Levels of the Vsr Endonuclease Do Not Regulate Stationary-Phase Reversion of a Lac⁻ Frameshift Allele in Escherichia coli

Abstract

Vsr endonuclease, which initiates very short patch repair, has been hypothesized to regulate mutation in stationary-phase cells. Overexpression of Vsr does dramatically increase the stationary-phase reversion of a Lac⁻ frameshift allele, but the absence of Vsr has no effect. Thus, at least in this case, Vsr has no regulatory role in stationary-phase mutation, and the effects of Vsr overproduction are likely to be artifactual.

When Escherichia coli and other microorganisms are subjected to nonlethal selective pressure, mutations can arise even though the cells are not dividing (2). This phenomenon has been called directed, adaptive, or stationary-phase mutation (reviewed in reference 6). Recently, it has been shown that in one case of stationary-phase mutation, reversion of an episomal frameshift mutation in E. coli, the mutations are not truly adaptive but may arise in a subpopulation of cells experiencing a transient state of enhanced mutation (7, 20). Methyl-directed mismatch repair (MMR), which is an important contributor to replication fidelity in dividing cells (17), is equally important in reducing stationary-phase mutations; thus, MMR activity cannot be globally limiting in nutritionally deprived or stationary-phase cells (reviewed in reference 8). Nonetheless, it has been suggested that MMR could become limiting in a subpopulation of cells and that these cells could then give rise to stationary-phase mutations (10).

Very short patch repair (VSR) is a DNA repair pathway that corrects T · G mismatches to C · G at CC(A/T)GG and similar sequences (reviewed in reference 13). The 5 position of the internal C’s of CC(A/T)GG sites in E. coli is methylated by the Dcm methylase, which is encoded by the dcm gene (reviewed in reference 16). Because deamination of 5-methylcytosine produces thymine, Dcm recognition sites are hot spots for mutation (3, 5). Although it is not known why E. coli methylates CC(A/T)GG sites, it is clear that VSR prevents G · C-to-A · T mutations that can result from this methylation (12, 18).

VSR is initiated by the Vsr endonuclease, which is encoded by the vsr gene. When Vsr is overproduced, a wide variety of mutations result independently of Dcm methylation (4, 15). Efficient VSR requires MutS and MutL, two proteins that participate in MMR (11); excess Vsr appears to deplete MMR proteins, particularly MutL, resulting in a general mutational state (15). VSR is active in stationary-phase cells (14), and it is possible that stationary-phase mutations are due to reduction of MMR caused by upregulation of VSR in a subpopulation of cells (10, 15). The experiments reported here refute this hypothesis by showing that in the complete absence of Vsr, cells experience a normal level of stationary-phase mutation.

The mutational target for stationary-phase mutation in our strain, FC40, is a +1 frameshift affecting the lacZ gene carried on an F′ episome (1). When plated on lactose minimal medium, FC40 does not divide but does produce Lac⁺ mutations at a constant rate for several days. Unlike growth-dependent (preplating) mutations, stationary-phase (postplating) Lac⁺ mutations are dependent on recombination functions and consist mainly of −1-bp frameshifts in runs of iterated bases (reviewed in reference 8).

When FC40 was transformed with a plasmid overproducing Vsr, both the preplating and the postplating mutation rates increased about 10²-fold (Table 1, experiment 1). This increase is equivalent to that produced by loss of MMR (1). As shown in Table 1, experiment 2, Vsr overproduction had no effect on stationary-phase mutation in a mutL mutant strain, supporting the hypothesis that Vsr overproduction depletes MMR activity. Interestingly, unlike normal postplating mutation in this strain, the Vsr-induced Lac⁺ mutations were not totally dependent on recA (Table 1, experiment 3).

TABLE 1.

Effect of overproduction of Vsr⁺ on stationary-phase reversion of a Lac⁻ frameshift allele^a

Expt and strain	Relevant allele(s)	Plasmid	nb	Mean no. of Lac+ revertants/108 cells plated ± SEM
				Day 2c	Days 3–5d
1
FC899	Wild type	Control	5	13 ± 2	51 ± 4
FC900	Wild type	Vsr+	5	819 ± 368	3,218 ± 324
2
FC1150	mutL	Control	4e	124 ± 79	1,956 ± 251
FC1151	mutL	Vsr+	5	280 ± 63	1,874 ± 195
3
FC1154	recA	Control	5	0.3 ± 0.2	0.5 ± 0.2
FC1155	recA	Vsr+	5	200 ± 20	56 ± 12
FC1159	recA mutL		5	86 ± 30	15 ± 4

^aCells (10⁸ or 10⁶) of the revertible strains were mixed with cells (10⁹) of a nonrevertible scavenger strain and plated in 2.5 ml of top agar on M9 lactose plates as previously described (1). The Vsr⁺ plasmid was pKKV, and the control plasmid was pKK223-2 (4; obtained from C. G. Cupples, Concordia University). The scavenger strain carried the control plasmid where appropriate. mutL211::Tn5 was transduced from NR9360 (obtained from R. Schaaper, National Institute of Environmental Health Sciences); the recA allele was recA938::Tn9 (obtained from the E. coli Genetic Stock Center). Except for the relevant alleles, the strains are isogenic, with the following two exceptions, neither of which influences mutation rates: (i) all but the mutL strains are rifampin resistant; and (ii) the episomes in FC1154 and FC1155 carry a Tet^s dTn10 element (7). 

^bNumber of independent cultures in the experiment. 

^cMean number of Lac⁺ mutants appearing 2 days after plating; these are due to growth-dependent mutations. 

^dMean number of Lac⁺ mutants appearing each day from day 3 to day 5; these are due to stationary-phase mutations. 

