MutS2 Promotes Homologous Recombination in Bacillus subtilis

ABSTRACT

Bacterial MutS proteins are subdivided into two families, MutS1 and MutS2. MutS1 family members recognize DNA replication errors during their participation in the well-characterized mismatch repair (MMR) pathway. In contrast to the well-described function of MutS1, the function of MutS2 in bacteria has remained less clear. In Helicobacter pylori and Thermus thermophilus, MutS2 has been shown to suppress homologous recombination. The role of MutS2 is unknown in the Gram-positive bacterium Bacillus subtilis. In this work, we investigated the contribution of MutS2 to maintaining genome integrity in B. subtilis. We found that deletion of mutS2 renders B. subtilis sensitive to the natural antibiotic mitomycin C (MMC), which requires homologous recombination for repair. We demonstrate that the C-terminal small MutS-related (Smr) domain is necessary but not sufficient for tolerance to MMC. Further, we developed a CRISPR/Cas9 genome editing system to test if the inducible prophage PBSX was the underlying cause of the observed MMC sensitivity. Genetic analysis revealed that MMC sensitivity was dependent on recombination and not on nucleotide excision repair or a symptom of prophage PBSX replication and cell lysis. We found that deletion of mutS2 resulted in decreased transformation efficiency using both plasmid and chromosomal DNA. Further, deletion of mutS2 in a strain lacking the Holliday junction endonuclease gene recU resulted in increased MMC sensitivity and decreased transformation efficiency, suggesting that MutS2 could function redundantly with RecU. Together, our results support a model where B. subtilis MutS2 helps to promote homologous recombination, demonstrating a new function for bacterial MutS2.

IMPORTANCE Cells contain pathways that promote or inhibit recombination. MutS2 homologs are Smr-endonuclease domain-containing proteins that have been shown to function in antirecombination in some bacteria. We present evidence that B. subtilis MutS2 promotes recombination, providing a new function for MutS2. We found that cells lacking mutS2 are sensitive to DNA damage that requires homologous recombination for repair and have reduced transformation efficiency. Further analysis indicates that the C-terminal Smr domain requires the N-terminal portion of MutS2 for function in vivo. Moreover, we show that a mutS2 deletion is additive with a recU deletion, suggesting that these proteins have a redundant function in homologous recombination. Together, our study shows that MutS2 proteins have adapted different functions that impact recombination.

INTRODUCTION

Many processes contribute to maintaining genetic information and generating genetic variation in bacterial cells. One process, critical for the repair of DNA breaks and horizontal gene transfer, is homologous recombination. Homologous recombination has been well defined in organisms ranging from bacteria to mammals (for a review, see reference 1). The Gram-positive soil bacterium Bacillus subtilis is a naturally competent organism that can take up DNA from its surroundings and, if sufficient homology exists, incorporate the exogenous DNA into its chromosome (for a review, see reference 2). In B. subtilis, a number of proteins have been shown to function in promoting homologous recombination either by assaying for DNA uptake and integration or through the study of DNA damaging agents, called clastogens, that require homologous recombination for repair (for a review, see references 3 and 4). These approaches have helped define the RecA-dependent homologous recombination and the RecA-independent nonhomologous end-joining (NHEJ) pathways in B. subtilis (3, 5–7).

In addition to promoting recombination, many bacteria and eukaryotes contain antirecombination pathways. In general terms, antirecombination suppresses the formation of deleterious crossover species, which have the potential to decrease cell viability (for a review, see references 8–10). In eukaryotes, RecQ family helicases can unwind recombination intermediates, thereby suppressing aberrant crossover formation (11). In Escherichia coli, deletion of the helicase UvrD and the helicase RuvA from the RuvABC Holliday junction endonuclease causes the formation of toxic recombination intermediates, referred to as bimolecular recombination intermediates (12). In addition to UvrD and RuvA, the MMR pathway has been shown to suppress recombination between nonidentical DNA sequences, often referred to as homeologous recombination (13). Such an activity has been shown to provide a barrier to horizontal gene transfer by conjugation between E. coli and Salmonella enterica serovar Typhimurium (14). Biochemical evidence shows that MutS interferes with RecA-mediated strand exchange, inhibiting recombination intermediates between nonidentical or methylated substrates (15, 16), and can recruit UvrD to unwind recombination intermediates (17). Therefore, in bacteria a number of different proteins have been shown to limit hyperrecombination between identical sequences or homologous recombination between nonidentical sequences.

