INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Cells with circular chromosomes and homologous recombination systems must be able to resolve chromosome dimers, or higher-order multimeric forms, that are generated by an odd number of recombination events between sister chromosomes during DNA replication. Failure to resolve chromosome dimers into monomers will prevent the proper partitioning of genomic material to newly forming daughter cells. A model for the coordination of chromosome dimer resolution and cell division has been elaborated in Escherichia coli based on a substantial accumulation of in vivo and in vitro data. In E. coli, two tyrosine family site-specific recombinases, XerC and XerD, act in concert at a site near the terminus of chromosome replication known as dif to resolve chromosome dimers into monomers during the process of cell division (9, 6, 12, 22). Deletion of Ecdif or mutations in xerC or xerD result in the development of a subpopulation of filamentous cells containing abnormally partitioned nucleoids. In addition, it has been demonstrated that the FtsK protein must be located at the constricting septum for the Xer-mediated resolution of chromosomes to occur (26, 37, 39). Similarly, plasmids containing the 28-bp minimal Ecdif site show dependence on FtsK for Xer-mediated inter- and intramolecular site-specific recombination in vivo (33). Dimeric chromosomes are thought to arise primarily as a by-product of homologous recombination events that enable bacteria to re-initiate replication at stalled replication forks. Although sister chromatid exchanges that generate dimers have been estimated to occur in approximately 15% of cells within a growing population, it is currently thought that recombinational DNA repair of stalled replication forks is a major housekeeping event in bacterial cells (38, 15).

A site-specific recombination system involved in chromosome partitioning has recently been described for the gram-positive species Bacillus subtilis (34). The B. subtilis homologues of the Xer proteins, CodV and RipX, share 35 and 44% identity with the E. coli XerC and XerD recombinases, respectively. CodV and RipX both possess the conserved amino acid residues indicative of the tyrosine family site-specific recombinases (16, 28, 34, 36). In vitro, RipX exhibited significant binding and catalytic activity on synthetic substrates containing the E. coli dif site. While CodV binding was significantly lower than that seen for RipX, cooperative interactions between the two recombinases were demonstrated (34).

Mutations in ripX resulted in the development of a subpopulation of cells that were either anucleate or contained aberrant nucleoids. The most likely interpretation of this partitioning defect is that chromosome dimers cannot be resolved in a ripX strain. Probably as secondary consequences of the partitioning failures seen in ripX mutants, cell division, competence, and sporulation deficiencies were also reported (34). The competence and sporulation defects highlight the concept that normal chromosome physiology in B. subtilis is required for successful progress through developmental pathways (20, 21, 31). Strikingly, the ripX phenotypes were not suppressed by the introduction of a recA mutation, in contrast to the case for E. coli xerC recA and dif recA double mutants (9, 34). Rather, ripX recA double mutants appeared to present a unique range of nucleoid phenotypes and a greater sporulation deficiency than that seen in either the ripX or the recA mutant.

In this study we present genetic and biochemical evidence that a Bsdif site is located at approximately 166° on the B. subtilis chromosome, 6° counterclockwise from the B. subtilis terminus of replication (23). Bsdif is utilized by the CodV and RipX recombinases to ensure that normal chromosome partitioning occurs in advance of the completion of cell division. We also show that the SP bacteriophage-encoded gene yopP, whose protein product shows limited homology to CodV and RipX, affects the penetration of the resolution-related phenotypes observed in codV, ripX, and dif mutants.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains. The B. subtilis parental strain used in this study (except where noted) was BR151 (trpC2 lys3 metB10). Strain SL7513 (trpC2 metB10 xin-1 SP), which was used to evaluate prophage contribution to the chromosome partitioning phenotypes, was obtained from the Bacillus Genetic Stock Center (YB886). See Table 1 for a complete listing of all B. subtilis strains and plasmids used. E. coli strain DH5 [F endA1 hsdR17 (r_K m_K⁺) supE44 thi-1 recA1 gyrA96 relA1 (lacZYA-argF)U169 80dlacZM15; Bethesda Research Laboratories] was used for cloning recombinant plasmids designed to inactivate B. subtilis genes. E. coli JC8679, a recBC sbcA derivative of AB1157 (42), was used to propagate Bsdif-containing plasmids and control plasmids used in recA integration assays. The E. coli strain used to overexpress the CodV and RipX maltose-binding protein (MBP) fusions was DS9009, a recF xerC::cat xerD::km derivative of AB1157 (34).

