Correspondence: Shu-Lin Liu, Genomics Research Center, Harbin Medical University, 194 Xuefu Road, Harbin 150081, China. Tel.: +864 518 666 9150; fax: +864 518 750 6380; e-mail: firstname.lastname@example.org
Bacterial adaptation to changing environments can be achieved through the acquisition of genetic novelty by accumulation of mutations and recombination of laterally transferred genes into the genome, but the mismatch repair (MMR) system strongly inhibits both these types of genetic changes. As mutation and recombination do occur in bacteria, it is of interest to understand how genetic novelty may be achieved in the presence of MMR. Previously, we observed associations of a defective MMR genotype, 6bpΔmutL, with greatly elevated bacterial mutability in Salmonella typhimurium. To validate these observations, we experimentally converted the mutL gene between the wild-type and 6bpΔmutL in S. typhimurium and inspected the bacterial mutability status. When 6bpΔmutL was converted to mutL, the originally highly mutable Salmonella strains regained genetic stability; when mutL was converted to 6bpΔmutL, the mutability was elevated 100-fold. Interestingly, mutL cells were found to grow out of 6bpΔmutL cells; the new mutL cells eventually replaced the original 6bpΔmutL population. As conversion between mutL and 6bpΔmutL may occur readily during DNA replication, it may represent a previously unrecognized mechanism to modulate bacterial mutability at the population level, allowing bacteria to respond rapidly to changing environments while minimizing the risks associated with persistent hypermutability.
Previously, we observed the peculiar phenomenon of genome diversification, i.e. mutants of Salmonella typhimurium LT7 stored at room temperature kept changing the physical structure of the genome (Liu et al., 2003). We have since concentrated on identifying the genetic basis responsible for this phenomenon, with a focus on mismatch repair (MMR) genes including mutL, mutS and mutH. Our initial work showed that all screened S. typhimurium LT7 mutants had intact mutS and mutH genes, but a deletion was found in mutL; this genotype was designated 6bpΔmutL (Gong et al., 2007).
MutL has been suggested to function as a master coordinator or molecular matchmaker in the MMR system. It has a weak ATPase function, binds to DNA, interacts directly with MutS, MutH and UvrD, and is required for initiation as well as subsequent steps in MMR processes (Ban & Yang, 1998; Hall et al., 1998; Ban et al., 1999; Spampinato & Modrich, 2000). The deleted sequence that we identified in mutL is one of three tandem 6-bp repeats, GCTGGC GCTGGC GCTGGC. Similar repeats were reported in Escherichia coli, although they were presented as G CTGGCG CTGGCG CTGGCG (Shaver & Sniegowski, 2003). The sequence (G)CTGGC GCTGGC GCTGGC C in Salmonella and (G) CTGGCG CTGGCG CTGGCG in E. coli both code for the amino acid sequence LALALA, which lies in a region of MutL that forms the lid of the ATP-binding pocket (Ban & Yang, 1998; Ban et al., 1999; Yang, 2000; Yang et al., 2000). In our case, the deletion of one of the 6-bp repeats will lead to one of the three LA sets missing in the ATP lid structure of MutL and may thus impair ATPase activity. As the repeat structure would facilitate deletion or duplication via slipped-strand mispairing (Streisinger et al., 1966; Levinson & Gutman, 1987), one can imagine that MMR may cease functioning when the 6-bp repeats of mutL decrease or increase one or more copies and resume functioning when the copy number again becomes three, both by ‘errors’ during replication. In this way, MMR may modulate bacterial mutability rapidly, inhibiting or facilitating genetic changes in response to environmental changes.
To validate the potential role of mutL as a genetic switch experimentally, through allele conversion, we converted mutL between the wild-type and 6bpΔmutL alleles using gene replacement techniques and examined changes of bacterial mutability after the manipulations. Here, we report our findings and discuss the significance of conversion between mutL and 6bpΔmutL in bacterial adaptation at the population level.
Materials and methods
Bacterial strains, culture media and growth conditions
The bacterial strains used in the study are listed in Table 1 and were cultured as described previously (Gong et al., 2007). M9 minimal medium, supplemented with proline (100 μg mL−1), tyrosine (100 μg mL−1), leucine (100 μg mL−1), lysine (100 μg mL−1), methionine (100 μg mL−1) or streptomycin (100 μg mL−1), was used for transduction and conjugation experiments.