Interestingly, some bacteria contain a MutS paralog, MutS2, which has been shown to possess antirecombination activity (18–20). MutS2 is composed of an N-terminal region involved in DNA binding, followed by a central ABC ATPase domain, which also participates in dimerization (21, 22). The MutS2 C-terminal domain contains the small MutS-related (Smr) region, harboring a DNase I-like endonuclease domain (22). MutS2 in Helicobacter pylori (HpMutS2) and Thermus thermophilus contributes to antirecombination (19, 21). H. pylori and T. thermophilus strains lacking mutS2 are more efficiently transformed with DNA, and the T. thermophilus ΔmutS2 strain is more resistant than the wild type (WT) to mitomycin C (MMC), an antibiotic that damages DNA and requires homologous recombination for repair (19, 22). Biochemical studies have shown that HpMutS2 and T. thermophilus MutS2 (TtMutS2) both have ATPase activity that is stimulated by recombination intermediates, and TtMutS2 has endonuclease activity toward recombination intermediates, including Holliday junctions (19, 22). These data support the hypothesis that MutS2 proteins suppress homologous recombination in bacteria. Therefore, although homologous recombination is critical for DNA repair and horizontal gene transfer, bacteria also contain pathways that suppress recombination, because hyperrecombination can lead to genome instability and cell death. It is not clear, however, if the antirecombination activity shown for H. pylori and T. thermophilus MutS2 is representative of how MutS2 functions in other bacteria.

The Gram-positive bacterium B. subtilis has a mutS2 gene (23). The function of B. subtilis MutS2 (BsMutS2) in genome maintenance is unknown. We initially hypothesized that B. subtilis MutS2 participated in antirecombination, as demonstrated for H. pylori and T. thermophilus MutS2 (19, 22). Instead, here we report that MutS2 promotes homologous recombination in B. subtilis. We show that B. subtilis cells with mutS2 deleted have a lower efficiency of plasmid and chromosomal DNA transformation. In addition, deletion of mutS2 results in sensitivity to the DNA-damaging agent MMC, and we provide genetic evidence that MutS2 functions redundantly with the Holliday junction endonuclease RecU. Therefore, our results support a new function for the MutS2 endonuclease in bacteria where it functions as a Holliday junction endonuclease. In addition to our study of MutS2, we also describe a new, efficient single plasmid-based CRISPR/Cas9 genome editing system for use in B. subtilis. Our method allows for efficient removal of the plasmid after genome editing, providing a significant advancement in markerless genetic manipulation of B. subtilis.

RESULTS

Bacillus subtilis MutS2 is a MutS paralog.

In B. subtilis, MutS2 is encoded by the yshC (or mutSB) gene, referred to here as mutS2 (23). MutS2 is a MutS paralog of 785 amino acids (Fig. 1A) (18, 22, 24). Domain predictions suggest that the N-terminal region of MutS2 corresponds to DNA binding, followed by a central ABC ATPase domain. The MutS2 C-terminal domain harbors the Smr region containing endonuclease activity (18, 22, 24). Based on sequence alignment, B. subtilis MutS2 lacks the domain involved in mismatch recognition and the C-terminal unstructured region involved in β-clamp (DnaN) binding, two functions well characterized in B. subtilis MutS that are important for mismatch repair (MMR) (Fig. 1A) (25–27).

FIG 1.

MutS2 is a MutS paralog expressed throughout exponential growth and has no role in mutagenesis. (A) A predicted domain alignment of MutS and MutS2 proteins. (B) Western blot of MutS2 protein levels throughout exponential growth. Cultures were grown at 37°C, and samples of equivalent OD₆₀₀ values were taken at increasing optical densities (OD₆₀₀ from 0.8 to 5.9). Lane 1, molecular mass marker (M); 2, WT (OD₆₀₀ of 5.9); 3, ΔmutS2 control (OD₆₀₀ of 5.6); 4 to 9, samples taken at increasing OD₆₀₀ (from 0.8 to 3.8). An antibody for DnaN was used as a loading control. The lower panel is a growth curve of B. subtilis at 37°C. Data are the means from six independent replicates from two experiments, and the error bars are the standard errors of the means. (C) A trimethoprim resistance assay to measure mutagenesis. Data were plotted as averages from three independent experiments, and error bars represent the standard errors of the means.

Prior work initially characterizing B. subtilis mutS2 showed that mutS2 transcript abundance was constant during the exponential phase and decreased toward stationary phase in rich medium, using a transcriptional fusion reporter (23). To begin, we monitored MutS2 protein abundance in vivo using antiserum raised against purified MutS2 (see Materials and Methods). In rich medium, MutS2 protein abundance was stable throughout exponential growth and to the onset of the stationary phase (Fig. 1B). As a control, we show that the immune-reactive species is absent from a ΔmutS2-bearing strain. Therefore, we find that MutS2 protein abundance remains constant throughout exponential phase and into stationary phase (Fig. 1B).

We mentioned above that MutS2 lacks a recognizable mismatch binding domain and β-clamp interaction site (23) (Fig. 1A). Prior work showed that MutS2 has no impact on spontaneous mutagenesis using rifampin resistance as an indicator for mutation rate in the presence or absence of the MMR genes mutS and mutL (23). Interestingly, however, this study also reported a modest increase in transversion mutations in a strain lacking mutS2 after sequencing of the reporter. Although it seems clear that MutS2 does not function in MMR, it could impact mutagenesis through a different pathway (23). Assessing spontaneous rifampin resistance only provides a limited sampling of mutations due to the restricted group of point mutations in the rpoB gene that are able to confer rifampin resistance (28–30). As a result, the number of measurable transversion mutations is limited. Additionally, MutS homologs in eukaryotes have been shown to have more specialized roles in recognizing insertions and deletions, two mutations that are not detected in rifampin resistance assays (31).