Genetic manipulations. B. subtilis transformations were performed as described previously (44) with selection on Luria-Bertani (LB) agar (Difco; 1.5%) containing chloramphenicol at 5 µg ml¹, neomycin at 12 µg ml¹, and spectinomycin at 100 µg ml¹ where appropriate. Erythromycin selection was done on LB agar containing erythromycin and lincomycin at 1 and 25 µg ml¹, respectively. Knockout mutations of ripX, recA, and spoIIIE have been described previously (34). Knockout mutations of codV, yomM, yopP, and ytpT were constructed by ligating antibiotic resistance cassettes within flanking chromosomal DNA. Antibiotic cassettes used in the construction of these knockout strains are oriented so that their transcription is in the same direction as the gene they were placed in. All knockout mutations were confirmed by PCR. Construction and purification of CodV and RipX MBP fusions was as previously described (34). The Bsdif site lies at the coordinates 1,941,798 to 1,941,825 (http://genolist.pasteur.fr/SubtiList). The 28-bp Bsdif sequence was verified by sequencing reactions on both DNA strands from strains BR151 and 168 (the strain whose genome has been sequenced). The Bsdif site and flanking BR151 DNA totaling 1,286 bp (coordinates 1,940,968 to 1,942,254) was amplified by PCR using the Pfu polymerase (Stratagene) and cloned into pPP390 at the SmaI site to create pSAS90-4. NdeI digestion of pSAS90-4 removed a 679-bp (coordinates 1,941,270 to 1,941,949) fragment containing the Bsdif site. A neo cassette was ligated to the NdeI ends of pSAS90-4, resulting in pSAS97. pSAS97 was then used to construct a chromosomal deletion of the dif site by transformation.

Growth conditions. In LB medium, experimental cultures were prepared by growing strains overnight at 37°C in the presence of the appropriate antibiotics at one-half of the concentrations used in agar plates (see above for antibiotic concentrations in agar). Overnight cultures were subsequently diluted 1:20 in fresh, prewarmed medium in the absence of antibiotics. When exponential growth was achieved, cultures were diluted a second time, 1:10, in fresh, prewarmed medium. All analyses were performed after this second dilution. The second (1:10) dilution reliably yielded an exponentially growing culture without a lagphase. Sporulation was induced by growing cells in modified Shaeffer's sporulation medium (MSSM; 16 g of Nutrient Broth [Difco], 2 g of KCl, 0.5 g of MgSO₄ · 7H₂O, 1 ml of CaNO₃ [1 M], 1 ml of MnCl₂ [0.1 M], 1 ml of FeSO₄ [0.1 M] per liter). In MSSM, experimental cultures were prepared in the same manner as for LB medium except that antibiotics were used at one-fourth of the concentrations used in agar (see above). All liquid growth was performed in conical flasks at 37°C, with rotary shaking at 150 rpm. Cultures occupied approximately 6 to 9% of the total flask volume.

Gel retardation and in vitro recombination assays. The methods used were those of Blakely et al. (4, 6). Each reaction contained recombinase at a final concentration of 1 µM with approximately 0.1 pmol of radiolabeled DNA. Binding reactions were performed in 50 mM NaCl, 20 mM Tris (pH 8), 1 mM EDTA, 10% glycerol, and 100 µg of poly (dI-dC) per ml at 37°C for 10 min before electrophoresis through 6% polyacrlyamide at 4°C. Suicide cleavage reactions used binding buffer as described above but were incubated at 37°C for 1 h before electrophoresis through 6% polyacrylamide containing 0.1% sodium dodecyl sulfate (SDS).