Wild-type or defective mutL was PCR-amplified from S. typhimurium LT7 strains with primers F1, CGGAATTCCGAACAGCGAAATGGCAAAC (EcoRI site underlined), and R1, GGATCCGCGGGTCAATCTCCAGATACAG (BamHI site underlined). PCR products were purified from agarose gels with QIAquick gel extraction kits (Qiagen) and an A-tailing nucleotide was added with Taq DNA polymerase (New England Biolabs) before cloning into pGEM-T (Promega) and introduction into chemically competent E. coli DH5α cells. Wild-type or defective mutL gene fragments were subcloned into EcoRI- and BamHI-digested pHSG415, which is a temperature-sensitive plasmid used for allele replacement via homologous recombination (White et al., 1999). Recombinant pHSG415 plasmids were first amplified in E. coli DH5α cells; after purification, these plasmids were transferred into S. typhimurium LT7 strains by transformation. The allelic-exchange experiments were carried out as described by White et al. (1999). PCR was used to screen colonies for bacterial cells bearing successful allele replacements. PCR products amplified with primers F2 (ATATCGACATCGAGCGTGGCGGCG) and R2 (GCTTTCGAGTCGTCAAGCGAGGCG) were resolved by agarose gel electrophoresis.
mutL gene knockout
The primer pair GK A1 (GGAATTCAACAGCGAAATGGCAAACT, EcoRI site underlined) and GK A2 (GCTTACAGAAATCTCCTTAATTCGC) was used to amplify a segment upstream of mutL, and the primer pair GK B1 (AGGAGATTTCTGTAAGCAAGGCGAG) and GK B2 (CGGATCCCAACGCCTCCCATCCAAG, BamHI site underlined) was used to amplify a segment downstream of mutL. The two amplified fragments were joined together by PCR and cloned into pHSG415, leading to pHSGAB, which contains the segments flanking mutL and was used to knock out the mutL gene by gene replacement (White et al., 1999).
Determination of mutation rates
Mutation rates were estimated by determining the frequency of spontaneous mutants resistant to rifampicin (Rif). Dilutions of overnight cultures grown in Luria–Bertani (LB) were spread on LB plates containing 100 μg mL−1 Rif and incubated at 37 °C. Dilutions of the samples were also plated on LB plates without antibiotics to determine the total number of CFUs. The colonies were scored for Rif resistance 24 h later. Mutation rates were determined as described by Foster (2006).
Bacteriophage P22-mediated transduction was used to inactivate proB, tyrA, leu, lysA, or metC in S. typhimurium LT7 and its 6bpΔmutL derivatives by transferring Tn10 insertions from S. typhimurium LT2, as described (Liu et al., 1993; Liu, 2007). For phenotype tests, 100-μL aliquots of overnight cultures were plated on M9 minimal media with or without the corresponding nutrients. We used phage P22 grown on Salmonella typhi Ty2 (Liu & Sanderson, 1995) as the donor for transduction frequency tests. For each transduction, 100 μL of recipient cells grown to 5 × 108 CFU mL−1 were infected with 10 μL of phage lysate diluted to yield a phage/bacteria ratio of 1 : 10. Bacterial cultures and phage lysates were mixed directly on M9 minimal medium plates containing glucose (8 mg mL−1) and incubated at 37 °C for 18 h. The transduction frequency was calculated by determining the number of cells growing on M9 plates divided by the total number of CFUs from three independent experiments.
Conjugational crosses and recombination frequencies
We used E. coli Hfr 3000 (leuD+; see Table 1) as the donor. Spontaneous mutants of S. typhimurium cells resistant to streptomycin (StrR) were isolated and made leuD− by Tn10 insertion inactivation for use as the recipients. Donor and recipient cells were separately grown in LB broth to 2 × 108 cells mL−1, mixed (1 : 1) and incubated for 40 min at 37 °C. LB (0.5 mL) was added and the mating mix was incubated for an additional 1 h. The culture mixture was plated on M9 containing streptomycin (100 μg mL−1), thiamine (30 μg mL−1) and glucose (8 mg mL−1). The Hfr donor cells were counter-selected by streptomycin and the recipient cells were unable to grow in the absence of leucine. Recombination frequencies were expressed as the number of recombinants per Hfr donor.
Computational modeling for functional prediction of the protein product encoded by 6bpΔmutL in comparison with the wild-type MutL
To elucidate the role of 6bpΔmutL in bacterial mutability dynamics, we first needed to determine whether 6bpΔmutL-encoded protein might still have a certain level of function or is entirely nonfunctional, especially considering that the 6-bp deletion results only in the deletion of two amino acids, L and A, without frame shifting or protein truncation. We thus carried out computational modeling, which showed that the LA deletion fell in the ATP-binding region and so would disrupt the conformation of the region, making ATP binding impossible (Fig. 1). To validate the bioinformatic prediction experimentally, we knocked out the whole mutL gene and made comparisons among wild-type mutL, 6bpΔmutL and mutL deletion strains. We found that 6bpΔmutL and mutL deletion strains had similar levels of mutability, demonstrating that 6bpΔmutL completely lost function.