We chose to empirically test whether the absence of MutS2 could impact mutagenesis using an assay that can detect a broader range of mutations. We thus used an assay for spontaneous trimethoprim resistance, which samples a much wider range of base pair substitutions while also sampling insertion and deletion mutations that occur in the thyA gene (32). We found that the mutation rate of a ΔmutS2 mutant strain was identical to that of the wild type, and if we combined the ΔmutS2 allele with a ΔmutL mutation, we found no difference relative to the ΔmutL strain alone (Fig. 1C). Similar results using a mutational reporter were observed in H. pylori (19). If loss of mutS2 impacted the number of transversions, we should have observed some difference in the mutation rate using trimethoprim resistance as an indicator. With these data, we conclude B. subtilis MutS2 does not function in suppressing mutagenesis.

The C-terminal Smr domain is necessary but not sufficient for mitomycin C tolerance.

Having excluded MutS2 from suppressing mutagenesis, we screened ΔmutS2 cells through several different DNA-damaging agents to determine if MutS2 was important for survival or resistance to damage (data not shown). We found that ΔmutS2 cells were reproducibly more sensitive to mitomycin C (MMC), a DNA cross-linking/alkylating agent (33–36), than the wild-type control (Fig. 2A). An MMC titration revealed that ΔmutS2 cells are approximately 10-fold more sensitive than wild-type cells during a chronic exposure (Fig. 2A). To better understand whether the sensitivity is a result of growth inhibition or survival, we used quantitative plating efficiency and survival assays. We found that during chronic exposure to MMC, the strain lacking mutS2 was 10-fold more sensitive than the wild type (Fig. 2B), whereas there was no difference in survival following acute treatment (Fig. 2C). From these data, we conclude that the absence of mutS2 results in a 10-fold sensitivity to MMC, which results from growth inhibition. To our knowledge, our observation of MMC sensitivity is the first phenotype shown for a ΔmutS2 deletion in B. subtilis. A previous study of Thermus thermophilus mutS2 found that deletion of mutS2 resulted in resistance to MMC treatment relative to the wild type, consistent with the inhibitory effect of TtMutS2 on recombination (22). Given that deletion of B. subtilis mutS2 has the opposite effect of TtΔmutS2, these results suggest that B. subtilis MutS2 promotes homologous recombination, a novel result for a bacterial MutS2 protein.

FIG 2.

ΔmutS2 cells are sensitive to mitomycin C chronic exposure. (A) Spot-titer assay of WT and ΔmutS2 cells. Cultures were normalized via OD₆₀₀ and serially diluted, and the indicated dilutions were spotted on an untreated control plate and the indicated concentrations of mitomycin C (MMC). (B) Plating efficiency assay. Cultures were normalized via OD₆₀₀, serially diluted, and plated on either LB or LB with 30 ng/ml MMC to determine CFU. Plating efficiency is the percent CFU on MMC relative to the LB control. A Mann-Whitney U test was used to determine statistical significance. An asterisk indicates a one-sided P value of <0.05. (C) Survival assay. WT and ΔmutS2 cultures were normalized via OD₆₀₀ and then treated with the indicated concentration of MMC for 30 min in saline, followed by plating for viable cells on LB. Percent survival is the percentage of CFU from the treatment relative to the untreated control.

We next asked whether MMC sensitivity could be complemented via ectopic expression of mutS2 from the chromosome. Indeed, expression of full-length MutS2 results in complementation of the MMC sensitivity observed with the deletion (Fig. 3B and C). We took advantage of the ΔmutS2 complementation assay to determine functional domains of MutS2 important for MMC tolerance (Fig. 3A). We found that ectopic expression of a MutS2 variant lacking the C-terminal Smr domain (mutS2ΔC) failed to complement the observed MMC sensitivity (Fig. 3B and C). A Western blot confirmed that MutS2ΔC was stably expressed (Fig. 3D). Conversely, expression of an N-terminal truncation (mutS2ΔN) leaving the Smr domain intact was able to partially complement the ΔmutS2 MMC phenotype (Fig. 3B and C), and Western blot analysis confirmed stable expression of the MutS2ΔN variant (Fig. 3D). From these data we conclude that the Smr domain is necessary but not sufficient to allow for wild-type growth in the presence of MMC, indicating that the Smr domain requires the N-terminal portion of MutS2 for full function in vivo.

FIG 3.

