Barriers to recombination between closely related bacteria: MutS and RecBCD inhibit recombination between Salmonella typhimurium and Salmonella typhi

Abstract

Previous studies have shown that inactivation of the MutS or MutL mismatch repair enzymes increases the efficiency of homeologous recombination between Escherichia coli and Salmonella typhimurium and between S. typhimurium and Salmonella typhi. However, even in mutants defective for mismatch repair the recombination frequencies are 10²- to 10³-fold less than observed during homologous recombination between a donor and recipient of the same species. In addition, the length of DNA exchanged during transduction between S. typhimurium and S. typhi is less than in transductions between strains of S. typhimurium. In homeologous transductions, mutations in the recD gene increased the frequency of transduction and the length of DNA exchanged. Furthermore, in mutS recD double mutants the frequency of homeologous recombination was nearly as high as that seen during homologous recombination. The phenotypes of the mutants indicate that the gene products of mutS and recD act independently. Because S. typhimurium and S. typhi are ≈98–99% identical at the DNA sequence level, the inhibition of recombination is probably not due to a failure of RecA to initiate strand exchange. Instead, these results suggest that mismatches act at a subsequent step, possibly by slowing the rate of branch migration. Slowing the rate of branch migration may stimulate helicase proteins to unwind rather than extend the heteroduplex and leave uncomplexed donor DNA susceptible to further degradation by RecBCD exonuclease.

Salmonella typhimurium and Salmonella typhi are closely related enteric bacteria (1). Based upon hybridization studies (2) and comparisons of known DNA sequences from the two organisms, the average DNA sequence similarity between S. typhimurium and S. typhi is about 98–99%. S. typhi Ty2 lacks restriction barriers (refs. 3–5; J. Neitzer, T.C.Z., and S.M., unpublished results), and DNA fragments can be readily transferred from S. typhimurium to S. typhi by transduction or conjugation (6). However, despite the high DNA sequence similarity and the ease of DNA transfer, the frequency of homeologous recombination between S. typhimurium and S. typhi is low (6). The inefficiency of recombination is not merely the result of rearrangements or insertions present in the S. typhi chromosome relative to S. typhimurium (7), because the frequency of transduction of DNA from all regions of the S. typhimurium chromosome into S. typhi is low (T.C.Z., J. Valenzuela, G. Mora, and S.M., unpublished data). The frequency of homeologous recombination following DNA transfer between less closely related bacteria such as Escherichia coli and S. typhimurium (8–11), as well as between species of Bacillus (12–15) and species of Streptococcus (16) is also low. Hence, differences in DNA sequence impede genetic recombination between related species. Such barriers to gene exchange may be effective in nature.

The mismatch repair system encoded by mutHSL genes is one barrier to homeologous recombination. Inheritance of chromosomal markers transferred from Escherichia coli occurs about 10²- to 10³-fold less frequently in wild-type S. typhimurium recipients than in isogenic mutS or mutL derivatives (11). The average DNA sequence similarity between E. coli and S. typhimurium is ≈85% (17). Similar results were observed for the inheritance of chromosomal markers transferred from S. typhimurium into S. typhi (6). Thus, the MutSL proteins inhibit homeologous recombination even when there are as few as 1–2 mismatches per 100 bp. These studies indicate that the mismatch repair system acts as a barrier to recombination even between very closely related species.

Although the mechanism of mismatch repair during DNA replication has been extensively studied, little is known about how mismatch repair inhibits recombination between related species. The minimum length of sequence identity (or “minimum efficient processing segment”, MEPS) required for efficient initiation of recombination by RecA is 23–27 bp (18). Thus, the 1–2% DNA sequence divergence between S. typhimurium and S. typhi is not likely to affect the initial strand invasion. Rather, base pair mismatches and their recognition by MutS and MutL may act at a later step in recombination, branch migration. In vitro studies support this idea. Although there is only 3% divergence between phage fd and M13, RecA-mediated branch migration within a fd/M13 heteroduplex is blocked by MutS and MutL (19). Furthermore, RecA-mediated branch migration of the fd and M13 heteroduplex DNA is inhibited in the absence of MutS or MutL proteins (20), suggesting that mismatches alone affect the rate of branch migration.