DAPI staining and microscopy. Nucleoid staining was performed using DAPI (4', 6'-diamidino-2-phenylindole). Cells were fixed prior to staining in 0.37% formalin. First, 20 µl of fixed cell samples was adsorbed onto 0.1% (wt/vol) poly-L-lysine-treated coverslips for 5 min before placing the coverslips onto 20-µl pools of DAPI at a concentration of 1 µg ml¹ for 30 min. The coverslips were then placed upon a slide containing a single drop of Slow-Fade (Molecular Probes) and sealed. Fluorescent observations were made with a Zeiss Axioskop Fluorescent microscope using standard DAPI filter sets. Images were photographed with a Sony DKC-5000 digital camera and acquired with Adobe Photoshop Software v4.0. Software processing of photographs was restricted to brightness and contrast adjustments only. Four different nucleoid phenotypes were scored as indications of abnormal chromosome partitioning in this study. These four phenotypes are indicated with arrows in Fig. 1 and are briefly described in the accompanying legend.

View larger version (150K): [in this window] [in a new window]   FIG. 1.   Micrographs of DAPI-stained cells fixed during mid-exponential-phase growth showing the mutant strain nucleoid abnormalities discussed in this study. The parent strain (BR151) (A), dif mutant (B), and codV mutants (C, D, and E) are shown. The numbered arrows point to four different signs of partitioning abnormalities: i, long, continuous nucleoid; ii, anucleate cells; iii, unpartitioned nucleoids occupying the center of a long and otherwise nucleoid-less cell; iv, nucleoids similar to those indicated by arrow iii, except that in these cases the nucleoids occupy a particular side of the cell, while the remaining area of the cell is nucleoid-less, can also be seen. ripX cells have been described previously (34). See Table 3 for a cumulative scoring of these abnormalities. Scale bar, 10 µm.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Cloned and synthetic dif DNA integrates into recA backgrounds at high frequency. The candidate B. subtilis dif (Bsdif) sequence used in this work was chosen for study based on three criteria: its proximity to the terminus of chromosome replication, its partial sequence identity with the E. coli dif site, and its proximity to a tRNA gene. The significance of the site's proximity to a tRNA gene lies in the observation that several of the bacteriophage that utilize tyrosine site-specific recombination to integrate into host bacterial chromosomes do so near to tRNA genes. It has been speculated that these localities protect the phage from chromosomal deletion events (http://members.home.net/domespo/trhome.html). It should also be noted that the B. subtilis dif site, like E. coli dif, is located intergenically.

Integration of Bsdif-containing plasmids is dependent on the presence of chromosomal dif, RipX, and CodV. The ability of the 28-bp sequence to support integration into recA strains of B. subtilis suggested that the mode of entry into the chromosome was mediated by site-specific recombination. Several studies have shown low-frequency integration of plasmids into various chromosomal regions in B. subtilis recA mutants (18, 19, 43). To determine if the RipX and/or CodV recombinases were specifically participating in the catalysis of the integration events at the chromosomal Bsdif site, we evaluated strains that were mutated at the ripX or codV loci or else deleted of chromosomal Bsdif for their ability to support integration into a recA strain. The results showed that in the absence of either RipX, CodV, or the chromosomal Bsdif site, integration of the Bsdif-containing plasmids was abolished (Table 2).

Deletion induced filamentation. If the Bsdif site discussed above is a unique site where CodV and RipX mediate chromosome dimer resolution, then deletion of the Bsdif site would be expected to result in chromosome partitioning failures and cellular filamentation in a portion of the cell population. To test this experimentally, we constructed a strain in which the Bsdif sequence DNA was replaced with a neomycin resistance cassette (SL8420). The dif-deleted cells were grown in LB medium and fixed during mid-exponential phase. DAPI staining of the fixed cells showed that about 25% of the sampled cell population contained evidence of chromosome partitioning failure (Fig. 1 and Table 3). The appearance and penetration of the dif-deleted phenotype was very similar to that seen in ripX mutants (Table 3) (34). Additionally, as is the case for ripX and codV mutants, cellular growth rates and competence were diminished in the dif deletion mutant (data not shown). The dif deletion strain used in this study also deletes the 3' portions of the flanking ynfE and ynfF genes. A strain constructed containing mutations in both ynfE and ynfF, but retaining the Bsdif site (SL8633), did not display significant partitioning defects (<1%, data not shown).