C-terminal Smr domain is necessary but not sufficient for MMC tolerance. (A) MutS2 constructs used for complementation experiments in panels B, C, and D. (B) A spot-titer assay was done as described for Fig. 2. Complementation constructs were under the control of the IPTG (isopropyl-β-d-thiogalactopyranoside)-inducible P_spac promoter. (C) Plating efficiency assay performed as described for Fig. 2 with 1 mM IPTG or without IPTG. The Mann-Whitney U test was used to determine statistical significance. ns, not significant; *, one-sided P value of <0.05. (D) Western blot using a MutS2 antibody and a DnaN antibody as a loading control. Cultures of the indicated genotypes were grown in the absence (lanes 1 to 5) or presence (lanes 6 to 8) of IPTG. The cultures were normalized via OD₆₀₀ and analyzed via Western blotting.

Genetic analysis suggests MutS2 participates in homologous recombination.

To begin to understand the contribution of MutS2 to tolerating MMC exposure, we performed a genetic analysis of the ΔmutS2 phenotype. B. subtilis contains a prophage, PBSX, that is induced following DNA damage (37, 38) and has the ability to cause cell lysis (39, 40). One potential explanation for the MMC sensitivity of the ΔmutS2 strain is that in the absence of MutS2, SOS induction has increased and PBSX prophage induction results in cell lysis, conferring the observed sensitivity. To test this idea, we developed a CRISPR/Cas9 genome editing system to delete the PBSX lysis genes (ΔxlyAB ΔxhlB) (41) or the entire 30.5-kb PBSX prophage in an otherwise wild-type strain or in a ΔmutS2 strain (see Fig. S1 to S3 in the supplemental material). The spot plate analysis showed that in both the ΔPBSX and ΔxlyAB ΔxhlB deletion strains, ΔmutS2 still conferred an ∼10-fold sensitivity to MMC (Fig. S4). We conclude that the growth inhibition of ΔmutS2 cells to DNA damage is independent of the PBSX prophage.

We asked whether MutS2 functioned in the pathways responsible for repair of MMC-induced DNA damage. MMC adducts are repaired by the combined action of nucleotide excision repair (NER) and homologous recombination (for a review, see references 42 and 43). We combined ΔmutS2 with a uvrA deletion to determine if MutS2 contributed to the NER pathway. Because uvrA-deficient strains are very sensitive to MMC, we used a lower dose than that used for the experiment shown in Fig. 3. If MutS2 functions in NER, we would expect a strain lacking uvrA (ΔuvrA::spc) to have the same sensitivity as a strain carrying both the ΔuvrA::spc (44) and ΔmutS2 alleles. Instead, we found that strains with ΔmutS2 and ΔuvrA::spc are more sensitive to MMC than strains with the ΔuvrA::spc allele alone (Fig. 4A). These data support the hypothesis that ΔmutS2 sensitivity to MMC is independent of NER.

FIG 4.

ΔmutS2 sensitivity is independent of uvrA and recU and shows no additive effects with recA. (A to C) Plating efficiency assays performed as described for Fig. 2 with the indicated genotypes and the indicated concentrations of MMC. The Mann-Whitney U test was used to determine statistical significance. ns, not significant; *, one-sided P value of <0.05. Note that lower concentrations of MMC were used for this experiment and ΔmutS2 cells do not show a phenotype.

We then tested if ΔmutS2 was epistatic or additive with a recA deletion (ΔrecA::loxP). We found that the ΔmutS2 ΔrecA::loxP double mutant was as sensitive to MMC as the ΔrecA::loxP single mutant (Fig. 4B). In this experiment we used an even lower dose of MMC, because recA-deficient strains are exquisitely sensitive to MMC. Taken together, these results suggest that during repair of MMC adducts, MutS2 contributes to a step that is recA dependent and that MutS2 does not participate in a step of MMC repair involving NER. Given that MutS2 proteins in other organisms have been shown to bind Holliday junctions (19, 22, 45), we chose to test whether there was a genetic interaction with the primary Holliday junction endonuclease RecU (46). To this end, we tested a strain either lacking recU (ΔrecU::erm) or lacking both mutS2 and recU in the plating efficiency assay. In contrast to the results with recA, we found that deletion of mutS2 in a strain lacking recU increased sensitivity to MMC (Fig. 4C). Taken together, our data suggest that MutS2 functions in a pathway dependent on RecA yet independent of RecU.

MutS2 is important for DNA transformation with plasmid and chromosomal DNA.

To further test if MutS2 promotes homologous recombination, we employed a transformation efficiency assay. Transformation of naturally competent B. subtilis cells with plasmid or chromosomal DNA is dependent on the function of several homologous recombination proteins, although the requirements for chromosomal DNA integration relative to plasmid assembly and maintenance are not identical (4, 47). Thus, if MutS2 functions by promoting homologous recombination, as the MMC phenotype suggests, we would expect to observe decreased transformation efficiency using plasmid and/or chromosomal DNA. Indeed, we found that a strain lacking mutS2 had decreased transformation efficiency with both plasmid (Fig. 5A) and chromosomal DNA (Fig. 5B). In order to determine whether the transformation efficiency phenotype was specific to mutS2 or both mutS1 and mutS2, we tested transformation efficiency in a strain lacking mutS1 (ΔmutS) and a strain lacking both mutS genes. We found that cells with ΔmutS1 had decreased plasmid transformation efficiency that is independent of mutS2 (Fig. S5). In contrast, we found that in the absence of mutS1 there is no change in chromosomal DNA transformation efficiency, and deletion of mutS1 in a mutS2 deletion strain resulted in a transformation efficiency indistinguishable from that of a mutS2 single mutant. As a result, it appears that decreased chromosomal transformation efficiency is specific to mutS2.

