Impact of mutS Inactivation on Foreign DNA Acquisition by Natural Transformation in Pseudomonas stutzeri

Abstract

In prokaryotic mismatch repair the MutS protein and its homologs recognize the mismatches. The mutS gene of naturally transformable Pseudomonas stutzeri ATCC 17587 (genomovar 2) was identified and characterized. The deduced amino acid sequence (859 amino acids; 95.6 kDa) displayed protein domains I to IV and a mismatch-binding motif similar to those in MutS of Escherichia coli. A mutS::aac mutant showed 20- to 163-fold-greater spontaneous mutability. Transformation experiments with DNA fragments of rpoB containing single nucleotide changes (providing rifampin resistance) indicated that mismatches resulting from both transitions and transversions were eliminated with about 90% efficiency in mutS⁺. The mutS⁺ gene of strain ATCC 17587 did not complement an E. coli mutant but partially complemented a P. stutzeri JM300 mutant (genomovar 4). The declining heterogamic transformation by DNA with 0.1 to 14.6% sequence divergence was partially alleviated by mutS::aac, indicating that there was a 14 to 16% contribution of mismatch repair to sexual isolation. Expression of mutS⁺ from a multicopy plasmid eliminated autogamic transformation and greatly decreased heterogamic transformation, suggesting that there is strong limitation of MutS in the wild type for marker rejection. Remarkably, mutS::aac altered foreign DNA acquisition by homology-facilitated illegitimate recombination (HFIR) during transformation, as follows: (i) the mean length of acquired DNA was increased in transformants having a net gain of DNA, (ii) the HFIR events became clustered (hot spots) and less dependent on microhomologies, which may have been due to topoisomerase action, and (iii) a novel type of transformants (14%) had integrated foreign DNA with no loss of resident DNA. We concluded that in P. stutzeri upregulation of MutS could enforce sexual isolation and downregulation could increase foreign DNA acquisition and that MutS affects mechanisms of HFIR.

In all organisms repair of DNA is an essential process for maintaining DNA integrity and for avoiding mutations. Misincorporated bases remaining after DNA replication or spontaneous base modification in DNA produce mismatches which are repaired by an excision-type mechanism provided by the mismatch repair (MMR) system (11, 56). MMR proteins were first identified in Streptococcus pneumoniae, and the best-characterized MMR proteins are those in Escherichia coli (11, 55). In prokaryotes MutS initiates the repair by recognition and binding to the mismatch; this is followed by a search together with MutL on both sides of the mismatch for a site from where excision can be initiated (30). The ubiquity of MutS homologs in all three biological kingdoms (19) underscores the importance of MutS in maintaining the genetic stability of cellular genomes (26). In bacteria loss of the mismatch repair capacity results in mutator strains with increased spontaneous mutation frequencies (23). Such mutator strains have been found among environmental, commensal, and pathogenic isolates of Pseudomonas aeruginosa (59), E. coli (51), Salmonella enterica (42), and Neisseria meningitidis (63). Higher spontaneous mutation rates accelerate genetic adaptation and are thought to contribute to diversification and speciation (76). Although the mutator phenotype can be temporarily advantageous for producing diversity, it can measurably impair the overall fitness, and a return to MMR proficiency is important for stabilizing the specific fitness gained by the adapted subpopulation. The hypothesis of adaptive evolutionary progress mediated by loss and reacquisition of mutS is supported by the greater mosaic structure of mutS genes than of housekeeping genes (7, 8, 14).

Mismatches arising by the formation of heteroduplex DNA during homologous recombination between DNA molecules which differ at single nucleotides are also targets of MMR (11, 56). In bacterial interspecific DNA transfer, with increasing sequence divergence between donor and recipient, homologous recombination is inhibited by the requirement to find sufficiently long stretches of conserved sequences for recombination (35, 64, 70, 93). In addition, homologous recombination between diverged nucleotide sequences (homeologous recombination) is specifically suppressed by MMR, as indicated by the increased homeologous recombination frequencies in mutS mutants during interspecific conjugation (62, 82), transduction (91, 92), and natural transformation (47, 49). Thus, the sexual isolation of strains defective in mismatch repair is decreased (13, 46). The antirecombination effect of the MMR system seen in vivo was explained by the abortion of strand exchange reactions with MutS and MutL demonstrated in vitro (87, 88). Therefore, the MMR system controls not only the level of spontaneous mutability of cells but also the level of recombinative acquisition of new genetic information by homologous recombination (9, 81).

Natural transformation has been detected in many prokaryotic species (18, 45). Pseudomonas stutzeri is widely present in the environment, including marine, sediment, and soil habitats (65, 72), and many members of this species are naturally transformable (10, 44, 73). P. stutzeri strains have high genotypic diversity. These organisms were grouped by DNA-DNA hybridization into nine distinct genomic groups termed genomovars (65, 69), which, however, have no phenotypic characteristics worthy of species status (85). Often members of several genomovars are present in the same habitat (72). The P. stutzeri ATCC 17587 strain used in this study is naturally transformable with a variety of broad-host-range plasmids (Meier and Wackernagel, unpublished data) and genomic DNA (44). Recently, a novel gene acquisition mechanism, homology-facilitated illegitimate recombination (HFIR) (17, 60) during natural transformation, was described for this strain (53). By using HFIR the strain can integrate into its genome long stretches of fully heterologous DNA when they are linked on one side to a short homologous piece of DNA which serves as a recombinational anchor and thereby strongly facilitates illegitimate fusion of the heterologous parts of the molecules to resident DNA. Homologous anchor sequences as short as 311 bp effected HFIR. The illegitimate recombination occurred mostly at sites where there were three to six identical nucleotides in donor and recipient DNAs (microhomologies), but sites without identical nucleotides were also found (53).

We asked in which direction and to what extent the absence of the MMR component MutS influences foreign DNA acquisition by homeologous recombination and HFIR during natural transformation of P. stutzeri. For these studies we identified and inactivated the mutS gene of P. stutzeri and examined the influence of MMR on the spontaneous mutability and integration of transition and transversion mutations during autogamic transformation. We then studied transformation by heterologous DNA with increasing sequence divergence and the acquisition of foreign DNA by HFIR. While the absence of MutS increased autogamic and heterogamic transformation, the expression of mutS⁺ from a multicopy plasmid had a suppressive effect on both types of transformation. Moreover, in the mutS mutant a number of qualitative changes in the outcome of HFIR were observed.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

The bacterial strains used are listed in Table 1. Most P. stutzeri and E. coli strains were grown at 37°C on Luria-Bertani (LB) agar plates or in LB liquid medium (67); the exceptions were strains 28a21 and 28a28, which were grown in SW-LB (73) at 28°C. If necessary, LB medium was supplemented with rifampin (40, 100, or 130 μg ml⁻¹), gentamicin (5 μg ml⁻¹), streptomycin (100 or 1,000 μg ml⁻¹), chloramphenicol (250 μg ml⁻¹), ampicillin (150 μg ml⁻¹), or kanamycin (50 μg ml⁻¹). E. coli DH5α containing pBlue-km1, pBluescript II SK(+), or pUCPKS and derivatives of these strains were used for preparation of plasmid DNA with a Nucleobond kit (Macherey and Nagel, Düren, Germany). Spontaneous Rif^r mutants of P. stutzeri strains were obtained by plating 300 μl of an overnight culture on agar plates containing 40 μg of rifampin ml⁻¹.