Integration of Bsdif-containing plasmids does not require the E. coli FtsK homologues SpoIIIE or YtpT. In E. coli it has been established that, in addition to the Xer recombinases, at least one other protein, FtsK, is required for resolution of chromosome multimers in vivo (10, 37). B. subtilis SpoIIIE and YtpT are, respectively, 48 and 49% identical to the C terminus of the E. coli FtsK protein. Unlike FtsK, however, SpoIIIE is normally not required during vegetative growth. Instead, SpoIIIE's principal function is during the specialized development of a spore, where it is required for the movement of chromosomal DNA into the small prespore compartment (35, 45). At present we are not aware of any published data on ytpT. Our observations of a ytpT deletion mutant and a spoIIIE mutant did not reveal a substantial nucleoid phenotype. However, inspection of nucleoids in a ytpT spoIIIE double mutant revealed a slight partitioning defect (Table 3). Therefore, it remained possible that YtpT and/or SpoIIIE contributed to the chromosome resolution reactions mediated at Bsdif. However, integration of Bsdif-containing plasmids (pSAS141, 142, 143, and 144) still occurred in the absence of SpoIIIE and/or YtpT (Table 4). These results reinforce the conclusion that SpoIIIE and YtpT are not required in CodV- and RipX-mediated site-specific recombination reactions at Bsdif.

CodV and RipX bind to the B. subtilis dif site. Our previous in vitro experiments demonstrated that CodV and RipX were capable of binding the E. coli dif site and that they could interact with XerC and XerD despite the great evolutionary divergence of the two organisms from which the proteins were derived. Additional evidence that CodV and RipX are catalytic partners was provided by the efficient resolution of a preformed, artificial Ecdif Holliday junction in vitro. This experiment also clearly indicated that a productive synapse was only formed in the presence of both recombinases (34).

View larger version (33K): [in this window] [in a new window]   FIG. 2.   Gel retardation analysis of recombinase binding to Bsdif DNA. The aligned recombination site sequences are Bsdif from B. subtilis, Hidif from H. influenzae, Ecdif from E. coli; cer from ColE1; and psi from pSC101. Central region sequences are in boldface type. The bottom panel shows binding of CodV- and RipX-MBP fusions to 32P-radiolabeled (*) Bsdif substrate. The composition of each retarded complex is indicated adjacent to the autoradiogram. Recombinases present in each reaction are shown below the appropriate lane.

View larger version (42K): [in this window] [in a new window]   FIG. 3.   Gel retardation analysis of CodV-MBP and RipX-MBP binding to half-sites of Bsdif. Each panel shows an autoradiogram of recombinase binding to 32P-labeled (*) half-sites of Bsdif. The sequence of each half-site is indicated above the appropriate panel, while the composition of each complex is indicated beside the autoradiograms. The recombinase present in each reaction is shown below the appropriate lane.

CodV and RipX mediated cleavage of Bsdif DNA in vitro. The catalytic function of a tyrosine recombinase monomer is the cleavage and subsequent exchange of one DNA strand between two synapsed recombination sites. Strand cleavages can be assayed by trapping recombinase-DNA covalent complexes using linear "suicide" substrates that contain a nick three nucleotides on the 3' side of the phosphodiester bond that is cleaved by the recombinase. Cleavage of the substrate generates a three-nucleotide fragment that is free to diffuse from the complex, thus preventing religation because the 5' OH that acts as the nucleophile is no longer present. Based on the sequence similarities between the putative B. subtilis dif site and the sites from E. coli and H. influenzae, we constructed suicide substrates with nicks in the middle of the central region on either the top strand (TS) or the bottom strand (BS; Fig. 4). By convention, the top strand is the first XerC-mediated strand exchange during site-specific recombination at the psi site (13).

