Genetic Recombination in Bacillus subtilis 168: Effects of recU and recS Mutations on DNA Repair and Homologous Recombination

Bacterial strains and plasmids.

The E. coli JM103 strain was used (23). All B. subtilis strains listed in Table 1 are isogenic with strain YB886 (rec⁺ control). The ypbC (National Biomedical Research Foundation-Protein Identification Resources [NBRF-PIR] and Swiss-Prot accession no. L47648.Gb_Ba) and yppB (L47838.Gb_Ba) genes were PCR amplified with specific oligonucleotides and cloned (Table 1). The transfer of the recU:cat deletion/insertion allele (termed recU1) and recS:cat deletion/insertion allele (termed recS1) from pCB182 and pCB176, respectively, into the B. subtilis chromosome by a double-crossover event was performed as previously described (5).

Chemical treatment.

Methyl methanesulfonate (MMS) was purchased from Eastman Kodak, Rochester, N.Y. 4-Nitroquinoline-1-oxide (4NQO) was purchased from Sigma Chemical Co., St. Louis, Mo. The chemical treatment of the mutant strains was performed essentially as previously described (1).

Transformation of bacteria.

Competent cells were prepared as described by Rottländer and Trautner (25) and Maniatis et al. (23) for B. subtilis and E. coli, respectively. Plasmid transformants were selected on tryptone-yeast agar medium (25) containing chloramphenicol or erythromycin at 5 μg/ml. Met⁺ recombinants were selected by plating on minimal agar containing all nutritional requirements except methionine (25).

Recombination frequencies.

Relative transformation frequencies were used as a measure of recombination. B. subtilis competent cells (about 5.0 × 10⁷ cells/ml) were transformed according to the method of Rottländer and Trautner (25). The yield of Met⁺ or plasmid transformants was corrected for DNA uptake and cell viability as described by Alonso et al. (1). DNA uptake, which is taken as a measurement of competence, was monitored as described previously (25).

Replacement of B. subtilis recU and recS genes by use of a selectable marker.

Recently by molecular cloning, DNA sequencing, and computer homology comparisons, a B. subtilis open reading frame (ORF), ypbC, was identified as coding for a putative recombinational helicase, the gene for which is “recQ-like” (coordinates 4291 to 5781 in the sequence with accession no. L47648.Gb_Ba) (27). Furthermore, a B. subtilis region (plasmid pMP29-42) able to rescue the recG40 mutation (24) has been sequenced. The pMP29-42-borne DNA segment bears two truncated ORFs, termed “yppB” and “yppC” (27) (accession no. L47838.Gb_Ba).

To identify and clone the recG gene of B. subtilis, a PCR-amplified DNA with coding capacity for yppB or yppC was used to transform the recG40 strain. Upon transformation of the MMS-sensitive (MMS^s) recG40 strain with purified DNA fragments, we learned that only the yppB ORF fully complements the recG40 defect, since yppB encodes a protein unrelated to the E. coli recG product. To avoid confusion in the nomenclature and for the sake of simplicity, we have renamed this ORF as “recU” and the mutant allele as “recU40.”

To learn whether the putative ypbC gene codes for a product involved in homologous recombination and recombinational repair and to challenge the grouping of the B. subtilis rec genes (described above), we cloned the chloramphenicol acetyltransferase gene (cat) 239 and 415 bp downstream of the ATG codons of the ypbC and recU genes, respectively. The resulting alleles were used to replace the ypbC and recU genes.

Strains containing the expected gene substitution were obtained by natural transformation of the B. subtilis rec⁺ strain and its isogenic rec-deficient derivatives with linearized plasmid-borne recU:cat (recU1) and ypbC:cat (Table 1). The presence of the desired replacement was confirmed by PCR amplification and nucleotide sequence analysis (data not shown). In all cases, integration had occurred by a double-crossover event, as predicted for transformation of competent cells with linear plasmid DNA (reviewed in reference 13).

To analyze if the constructs were impaired in recombinational repair processes, we exposed the mutant strains to the lethal effect of MMS. Both recU1 and ypbC:cat strains were unable to form colonies on plates containing 300 (2.7 mM) and 100 (0.9 mM) μg of MMS per ml, respectively, whereas the wild-type strain formed colonies even in plates containing 450 μg (4 mM) of MMS per ml. The recU1 and ypbC:cat strains, exposed to the killing action of 10 mM MMS, were more sensitive than the otherwise rec⁺ strain (Fig. 1A and Fig. 2). The same results were obtained when the cells were challenged with 0.1 mM 4NQO (data not shown). It is likely, therefore, that the ypbC gene product is involved in DNA repair. We have renamed the recS gene and its mutant allele “recS1” (ypbC:cat). As revealed in Fig. 1A, the cells carrying the recU1 null allele are slightly more sensitive to the killing action of MMS than those of the recU40 strain, and the defect can be complemented by a plasmid-borne recU gene (coordinates 2536 to 2874 in the sequence with accession no. L47838.Gb_Ba).