FIG 5.

MutS2 is important for plasmid transformation, and mutS2 deletion is additive with a recU deletion in chromosomal DNA (chrDNA) transformation. (A) Transformation efficiency assay using plasmid DNA. (B) Transformation efficiency using chromosomal DNA. (C) Transformation efficiency assay using chromosomal DNA in the presence of 1 mM IPTG. Transformation efficiency is the ratio of transformants (CFU) to viable cells (CFU). The Mann-Whitney U test was used to determine statistical significance. ns, not significant; *, two-sided P value of <0.05.

To further examine the role of MutS2 in transformation, we tested chromosomal DNA transformation in a strain lacking the Holliday junction endonuclease gene recU and the recU mutS2 double mutant. Our hypothesis that MutS2 functions in a pathway independent of RecU predicts that these mutants should have an additive effect on transformation efficiency. Indeed, we found that deletion of mutS2 or recU resulted in a 4.9-fold or 11.6-fold decrease in transformation efficiency, respectively (Fig. 5B). An additive effect predicts an approximately 57-fold decrease in transformation efficiency, and strikingly the double mutant had a 57-fold decrease in transformation efficiency (Fig. 5B). These data further support the hypothesis that MutS2 functions independently of recU, potentially as a Holliday junction endonuclease (see Discussion).

As a control, we examined whether the complementation analysis performed in Fig. 3 would yield similar results in the chromosomal DNA transformation efficiency assay. Indeed, expression of full-length MutS2 resulted in a transformation efficiency that was not significantly different from that of the WT (Fig. 5C). We also found that expression of the variant lacking the C terminus was not significantly different from the mutS2 deletion strain (Fig. 5C). Further, expression of the C-terminal Smr domain alone resulted in a partial complementation similar to the results obtained in Fig. 3 (Fig. 5C). Together, these results indicate that the C-terminal Smr domain requires the N-terminal ATPase domain to fully function in vivo.

MutS2-GFP is diffusely distributed in vivo.

Our previous research investigating MutS1 has shown that B. subtilis MutS1-GFP forms foci that are recruited to the site of DNA synthesis (25–28, 48). Further, homologous recombination proteins and nucleotide excision repair proteins have been shown to form foci and localize to the nucleoid, respectively, following MMC treatment (44, 49). We fused a monomeric version of green fluorescent protein (GFP) (gfpmut3) to mutS2, resulting in a gfp-mutS2 allele, which is functional in the spot-titer assay for MMC sensitivity (Fig. 6A), expressed from the native locus as the only source of MutS2 in vivo (Fig. 6B). Fluorescence microscopy showed that MutS2-GFP was diffusely distributed in cells and did not show focus formation during normal growth or following growth in the presence of MMC (Fig. 6C). We conclude that MutS2 is diffusely distributed in B. subtilis cells, and that MutS2 is not specifically recruited to sites of DNA damage in quantities detectable using bulk fluorescence.

FIG 6.

Functional GFP-MutS2 fusion has a diffuse cytosolic distribution. (A) A spot-titer assay showing that GFP fused to the N terminus of MutS2 expressed from the native locus is functional. (B) Western blot using a MutS2 antibody shows that GFP-MutS2 is expressed and we do not detect degradative products, indicating MutS2 is intact. (C) Fluorescence microscopy to visualize GFP-MutS2 under vehicle control and 100-ng/ml MMC treatment conditions (red, membranes stained with FM4-64; green, GFP-MutS2; scale bar is 2 μm).

DISCUSSION

Our investigation of B. subtilis MutS2 function led to the observation that cells lacking MutS2 are growth inhibited by the natural antibiotic mitomycin C (MMC). Although the sensitivity is a modest 10-fold during chronic exposure, it is highly reproducible and can be complemented via ectopic mutS2 expression, suggesting the sensitivity is indeed dependent on the absence of mutS2. This observation is a novel phenotype for a bacterial mutS2 mutant and is the opposite of the phenotype observed in T. thermophilus (22), suggesting that MutS2 promotes recombination in B. subtilis.