TABLE 1.

Strains and accession numbers for rpoB and mutS sequences

Strain or plasmid	Relevant genotype or characteristic(s) (accession no.)	Reference(s) or source
E. coli K-12 strains
DH5α		67
AB1157		29
WA753	Like AB1157, but mutS215::Tn10	G. Walker via M. Radman
P. stutzeri strains
ATCC 17587	mutS+ (AJ783431)	75; this study
	rpoB+ (AJ783424)	This study
ME350	Like ATCC 17587, but mutS::aac	This study
ATCC 14405	rpoB+ (AJ783426)	75; this study
F8-5	rpoB+ (AJ783428)	6; this study
ATCC 17595	rpoB+ (AJ783425)	75; this study
19smn4	rpoB+ (AJ783427)	65; this study
28a21	rpoB+ (AJ783961)	73
28a28	rpoB+ (AJ783962)	73
JM300	mutS+ (AJ783432)	10; this study
ME255	Like JM300, but mutS::aac	This study
P. putida DSM291T		DSMZa
Plasmids
pBlue-km1	pBluescript II SK(+) nptII+ Apr	17
pMR7	ΔnptII (10 bp internal) Smr Cmr	16
pMR30	ΔnptII (51 bp C terminal) Smr Cmr	17
pUCPKS	Apr	84
pPM8	pBluescript II SK(+) mutSput::aac Apr	This study
pPM17	pUCPKS; mutS (ATCC 17587) Apr	This study
pPM18	pPM17; mutS::aac Apr	This study

^aDSMZ, Deutsche Sammlung für Mikroorganismen und Zellkulturen.

Plasmids.

The plasmids derived from RSF1010 were pMR7 (16) and pMR30 (17). These and other plasmids were introduced into P. stutzeri ATCC 17587 and the mutS::aac mutant by electroporation. Plasmid pUCPKS is an E. coli-Pseudomonas shuttle vector (84).

Construction of mutS::aac alleles.

The Pfx polymerase (Invitrogen, Karlsruhe, Germany) was used for PCR amplification. A part of mutS of Pseudomonas putida DSM291^T was amplified with primers F-put (5′-CTACATCCGCCGCCAGACCC-3′) and R-put (5′-GGCAGGCTGGTGGTTTCC-3′) (annealing temperature, 55°C) and cloned into the EcoRV site of pBluescript II SK(+) (Stratagene), yielding pPM7. pPM8 was a derivative of pPM7 in which an aac cassette (Gm^r) was inserted into the NaeI site in the middle of the partial mutS gene. The aac cassette was amplified by PCR (annealing temperature, 52°C) with the following primers having overhangs (boldface type) with MluI sites (underlined) for detection of positive clones by restriction analysis with MluI: F-Gm1 (5′-CGACGCGTCTTGCGTATAATATTTGCCC-3′) and R-Gm1 (5′-GCACGCGTCAATTTACCGAACAACTCCG-3′) from pBBR1 MCS-5 (36).

The mutS gene of P. stutzeri with its ribosomal binding site was amplified by PCR by using primers F-mutS (5′-ATCTCGCCCATCAACGAGCCGCAC-3′) and R-mutS (5′-ATGCGCGCGATTCTAGCAGC-3′) (annealing temperature, 57°C). The F-mutS primer contained one-half of an EcoRV site (underlined), which completed the EcoRV cloning site of pUCPKS into which the 2.64-kb fragment was ligated. In this plasmid (pPM17) the mutS gene is expressed from the lac promoter of pUCPKS. From pPM17 plasmid pPM18 was derived, which contained the aac cassette in the BclI site of mutS (about two-thirds of the way downstream from the start codon). The cassette was amplified by PCR (annealing temperature, 60°C) by using the following primers with BamHI sites (underlined; compatible with BclI sites) in their overhangs (boldface type): F-Gm2 (5′-CGCGGATCCCTTGCGTATAATATTTGCCC-3′) and R-Gm2 (5′-GCGGGATCCCAATTTACCGAACAACTCCG-3′). The PCR product was digested with BamHI before ligation.

Part of the mutS gene (1,147 bp; positions 667 to 1813) of P. stutzeri JM300 was amplified by PCR (annealing temperature, 58°C) with primers F-pmutS (5′-CTGAAAGGCTTCGGTTGCG-3′) and R-pmutS (5′-TCAGGTCGTTGGCGACGAAG-3′) and cloned into the EcoRV site of pUCPKS, yielding pPM9. Plasmid pPM10 was derived from pPM9 by insertion of the aac cassette into the SphI site in about the middle of mutS. The resistance cassette was amplified by PCR (annealing temperature, 56°C) with the following primers containing SphI sites (underlined) in their overhangs (boldface type): F-Gm3 (5′-ACGGCATGCCTTGCGTATAATATTTGCCC-3′) and R-Gm3 (5′-ACTGCATGCCAATTTACCGAACAACTCCG-3′).

Determination of spontaneous mutability.

Three-milliliter cultures were grown overnight from single colonies. The mutant cell titer was determined on LB agar plates containing different antibiotics, and the total cell titer was determined on LB agar plates. In some experiments (e.g., complementation experiments) both media contained an additional antibiotic to select for plasmids in the cells. Spontaneous mutation frequencies were expressed as the number of mutants per viable cell.

Amplification of part of the rpoB gene.

An approximately 1.5-kbp part of the rpoB gene of the various P. stutzeri strains was amplified at an annealing temperature of 61°C with primers rpoB-F1a (5′-CCTTACCGCGGTTCCTG-3′) and rpoB-R_int (5′-CGGTTGGCGTCGTCGTGCTC-3′) (44) covering the sequence encoding amino acids 178 to 678 of the corresponding E. coli RpoB protein. The amplification products were purified with a QIAGEN PCR purification kit (QIAGEN, Hilden, Germany).

Sequence analyses.

Manual alignment of mutS and partial rpoB sequences was performed by using BioEdit (version 5.0.9) (25). Sequence divergences were calculated as p-distances by using Mega 2.1 (38).

Natural transformation.