Although the mismatch repair system is partly responsible for the barrier to homeologous recombination, another barrier exists as well. For example, transduction from S. typhimurium into mutS derivatives of S. typhi is 10³-fold higher than transduction into mutS⁺ derivatives, but the frequency remains 10³-fold lower than that of corresponding transductions between strains of S. typhimurium (6). The results presented here indicate that the additional recombination barrier is due to mismatches in the recombining DNA and to the recD-dependent exonuclease activity of RecBCD. We propose a model to explain how base pair mismatches, MutSL, and RecBCD inhibit homeologous recombination.

MATERIALS AND METHODS

Media and Antibiotics.

Rich medium (NB) contained 0.8% Difco nutrient broth and 0.5% NaCl. Minimal medium contained E salts (21) and 0.2% glucose. Because S. typhi is a natural cysteine and tryptophan auxotroph, minimal medium was supplemented with 1.05% cysteine and 0.41% tryptophan. To select for antibiotic-resistant transductants, 30 μg/ml ampicillin (Amp), 50 μg/ml kanamycin sulfate (Kan), 10 μg/ml tetracyline (Tet), or 20 μg/ml chloramphenicol (Cam) were added to rich medium. Solid medium contained 1.5% Difco Bacto-agar.

Bacteria and Strain Constructions.

All strains used in this study are listed in Table 1. The strains are derivatives of either wild-type S. typhimurium LT2, wild-type S. typhi Ty2. S. typhi mutants were constructed by transposition (34) or transduction of mutations from S. typhimurium into S. typhi (6). Derivatives containing deletions in mutS were constructed by transducing the mutS121::Tn10 insertion into the recipient and then selecting for Tet^s (35, 36). Phage P22HT105/1 int-201 was used for all transductions (37). Preparation of phage lysates and transductions were done as described (38). Phage lysates used for transduction into S. typhimurium or S. typhi were typically diluted to give a multiplicity of infection (moi) between 0.01 and 0.1. Lysogen-free transductants of S. typhimurium were obtained by streaking for single colonies on Evans blue-uranine plates (39) and cross-streaking against a c2 mutant of phage P22. Transductants of S. typhi were not screened for P22 lysogeny, because P22 does not lysogenize S. typhi due to a defect in early gene expression (A. Hund and S.M., unpublished results).

Table 1.

Bacterial strains used in this study

Strain	Relevant genotype*	Ref. or source
S. typhimurium
LT2	Wild type	J. Roth
AK3042	zjj-3042::Tn10dTet linked 99% to serB+	22
AK3112	zjj-3112::Tn10dTet linked 78% to serB+	22
SA965	leuBCD39 ara-7 HfrK17(pro-metA-ilv … purE)	23
TE958	leuA414(Am) hsdL (r−m+) serB1465::MudJ	24
TT460	pyrB692::Tn10	25
TT2661	zcc-7::Tn10 linked 80% to put610	26
TT8371	thr-469::MudA	24
TT16813	recD542::Tn10dCam	27
MST1503	zcc-5::Tn10 linked 50% to putP1123	26
MST1504	zcc-4::Tn10 linked 20% to putA1146	28
MST1933	mutS121::Tn10	29
MST3063	leuA414 (Am) hsdL (r−m+) Fels−mutS121::Tn10	This study
MST3433	ΔmutS280 recD542::Tn10dCam	This study
MST3434	ΔmutS281	This study
MST3516	zjj-3042::Tn10dTet serB1465:MudJ	This study
MST3517	zjj-3112::Tn10dTet serB1465::MudJ	This study
S. typhi
Ty2	Wild type	G. Mora
TYT1148	recD542::Tn10dCam	This study
TYT1258	ΔmutS153	This study
TYT1512	ΔmutS153 recD542::Tn10dCam	This study
TYT1513	serB1465::MudJ	This study
TYT1681	serB1465::MudJ thr-469::MudA	This study
TYT1685	serB1465::MudJ thr-469::MudA pyrB692::Tn10	This study
TYT1932	Hybrid (thr-pyrB substitution from S. typhimurium)	This study
TYT1933	ΔmutS284 hybrid (thr-pyrB substitution from S. typhimurium)	This study
TYT1934	recD542::Tn10dCam hybrid (thr-pyrB substitution from S. typhimurium)	This study
TYT1935	ΔmutS284 recD542::Tn10dCam hybrid (thr-pyrB substitution from S. typhimurium)	This study