Experimental conversion between mutL and 6bpΔmutL alleles
To rule out the possibility that defects other than 6bpΔmutL might complicate the mutability studies, we experimentally converted mutL between the wild-type and the 6bpΔmutL alleles and examined the mutability status of the bacteria after the conversion, starting with S. typhimurium LT7 mutant strain 8608F2 (Table 1), which was described previously (Liu et al., 2003). Having confirmed by sequencing that 8608F2 had the 6bpΔmutL genotype, we converted the allele into the wild-type mutL and obtained 8608F2mutL. In a parallel series of experiments, we converted the mutL of S. typhimurium LT7 strain SGSC1417 into 6bpΔmutL and obtained SGSC14176bpΔmutL. We also converted the 6bpΔmutL allele of strains 8111C and 9052D142332 into mutL and confirmed the genotypes of the strains by sequencing after the conversion experiments.
Elevated mutation rates of bacterial cells containing 6bpΔmutL
To test correlations between high mutability and the 6bpΔmutL genotype, we measured the frequency of spontaneous mutants resistant to rifampicin (RifR) in 8608F2, 8111C and 9052D142332 (Table 1); they all had mutation rates of approximately 10−6 per cell generation. Notably, the mutL-knocked SGSC1417 (SGSC1417ΔmutL) and SGSC1417 with the 6-bp deletion (SGSC14176bpΔmutL) had similar levels of mutation rates, comparable to those of 8608F2, 8111C and 9052D142332 (Fig. 2), implying total loss of function of MutL encoded by 6bpΔmutL. In parallel experiments, SGSC1417 (S. typhimurium LT7 with the wild-type mutL) and 9052D1a (wild-type mutL derivative of the 6bpΔmutL strain 9052D1; Gong et al., 2007) had mutation rates of approximately 10−8 per cell generation. After replacement of 6bpΔmutL with mutL, 8608F2, 8111C and 9052D142332 became 8608F2mutL, 8111CmutL and 9052D142332mutL, respectively, and their mutation rates dropped 100-fold to 10−8 per cell generation (Fig. 2).
Elevated recombination frequencies in 6bpΔmutL cells
Next, we estimated and compared homologous recombination frequencies of 6bpΔmutL and mutL cells by transduction of DNA from S. typhi. We transferred Tn10 in proB, tyrA, leuD, lysA and metC from S. typhimurium LT2 to S. typhimurium LT7 derivatives, including SGSC1417, SGSC14176bpΔmutL, 8608F2 and 8608F2mutL, and confirmed the auxotropic phenotypes of the transductants. We then used P22 lysates prepared on S. typhi Ty2 to transduce the S. typhimurium LT7 mutants carrying the Tn10 insertions and screened the M9 plates for proB+, tyrA+, leuD+, lysA+ or metC+ transductants. For SGSC1417 and 8608F2mutL, the transduction efficiencies for all five Tn10 insertion mutants were <10−8; by contrast, for SGSC14176bpΔmutL and 8608F2, the transduction efficiencies for all five Tn10 insertion mutants were two orders of magnitude higher, similar to the levels of the ΔmutL strain of SGSC1417 (Table 2; only SGSC1417 data are given for conciseness). These data are consistent with previous findings on mutL (e.g. Zahrt et al., 1994); here, we repeated the same kind of experiments to demonstrate the association of greatly elevated mutability with the specific 6-bp lesion of mutL, which will be the basis for the proposal of spontaneous conversion between mutL and 6bpΔmutL as a genetic switch in bacterial evolution. In particular, these experimental results support the presumption that the 6bpΔmutL genotype facilitates homologous recombination and thus provides the bacteria many more chances to incorporate beneficial DNA of foreign sources.
Table 2. Representative results of transduction frequencies among mutL and 6bpΔmutL Salmonella typhimurium LT7 strains
4–5 × 10−6
5–6 × 10−6
1–2 × 10−6
1–2 × 10−6
2–3 × 10−6
3–4 × 10−6
2–3 × 10−6
2–3 × 10−6
1–2 × 10−6
1–2 × 10−6
Recombination frequency differences estimated by conjugation
To further confirm differences in homologous DNA recombination efficiency between 6bpΔmutL and mutL cells, we carried out crosses between S. typhimurium LT7 (SGSC1417 and 8608F2 series) and E. coli Hfr cells by conjugation, using E. coli Hfr 3000 as the donor (Low, 1973; Theze et al., 1974). With SGSC1417 or 8608F2mutL as the recipient, the conjugation frequencies were <10−8 per Hfr; with SGSC14176bpΔmutL, 8608F2 or the ΔmutL strain of SGSC1417 as the recipient, the conjugation frequencies were >10−6 per Hfr (Fig. 3).