Plate transformation of P. stutzeri strains was carried out as previously described (52). A cell suspension (20 μl) was mixed with 250 ng of purified rpoB PCR fragments or with 10 or 50 ng of genomic DNA or linear plasmid DNA. Bacterial genomic DNA was isolated by using a genomic DNA kit (QIAGEN). pBlue-km1 DNA was extensively treated with DraI and SspI, and the main DNA fragment was recovered after electrophoretic separation on a gel. The total cell titer was determined on LB agar plates, and the transformant cell titer was determined on selective LB agar plates. In some experiments the LB agar plates and the selective LB agar plates contained an antibiotic for maintenance of plasmids. The transformation frequencies are expressed below as the number of transformants per viable cell. The transformation efficiencies are expressed as the number of transformants per nptII.

Phenotypic characterization of transformants after HFIR.

Km^r transformants obtained after natural transformation of P. stutzeri harboring pMR30 with pBlue-km1 DNA were checked by two tests to make sure that they did not result from cointegrate formation with reconstituted donor plasmids (53). First, they were screened for the absence of ampicillin resistance, and then they were screened by PCR amplification to obtain no product with two primers, one of which annealed within the donor DNA upstream of nptII⁺ and one of which annealed within the recipient downstream of nptII′ (53). Any Ap^r transformants also gave a PCR product. The presence of functional strAB genes was tested on medium with streptomycin (1,000 μg ml⁻¹). The presence of blm (bleomycin resistance) was determined by growth in LB broth with bleomycin (5 μg ml⁻¹).

Nucleotide sequence accession numbers.

The nucleotide sequences reported in this paper have been deposited in the EMBL nucleotide sequence database under the accession numbers listed in Table 1.

RESULTS

Identification and structure of mutS.

The search for the mutS gene of P. stutzeri ATCC 17587 was carried out by using an indirect strategy because direct amplification of part of the mutS gene with degenerate primers designed to fit putatively conserved regions of mutS was not successful. P. putida strain DSM291^T was employed to amplify part of the mutS gene with primers matching the previously published sequence of mutS of P. putida ATCC 33015 (accession number AB039965) (39), which resulted in a 939-bp fragment (encoding amino acids 193 to 506 in the P. putida sequence deposited under accession number AB039965). This fragment was cloned in pBluescript II SK(+), and then a PCR-amplified aac gene cassette (gentamicin resistance [Gm^r]) was inserted into the NaeI site to obtain plasmid pPM8. After natural transformation of P. stutzeri with linearized pPM8, three Gm^r transformants were obtained. All of these transformants showed increased spontaneous mutability, suggesting that there was insertional inactivation of the mutS gene by homeologous recombination with the P. putida mutS::aac fragment. Partial digestion of chromosomal DNA of one of the transformants (ME137) with PstI and SalI provided DNA fragments which after ligation into the vector pUCPKS gave Gm^r transformants of E. coli. The nucleotide sequences of the chromosomal DNA fragments cloned in this way indicated that the mutS gene in ME137 had a mosaic structure consisting of an internal part (about 840 bp) of P. putida DNA (with the aac insertion) flanked on both sides by P. stutzeri mutS sequences. One PstI clone covered the C terminus of the mutS gene of P. stutzeri, and one SalI clone covered the N terminus. The sequences located upstream and downstream of mutS were used to design primers to amplify the whole mutS gene from genomic DNA, including the ribosomal binding site, by PCR for cloning and sequencing.

The P. stutzeri ATCC 17587 mutS open reading frame was 2,580 bp long and had high levels of nucleotide sequence similarity to mutS genes of other gram-negative bacteria, including P. aeruginosa (84.3% similarity; accession number AE004782), Pseudomonas syringae (79.9% similarity; accession number AE0016870), Azotobacter vinelandii (81.8% similarity; accession number AAAUO2000003), and P. putida (83.0% similarity; accession number AB039965). The high levels of sequence similarity probably facilitated the homeologous recombination exploited for the in vivo gene inactivation described above. The deduced protein had 859 amino acids (95.3 kDa) and exhibited high levels of amino acid sequence identity with the MutS proteins of P. aeruginosa (87.7% identity), P. putida (83.0% identity), P. syringae (86.7% identity), A. vinelandii (86.9% identity), and E. coli (59.2% identity; accession number U000096). A search for conserved domains with rpsblast by using clusters of orthologous groups of proteins (77, 78) revealed the typical domains of MutS proteins (40), including domain I (DNA mismatch binding; amino acids 12 to 124), domain II (connector domain; amino acids 132 to 257), domain III (core; amino acids 266 to 430 and 523 to 565), domain IV (clamp domain; amino acids 431 to 522), and domain V (ATPase motif; amino acids 572 to 801). The typical mismatch binding motif Phe-X-Glu was found at positions 37 to 39 of the protein (positions 36 to 38 in E. coli [68]). Promoter and terminator sequences of mutS of P. stutzeri were not identified. Downstream of mutS and transcribed in the same direction, we identified the fdxA gene, which is located in corresponding positions in the genomes of P. aeruginosa (accession number AE004782), P. putida (39), and A. vinelandii (41).

Isolation and characterization of a mutS::aac (Gm^r) mutant.

Natural transformation of strain ATCC 17587 with linearized plasmid pPM18 (Table 1) carrying the complete mutS gene inactivated by an aac cassette insertion was used for chromosomal allelic exchange. All eight Gm^r transformants tested showed increased mutability, and one transformant (strain ME350) was characterized further. The spontaneous mutability of ME350 was about 20- to 160-fold higher than that of the wild type (Table 2), which corresponded to observations for other bacterial mutS mutants, including mutants of E. coli (23), P. putida (39), Acinetobacter sp. (89), S. pneumoniae (80), and P. aeruginosa (58). The growth of the mutS::aac strain did not differ from the growth of the parental strain in terms of the length of the lag phase, the growth rate during the log phase, and the final titer reached in the stationary phase (data not shown).

TABLE 2.

Spontaneous mutability of P. stutzeri ATCC 17587 and its mutS::aac derivative

Antibiotic	Concn (μg ml−1)	Frequency of spontaneous antibiotic-resistant mutantsa		B/A
		mutS+ (A)	mutS::aac (B)
Rifampin	40	(5.9 ± 2.3) × 10−8	(9.6 ± 1.7) × 10−6	163
Rifampin	130	(2.4 ± 1.8) × 10−8	(2.1 ± 1.3) × 10−6	85
Chloramphenicol	250	(2.8 ± 1.0) × 10−7	(1.4 ± 0.5) × 10−6	20
Streptomycin	100	(2.3 ± 1.5) × 10−9	(3.0 ± 2.8) × 10−7	125

^aThe data are means ± standard deviations for four independent determinations.

Isolation of a mutS::aac mutant of P. stutzeri JM300.