Our finding that a ΔmutS2 mutant is sensitive to MMC provided an assay to further characterize the genetic interactions of mutS2. Our initial genetic analysis resulted in the creation of a CRISPR/Cas9 genome editing system on a single, temperature-sensitive plasmid that results in markerless mutations that can be quickly introduced into many genetic backgrounds. Although this is not the first report of CRISPR/Cas9 genome editing in Bacillus subtilis (50, 51), the system described here does not require an inducer for Cas9 expression and still allows for efficient removal of the plasmid editing system. We further demonstrated that the novel mutS2 phenotype depends on recA and does not depend on uvrA or recU. These results suggested that MutS2 functions in homologous recombination in a pathway independent of RecU.

To further test the hypothesis that MutS2 promotes homologous recombination, we assayed transformation efficiency with both plasmid and chromosomal DNA. Our hypothesis predicts a decrease in transformation, and indeed we observed decreased transformation efficiency with plasmid and chromosomal DNA. We further demonstrate that the Smr endonuclease domain is necessary but not sufficient for chromosomal transformation and MMC tolerance. These results indicate that the C-terminal Smr domain requires the N-terminal portion of the protein for complete function in vivo. Additionally, we performed an epistasis analysis with recU using chromosomal DNA transformation. Similar to our results with MMC sensitivity, the double mutant had a more severe reduction in transformation efficiency than either single mutant. In fact, this assay revealed an additive effect of recU and mutS2 deletions. With these results we suggest that MutS2 functions as a back-up Holliday junction endonuclease, although other mechanisms could also explain our results.

Previous studies of bacterial mutS2 have observed a role in inhibiting recombination. Our findings suggest that different lineages of bacteria are capable of adapting specific proteins to their repair requirements. Interestingly, although a BLAST search finds 99% and 83% coverage in comparisons of B. subtilis MutS2 to T. thermophilus and H. pylori, respectively, the percent identity is only 35 and 28, respectively (data not shown).

Despite the sequence divergence, we speculate that MutS2 performs a similar function in these bacteria: a nuclease that is capable of processing recombination intermediates. HpMutS2 has been shown to have ATPase activity that is stimulated by a four-way junction (Holliday junction) and by a fork or “Y” structure (19). A more recent study found that HpMutS2 has nuclease activity toward both single-stranded DNA and Holliday junctions that is dependent on two nuclease sites in the protein: the Smr domain and an N-terminal LDLK motif (45). Interestingly, the LDLK motif is not conserved in B. subtilis or T. thermophilus. Studies using TtMutS2 have found that MutS2 has an Smr domain-dependent nuclease activity (52). Further analysis revealed that TtMutS2 recognizes Holliday junctions and D-loops (22). Unfortunately, we were unable to obtain active preparations of BsMutS2 to gain a more mechanistic understanding of its role in promoting recombination. Based on our genetic results, we suggest that BsMutS2 has activity toward Holliday junctions or D-loops and MutS2 functions as a secondary Holliday junction endonuclease in addition to RecU.

MATERIALS AND METHODS

Bacteriological methods and chemicals.

All B. subtilis strains used in this study are isogenic derivatives of PY79. All strains, plasmids, and primers used in this study are listed in the supplemental material (see Tables S1 to S3 in the supplemental material). Construction of all strains and plasmids is detailed in the supplemental methods. For all experiments with B. subtilis, strains were struck out from frozen stocks on LB agar plates or LB agar plates containing the appropriate antibiotics and incubated at 30°C overnight. Antibiotics were used at the following concentrations: spectinomycin, 100 μg/ml; chloramphenicol, 5 μg/ml; and erythromycin, 0.5 μg/ml. Mitomycin C was obtained from Fisher Scientific (Fisher BioReagents) and used at the final concentrations indicated in the figures.

Trimethoprim resistance assay.

Trimethoprim resistance assays were performed essentially as described previously (32). Briefly, an LB (10 g/liter NaCl, 10 g/liter tryptone, 5 g/liter yeast extract) culture for each strain was inoculated and grown at 37°C until an optical density at 600 nm (OD₆₀₀) of about 1.0. Cultures were back diluted 1:500 into LB with 200 μM thymidine and grown for 4 h at 37°C. Cultures were then serially diluted and plated on LB with 200 μM thymidine for viable cells, and minimal medium plates containing trimethoprim (1× S7₅₀ salts [53, 54], 1% glucose, 0.1% glutamate, 0.1 μM tryptophan, 0.1 μM phenylalanine, 200 μM thymidine, 1× metals [53, 54], 0.2% Casamino Acids, 1.8% agar, and 34 μM trimethoprim) were used to select for thyA mutants. For viable cells, 100-μl aliquots of the 10⁻⁶ dilution were incubated on plates at 30°C overnight. For trimethoprim plates, 100 μl of the 10⁰ (PY79 and PEB11) or 10⁻¹ (PEB112 and PEB118) dilution was plated and incubated at 45°C overnight. A total of 17 independent cultures were used for each strain. Mutation rate was calculated via fluctuation analysis using the online tool FALCOR at http://www.mitochondria.org/protocols/FALCOR.html as described previously (55, 56).

Western blotting.