View larger version (31K): [in this window] [in a new window]   FIG. 4.   Recombinase-mediated strand cleavage of Bsdif DNA. 32P-radiolabeled (*) suicide substrates based on the Bsdif sequence were constructed with a nick (black triangles) either in the top strand (TS) or the bottom strand (BS) of the central region (boldface letters). The proposed positions of strand cleavage for CodV and RipX are indicated at either end of the central region by arrows. The bottom panels show autoradiograms of cleavage reactions analysed by denaturing gel electrophoresis in the presence of 0.1% SDS. The composition of each covalent complex is indicated adjacent to the autoradiogram, while the recombinases present in each reaction are indicated below the appropriate lane.

The SP phage gene yopP affects chromosome dimer resolution in vivo. The SP prophage is maintained near the terminus of almost all strains of B. subtilis 168 (24, 47). The presence of two genes in the SP prophage whose products show weak similarity to CodV and RipX stimulated our investigation into the possibility that the relative lack of chromosome abnormalities seen in codV mutants was the result of an SP prophage recombinase substituting for CodV in recombination reactions. The YomM and YopP proteins coded by the SP prophage share, respectively, 19 and 22% identity with the CodV recombinase, and each possesses several highly conserved residues found in the tyrosine family of site-specific recombinases (no Bsdifhomologs were found in the SP sequence using computer searches allowing up to 8-bp of mismatch).

Sporulation is diminished in mutants implicated in chromosome dimer resolution. Successful development of a spore requires that a chromosome be packaged into the small prespore compartment. The small architecture of the prespore compartment results from the construction of a highly asymmetric spore septum early in the developmental process (31). Under normal circumstances mature B. subtilis spores contain a single, completed chromosome (11). The presence of a dimerized chromosome therefore would present a developing spore with a difficult challenge in terms of packaging twice the normal amount of DNA into a spatially confined location. Thus, we have utilized the sporulation program of B. subtilis as a second, indirect measure of partitioning difficulties in the various mutant strains examined previously by nucleoid appearance. Strains were grown in MSSM to promote sporulation, followed by examination using phase-contrast microscopy to determine the efficiency of sporulation. The codV, ripX, and dif mutants exhibited decreased sporulation frequencies relative to their parent strain (Table 3). The decreases in sporulation frequency observed among the various mutants showed some correlation with the increases in abnormal nucleoid phenotype recorded during vegetative growth (Table 3).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

This report illuminates three prominent features of chromosome partitioning in B. subtilis. First, we have identified and characterized a B. subtilis dif site. Second, we show that the B. subtilis FtsK-like homologues SpoIIIE and YtpT are not required for recombination at Bsdif. Third, we have detailed an exacerbation of chromosome partitioning defects associated with the absence of the SP phage gene product YopP. The absence of YopP or the SP phage had its most severe effect in the codV mutant.

The B. subtilis dif site is located at approximately 166° of the chromosome, at bp 1,941,798 to 1,941,825. Three lines of experimental evidence have been provided to substantiate the authenticity of this site. First, integration of nonautonomously replicating plasmids carrying either cloned Bsdif DNA or a synthesized Bsdif-oligomer occurs at a high frequency in recA backgrounds. The integration of Bsdif-containing plasmids was dependent on the presence of RipX, CodV, and the chromosomal dif site. Second, deletion of the Bsdif site from the chromosome resulted in the development of a subpopulation of cells with aberrantly partitioned nucleoids that closely resembled in appearance and frequency those seen in ripX mutants. Third, the RipX and CodV proteins demonstrated specific binding to, and cleavage of, synthetic Bsdif DNA in vitro.

In E. coli the septal localization of FtsK is required for chromosome dimer resolution and site-specific recombination at other ectopic dif sites, possibly through a role in altering Holliday junction conformation after the first XerC-mediated strand exchange. Inhibition of cell division at a nonpermissive temperature in an ftsZ temperature-sensitive mutant of E. coli leads to a mild segregation defect that has been ascribed to the failure of dimer resolution, presumably as a consequence of FtsK failing to localize (33). Recchia and Sherratt have since proposed that all Eubacteria with circular chromosomes and Xer homologues also have FtsK homologues and suggest that this demonstrates a functional interaction (32).