We wanted to test by complementation whether MutS of P. stutzeri ATCC 17587 (genomovar 2) also functions in a member of a different P. stutzeri genomovar. During the cloning and characterization of a chromosomal fragment with recA of P. stutzeri JM300 (genomovar 8), we determined the sequence of the neighboring mutS gene. The length of mutS was 2,580 bp, which coded for 859 amino acids (98 kDa). The deduced amino acid sequence gave identity values of 93% with the MutS protein of P. stutzeri ATCC 17587, 87.8% with the MutS protein of P. aeruginosa, 86.5% with the MutS protein of A. vinelandii, and 59.6% with the MutS protein of E. coli. The typical conserved domains of MutS (see above), including the mismatch binding motif (at positions 37 to 39), were also present in the protein of P. stutzeri JM300.

Natural transformation of strain JM300 with linearized pPM10 carrying part of the JM300 mutS gene with the inserted aac cassette was carried out for chromosomal allelic exchange. The Gm^r transformants showed a 70-fold-higher spontaneous mutation frequency toward rifampin resistance (in medium with 130 μg of rifampin ml⁻¹) than the wild type. One mutS::aac mutant of JM300 (ME255) was chosen for complementation experiments.

Complementation of mutS mutants of P. stutzeri and E. coli.

Plasmid pPM17 (Table 1) containing mutS⁺ strongly decreased the high spontaneous mutation frequency of strain ME350 to a level that was about 10-fold below the level of the mutS⁺ strain with the vector plasmid (Fig. 1A). We noticed an approximately 10-fold-higher spontaneous Rif^r mutation frequency of P. stutzeri ATCC 17587 and its mutS mutant ME350 when they contained the vector plasmid pUCPKS (Table 2 and Fig. 1A). The reason for this phenomenon is not known. The multicopy mutS⁺ gene apparently also eliminated this extra mutability along with that caused by the mutS mutation. The observed complementation indicated that there was functional expression of mutS⁺ from the plasmid. In the mutS mutant of the genomovar 8 strain JM300, ME255, the plasmid also decreased the spontaneous mutation frequency, but not quite to the value observed for the mutS⁺ strain (Fig. 1B). Still, the heterologous complementation was rather effective considering that the difference between the melting temperature of the DNA of JM300 and the melting temperature of the DNA of a close relative of ATCC 17587, strain ATCC 14405, is 7.3°C (values higher than 5°C are used for species circumscription [66]) and that their 16S rRNA sequences differ by 1.4% (72). In contrast, there was no heterologous complementation of the E. coli mutS mutant (Fig. 1C). This observation can be explained by assuming that MutS of P. stutzeri is unable to cooperate with another component(s) of the MMR system (e.g., MutL) of a rather distantly related species. Similarly, mutS mutants of E. coli were not complemented by the MutS analog of S. pneumoniae, HexA, (61), and MutS of P. aeruginosa (59). Only the small protein encoded by an incomplete mutS gene of P. putida could partially complement a mutS mutant of E. coli (39).

FIG. 1.

Complementation of mutS mutants of P. stutzeri ATCC 17587 (ME350) (A), P. stutzeri JM300 (ME255) (B), and E. coli AB1157 (WA753) (C) by the mutS⁺ gene of P. stutzeri ATCC 17587 present on plasmid pPM17. The bars indicate the frequencies of spontaneous Rif^r mutations (130 μg of rifampin ml⁻¹), and the error bars indicate standard deviations (n = 3). Open bars, wild-type cells with vector plasmid pUCPSK; striped bars, mutS mutant cells with vector; gray bars, mutS mutant cells with pPM17. For details see the text.

Repair of transitions, transversions, and loops during natural transformation.

In natural transformation only a single strand of the donor DNA enters the cytoplasm, where it can hybridize with the resident DNA. Therefore, the in vivo recognition of different mismatches by MutS was tested by transformation of cells with DNAs of various spontaneous mutants having point mutations in the rpoB gene that resulted in a rifampin-resistant phenotype. Part of the rpoB gene (1.5 kbp; rpoB fragment) was amplified by PCR from 24 Rif^r mutants of strain ATCC 17587. The sequences indicated that 23 of these mutants had single base pair changes in cluster I (covering amino acids 507 to 533 encoded by the rpoB gene of E. coli [34]). One of these 23 mutants was slow growing, and the other 22 mutants represented six different locations, the most frequent of which was at nucleotide position 1592 (14 of 23 mutants) compared to the E. coli rpoB gene (34). The rpoB fragments of the six alleles (four transitions and two transversions) with mutations at five different positions were used as donor DNA (Table 3). The transformation frequencies of the mutS mutant with the rpoB fragments were 4- to 16-fold higher than those of the wild type (Table 3). The mismatches that emerged from transformation with donor allele 1 (CA and GT) are the same mismatches that occurred with alleles 2, 5, and 6, and the frequencies of transformation in mutS were similar. The origin of the C or G in the mismatch, whether it was derived from recipient DNA (alleles 1 and 6) or donor DNA (alleles 2 and 5), did not influence the transformation frequencies. Taken together, mispaired bases resulting from transitions and transversions were corrected with about equal efficiencies.

TABLE 3.

Transformation of P. stutzeri ATCC 17587 and its mutS::aac derivative with 1.5-kbp PCR fragments from different rpoB mutant alleles of ATCC 17587 conferring rifampin resistance

RifrrpoB allelea	Positionb	Parental base pair (recipient)	Mutant base pair (donor)	nc	Rifr transformation frequencyd		B/A	Type of mutation
					mutS+ (A)	mutS::aac (B)
1	1592	CG	TA	14	(8.7 ± 2.9) × 10−6	(8.5 ± 3.1) × 10−5	10	Transition
2	1538	AT	GC	2	(1.1 ± 0.6) × 10−5	(1.2 ± 0.5) × 10−4	11	Transition
3	1592	CG	AT	1	(8.6 ± 0.8) × 10−6	(8.9 ± 2.3) × 10−5	10	Transversion
4	1574	CG	GC	1	(1.3 ± 0.4) × 10−5	(4.8 ± 2.1) × 10−5	4	Transversion
5	1547	AT	GC	1	(3.5 ± 1.8) × 10−5	(3.4 ± 1.4) × 10−5	10	Transition
6	1546	GC	AT	3	(3.9 ± 1.5) × 10−6	(6.4 ± 3.6) × 10−5	16	Transition

^aThe alleles were chosen from spontaneous Rif^r mutants; Rif^r transformants were determined on plates with rifampin (40 μg ml⁻¹).

^bNucleotide position in E. coli.

^cOccurrence among 22 spontaneous Rif^r mutants.

^dThe amount of PCR product (1.5 kbp) used in each transformation assay was 250 ng; the data are means ± standard deviations for three independent determinations.