^*Tn10dTet refers to the transposition-defective derivative of Tn10, Tn10Δ16Δ17 (30). MudJ is an abbreviation for MudI1734, a Kan^r mini-Mu derivative that forms stable lac operon fusions (31). MudA is an abbreviation for Mud1–8, an Amp^r Mu derivative that forms stable lac operon fusions (32). (thr-pyrB substitution from S. typhimurium) refers to hybrid strains of S. typhi that contain a replacement of the thr-pyrB region of DNA from S. typhimurium. Genetic nomenclature is derived from the genetic map of S. typhimurium edition VIII (33). 

Construction of S. typhi Hybrids.

S. typhi hybrids containing a large region of the S. typhimurium chromosome were constructed by Hfr matings. Initially, transposon insertions in pyrB (Tn10dTet), serB (MudJ), and thr (MudA) were transduced into wild-type S. typhi Ty2. S. typhi DNA spanning thr through pyrB was replaced with the corresponding DNA from S. typhimurium by Hfr matings selecting for prototrophy on minimal medium supplemented with cysteine and tryptophan. Strains were grown overnight, subcultured, and regrown to ≈5 × 10⁸ cells/ml in rich medium. Donor and recipient cells were mixed in a 1:20 ratio and shaken slowly at 37°C for 6 h. Cells were vortexed mixed vigorously to break up mating pairs and then washed twice in 0.85% NaCl. Dilutions of the mating mix were plated onto supplemented minimal medium and incubated for 48 h at 37°C. Prototrophic transconjugants were then scored for loss of Tet^r, Kan^r, and Amp^r, indicating the replacement of a contiguous track of S. typhi DNA with S. typhimurium DNA (40). As expected, the frequency of recombination following conjugational matings between S. typhimurium and S. typhi was low.

Determination of Single Marker Transduction Frequency.

The transduction frequency of single markers was determined using phage lysates made on strains of S. typhimurium carrying Tn10 or Tn10dTet insertions. For each transduction 0.1 ml of recipient cells grown to 5 × 10⁸ cells/ml were infected with 0.1 ml of phage lysate diluted to yield an moi between 0.01 and 0.1. Cells and phage were mixed directly on rich medium containing Tet, and incubated at 37°C for 18 h. The frequency of transduction was calculated by determining the number of Tet^r transductants per viable cell per phage. The number of viable cells was calculated prior to infection; however, the number of viable cells did not decrease following infection.

Determination of Coinheritance Frequency.

Donor strains used for assaying coinheritance contained a selectable and a scorable marker that were cotransducible by P22HT. For the transductions shown, selection was for Tn10dTet markers linked to serB1465::MudJ. The physical distance separating these markers was determined according to the modified Wu formula (41, 42); zjj-3112::Tn10 is ≈0.5 kb from serB1465::MudJ, and zjj-3042::Tn10 is ≈3.0 kb from serB1465::MudJ. Unless otherwise indicated, 100 colonies from individual transductions were picked and patched onto rich medium containing Tet to avoid potential abortive transductants. When coinheritance of serB1465::MudJ was scored, the Tet^r colonies were replica plated onto rich medium plates supplemented with Kan. The frequency of coinheritance was expressed as the number of colonies that inherited the scored marker (Kan)/number of colonies that inherited the selected marker (Tet).

RESULTS

RecBCD Inhibits Transduction Between S. typhimurium and S. typhi.

