Reduction of GC → TA Transversion Mutation by Overexpression of MutS in Escherichia coli K-12

Abstract

Overexpression of the MutS repair protein significantly decreased the rate of lacZ GC → TA transversion mutation in stationary-phase and exponentially growing bacteria and in mutY and mutM mutants, which accumulate mismatches between 8-oxoguanine (8-oxoG) and adenine residues in DNA. Conversely, GC → TA transversion increased in mutL or mutS mutants in stationary phase. In contrast, overexpression of MutS did not appreciably reduce lacZ AT → CG transversion mutation in a mutT mutant. These results suggest that MutS-dependent repair can correct 8-oxoG:A mismatches in Escherichia coli cells but may not be able to compete with mutation fixation by MutY in mutT mutants.

The MutS homodimer binds to mismatched bases, small bulge loops, and damaged bases as an initial step in methyl-directed DNA mismatch repair in Escherichia coli (2, 4, 7, 22, 26, 27, 29, 30, 34). MutS is also thought to play roles in methyl-independent repair pathways, such as very-short-patch repair (18). The amount of MutS is strongly down regulated as E. coli cells enter stationary phase (9, 31). This regulation is partly mediated at the level of mRNA stability by the RNA chaperone protein Hfq, which acts to destabilize the mutS transcript (31). Down regulation of the amount of MutS in stationary-phase cells may help to coordinate repair capacity with decreased DNA replication (9, 31). In addition, decreased amounts of MutS could contribute to increased mutagenesis or homeologous recombination in stationary-phase cells or in bacterial cells that resume growth after stationary phase. Transfection experiments of phage DNA containing heteroduplexes suggest that mismatch repair capacity is indeed decreased in cultures grown overnight and that this reduced repair capacity can be reversed by overexpression of MutS but not MutL (5).

To investigate further possible effects of the amount of MutS on mutation rate, we determined mutation rates in E. coli cells that gradually enter stationary phase. Hall devised a papillation assay to determine the rates of mutation as the Cupples-Miller base substitution and frameshift tester strains enter and remain in stationary phase (8, 12). In this assay, bacteria are spread onto plates containing 0.01% glycerol, 0.1% lactose, and X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) at 30°C. Bacterial colonies stop growing when they exhaust the limited glycerol; however, blue lacZ⁺ papillae continue to appear gradually within colonies. In contrast to the abrupt cessation of growth after carbon source deprivation, which is used routinely to determine adaptive mutagenesis (10, 15), the total viable bacteria per day increased very slowly in this papillation assay (data not shown). Hall reported that of the possible transitions, transversions, and frameshifts, the most frequent mutation in stationary-phase bacteria was the GC → TA transversion in tester strain CC104 (12). We obtained results similar to those of Hall for CC104 and CC104 containing an empty cloning vector (pVector) (Fig. 1) (15).

FIG. 1.

Overexpression of MutS and to a lesser extent MutL decreased the number of lacZ GC → TA transversion mutants in E. coli strain CC104 in stationary-phase colonies. Experiments were done essentially by the method of Hall (12). CC104 containing the indicated plasmids described in reference 15 was grown overnight at 30°C with shaking in Luria-Bertani medium containing 100 μg of kanamycin per ml. Cells were collected by centrifugation and washed twice by resuspension in minimal A salts. About 200 cells of each strain were spread onto plates containing minimal A salts medium containing 0.01% (vol/vol) glycerol, 0.1% (wt/vol) lactose, 1 mM MgSO₄, 40 μg of X-Gal per ml, 50 μg of kanamycin per ml, and 1.5% (wt/vol) Bacto Agar (12). Plates were incubated at 30°C, and blue lacZ⁺ papillae were counted each day. The experiment was repeated several times with essentially the same results. Control experiments demonstrated that there was no detectable plasmid loss from the strains even after 8 days of incubation.

Unexpectedly, we found that overexpression of MutS, and to a lesser extent MutL, significantly decreased GC → TA transversion in CC104 under these experimental conditions (Fig. 1) (15). MutL is an ATPase that interacts with DNA, MutS, MutH, and UvrD during methyl-directed mismatch repair (3, 11, 13, 14, 28, 34, 35). Previously, we demonstrated that pMutS and pMutL overexpress MutS and MutL by about 60- and 20-fold, respectively, in growing and starved bacteria (15). In contrast to CC104, MutS or MutL overexpression did not significantly affect the appearance of lacZ⁺ papillae by frameshift tester CC107 (data not shown), which is extremely sensitive to defects in methyl-directed mismatch repair (8). Transition mutation testers CC102 (GC → AT) and CC106 (AT → GC), which are also sensitive to defects in methyl-directed mismatch repair (8), were not used in these assays. In our hands, the background color of CC102 colonies was too blue to detect papillae reliably (data not shown), and the number of lacZ⁺ papillae formed by CC106 colonies was very low, even after prolonged incubation (12).