An overall tendency of spontaneous conversions from 6bpΔmutL to mutL alleles
As the 6bpΔmutL genotype significantly facilitated mutability, as shown above, and because the 6-bp tandem repeat structure easily leads to copy number changes through slipped-strand mispairing (Streisinger et al., 1966; Levinson & Gutman, 1987), we hypothesized that bacterial populations dominated with 6bpΔmutL cells under some kind of selective pressure may begin having mutL cells when the pressure is no longer present, and the mutL cells may continuously increase in number or even replace the original 6bpΔmutL population. As the S. typhimurium LT7 mutants had suffered from starvation during stock in sealed agar stab cultures under room temperature for over 40 years, we cultured the bacteria on LB plates to ‘remove’ such starvation pressures. We started with strain 9052D1, which was the first strain to have the MMR genes sequenced in our laboratory and was found to have the 6bpΔmutL genotype (Gong et al., 2007). During the first plating, a minority of colonies contained both 6bpΔmutL and mutL cells (Fig. 4a, lane 9). When such colonies were restreaked, most of them made only mutL cells (data not shown). The 6bpΔmutL cells continued making both 6bpΔmutL and mutL cells, but very few mutL cells made 6bpΔmutL cells (Fig. 4b), which demonstrates a much stronger tendency of the 6bpΔmutL allele to convert to mutL than in the opposite direction.
Bacteria use several strategies to increase mutability for acquiring genetic novelty in adaptation to changing environments, involving, in addition to allele conversion of the MMR genes as reported in this paper, the RpoS regulon, SOS responses, DinB error-prone DNA polymerase, RecA, etc., as has been documented widely (Bjedov et al., 2003; Friedman et al., 2005; Ponder et al., 2005; Finkel, 2006; Galhardo et al., 2007, 2009; Sundin & Weigand, 2007; Weigand & Sundin, 2009). Although these mechanisms undoubtedly all lead to high mutability state and hence contribute to bacterial evolution, only allele conversion of MMR genes by copy number changes of repeat units through slipped-strand mispairing is likely to work in a timely way. This is because such allele conversion occurs constantly as stochastic events among cells of the bacterial population: at any time point, there must be some cells that have the wild-type allele converted to the defective one, for example from wild-type mutL to 6bpΔmutL as demonstrated here. So the conversion could be regarded as spontaneous and MMR-negative cells should start accumulating genetic changes swiftly, with those that acquired the ‘right’ genetic traits becoming more fit and thus selected for. Mutation rate variation in relation to fitness differences has been reported (Saunders et al., 2003), and here we provide experimental evidence showing that levels of mutability, or variation of mutation rates, could be modulated by allele conversion of an MMR gene.
Tandem repeats in bacterial genomes have been widely studied in relation to pathogenicity or escape from the host immunity (Hollingshead et al., 1987; Madoff et al., 1996; Tonjum et al., 1998; Jordan et al., 2003), but to our knowledge we are the first to report tandem repeats in MMR genes, and playing important roles in modulating bacterial mutability. The repetitive structure within the sequence of mutL enables it to function as a genetic switch modulating bacterial mutability, to be turned on or off at the population level nearly spontaneously. Fluctuations in the frequency of 6bpΔmutL cells in the bacterial populations would provide the bacteria with exceptional adaptation potential. One point that we need to emphasize is that the genetic ‘switch’ is turned on or off at the population level by selection of existing cells possessing the favorable traits, rather than at a single-cell level by any induced adaptive mechanisms.
In conclusion, the 6bpΔmutL genotype will facilitate genetic variation and gradual divergence of the bacteria. It is important to note, however, that the scenario presumed to be occurring in the sealed agar stabs is a slow-played version of what might occur in nature, because, in the natural environment, where competition constantly works to renew survivors, cells will quickly become predominant in the populations if selected, or quickly disappear if counter-selected, leading to reasonably low frequencies of 6bpΔmutL cells most of the time. The basic idea of a spontaneous genetic switch model may be generalized to other bacteria, because all bacteria evolve and, when doing so, need a genetic switch to transiently allow the genome to accept foreign DNA or to accumulate mutations. Eventually, variants of the bacteria with beneficial genetic changes will be selected. However, the three 6-bp tandem repeats have been found only in Salmonella and very closely related bacteria (Shaver & Sniegowski, 2003), suggesting that other bacteria may use different molecular mechanisms for the genetic switch.
This work was funded by a CIHR grant to R.N.J.; a grant of the National Natural Science Foundation of China (NSFC30970078) and a grant of the Natural Science Foundation of Heilongjiang Province of China to G.-R.L.; a grant from Harbin Medical University, a 985 Project grant of Peking University Health Science Center, grants of the National Natural Science Foundation of China (NSFC30870098, 30970119), and Specialized Research Fund for the Doctoral Program of Higher Education (SRFDP, 20092307110001) to S.-L.L.
F.C., W.-Q.L. and Z.-H.L. contributed equally to this work. W.-Q.L. was a visiting student to Harbin Medical University.