Ten-base-pair loops are not recognized by the MutS protein of P. stutzeri ATCC 17587, as indicated by the results of transformation experiments with mutS⁺ and mutS strains, both of which harbored pMR7, which carried a defective nptII gene with an internal 10-bp deletion (16). When linearized pBlue-km1 was used as the donor DNA with nptII⁺ (16.7 μg ml⁻¹ in each assay mixture), the mutS⁺ strain gave a transformation efficiency of 1.3 × 10⁻⁴ ± 1.7 × 10⁻⁴ (n = 3), which was indistinguishable from the transformation efficiency observed with the mutS derivative (1.5 × 10⁻⁴ ± 1.8 × 10⁻⁴; n = 3). Insertions and deletions that were more than four bases long were not repaired in other transformable species in which MMR-deficient mutants were examined, including Acinetobacter species (89), S. pneumoniae (22), and Bacillus subtilis (48).

Mismatch repair deficiency increases heterogamic transformation.

To determine to what extent the MutS protein of strain ATCC 17587 decreases heterogamic transformation, the mutS⁺ and mutS strains were transformed in parallel with PCR-amplified 1.5-kbp rpoB fragments from spontaneous Rif^r mutants of various P. stutzeri strains with increasing sequence divergence at this locus (including some strains belonging to a different genomovar) (Table 4). The DNA for homogamic transformation came from the Rif^r mutant with allele 3 and gave the expected higher transformation frequency in the mutS mutant compared to the mutS⁺ strain. A single additional base pair change in the donor DNA did not decrease the transformation frequency in mutS⁺, while a 2-fold decrease occurred with DNA having up to 12 mismatches and a 25-fold decrease occurred with DNA having 46 mismatches. There was only a small further decrease with a total of 151 mismatches. The transformation frequencies were always significantly higher in the mutS mutant than in the mutS⁺ mutant. In the mutS mutant a notable decrease in transformation was seen with 46 mismatches (7-fold), and a greater decrease in transformation was seen with 151 mismatches (16-fold). The decreased transformation of the mutS strain with high sequence divergence reflected the decrease in homologous recombination with the level of relatedness between donor and recipient DNAs (64). The values for sexual isolation (Table 4) with the heterogamic DNAs were significantly higher in the mutS⁺ strain than in the mutS mutant when the nucleotide sequence divergence between donor and recipient DNAs was 0.8 to 10.1%. The data in Table 4 were used to estimate the contribution of the MutS function to sexual isolation as described by Cohan and coworkers (47, 49, 64). The slopes for the two regressions of relative transformation frequencies with sequence divergence gave an estimate of 14% higher sexual isolation in mutS⁺ than in mutS. An analysis of covariance indicated that there was no significant difference between the two sets of data.

TABLE 4.

Influence of mutS inactivation in P. stutzeri ATCC 17587 on heterogamic transformation with rpoB fragments with increasing sequence divergence from different Rif^r P. stutzeri strains

Donor (Rifr mutant)	Genomovara	% Divergence (no. of mismatches)	Rifr transformation frequency (10−6)b		B/A	Sexual isolationc
			mutS+ (A)	mutS::aac (B)		mutS+	mutS::aac
ATCC 17587	2	0 (0)	10.1 ± 3.7	89.0 ± 37.0	8.9	1	1
ATCC 14405	2	0.1 (1)	8.4 ± 2.2	72.3 ± 38.0	8.6	1.19 ± 0.26	1.29 ± 0.17
F8-5	2	0.8 (12)	5.5 ± 3.8	147.7 ± 73.3	27.3	2.53 ± 1.62	0.64 ± 0.12
ATCC 17595	4	3.1 (46)	0.4 ± 0.2	15.2 ± 8.3	37.5	26.80 ± 5.78	6.26 ± 1.09
19smn4	4	10.1 (151)	0.3 ± 0.1	5.5 ± 2.7	19.0	35.10 ± 3.35	16.55 ± 1.55

^aSee reference 72.

^bThe data are means ± standard deviations (n = 3). The donor DNA was the rpoB fragment (1.5 kbp) of each donor strain at a concentration of 250 ng per assay. The transformation frequencies were corrected for spontaneous mutations.

^cSexual isolation values were determined by dividing the frequency of homogamic transformation (obtained by using the DNA of the recipient's own Rif^r mutant) by the frequency of heterogamic transformation.

The wild type and the mutS mutant were also transformed with heterogamic chromosomal DNA (mean size, roughly 40 kbp) from the Rif^r derivatives of the four strains shown in Table 4 and two additional strains with even higher numbers of mismatches in the 1.5-kbp segment of rpoB (207 and 219 mismatches) (Table 5). These two DNA species could be included in the experiments because the transformation frequencies were generally higher with high-molecular-weight chromosomal DNA than with PCR fragments which increased the values for Rif^r transformation frequencies above the background values for spontaneous Rif^r mutants (2 × 10⁻⁶ for mutS). (The higher transformation frequencies with chromosomal DNA were not due to DNA restriction of the nonmodified DNA produced by PCR, because P. stutzeri ATCC 17587 is naturally restriction negative [4].) The results (Table 5) show three things. First, with chromosomal DNA the effect of sequence divergence was decreased in autogamic crosses. With DNA of strain 19smn4 the transformation decreased to 20% of the autogamic frequency (which was 3% with the rpoB fragment [Table 4]), and the even more divergent DNA of strain 28a21 still resulted in about 14% autogamic transformation. Second, the absence of MutS increased autogamic transformation with chromosomal DNA only 1.9-fold (compared to 8.9-fold with the PCR product [Table 4]), indicating that the level of marker rejection by MMR was inversely related to the molecular weight of the transforming DNA. Third, the heterogamic transformation with chromosomal DNA decreased with increasing sequence divergence less in the mutS strain than in the mutS⁺ strain, and the average increase in the transformation frequencies for the mutS strain compared to the mutS⁺ strain (3.2- ± 1.8-fold) was smaller than the average increase with the PCR fragment (20.3- ± 12.4-fold). These observations suggest that with large DNA fragments homologous recombination between diverged sequences is more effective and that interference of MMR with marker integration is suppressed.

TABLE 5.

Influence of mutS inactivation in P. stutzeri ATCC 17587 on heterogamic transformation with genomic DNA with increasing sequence divergence from different Rif^r P. stutzeri strains

Donor (Rifr mutant)	Genomovara	% Divergence (no. of mismatches)	Rifr transformation frequency (10−6)b		B/A	Sexual isolationc
			mutS+ (A)	mutS::aac (B)		mutS+	mutS::aac
ATCC 17587	2	0 (0)	5.9 ± 1.8	11.5 ± 0.7	1.9	1	1
ATCC 14405	2	0.1 (1)	5.0 ± 0.2	8.4 ± 2.2	1.7	1.18 ± 0.31	1.44 ± 0.46
F8-5	2	0.8 (12)	9.2 ± 1.1	9.8 ± 1.7	1.1	0.63 ± 0.11	1.20 ± 0.28
ATCC 17595	4	3.1 (46)	1.6 ± 0.1	5.5 ± 1.1	3.4	3.62 ± 0.78	2.15 ± 0.57
19smn4	4	10.1 (151)	1.2 ± 0.5	3.7 ± 0.1	3.1	4.92 ± 0.69	3.11 ± 0.07
28a28		13.8 (207)	0.8 ± 0.2	4.0 ± 3.2	5.0	41.67 ± 10.31d	20.84 ± 1.30d
28a21		14.6 (219)	0.8 ± 0.2	3.7 ± 3.4	5.9	41.62 ± 10.68d	20.26 ± 7.02d

^aSee reference 72. The dashes indicate that the strains do not group with established genomovars.