B. subtilis has two proteins, SpoIIIE and YtpT, that share substantial homology to the E. coli FtsK protein. However, integration of Bsdif-containing plasmids into recA backgrounds persisted in the absence of the SpoIIIE and/or YtpT proteins. In contrast, Xer-mediated plasmid integration at Ecdif is reduced by 100-fold in an ftsK strain of E. coli (33). These observations demonstrate that neither YtpT alone nor YtpT in combination with SpoIIIE has a required FtsK-like function in cell division or chromosome dimer resolution in B. subtilis. Additionally, it has also been previously shown that reducing the levels of FtsZ in outgrowing spores of B. subtilis did not lead to any obvious defects in timing or rate of chromosome segregation (29). Therefore, a possible conclusion from these combined observations is that the control of dimer resolution in B. subtilis is not intimately linked to septation.

Is there an analogue of FtsK that controls dimer resolution and/or links it to division somehow in B. subtilis? One potential candidate protein, PrfA, which has been implicated in homologous recombination (17), has been proposed by Pederson and Setlow (30). Mutations in prfA generate a subpopulation of cells with a chromosome segregation defect that resembles the RipX phenotype, i.e., aberrant nucleoids and anucleate cells. It is not yet clear, however, whether the prfA phenotype is a true partition defect or a consequence of aberrant recombination as seen in some recA strains of B. subtilis (34).

The dif sites from E. coli and H. influenzae have been characterized by the presence of two 11-bp half-sites that contain partial dyad symmetry separated by a 6-bp central region (5'-ATAA N6 TTAT) that delineate the positions of strand cleavage and exchange (4, 27). The right half-site of Bsdif is remarkably conserved compared to the gram-negative bacterial sites (9 of 11 bp match) considering their great evolutionary divergence. Analysis of the XerD crystal structure has suggested six amino acids in a helix-turn-helix motif that make base- and phosphate-specific contacts that define sequence recognition of the Ecdif right half-site (40). Five of these six residues are conserved in RipX.

The left half-site of Bsdif is much more divergent compared to other XerC binding sites. We particularly note the G residue in the dyad symmetry as being unusual; sequence changes at this position are usually indicative of plasmid-borne recombination sites (Fig. 2) (14, 41). The binding site sequence divergence is also reflected by differences between XerC and CodV in the residues implicated in base and phosphate contacts that determine binding specificity.

Phylogenetic analysis of the Xer recombinases suggests that CodV may have arisen from a gene duplication of an ancestral homologue of RipX and CodV, while XerC may have arisen from a duplication event that occurred at an earlier time (see http://members.home.net/domespo/tr/fam-xer.html). The evolutionary maintenance of two Xer recombinases in many different eubacteria suggests that the functional control provided is of great importance to the outcome of the recombination reaction and thus successful chromosome segregation. Having two recombinases provides two potential advantages: (i) correct site alignment is always ensured, thus avoiding a homology "testing" step after the first strand exchange, and (ii) precise control over catalysis, and consequently the order of strand cleavage and exchange, is also ensured. Previous in vitro work has shown that XerC cleaves Ecdif DNA more frequently than XerD, and as a consequence XerC promotes top-strand exchange first to form a Holliday junction that is subsequently resolved by XerD (2, 4, 5). We have shown that CodV cleaves linear Bsdif preferentially, and by inference we suggest that CodV performs the first strand exchange to generate a Holliday junction that is then a substrate for RipX. We also note that the central region sequence of Bsdif is very similar to Ecdif (5 of 6 bp matches) and suggest that maintenance of this sequence is of major functional importance. The ratio of purines to pyrimidines in the central region has been implicated in the folding preference of the Holliday junction formed by the first strand exchange and as such dictates which pair of recombinases will be catalytically active (3, 2). The postulated role for FtsK in E. coli is the alteration of the Ecdif Holliday junction conformation to enable the second strand exchange (33). The similarity between Ecdif and Bsdif central regions suggests that an analogue of FtsK, other than SpoIIIE and/or YtpT, responsible for controlling catalysis of CodV and RipX may exist in B. subtilis. The evolutionary conservation between Bsdif and Ecdif and the presence of Xer recombinases in many organisms led us to search the microbial genome databases for similar sequences in gram-positive bacteria. As an example of the possible conservation of dif sites, we detected a sequence in Staphylococcus aureus with a 27-of-28- bp match to Bsdif (5'ACTTCCTATAA TATATA TTATGTAAACT [http://www.tigr.org]) (Fig. 2). The functional importance of this sequence remains to be tested.