Mutations in the mutSL genes increase the frequency of homeologous recombination between S. typhimurium and S. typhi, but the frequency remains 10²- to 10³-fold less than the frequency of homologous recombination between strains of S. typhimurium (6). Because degradation of free ends of the double-stranded donor DNA by RecBCD exonuclease (26, 43-46) could account for the decreased recombination frequency observed, we tested the effect of recD mutations on homeologous recombination. recD mutants remain proficient in recombination but lack the exonuclease activity (47). The frequency of transduction of S. typhimurium DNA into S. typhi recD recipients was 10¹- to 10²-fold higher than the frequency of transduction into wild-type S. typhi (Table 2, row g). To determine whether the effect of recD was independent of mutS, mutS recD double mutants were also tested. The frequency of transduction into these derivatives was up to 10⁶-fold higher than into wild-type S. typhi and 10²- to 10³-fold higher than into mutants that lacked only mutS or recD (Table 2, row h). The observed frequency of homeologous recombination in these mutS recD derivatives of S. typhi was nearly as high as that observed during homologous recombination between strains of S. typhimurium.

Table 2.

Frequency of single marker transductions

Recipient strain	Relevant genotype	Frequency of transduction, × 105
		Tn10 insertions linked to put			Tn10dTet insertions linked to serB
		zcc-5	zcc-4	zcc-7	zjj-3042	zjj-3112
S. typhimurium
a	Wild type	98.5	712.0	545.0	1040.0	318.0
b	mutS	696.0	1392.0	819.0	1470.0	944.0
c	recD	5420.0	3560.0	1310.0	4540.0	1540.0
d	mutS recD	1420.0	3370.0	1610.0	3180.0	2040.0
S. typhi
e	Wild type	<0.0001	0.011	<0.0001	<0.0001	<0.0001
f	mutS	1.550	7.970	1.70	14.20	3.41
g	recD	0.081	0.294	0.018	0.008	0.018
h	mutS recD	108.0	285.0	19.4	84.9	28.1
S. typhi hybrids
i	Wild type	<0.0001	0.021	<0.0001	109.0	21.9
j	mutS	0.148	1.50	0.196	85.2	18.0
k	recD	0.025	0.489	0.107	500.0	200.0
l	mutS recD	181.0	326.0	37.3	363.0	126.0

The mean from three independent transductions are shown. The standard deviation was typically <10% of the mean. Controls containing phage only or cells only did not give rise to Tet^r colonies. 

MutS and RecBCD Reduce Coinheritance of Linked Markers During Homeologous Recombination.

To determine whether the increase in single marker transduction from S. typhimurium into mutS recD derivatives of S. typhi was the result of an increase in the overall length of DNA exchanged, we assayed the frequency of coinheritance of linked markers (Table 3). The frequency of coinheritance of linked markers in S. typhi recipients was decreased relative to S. typhimurium recipients. The frequency of transduction into wild-type S. typhi was so low (Table 2, row e) that recombinants inheriting both markers were not detectable (Table 3, rows a and b). However, coinheritance was observed in mutS derivatives of S. typhi, where the transduction frequency was higher. For example, two markers that were 88% coinherited when transduced into a mutS derivative of S. typhimurium were coinherited 40% when transduced into a mutS derivative of S. typhi (Table 3, row c). This indicates that, although mutS inhibits homeologous recombination in S. typhi, inactivation of the MutS protein does not totally alleviate the reduced coinheritance frequency. Transduction into recD derivatives of S. typhi also resulted in an increase in the frequency of coinheritance compared with transduction into wild-type S. typhi (Table 3, compare rows e and f vs. a and b). In addition, transduction into mutS recD mutants increased the frequency of coinheritance to levels higher than those observed in wild-type or mutS derivatives of S. typhi (Table 3, compare rows g and h vs. a and b and rows c and d). Thus, increasing the frequency of transduction of S. typhimurium markers in S. typhi by inactivating mutS and/or recD increases the frequency at which linked markers are coinherited.

Table 3.

Coinheritance frequencies of linked genetic markers transduced from S. typhimurium

Row	Recipient genotype	Selected marker	Coinheritance frequency,* %
			S. typhimurium	S. typhi	S. typhi hybrid
a	Wild type	zjj-3112::Tn10dTet	89	ND	42
b		zjj-3042::Tn10dTet	47	ND	5
c	mutS	zjj-3112::Tn10dTet	88	40	36
d		zjj-3042::Tn10dTet	44	7	7
e	recD	zjj-3112::Tn10dTet	92	83	68
f		zjj-3042::Tn10dTet	60	75	15
g	mutS recD	zjj-3112::Tn10dTet	92	66	76
h		zjj-3042::Tn10dTet	49	29	27

^*Coinheritance of markers in S. typhi hybrids occurred between homologous sequences. The unselected marker was serB1465∷MudJ. Frequencies represent the mean value from several independent transductions. The standard deviation was typically <5%. ND, not detectable. 