For Western blot analysis, cultures were grown in LB at 37°C to an OD₆₀₀ of ∼1 to 3 or as indicated for Fig. 1B. A 1-ml volume with an OD₆₀₀ of 1 from each strain was pelleted via centrifugation (10,000 × g for 5 min at room temperature [RT]). Cell pellets were resuspended in 50 μl RE buffer (50 mM Tris-Cl, pH 7.5, 150 mM NaCl, 50 mM EDTA, 1× Roche Complete EDTA-free protease inhibitor cocktail [04693132001], and 10 mg/ml lysozyme) and incubated at 37°C for 20 min. The samples were then iced briefly (2 to 3 min), followed by the addition of SDS to a final concentration of 1% and SDS loading dye to 1×. Samples were incubated at 100°C for 10 min, followed by SDS-PAGE (8% for Fig. 1B and S5B; 4 to 15% gradient gel for Fig. 3C). Proteins were transferred to a nitrocellulose membrane using the Bio-Rad Trans-Blot turbo apparatus using the manufacturer's kit and manufacturer's recommendations. Membranes were blocked using 5% milk in Tris-buffered saline with Tween 20 (TBST) overnight at 4°C or for 1 h at RT. Primary antibodies were used at a 1:5,000 dilution (both MutS2 and DnaN) in 2% milk in TBST for 1 h at RT. Secondary antibodies (LI-COR goat anti-rabbit conjugated to an infrared dye) were used at a 1:15,000 dilution in 2% milk in TBST for 1 h at RT. The membranes were then visualized using the LI-COR Odyssey infrared imaging system. For Fig. 3, cells were lysed via sonication in RE buffer without lysozyme on ice. All Western blot analyses were performed from two independent replicates.

Spot-titer assays.

A single colony was used to inoculate an LB culture and grown to an OD₆₀₀ of ∼3. Cultures were normalized via OD₆₀₀ and serially diluted, and 5-μl aliquots of the 10⁻¹ to 10⁻⁶ dilutions were spotted onto LB agar plates or plates with the indicated concentration of MMC. LB agar with a vehicle control was used as well and showed results identical to those of the untreated control (data not shown). Dilutions were spotted using a BioTek precision XS pipetting robotics system. Plates were incubated at 30°C overnight (16 to 20 h). All spot-titer assays were performed at least twice.

Plating efficiency assays.

A single colony was used to inoculate an LB culture grown to an OD₆₀₀ between 0.5 and 1. Cultures were normalized via OD₆₀₀, and a scorable dilution resulting in approximately 30 to 300 colonies was plated on LB for viable cells and on plates containing the concentration of MMC indicated in the figures. Plating efficiency was determined by dividing the number of CFU on MMC plates by the number of CFU on LB plates and multiplying by 100 to obtain a percentage. All experiments were performed in biological triplicate, and each triplicate was performed at least twice. The data plotted are the means of the pooled data, and the error bars represent standard errors of the means.

Acute MMC survival assays.

A single colony was used to inoculate an LB culture and grown to an OD₆₀₀ between 0.5 and 1. An equivalent OD₆₀₀ was taken and cells were washed in 0.85% NaCl. Cells were then resuspended in saline as a no-treatment control or the concentration of MMC shown in Fig. 2C and treated at 37°C for 30 min. Cells were pelleted and the supernatant removed and resuspended in saline. As described above, a scorable serial dilution was then plated on LB to determine the number of surviving CFU. The percent survival is the number of CFU from the indicated MMC treatment divided by the number of CFU from the no-treatment control multiplied by 100. Each experiment was performed in triplicate at least twice. The data plotted are the means of the pooled data, and the error bars represent the standard errors of the means.

Transformation efficiency assays.

Transformation efficiency assays were performed by transforming cultures grown to natural competence. Competent cell cultures were prepared by inoculating a 2-ml culture of LM medium (LB with 3 mM MgSO₄) with a single colony and grown at 37°C to an OD₆₀₀ between 1 and 1.5 and then transferring 20 μl of the LM culture into 500 μl prewarmed MD medium (1× PC buffer [10× PC buffer is 107 g/liter K₂HPO₄, 60 g/liter KH₂PO₄, 10 g/liter trisodium citrate · (H₂O)₅], 2% glucose, 50 μg/ml tryptophan, 50 μg/ml phenylalanine, 11 μg/ml ferric ammonium citrate, 2.5 mg/ml potassium aspartate, 3 mM MgSO₄) and incubated at 37°C for 4 h. Ten nanograms of pPB153 plasmid DNA (purified from E. coli strain MC1061, which is recA⁺, resulting in oligomeric plasmid preparations, using the Qiagen plasmid HiSpeed Midi kit with DNA eluted in double-distilled water) or 25 ng of chromosomal DNA (purified using a spin column method; see the supplemental methods) from PEB43 was added to cultures, followed by incubation at 37°C for 1.5 h. The cultures were serially diluted, and 100-μl aliquots of the 10⁻⁶ dilution were plated for viable cells on LB agar. For transformants, 100 μl of a scorable serial dilution as described above was plated on LB with 100 μg/ml spectinomycin (plasmid transformations) or LB with 5 μg/ml chloramphenicol (chromosomal DNA transformations), followed by incubation at 30°C overnight. Transformation experiments were performed in at least biological triplicate, and each experiment was performed three times.