^eOne culture with a jackpot of Lac⁺ mutants was eliminated. 

So far, our results confirm for stationary-phase mutations the results obtained previously for growth-dependent mutations (15). However, the hypothesis that Vsr normally determines the rate of stationary-phase mutation predicts that stationary-phase mutations should decline or disappear in a vsr mutant. To test this prediction, we moved into our strain by P1 transduction a deletion that includes the overlapping vsr and dcm genes. The donor was a strain (CC221/F′) in which this deletion is linked to a dTn10 element (4), and the recipient was the F⁻ parent of FC40, FC36 (1). After selection for tetracycline resistance (Tet^r, encoded by the dTn10 element), Tet^r isolates were screened for loss of motility (Mot⁻), which is also conferred by the deletion. The Lac⁻ frameshift-carrying episome was then mated into Mot⁻ and Mot⁺ isolates, creating FC1106 and FC1107 (the latter control demonstrated that the transposon itself had no effect on mutation; see Table 2, experiment 4). To confirm that the dcm gene was, indeed, lost in Mot⁻ transductants, the strains were transformed with a plasmid and plasmid DNA was isolated and digested with endonuclease EcoRII, which is blocked by Dcm methylation (16). Only when isolated from FC1106 was plasmid DNA sensitive to cleavage by EcoRII (data not shown). Although only one pair of strains was used for the experiments presented here, several other paired isolates, produced as described above or by backcrossing from FC1106, were shown to have the same phenotypes.

TABLE 2.

Loss of vsr and dcm does not affect stationary-phase reversion of a Lac⁻ frameshift allele^a

Expt and strain	Relevant allele	Plasmid	nb	Mean no. of Lac+ revertants/108 cells plated ± SEM
				Day 2c	Days 3–5d
4
FC40	Wild type		10	1 ± 0.3	20 ± 2
FC1107	Wild type		10	2 ± 0.3	25 ± 2
FC1106	Δ(vsr dcm)		10	2 ± 0.6	22 ± 1
5
FC1141	Wild type	Control	5	4 ± 2	21 ± 2
FC1142	Wild type	Vsr+	5	316 ± 76	2,111 ± 268
FC1139	Δ(vsr dcm)	Control	4e	3 ± 1	28 ± 2
FC1140	Δ(vsr dcm)	Vsr+	5	407 ± 57	3,008 ± 319
6
FC1147	Wild type	Control	5	6 ± 1	43 ± 2
FC1148	Wild type	Dcm+	5	10 ± 1	48 ± 3
FC1143	Δ(vsr dcm)	Control	4e	11 ± 2	34 ± 2
FC1144	Δ(vsr dcm)	Dcm+	5	6 ± 1	21 ± 1

^aExperimental protocols and the Vsr⁺ plasmid and its control were as described in Table 1, footnote a. In experiment 4, two scavenger strains were used, one wild type and one Δ(vsr dcm); this made no difference, so the results were combined. In experiment 6, the control plasmid was a Tet^s derivative of pBR322 (pSTL174; obtained from S. T. Lovett, Brandeis University) and the Dcm⁺ plasmid was pDcm21 (19; obtained from A. S. Bhagwat, Wayne State University). FC1107, FC1141, and FC1142 carry the same chromosomal dTn10 element as the Δ(vsr dcm) strains. 

^bNumber of independent cultures in the experiment. 

^cMean number of Lac⁺ mutants appearing 2 days after plating; these are due to growth-dependent mutations. 

^dMean number of Lac⁺ mutants appearing each day from day 3 to day 5; these are due to stationary-phase mutations. 

^eOne culture with a jackpot of Lac⁺ mutants was eliminated. 

As shown in Table 2, experiment 4, loss of vsr and dcm had no effect on stationary-phase reversion of the Lac⁻ frameshift allele. The stationary-phase Lac⁺ mutations in the Δ(vsr dcm) strain were normal in that they were totally dependent on recA. [In a small-scale experiment (9), the numbers of postplating Lac⁺ mutants per sector were 0.2 ± 0.2 for a Δ(vsr dcm) recA strain and 0.9 ± 0.3 for a vsr⁺ dcm⁺ recA control.] Because the mutagenic effect of Vsr overproduction is independent of Dcm methylation (15; see below), these results disprove, at least for the Lac⁻ frameshift allele, the hypothesis that the level of Vsr determines stationary-phase mutation.

We additionally tested whether Vsr or Dcm overproduction would affect mutation in the Δ(vsr dcm) background (Table 2). Overproduction of Vsr was equally mutagenic in the deletion strain and in the wild-type strain (Table 2, experiment 5), confirming that Vsr-induced mutation does not depend on dcm methylation (4). Overproduction of Dcm had a small inhibitory effect on Lac⁺ mutation in the Δ(vsr dcm) strain (Table 2, experiment 6) (although it was within the range of variation of our experiments, we have confirmed this result in several additional experiments). This might be due to loss of viability, but, at least during the course of this experiment, we detected no difference in viability between the Δ(vsr dcm) and wild-type strains.

In conclusion, overproduction of Vsr is as powerful a mutator in stationary-phase cells as it is in growing cells. In addition, our data suggest that Vsr overproduction may induce some postplating Lac⁺ mutations by a recA-independent pathway. In contrast, overproduction of Dcm in the absence of Vsr has a small inhibitory effect on stationary-phase mutation. However, these effects appear to be artifacts of overproduction because loss of vsr and dcm does not prevent the normal level of Lac⁺ reversion in stationary-phase cells. Thus, at least in this case, Vsr has no regulatory role in stationary-phase mutation.