To determine if the reduced coinheritance in S. typhi was specific to the serB region, we measured the frequency of coinheritance of several genetically linked markers from various regions of the S. typhimurium chromosome. For each pair of markers tested, coinheritance was lower after transduction into mutS derivatives of S. typhi than after transduction into S. typhimurium recipients. For example, markers that were separated by ≈10 kb were coinherited ≈50% in transductions between strains of S. typhimurium but were coinherited <5% when transduced into mutS derivatives of S. typhi. Hence, linked markers transduced from S. typhimurium into mutS derivatives of S. typhi are rarely coinherited regardless of their genetic location. These results suggest that shorter regions of DNA are inherited during interspecies transductions.

The Low Frequency of Transduction and Coinheritance Is Due to Base Pair Mismatches Between the Homeologous DNA Sequences.

To determine whether the inefficient exchange of genetic markers from S. typhimurium into mutS⁺ recD⁺ derivatives of S. typhi was a direct consequence of mismatches in the resulting heteroduplex, we tested the efficiency of transduction and coinheritance in S. typhi hybrids. There was a substantial increase in the frequency of transduction of single alleles and coinheritance of linked markers when the selected recombination events occurred between homologous DNA from the S. typhimurium donor and S. typhi hybrid recipient. For example, the zjj::Tn10dTet markers linked to serB1465::MudJ were transduced up to 10⁶-fold more frequently when the exchange occurred between the homologous sequence present in the S. typhi hybrid (Table 2, compare rows e and i). The increased frequency of transduction was not an artifact of construction of the S. typhi hybrids because transduction of DNA from S. typhimurium into S. typhi hybrids remained inefficient when the selected markers lay outside of the region of homology. For example, the frequency of transduction of Tn10 insertions linked to put in S. typhi hybrid recipients remained as low as that observed in S. typhi (Table 2). In addition, the unselected serB1465::MudJ marker, which was never coinherited in wild-type S. typhi, was coinherited at a detectable frequency in the wild-type S. typhi hybrid (Table 3, rows a and b).

DISCUSSION

The genetic exchange of linked markers by homologous recombination is dependent on several factors. For transduction, these factors include the amount of DNA packaged into the phage head and the physical distance separating the markers (18, 48). P22 packages about 44 kb of DNA into a phage head (49); however, due to the requirement for flanking homologous DNA, coinheritance of two markers from a P22 transducing particle will occur only if they are separated by less than 44 kb. As the distance between markers decreases, the probability of recombination within the flanking DNA sequences increases. Thus, for transduction of DNA between strains of the same bacterial species, there is a simple, empirically verified, mathematical relationship among the size of a transducing fragment, the physical distance separating the markers, and the frequency of cotransduction (41, 42). However, for transduction of homeologous DNA, the frequency of inheritance is lower than predicted from this mathematical relationship. Because DNA sequence mismatches and their recognition by proteins from the mismatch repair system appear to affect branch migration of heteroduplexes in vitro, we reasoned that the inefficient exchange of chromosomal DNA between S. typhimurium and S. typhi may also be the result of an inhibition of branch migration in vivo.

Transduction of single markers and coinheritance of linked markers is dependent upon sequence identity between the donor and recipient. Transductions in which the donor DNA diverges slightly from the recipient, such as between S. typhimurium and S. typhi, results in a decrease in the frequency of transduction and the length of DNA recombined into the recipient genome. The decreased frequency of transduction could be partially alleviated if the recipient lacked the MutS protein (Table 2). This is not surprising because previous studies have shown that MutS and MutL can inhibit homeologous recombination in vivo (6, 11) and in vitro (19). However, even in a mutS or mutL recipient, the frequency of transduction and the length of DNA recombined from S. typhimurium into S. typhi remained significantly less than that observed during recombination between strains of S. typhimurium (Table 2).