In our previous examinations of ripX and codV mutants we saw little effect of a codV mutation on nucleoid morphology (34). In experiments presented here, where a broader range of partitioning difficulty indicators was assessed, codV mutants revealed a weak partitioning phenotype (Table 3). We speculated, therefore, that B. subtilis might retain a protein that is redundant for CodV's function in chromosome dimer resolution reactions or that RipX itself might be able to perform the resolution reactions without a "XerC-like" partner recombinase. To address the first possibility, we examined two proteins in the SP bacteriophage that possess limited homology with the RipX and CodV proteins. YomM and YopP have between 19 to 25% identity with CodV and RipX in pairwise alignments, and each has an R...H-X-X-K...Y motif that can be aligned with the R...H-X-X-R...Y motif found in almost all tyrosine recombinases (16, 28) (note the substitution of the canonical second arginine with a lysine residue in the phage sequence).

An encouraging sign that one or both of these proteins might have a suppressing function in codV mutants came from an initial examination of codV mutants in strains lacking the SP phage. These SP() codV strains displayed a twofold increase in the amount of cells exhibiting aberrant nucleoids over that seen in the SP(+) codV cognate strain (Table 3). Moreover, although the yomM codV double mutant did not display an increase in nucleoid phenotype over that seen in the codV single mutant, the yopP codV double mutant did display an increase in the number of cells with aberrant nucleoids over that seen in a codV mutant, again, by about twofold. By themselves, yopP (SL8081) and SP (SL7513) had little if any affect on nucleoid phenotype. The similar increases in aberrant nucleoid phenotype observed in the SP() codV mutant and yopP codV double mutant suggested that the absence of YopP was responsible for most, if not all, of the difference in nucleoid phenotype penetration between SP(+) and SP() codV strains.

However, three lines of evidence argue against YopP being able to partially substitute for CodV in site-specific recombination reactions at Bsdif. First, we have not observed integration of Bsdif-containing plasmids in codV recA mutants. Second, yopP mutations elevate not only the codV nucleoid phenotype but also the ripX and dif phenotypes, albeit to a lesser extent. Third, we have not been able to demonstrate binding of purified YopP to the Bsdif site either alone, or in combination with YomM, CodV, or RipX. The simplest explanation based on our accumulated data is that YopP does not have a role in chromosome dimer resolution per se but rather some facilitative role during chromosome partitioning in general. One possibility, for example, is that YopP acts at a second chromosomal site as a fail-safe mechanism should CodV-RipX activity at Bsdif be impaired in some way. This explanation is consistent with our observation that yopP mutations elevate not only the amount of codV aberrant nucleoid phenotype but also the phenotypes associated with ripX and dif. This explanation is also consistent with the penetration of phenotypes observed in the SP() background. The primary deficiency that we note in this model is that it leaves unexplained why the codV mutant nucleoid phenotype is substantially less prominent than those seen in ripX and dif strains. It is formally possible that RecA itself (absent in our integration assays to ensure that homologous recombination was eliminated) is able to facilitate some amount of recombination at or near Bsdif in codV strains, but not in ripX and dif strains, thereby reducing to some extent the apparent amount of aberrant nucleoids seen in the codV strains.

Finally, we note that the identification of a B. subtilis dif site has an experimental utility. By employing integrative vectors that contain the Bsdif site, genetic information can be introduced at a relatively high frequency into the chromosomes of recA strains.