^bThe data are means ± standard deviations based on two experiments. The donor DNA was chromosomal DNA of the Rif^r mutant of each donor at a concentration of 10 ng per assay. The transformation frequencies were corrected for mutations.

^cSexual isolation values were determined by dividing the frequency of homogamic transformation (obtained by using DNA of the recipient's own Rif^r mutant) by the frequency of heterogamic transformation.

^dThe amount of donor DNA was 50 ng per assay. The transformation frequency of ATCC 17587 at a DNA concentration of 50 ng per assay was 3.5 × 10⁻⁵ ± 1.3 × 10⁻⁵, and that of the mutS::aac mutant was 9.4 × 10⁻⁵ ± 6.7 × 10⁻⁵.

With chromosomal DNA the slopes of the regressions for sexual isolation versus sequence divergence differed for the mutS⁺ and mutS strains and indicated that the contribution of MMR to sexual isolation was 16%. As observed with the PCR products, with the large transforming DNA fragments the analysis of covariance of the data sets from both transformation series with the mutS⁺ and mutS strains did not reveal a statistically significant difference.

Expression of mutS⁺ from a multicopy plasmid eliminates autogamic transformation.

The titration of the MMR system by high numbers of mismatches was proposed to be the reason why heterogamic transformation with DNA having high sequence divergence compared to the recipient DNA is not strongly inhibited in mismatch repair-proficient S. pneumoniae (31). Starting from the titration model, we asked how expression of mutS⁺ from the multicopy plasmid pPM17 with about 15 copies per cell would affect autogamic and heterogamic transformation in P. stutzeri ATCC 17587. Transformation of cells containing multiple copies of mutS⁺ with the autogamic 1.5-kbp rpoB fragments carrying Rif^r allele 1 (Table 3) was decreased about 100-fold to a level similar to the spontaneous frequency of Rif^r mutations (Fig. 2A). Results similar to the results obtained with transition allele 1 were obtained with transversion allele 3 (data not shown). These observations suggest that in wild-type cells the mismatch-recognizing component MutS is a limiting factor. At its normal level MutS prevented about 90% of the potential transformation events with a single-base-exchange marker (as indicated by the approximately 10-fold-higher transformation frequency of the mutS mutant shown in Tables 3 and 4), and with multicopy mutS⁺ MutS prevented at least 99.9% of the potential transformation events. The 25-fold-lower transformation with the 3.1% divergent rpoB fragment in mutS⁺ (Table 4) was decreased further significantly (7-fold) by multiple copies of mutS⁺, resulting in frequencies that were not distinguishable from the background frequencies of spontaneous mutations (Fig. 2B). Thus, MutS is a limiting component in the prevention of recombinational integration of heterologous DNA. Finally, the transformation with autogamic chromosomal DNA (from Rif^r mutant 3 [Table 3]) was decreased about 20-fold in the mutS⁺ multicopy strain (Fig. 2C). Moreover, the alleviation of marker rejection in wild-type cells through the use of chromosomal DNA (Tables 3 and 4) was partially overcome by multiple copies of mutS⁺. In S. pneumoniae, overexpression of the mutS⁺ homolog hexA⁺ decreased transformation frequencies about threefold (31). Figure 2 also shows that multiple copies of mutS⁺ did not change the frequency of spontaneous Rif^r mutations, suggesting that there was strong limitation of MutS only in the elimination of transformants with single mismatch markers and not in the replication-associated correction of DNA synthesis errors.

FIG. 2.

Influence of mutS⁺ expression from multicopy plasmid pPM17 on the transformation of P. stutzeri ATCC 17587 by DNA covering the rpoB region with a Rif^r mutation. The recipient contained the vector pUCPSK (open bars) or pPM17 (striped bars). The transforming DNA was the 1.5-kb autogamic rpoB fragment of the mutant described in line 1 of Table 4 (A), the 3.1% divergent 1.5-kbp heterogamic rpoB fragment of the mutant described in line 4 of Table 4 (B), or autogamic chromosomal DNA (C). (D) No DNA applied. The bars indicate Rif^r transformation frequencies, and the error bars indicate standard deviations (n = 3).

MutS deficiency affects distribution of illegitimate fusion sites during homology-facilitated illegitimate recombination.

The HFIR mechanism was previously studied in P. stutzeri (53) by transformation of cells carrying plasmid pMR30 (which contains an nptII′ gene with a 51-bp deletion of the 3′ end) by using donor DNA with a complete nptII gene followed by a heterologous nucleotide sequence (linearized pBlue-km1 DNA). Homologous recombination in nptII′ and illegitimate fusion downstream resulted in a complete nptII⁺ product detected as a Km^r transformant. With this system (donor DNA concentration, 16.7 μg ml⁻¹) it was found that the frequency of HFIR events was the same in the mutS::aac strain (4.6 × 10⁻¹⁰  ± 1.6 × 10⁻¹⁰; n = 3) and the mutS⁺ strain (5.5 × 10⁻¹⁰ ± 2.0 × 10⁻¹⁰; n = 3). However, when 21 randomly chosen transformants of the mutS mutant were analyzed by PCR for the locations of the illegitimate fusion sites (Fig. 3A), three differences from the previous results obtained with mutS⁺ (53) were found. First, the fraction of fusions occurring within the first 500 bp following the anchor and having short segments of resident DNA replaced by short segments of donor DNA was smaller in the mutS strain (3 of 21 fusions) than in the mutS⁺ strain (13 of 24 fusions). Second, a hot spot site in the recipient DNA accumulated 8 of 21 illegitimate fusion events (38%; fusions 9, 10, 12, and 14 to 18). This site was employed only three times among 24 fusions (13%) in the mutS⁺ strain. The different distributions of fusion sites in the mutS and mutS⁺ strains were also reflected by the different proportions of phenotype classes among the transformants, which resulted from the presence or absence of a complete blm gene and inactivation of the resident str genes by fusion or maintenance of these genes (Fig. 3B). Third, there was a novel class of recombinants (14%) in which the illegitimate fusion had occurred within the homologous region (fusions 1, 2, and 3). These transformants gained heterologous DNA (one of them gained the complete blm gene) without any loss of resident DNA. In these transformants 63 to 468 bp of the terminus of nptII′ was duplicated. Overall, although the average length of acquired foreign DNA per transformant was not increased in the mutS strain (545 bp versus 595 bp in the mutS⁺ strain), the additional length of foreign DNA integrated in the transformants that had a net gain of DNA was greater in the mutS strain (average, 279 bp in 11 of 21 transformants) than in the mutS⁺ strain (103 bp in 16 of 24 transformants).