Effect of DNA-damaging agents in recU and recH mutant strains.

To date in B. subtilis, 13 genes involved in homologous recombination (recA, addA, addB, recF, recL, recR, recN, recH, recP, recB, recD, recU, and mfd) have been identified (reviewed in reference 6). Genetic analysis of double-rec mutants with chromosomal and plasmid transformation has led to the classification of the rec genes, other than recA, into four epistatic groups (3). The genes involved in the α (recF, recL, recR, and recN) and β (addA and addB) groups are well characterized (7, 8, 19), but those comprising the γ (recH and recP) and ɛ (recB, recD, and recU) groups are poorly defined.

Previously, we have shown that the recU40 (ɛ group) gene in combination with mutations in the addA addB (β epistatic group) genes was more sensitive to the killing action of MMS and 4NQO than the single-mutant parent, whereas with a mutation in the recF (α group) gene, it was as sensitive as the single parent (3, 5). However, no data are available about the effect of double mutations in the γ and ɛ groups. The recH342 strain was selected as a representative of the γ epistatic group. We have constructed a recU recH double mutant and challenged these cells with the killing action of 10 mM MMS. As revealed in Fig. 1B, the recH mutation did not affect the MMS sensitivity of the recU cells.

Effects of recS, recF, recH, recU, and addAB mutations on MMS sensitivity.

B. subtilis rec⁺ and its isogenic rec-deficient derivative strains (BG431, BG435, BG429, and BG433 [listed in Table 1]) were exposed to the killing action of 10 mM MMS. The recF, addAB, recH, and recU genes were selected as representatives of the α, β γ, and ɛ groups, respectively. As revealed in Fig. 2, various degrees of increased sensitivity were observed. The recS, addAB, and recH strains are moderately affected, whereas the recF and recU strains are very sensitive to the killing action of MMS compared to the rec⁺ control (Fig. 2). The recS mutation partially suppressed the sensitivity of addAB cells to MMS (Fig. 2A). The recS mutation moderately increased the MMS sensitivity of recH, recU, and recF cells (Fig. 2B, C, and D).

Together, these results indicate that (i) the recS product is needed for the removal of DNA damage; (ii) the recS genes contribute to different extents to the removal of DNA damage in recH, recU, and recF cells; and (iii) recS partially suppresses the defect of the addAB cells.

Effects of recF, addAB, recU, recH, and recS mutations on genetic recombination.

To study the effects of the recF, addAB, recH, recU, and recS mutations on genetic recombination, we analyzed the requirement for these functions in natural transformational recombination. We measured the frequency of transformation by using plasmid and chromosomal DNA. Transformation in B. subtilis involves the transfer of naked double-stranded DNA from the media to the recipient competent uninucleated cell (22, 26). During DNA uptake, a set of competence (com) gene products degrades one of the DNA strands and takes up the other one in a linear single-stranded form (reviewed in reference 13). Therefore, by the activity of the com genes, the donor DNA is presented to the recombinational machinery in a form such that it is ready for homology searching (synaptic stage) on the parental molecule (13, 22). Hence, in our analysis, we are studying the synaptic and postsynaptic stages but neglect the involvement of rec functions in the presentation of the substrate (presynaptic stage).

B. subtilis transformation with chromosomal DNA (chromosomal transformation) does not require replication for integration of donor markers (13, 22). In contrast, both replication and recombination functions are required for the establishment of plasmid DNA (13, 22). Canosi et al. (12) and de Vos et al. (12a) have proposed that after uptake of the oligomeric plasmid DNA molecule (a monomer is inactive in transformation) and synthesis of the complementary strand, pairing of one of the incoming single-stranded DNA ends with the newly replicated strand results in plasmid establishment. By measuring both chromosomal (intermolecular recombination) and plasmid (intramolecular recombination) transformation, we can examine different types of events. B. subtilis competent cells were transformed with 1 μg of homologous chromosomal DNA or plasmid DNA per ml to determine the transformation frequency of the Rec⁻ mutant strains. The frequency of appearance of met⁺ transformants in the single-Rec⁻ strain and in certain double-Rec⁻ strains has been previously reported (1, 3). Here, the experiments were performed in parallel for comparison with other strains (Table 2).

Except for recA, the frequency of chromosomal transformation (intermolecular recombination) in the singly rec-deficient strain did not change more than fourfold relative to the rec⁺ value (3) (Table 2). The same results were found for plasmid transformation.

The recU null allele (recU1) and recU40 mutations reduce the chromosomal transformation less than threefold (1) (Table 2). As previously reported, plasmid transformation is marginally affected by the recU40 mutation, but it is reduced 25-fold and 100-fold in the recU1 and recS1 cells, respectively (Table 2).

