl of > 1:10 diluted cDNA. Primers used for amplification of MG_339 transcripts were 5′RTMG_339 and 3′RTMG_339 (Table S1). PCR conditions were the same as those used for the reference gene (16S ribosomal RNA), which were previously described in the mgpB/C gene variation assay (see above). Arbitrary quantification of target and reference genes was determined from standard curves generated by five serial dilutions of the cDNA. The relative abundance of MG_339 transcripts for each sample was then normalized to the 16S ribosomal RNA.

Haemadsorption (HA) assay

G37C, ΔrecA and ΔrecA+recA strains were serially passaged as described above for the mgpB/C gene variation assay. Serial dilutions of cells from the first and 10th passage were grown on SP-4 plates and the resulting colonies were tested for their capacity to adsorb sheep erythrocytes (Remel) as previously described (Pich et al., 2006b). Colonies were visualized by light microscopy and the frequency of spontaneous HA-deficient variants was calculated by dividing the number of colonies that failed to adsorb erythrocytes by the total number of colonies analysed. Approximately 3000 colonies per passage were analysed for G37C and ΔrecA+recA strains, whereas ∼ 15 000 colonies were examined for the ΔrecA mutant.

UV and mitomycin C sensitivity assays

Mycoplasma genitalium G37 and mutant strains were grown to exponential phase in 20 ml of SP-4 broth. For the UV sensitivity assay, cells were washed three times in phosphate-buffered saline (PBS), scraped off the tissue culture flasks and diluted in 10 ml of PBS to obtain approximately 1 × 108 cfu ml−1. One millilitre of this cell suspension was then dispensed into several 35 × 10 mm culture dishes, which were separately exposed to increasing doses of UV radiation using a CL-1000 Ultraviolet Crosslinker (UVP). After UV exposure, cells were serially diluted and 10 µl of each dilution was spotted in duplicate onto SP-4 plates. Although no photolyase enzymes have been identified in M. genitalium, light was minimized during the whole procedure. After incubation for 14 days, colonies were counted and survival rates were calculated as compared to untreated samples.

For the MMC sensitivity assay, cells were scraped off the tissue culture flasks, suspended in SP-4 broth and aliquoted into tubes containing serial twofold dilutions of MMC (Sigma). After 24 h of MMC exposure at 37°C under 5% CO2, serial dilutions of each treatment were plated in duplicate onto SP-4 plates and survival rates were analysed as described above for the UV sensitivity assay.


This work was supported by NIH Grant R56 AI071175 ARRA to P. A. T. and the US Department of Energy Cooperative Agreement No. DE-FC02-02ER63453 at the J. Craig Venter Institute. We thank Nacyra Assad-Garcia and Carole Lartigue for assistance making the ΔrecA mutant and Stefanie Iverson-Cabral for helpful discussions and critical reading of the manuscript.