Control experiments indicated that CC104(pVector), CC104(pMutS), and CC104(pMutL) formed the same number of total viable cells per plate per day to 4 days of incubation after which lacZ⁺ papillae formation became significant for CC104(pVector) (Fig. 1; data not shown). We used CC101, which is the tester for AT → CG transversion (8), to determine the total number of viable cells per plate to 8 days, because CC101, CC101(pVector), CC101(pMutS), and CC101(pMutL) are isogenic, except for the lacZ allele, grew at rates similar to those of the corresponding CC104 derivatives, and formed a negligible number of lacZ⁺ papillae in this assay (data not shown) (12). We did not use a P90C [F′ Δ(lacI lacZ)] strain, which should have been isogenic to CC101 and CC104, because our isolate of this strain exhibited different growth characteristics from those of CC101 and CC104 on these media (data not shown). The number of total viable cells per day was used to convert data such as those in Fig. 1 into mutation rates by the method of Hall (12). The rate of lacZ GC → TA transversion was decreased by about 4.3- or 1.6-fold by MutS or MutL overexpression, respectively (Fig. 2A), by a mechanism that was independent of recA function (data not shown). In contrast, the form of adaptive mutagenesis studied in most detail is strictly dependent on recA function (10, 15).

FIG. 2.

MutS and to a lesser extent MutL overexpression reduced the rates of lacZ GC → TA transversion in bacterial colonies in stationary phase on plates (A) and in exponentially growing liquid cultures (B). (A) Rates of mutation were determined by the method of Hall (12) using time courses, such as those shown Fig. 1, that were normalized for cell numbers using isogenic E. coli CC101 derivatives, which grew identically to CC104 derivatives on these media and formed a negligible number of lacZ⁺ papillae (see text). Mutation rates are averages for days 5, 6, and 7. The experiment was performed three times, and standard errors of the mean are indicated by the error bars. (B) At least 30 independent cultures of each strain were grown at 30°C to mid-exponential phase in minimal A salts containing 0.3% (vol/vol) glycerol, 1 mM MgSO₄, and 50 μg of kanamycin per ml. Cells were collected by centrifugation, washed twice in minimal A salts, and added to molten soft agar that was poured onto plates containing minimal A salts medium containing 1 mM MgSO₄, 50 μg of kanamycin per ml, 1.5% (wt/vol) Bacto Agar, and either 0.1% (wt/vol) lactose (for lacZ⁺ revertants) or 0.2% (wt/vol) glucose (for total viable cells). Plates were incubated at 30°C for 48 h before bacterial colonies were counted. Reconstitution control experiments indicated that 48 h was sufficient for the appearance of lacZ⁺ colonies that arose in exponentially growing cultures before plating, whereas adaptive mutants that arose on the plates began to appear after 72 h of incubation. Mutation rates were calculated by the method of Lea and Coulson (17). The experiment was performed three times, and standard errors of the mean are indicated by the error bars.

The rate of lacZ GC → TA transversion was determined by the method of Lea and Coulson (17) in cells growing exponentially at 30°C in minimal A salts medium containing 0.3% glycerol (Fig. 2B). MutS overexpression reduced the rate of lacZ GC → TA transversion by 2.9-fold, whereas the smaller 1.4-fold reduction by MutL overexpression was statistically marginal. Consistent with Hall's results (12), the mutation rate of the CC104(pVector) strain was about sixfold higher in stationary-phase cells in the papillation assay than in exponentially growing cells. We obtained essentially the same rate of GC → TA transversion for exponentially growing CC104 by using the Lea and Coulson method as Hall did using the Mutants C method (Fig. 2B) (12).

To examine whether down regulation of the amount of MutS might contribute to the increase in GC → TA transversion in E. coli CC104, we determined the mutation rates of CC104 hfq-1 and CC104 hfq-2 mutants, which contain inactive or active Hfq protein, respectively (31, 32). We showed previously that the amount of MutS did not decrease in stationary-phase cultures of hfq-1 mutants grown overnight (31). We did not detect a significant difference in the rate of lacZ GC → TA transversion in CC104 hfq⁺ and either hfq mutant (data not shown). However, the hfq-1 mutants exhibit defective rpoS expression (31), and CC104 hfq-1 showed a significant lag in the formation of papillae and in growth yield in these experiments, which lasted for several days (Fig. 1). Therefore, while not completely conclusive, these experiments suggest that down regulation of MutS amount may not play a major role in the accumulation of GC → TA transversion mutations under these physiological conditions. Consistent with this interpretation, MutS overexpression did not influence the rate of frameshift mutation of CC107, which is very sensitive to defects in mismatch repair (8), at least in this papillation assay (see above).