Since (i) the recH (epistatic group γ) mutation blocked (>5,000-fold reduction) chromosomal transformation of recF (group α) and addAB (group β) cells (3, 5) and reduced about 250-fold that of recS cells (Table 2) but did not affect chromosomal transformation of recU cells (group ɛ), and (ii) the recH mutation blocks plasmid transformation of recF addAB cells (5) and reduced more than 1,000-fold that of recS cells (Table 2) but did not affect (2-fold) plasmid transformation of recU cells (Table 2), we can classify recH and recS into two different epistatic groups by using transformational recombination, and confirmed our previous classification of recH and recU within different epistatic groups (3). The classification of recH within group γ and recU within ɛ was based on the following facts. (i) recH cells are only moderately sensitive to DNA-damaging agents, whereas recU cells are very sensitive to these agents (1, 3). (ii) Double-mutant strains impaired in the α plus γ or β plus γ epistatic groups are blocked in chromosomal recombination, whereas the α plus ɛ or β plus ɛ groups are moderately affected (1, 3). (iii) Transductional intramolecular recombination is modestly affected in the double α plus γ mutants but is highly impaired in the α plus ɛ mutants (4). (iv) Chromosomal transformation is blocked in hbs recH cells but is reduced only 11-fold in hbs recU40 cells (15). To reconcile these differences with the data presented in this report, we have to assume that certain gene products of the γ and ɛ epistatic groups cooperate and that recH and recU operate at different stages in the recombination pathway. Both the γ and ɛ groupings will tentatively be maintained until more information becomes available.

The frequency of chromosomal transformation of the recS null allele was reduced about 4-fold, whereas plasmid transformation was reduced about 100-fold (Table 2). The recS mutation reduced chromosomal transformation in recF, recU, and addAB cells about 10- to 20-fold and that in recH cells about 250-fold (Table 2). Plasmid transformation was reduced by the recS mutation more than 1,000-fold in strains representative of the β, γ, and ɛ epistatic groups and was reduced about 200-fold in the α epistatic group.

On the basis of these and previous results, we conclude that no interaction exists between the RecU protein and the addAB and recF gene products, whereas a significant interaction may exist between the RecU and RecH proteins. No interaction was observed between the recS and the addAB, recH, recU, and recF functions. We tentatively place the recS mutation within a different epistatic group, termed “ζ.”

RecU and RecS proteins have no functional counterparts in E. coli.

A homology search of the RecU and RecS proteins was performed with the protein sequences available in the NBRF-PIR (release 50) and Swiss-Prot (release 34) databases. Significant identity (about 50%) was observed between the 23.9-kDa RecU protein (database accession no. I40528) and an equivalent 23.1-kDa product of Streptococcus pneumoniae (27) (accession no. ypoa_strpn, ypoa_stror), and 34% identity was observed with a 19.3-kDa protein of Mycoplasma genitalium (17) (accession no. I64238). A protein similar to RecU has not been detected so far in gram-negative bacteria. Both E. coli and Haemophilus influenzae seem to lack this protein (10, 16).

The 56.5-kDa RecS protein (accession no. L47648.GB_Ba) shares 34 to 36% identity with E. coli RecQ (27) (accession no. BVECRQ) and the putative H. influenzae RecQ (accession no. U32756_D) and B. subtilis “RecQ” (accession no. AF027868.Gb_Ba) proteins. The degree of identity of RecS with both E. coli RecQ and the putative B. subtilis “RecQ” protein could be enhanced up to 43 and 40%, respectively, if only the regions containing the seven-amino-acid motifs of DExH-box DNA helicases (first 330 residues) were used for the alignment. The putative B. subtilis “RecQ” protein (yocI) shares 40% identity with the E. coli RecQ protein.

Unlike E. coli containing only one recQ gene product with DNA helicase activity (28), B. subtilis could possess more than one such putative helicase (21). This is consistent with the fact that human cells contain genes that code for three related putative helicases (RecQL [human RecQL], accession no. A55311), HumBS [Bloom’s syndrome, accession no. A57570], and HumWS [Werner’s syndrome, accession no. L76937]) (see reference 14). Although the B. subtilis “recQ” gene has not yet been tested, the gene seems to be nonessential for cell viability (reviewed in references 14 and 20).

It is generally accepted that the role of E. coli RecQ (and probably B. subtilis “RecQ”) protein, alone or in combination with the E. coli RecJ protein, is to generate a 3′-terminal single-stranded DNA to which RecA could bind in a step proceeding to the homologous pairing. With chromosomal and plasmid transformation, we are measuring the synaptic and postsynaptic stages of recombination (13, 22). In this report, we show that the RecS protein is required in either the synaptic stage, postsynaptic stage, or both stages. It remains to be documented if RecS is a bona fide DNA helicase and if the helicase activity participates directly in recombination. We also report here that deletion of the recS gene partially suppressed the defect of addAB cells. The AddAB protein is a multifunctional enzyme associated with nuclease and helicase activities (19, 20); however, the defects generated by the addA5 and addB71 mutations in this genetic background remain to be characterized.