In addition to mutSL, the recD-dependent exonuclease activity of RecBCD inhibited homeologous recombination between S. typhimurium and S. typhi (Table 2). S. typhi derivatives containing both mutS and recD mutations showed a synergistic increase in transduction frequency relative to S. typhi recipients that lacked only MutS or the RecD subunit. This result suggests that RecBCD exonuclease and MutS act at different steps of homeologous recombination. The increase in transduction of single markers observed in recD derivatives of S. typhi corresponded to an increase in the overall length of DNA exchanged, suggesting that the RecBCD exonuclease decreases the length of DNA donated in transduction between homeologous sequences by degrading free ends of donor DNA.

To determine whether the decrease in the amount of DNA exchanged was a direct result of base mismatches between the homeologous DNA sequences, and to rule out the possibility that S. typhi simply had a hyper-RecBCD exonuclease, we determined the transduction and coinheritance frequencies of S. typhimurium markers in S. typhi hybrid recipients which contained an extended region of DNA homologous to the donor. Transduction of homologous DNA from S. typhimurium into mutS⁺recD⁺ S. typhi hybrids occurred at high frequencies, characteristic of that observed between strains of S. typhimurium (Tables 2 and 3). However, when mutS⁺ recD⁺ S. typhi hybrid derivatives were transduced with DNA from outside the region of donor homology, transduction of homeologous DNA occurred at low frequencies, characteristic of that observed between S. typhimurium and S. typhi (Table 2). Thus, the decreased transduction and coinheritance frequencies caused by both mutS and recD is a direct consequence of base pair mismatches between the recombining DNA sequences. Furthermore, although mutations in recD or mutS recD had only a small effect on the transduction and coinheritance frequencies of homologous DNA in S. typhimurium and in the S. typhi hybrid, they had a large effect on transduction and coinheritance frequencies of homeologous DNA in S. typhi (Tables 2 and 3, compare rows a and b vs. e and f and rows g and h). This indicates that the recD-dependent exonuclease activity of RecBCD is a major barrier to homeologous recombination.

The effect of mismatches on homeologous recombination between closely related bacteria has been extensively studied in vivo (reviewed in ref. 50). For example, DNA sequence divergence is directly responsible for the inefficient genetic exchange observed between E. coli and S. typhimurium (51, 52). In such divergent species, the effect may be accounted for by the reduced probability of encountering the 23- to 27-bp region of sequence identity (MEPS) required for efficient initiation of recombination by RecA (18). For genetic exchanges between more closely related species such as S. typhimurium and S. typhi, the extent of sequence identity between the recombining substrates ensures a high probability of encountering the MEPS. Instead, the extent of DNA exchange may depend on a subsequent step in recombination, such as branch migration. Although the initiation of branch migration is mediated by the RecA protein, later steps in branch migration are catalyzed by other proteins including the helicase proteins RuvABC (53–56) and RecG (57–60).

The effect of mismatches on branch migration in vitro has been studied previously. Spontaneous branch migration through regions of heterologous DNA is inefficient (61). Furthermore, RecA-mediated branch migration appears to stall in regions of heterology (19, 20). For example, nucleotide mismatches alone, and their recognition by MutS and MutL, inhibit strand exchange between DNA from phages fd and M13. Branch migration through heterologous DNA sequences has been observed in reactions catalyzed by RuvAB in vitro (53–56). However, it is unclear whether RuvAB or even RecG-mediated branch migration through homeologous DNA is blocked or slowed in vivo. Our results suggest that branch migration mediated by these proteins is inhibited by mismatches, or by mismatch recognition by the MutS protein. Stalled branch migration may favor the unwinding of heteroduplexed DNA and subsequent expulsion of the donor strand. The expelled donor DNA would be a substrate for subsequent rounds of degradation by the RecBCD exonuclease.

The model shown in Fig. 1 may explain the decreased frequency of transduction and inability to introduce large regions of DNA during transduction between S. typhimurium and S. typhi. Transduction of DNA from S. typhimurium into wild-type strains of S. typhi is inefficient because both DNA sequence mismatches, and their recognition by MutS, act as barriers to branch migration. Inactivation of MutS in S. typhi recipients partially alleviates the barrier to homeologous recombination by allowing branch migration to proceed through some regions of heterology. However, DNA sequence mismatches inhibit branch migration in the absence of mismatch repair enzymes. Inactivation of the exonuclease activity of RecBCD in S. typhi recipients further alleviates the barrier to homeologous recombination, presumably by preserving the ends of donor DNA which may be susceptible to further rounds of degradation. Thus, when both MutS and RecBCD exonuclease are inactivated, the efficiency of homeologous recombination approaches the efficiency observed during homologous recombination.