FIG. 3.

(A) Illegitimate fusions between donor DNA (pBlue-km1) (upper line) and recipient DNA (pMR30) (lower line) in 21 transformants. Different lines correspond to the specific transformant phenotypes. The dotted lines (class I) indicate Bm^s Sm^r transformants, the dotted and dashed lines (class II) indicate Bm^r Sm^r transformants, the solid lines (class III) indicate Bm^s Sm^s transformants, and the dashed lines (class IV) indicate Bm^r Sm^s transformants. The small arrows indicate primer positions used in PCR amplification for the localization of fusion sites. The boxes with arrow ends indicate functional antibiotic resistance genes (nptII, kanamycin resistance; blm, bleomycin resistance; strA and strB, streptomycin resistance). The box without an arrow end indicates the deleted nptII gene (nptII′). The area of homology is gray. (B) Distribution of nptII⁺ transformants of the mutS strain in the four phenotype classes. The values in parentheses are the values obtained with the mutS⁺ strain (53). A total of 56 transformants were analyzed for the mutS::aac strain, and 42 transformants were analyzed for the wild type.

Illegitimate recombination sites in the mutS mutant.

The sequenced sites of the 21 transformants (Fig. 4) were all different. Nine of them had overlapping identical sequences between donor and recipient DNAs consisting of 3 to 10 bp (microhomologies) and having a higher G+C content (63.5%) than the donor DNA (55.5%) or the recipient DNA (52.3%) over the 2.5 kbp following the anchor sequence. It is remarkable that large (10-bp) microhomologies like those of transformants 2 and 19 (ΔG°₃₇ = −13.95 and ΔG°₃₇ = −13.17 kcal/mol, respectively) were not found among 44 illegitimate events involving the same donor and recipient sequences in mutS⁺ strains of P. stutzeri and Acinetobacter sp. strain BD413 (17, 53). In the mutS mutant the fraction of sites not located in microhomologies (i.e., the sites having no overlap or only a single-base-pair overlap) was 52%, and this was higher than the fraction in the mutS⁺ strain (30%) (53). The eight fusions at the hot spot site had the same sequence (5′-GAAC-3′) in the recipient, and related sequences were present in the fusions in donor and/or recipient DNA in transformants 1, 3, 4, 5, and 8. Ten of these 13 fusions did not contain a microhomology. These observations suggest that the fusion mechanism that is particularly active in the mutS strain is sequence specific to some extent. The different sets of fusion sites employed in the mutS mutant and the mutS⁺ strain are also reflected by the different local distributions of the sites (see above).

FIG. 4.

Nucleotide sequences around the illegitimate fusion sites present in the 21 transformants shown in Fig. 3 (same order). The sequences of the transformants are indicated by boldface type. D, donor DNA; R, recipient DNA. Identical nucleotides in the DNAs of the donor and recipient are indicated by dots, and at the fusion sites overlapping nucleotides are enclosed in boxes.

DISCUSSION

The lengths of the amino acid sequences deduced from the mutS genes of the genomovar 2 (ATCC 17587) and 8 (JM300) strains of P. stutzeri are rather similar to each other and to that of E. coli, and the typical domains found in the E. coli protein, including the mismatch binding motif, are present in both of these sequences. The functional equivalence of the mutS genes to the genes of other gram-negative and gram-positive bacteria was established by the highly increased spontaneous mutability of mutS::aac insertion mutants and by the intraspecific complementation through cloned mutS⁺. MutS of P. stutzeri ATCC 17587 recognized single-base mismatches formed during natural transformation with about equally high efficiency whether they resulted from transition or transversion mutations. This led to approximately 10-fold-higher transformation frequencies in a mutS mutant. These observations for P. stutzeri are different from the generally lower level of repair of transversion mismatches during natural transformation of S. pneumoniae (11, 12) and Acinetobacter sp. strain BD4 (89). Transversion allele 4 (Table 3), which can result in CC and GG mismatches in the transformants, showed only a fourfold-higher transformation frequency in the mutS mutant than in the mutS⁺ strain. It is likely that this intermediate value resulted from efficient correction of the GG mismatches and the absence of recognition of the CC mismatches. There is no evidence for correction of CC mismatches in S. pneumoniae (11), E. coli (55), and Saccharomyces cerevisiae (37). As in other prokaryotic MMR systems, small 10-bp loops were not repaired in P. stutzeri. A corresponding mutL gene has not been detected in P. stutzeri but has been identified in P. aeruginosa (58) and is present in the genomes of P. syringae and P. putida. It is likely that MMR in P. stutzeri is not methyl directed as it is in E. coli and other enterobacteria because dam methylation of GATC sequences is absent in P. stutzeri strains JM300 and ATCC 17587 (Meier and Wackernagel, unpublished data) and homologs of the dam and mutH genes are not present in the P. aeruginosa, P. syringae, and P. putida genomes.

The 89% suppression by MMR of homogamic transformation by a single mismatch marker on a 1.5-kbp PCR fragment (Table 4) was greatly alleviated by use of large chromosomal DNA fragments (Table 5). With homogamic PCR fragments that were only 1.1 kbp long transformation was suppressed by 93% (Meier and Wackernagel, unpublished data). These observations suggest that after mismatch recognition finding the strand interruption necessary for strand rejection is more successful when PCR products are used for transformation than when chromosomal DNA fragments are used. In the latter case much larger donor DNA segments are integrated into recipient DNA during transformation than with PCR fragments (92), yielding longer distances from the mismatch to a strand interruption. In S. cerevisiae a fully homologous approximately 610-bp sequence next to the mismatch was proposed to prevent the antirecombination action of MMR (20).

The 25-fold-lower transformation of the mutS⁺ strain with heterogamic PCR-amplified DNA fragments with 3.1% sequence divergence (46 mismatches in addition to the marker mismatch) than with autogamic DNA fragments was only partially due to inhibition of recombination by sequence divergence since in the mutS mutant the transformation was decreased hardly sevenfold. It has been proposed that larger numbers of mismatches increase the chance that any one of the mismatches is recognized, which in turn can lead to coelimination of the marker mismatch in the vicinity (11, 24). This proposal implies that mismatch recognition is a limiting step in marker rejection in P. stutzeri. This hypothesis is supported by the effect of multiple mutS⁺ genes. Such genes completely suppressed single mismatch marker transformation, irrespective of whether the marker resided on autogamic PCR-amplified DNA fragments or heterogamic PCR-amplified DNA fragments with 3.1% sequence divergence. Moreover, the single mismatch marker on large chromosomal DNA fragments, which was normally eliminated by MMR in only about 50% of the cases, was eliminated in cells with multiple mutS⁺ genes in about 95% of the cases. This observation suggests that the elevated MutS level not only increases mismatch recognition but also improves the tracking success of the MMR complex to the nearest DNA strand interruption. If the nearest strand interruption is beyond the mean tracking distance of the MMR complex, then the higher MutS level could increase the tracking attempts, resulting in a greater chance to reach distant strand interruptions. Repeated cycles of MutS binding to a mismatch and sliding away have been discussed as a basic process in MMR (50). There is not a similar effect of multiple mutS⁺ gene copies (like the strong decrease in homeologous recombination) on the spontaneous mutability of growing P. stutzeri, which is in accord with observations in other organisms (11, 28, 55, 95).