One mechanism, but certainly not the only possible mechanism, for the increased GC → TA transversion in E. coli CC104 in the papillation assay is oxidative damage of guanine bases to 8-oxoguanine (8-oxoG) (20, 21, 23). 8-oxoG, which can mispair with adenine residues in DNA and seems to accumulate in nongrowing bacteria (6), is repaired in E. coli by the GO repair pathway that includes the MutY and MutM DNA glycosylases (1, 19, 20, 21, 24). We tested whether MutS or MutL overexpression could suppress lacZ GC → TA transversion in mutY or mutM mutants, which are thought to accumulate GC → TA transversions through uncorrected A:8-oxoG mismatches (1, 24). Overexpression of MutS, but not MutL, decreased the rate of lacZ GC → TA transversion in a mutY mutant by 3.4-fold to the level in the CC104 mutY⁺(pVector) strain (Fig. 3). Overexpression of MutS also significantly decreased the frequency of lacZ GC → TA transversion in a mutM mutant as judged by visual inspection of formation of papillae (data not shown). These results suggest that MutS can recognize and lead to the correction of A:8-oxoG mismatches in vivo. Consistent with this hypothesis, we found that mutL or mutS mutants have a higher rate of lacZ GC → TA transversion than their parent in this papillation assay (Fig. 4).

FIG. 3.

MutS but not MutL overexpression decreased the rate of GC → TA transversion in an E. coli CC104 mutY mutant defective in GO repair. CC104 mutY::miniTn10(Tet^r) was constructed by generalized transduction with bacteriophage P1vir. Mutation rates were determined in stationary-phase cells by the papillation assay described in the legends to Fig. 1 and 2A and reference 12 and are the averages for day 5. The experiment was performed three times, and standard errors of the mean are indicated by the error bars.

FIG. 4.

The rate of GC → TA transversion increased in mutS and mutL mutants of E. coli CC104. CC104 mutL::Ω(Km^r; BsaAI) and CC104 mutS::Ω(Km^r; EcoRV) were constructed by generalized transduction with bacteriophage P1vir (31). Mutation rates were determined in stationary-phase cells by the papillation assay described in the legends to Fig. 1 and 2A and reference 12 and are the averages for days 5, 6, and 7. The experiment was performed three times, and standard errors of the mean are indicated by the error bars.

Taken together, the results reported here suggest that MutS-dependent repair can recognize and lead to the correction of A:8-oxoG mismatches in vivo in E. coli K-12. A role for MutS-dependent mismatch repair could account for the increase in GC → TA transversion observed in mutL or mutS mutants (Fig. 4). Consistent with our findings, it was recently reported that the Saccharomyces cerevisiae MSH2-MSH6 heterodimer, which is a homologue of E. coli MutS, binds to A:8-oxoG mismatches in vitro and likely corrects these lesions in vivo (25). Moreover, E. coli MutS was recently reported to bind to mismatches between 5-formyluracil, which is formed by oxidative damage of thymine, and guanine residues in DNA (30). In the latter study, it was mentioned that MutS did not seem to bind appreciably to A:8-oxoG mismatches in vitro (30). Our in vivo results suggest that this biochemical issue needs to be reappraised and that MutS-dependent repair may play a general role in the correction of mismatches that arise when bases are oxidatively damaged in bacteria and yeast.

Finally, we observed by visual inspection that MutS overexpression did not reduce the frequency of AT → CG transversion in an E. coli CC101 mutT mutant (data not shown). MutT participates in GO repair by converting 8-oxoGTP to 8-oxoGMP and thereby reducing 8-oxoGTP mispairing with adenine bases in DNA during replication (20). In mutT mutants, AT → CG transversions result from MutY removing adenines of 8-oxoG:A mismatches arising from 8-oxoGTP incorporation (20, 33). The lack of effect of MutS overexpression on AT → CG transversion in a mutT mutant could simply reflect more efficient mutation fixation of 8-oxoG:A mismatches by MutY than repair by a MutS-mediated pathway, which would not be detected as a AT → CG transversion. Alternatively, MutS-mediated repair may correct A:8-oxoG mismatches that arise in mutY and mutM mutants more efficiently than it corrects 8-oxoG:A mismatches that arise in mutT mutants, where the first base in each pair is contained in a newly replicated DNA strand. As a precedent, a strand-specific mechanism mediated by MutS homologs to correct oxidatively damaged bases in human cells has been proposed (16).