Figure 1.

Model for interspecies recombination between S. typhimurium and S. typhi. Donor DNA is indicated by the darker lines. The steps of homeologous recombination are shown from only one end of the donor DNA; a second recombination event from the opposite end must occur to incorporate double-stranded DNA into the recipient. Strand invasion is depicted from the 5′ end of donor DNA. (i) Double-stranded donor DNA is transduced into a S. typhi recipient. (ii) RecBCD recognizes a free end of double-stranded donor DNA and degrades it until a properly oriented chi site is encountered. (iii) Upon recognition of a chi sequence, the RecD subunit is lost and RecBC continues to function as a helicase, producing the single-stranded donor DNA substrate for RecA-mediated strand invasion. RecA must encounter 23–27 bp of sequence identity (MEPS) between the donor and recipient DNA for pairing and strand invasion to occur. (iv) Formation of the heteroduplex between homeologous DNA from S. typhimurium and S. typhi results in mismatches which may stall or slow the RecG-mediated rate of branch migration of the heteroduplex. (v) A race between RecG and RuvABC either extends the heteroduplex in favor of product formation or unwinds the heteroduplex to reconstitute substrates. Unwinding of the heteroduplex by RuvABC expels the donor DNA and leaves it susceptible to a second round of degradation by RecBCD exonuclease.

Our model argues that the extent of DNA exchange during homeologous recombination is determined by a race between helicase proteins to either extend the heteroduplex in the direction of product formation or to unwind the heteroduplex and reconstitute initial substrates. RuvABC and RecG appear to have such opposing roles during Lac⁺ adaptive reversion in E. coli (62, 63). An alternative model, devised to explain the conjugational defects between E. coli and S. typhimurium, argues that homeologous recombination is limited by a shortage of the MEPS required to initiate RecA-mediated strand exchange (51, 52). The 10–15% DNA sequence divergence between E. coli and S. typhimurium makes the two models difficult to distinguish from one another. However, based upon the expectation that RecA recognizes the MEPS the same way in these closely related bacteria, the insufficient MEPS model is unable to account for the transductional defects observed between S. typhimurium and S. typhi.

Inactivation of the nuclease activity of RecBCD improves genetic exchange between S. typhimurium and S. typhi during transduction, but not between E. coli and S. typhimurium during conjugation (51). Several possible reasons for this observation include the following: differences in the length of donor DNA available for recombination (transduction vs. conjugation), differences in DNA sequence divergence between recombining substrates (1–2% vs. 15%), or differences in the recombination pathways operating on the DNA introduced by these alternate methods. Of these possibilities, differences in the length of donor DNA provided during conjugation (typically >1,000 kb) (40) vs. transduction (44 kb) seems to be the most straightforward explanation. On the longer DNA substrate, multiple attempts to initiate strand invasion could occur before the degradation of donor DNA extended into the sequence being selected or scored during recombination. Consistent with this idea, we have observed only small increases in conjugation frequencies between S. typhimurium and a recD derivative of S. typhi (data not shown).

In summary, efficient exchange of DNA between S. typhimurium and S. typhi is inhibited by two barriers to transduction: the mismatch repair proteins encoded by mutS and mutL, and the recD-dependent exonuclease activity of RecBCD. S. typhi recipients that lack both MutS and RecBCD exonuclease are able to undergo efficient recombination following transduction with homeologous DNA from S. typhimurium. The gene products of mutS and recD may be barriers to recombination between other closely related species of bacteria, and thus may play an important role in bacterial evolution and speciation. Inactivation of these barriers to homeologous recombination may allow construction of useful hybrid strains of bacteria.

ABBREVIATIONS

MEPS

minimum efficient processing segment

Amp

ampicillin

Kan

kanamycin sulfate

Tet

tetracyline

Cam

chloramphenicol