The contributions of MMR to limitation of interspecific recombination are different in different microorganisms. In E. coli, MMR provides the main barrier and plays a much stronger role than sequence divergence (62, 82, 90, 91), while MMR contributes only a small part to sexual isolation in B. subtilis transformation (about 16%) (47). The MMR system of S. pneumoniae has an intermediate role, providing about 34% of the sexual isolation during transformation (49). In P. stutzeri the contribution of MMR to sexual isolation is also small, about 14 and 16% depending on whether transformation occurred by means of PCR-amplified DNA fragments or chromosomal DNA. Zawadzki et al. (93) showed previously that in B. subtilis the relationship between sexual isolation and sequence divergence is rather robust with respect to the variation in donor DNA length. Our data are consistent with this finding. In transformation of Acinetobacter sp. strain BD4 the extent of sexual isolation conferred by MMR is about 22%, as estimated from the data presented by Young and Ornston (89). The observed strong reduction in interspecific transformation of cells with multiple copies of the mutS⁺ gene also supports the view that normally MMR does not provide a strong barrier to genetic exchange in P. stutzeri. It can be hypothesized that sexual isolation could be enforced by elevation of the level of MMR enzymes, as has been demonstrated for conjugation between enterobacteria (82). Temporal upregulation of the MMR system under natural conditions has not been reported yet. However, there is strong evidence that genetic and physiological conditions that decrease the level of MMR enzymes are encountered by cells (28, 42, 43, 51, 59, 63) and could foster the production of diversity by high spontaneous mutability and enhanced recombinational foreign DNA integration.

In the mutS mutant the frequency of HFIR events was not increased. This observation resembles previous findings for S. pneumoniae, in which an effect of mutS deficiency on HFIR was not seen (60). However, in the P. stutzeri mutS strain the illegitimate fusion sites were more frequently shifted away from the anchor sequence into the heterologous region, frequently leading to increased foreign DNA acquisition compared to the deleted DNA. The shift of the fusion point can be explained by MutS protein providing an obstacle to heteroduplex extension into nonhomologous sequences. Such extensions are catalyzed by RecA and facilitated by RuvAB to cover stretches that are 1 kbp or more long (1, 33). MutS was shown to impede strand exchange and branch migration between divergent DNA sequences by RecA (87, 88). Similarly, branch migration through regions of heterology stimulated by RuvAB was prevented through the MutSL complex (21). The MutS protein homodimer can form a clamp around the DNA duplex (40, 57), which presumably provides a roadblock for branch migration processes. The convergent evolution of MutS proteins and topoisomerase II enzymes suggests that MutS can effectively clamp DNA crossovers and Holliday junctions (79). It is possible that the MutS clamp can act during branch migration at the transition point from the anchor sequence to the heterologous region (86). The MutS homologs MSH2 and MSH3 of S. cerevisiae recognize and bind branched DNA structures, loops, and Holliday junctions (27).

Although in the mutS::aac strain many illegitimate recombination events occurred at GC-rich microhomologies, the events were more frequently independent of microhomologies. Illegitimate fusions without a 3-to-10 bp microhomology were not observed in transformants of Acinetobacter sp. strain BD413 (17) and S. pneumoniae (60), suggesting that in these organisms a mechanism of illegitimate recombination that is independent of short sequence identities is absent or much less prominent. Two novel findings were the identification in the mutS mutant of a hot spot of illegitimate recombination in the recipient DNA and the finding that 13 of 21 transformants had similar sequence motifs (5′-G/CAAC/G-3′ or 5′-G/CTTC/G-3′) at their fusion sites which mostly lacked microhomology. The sequence motif was always fused to different sites in the recombination partner molecules. This suggests that these events resulted from topoisomerase function. Microhomology-independent illegitimate recombination is promoted by topoisomerases, when the nicking and cloning reactions are separated and act on unrelated DNA ends (2, 5, 32, 54). Although topoisomerases generally do not have strong sequence specificity (83), several of the topoisomerases and also RuvC have AA or TT at or close to their cleavage sites (3, 71, 74, 83, 94), as found here. Some topoisomerases also break and join single-stranded DNA (83), which could fuse the transforming single strand to resident DNA.

In contrast to the hot spot identified here, the hot spots of illegitimate recombination during HFIR recently observed in Acinetobacter sp. strain BD413 cells transformed with tobacco plastid DNA (15) were a different kind. In these cases each hot spot encompassed the same sites in donor and recipient DNAs. Moreover, the hot spots contained microhomologies and were located at the end of high-G+C-content islands, which may have helped to form a ligatable structure of donor and recipient DNAs (15).

In a novel and unexpected class of transformants the anchor sequence was employed for both the homologous and illegitimate recombination events. In HFIR the homologous strand transfer was proposed to be the first step, followed by branch migration into the heterology catalyzed by RecA plus RuvAB (17). Branch migration can reverse direction (1, 33). It is conceivable that the MutS clamp bound at the beginning of the nonmatching heteroduplex region (27, 79) blocks reversed branch migration. In the mutS mutant this block would be absent, which might stimulate illegitimate fusion within the anchor region. Other hypotheses for double use of the anchor sequence such that the homologous recombination and the illegitimate fusion occur in the two anchor regions shortly after passage of the replication fork are also possible. In any case, a major genetic effect of mutS inactivation is the acquisition of foreign DNA without a loss of resident DNA in 14% of the HFIR events. Such recombinants have not been observed previously in mutS⁺ strains of P. stutzeri (24 transformants analyzed [53]), Acinetobacter sp. strain BD413 (60 transformants [18; C. Rohde and J. de Vries, unpublished data]), and S. pneumoniae (more than 50 transformants [60]). Only in one study of the integration of tobacco plastid DNA into the genome of Acinetobacter sp. strain BD413 was one such case among 32 transformants identified (15). To sum up, it can be said that a lack of MutS activity not only increases the recombinational integration of nucleotide sequences having single and multiple base changes but also modulates HFIR, resulting in longer stretches of integrated foreign DNA in the cases with a net gain of DNA and more frequent integrations with no loss of resident DNA. Such effects of MMR deficiency on foreign DNA acquisition could add to the increased spontaneous mutability, accelerating generation of genetic diversity and thereby adaptation of cells to a changing environment.