CRISPR/Cas9 editing plasmid construction.

Construction of a CRISPR/Cas9 genome editing plasmid was done in two cloning steps (Fig. S1). First, the spacer was incorporated into the CRISPR array via restriction digestion and ligation cloning. The plasmid was digested with BsaI-HF (NEB) in CutSmart buffer at 37°C. Meanwhile, the spacer was prepared to be ligated into the plasmid. First the oligonucleotides, ordered with the correct overhangs (57), were annealed by mixing the oligonucleotides at 10 μM each in 1× annealing buffer (10 mM Tris-Cl, pH 7.5, 100 mM NaCl, and 0.1 mM EDTA). Annealed oligonucleotides were phosphorylated using T₄ polynucleotide kinase (PNK; NEB). A 50-μl reaction mixture was assembled with 1× T₄ DNA ligase buffer, 10 U T₄ PNK, and 1 μM annealed oligonucleotides and incubated at 37°C for 30 min. T₄ PNK was heat inactivated at 65°C for 20 min. A 20-μl ligation reaction mixture was assembled using 1× T₄ DNA ligase buffer, 40 to 100 ng plasmid DNA, 25 nM phosphorylated and annealed oligonucleotides, and 400 units T₄ DNA ligase (NEB). The ligase reaction was performed at room temperature for 2 to 3 h. The ligation was then used to transform chemically competent (CaCl₂ method) E. coli cells (Top10; Thermo Fisher Scientific): 10 μl of the ligation reaction mixture was used to transform 60 to 100 μl competent cells. The plasmid was verified via Sanger sequencing. The resulting plasmid (essentially a targeting plasmid) was then used to create the editing plasmid.

The editing plasmid was constructed using Gibson Assembly (58) of the following four (or more) PCR amplicons that were gel purified from an agarose gel: (i) the vector backbone of pPB41, amplified using oPEB217/oPEB218; (ii) the Cas9/CRISPR array containing the spacer incorporated in the targeting plasmid, amplified using oPEB232/oPEB234; (iii) the upstream portion of the editing template, which contains overlaps for pPB41 at oPEB217 and for the downstream portion of the editing template (see pPB50 and pPB51 for examples); and (iv) the downstream portion of the editing template, which contains overlaps for the upstream portion of the editing template and for pPB41 at oPEB232 (see pPB50 and pPB51 for examples). A 10-μl Gibson Assembly reaction mixture was assembled using 1× Gibson Assembly master mix, 40 to 100 ng of the vector backbone of pPB41, 40 to 100 ng of the Cas9/CRISPR array containing the spacer, and about 20 to 40 ng of each portion of the editing template. The Gibson reaction was performed at 50°C for 90 min. The reaction was then used to transform chemically competent MC1061 E. coli cells. Clones were verified using Sanger sequencing at the University of Michigan Core Facility. The result of this second step yields a complete editing plasmid that can be used to efficiently manipulate the B. subtilis genome.

CRISPR/Cas9 genome editing.

After construction of the editing plasmid, fresh B. subtilis competent cells (500 μl [see “Transformation efficiency assays” above]) were transformed with 200 to 600 ng of the editing plasmid that had been purified from a strain of E. coli that yields multimers (e.g., MC1061). Transformations were plated on LB agar plates with spectinomycin and incubated at 30°C overnight. The next day, 6 to 24 isolates were colony purified by restreaking on LB agar plates with spectinomycin. The editing plasmid contains a temperature-sensitive origin of replication (derived from pDR244). To evict the plasmid, isolates were restruck on LB and incubated at 42 to 45°C for 10 to 16 h. Isolates were then screened for loss of the plasmid by restreaking on LB agar plates with and without spectinomycin and incubating at 42 to 45°C overnight. The isolates that were found to be spectinomycin sensitive were further screened via PCR to determine if the isolate had the expected deletion.

Live-cell microscopy.

For live imaging experiments, a plate grown overnight at 30°C was washed using S7₅₀ minimal medium supplemented with 2% glucose essentially as described previously (25, 26, 59). The OD₆₀₀ was measured and cells were diluted to an OD₆₀₀ of 0.1. Cultures were incubated at 30°C for 2 h (OD₆₀₀ of 0.2). Cultures were treated with either a vehicle control or 100 μg/ml MMC and incubated at 30°C for an additional 2 h (OD₆₀₀ of ∼0.5). An aliquot of cells was removed, followed by incubation with the membrane stain FM4-64 at 1 μg/ml. Cells were then placed on 1% agarose pads made of 1× Spizizen's salts (53, 54). Cells were imaged using an Olympus BX61 microscope as described previously (25, 26, 59).