Competition between MutY and Mismatch Repair at A · C Mispairs In Vivo

Abstract

We show that the MutY protein competes with the MutS-dependent mismatch repair system to process at least some A · C mispairs in vivo, converting them to G · C pairs. In the presence of an increased dCTP pool resulting from the loss of nucleotide diphosphate kinase, the frequency of A · T→G · C transitions at a hot spot in the rpoB gene is 30-fold lower in a MutY-deficient derivative than in the wild type.

The MutY protein, the product of the mutY gene in Escherichia coli (18), is a glycosylase that plays an important role in the repair of oxidatively damaged DNA (11, 12). Loss of function in both chromosomal copies of the human gene encoding MutY leads to increased susceptibility to colon cancer in humans (1, 9). MutY removes A residues that are mispaired with 7,8-dihydro-8-oxoguanine (oxoG) (11, 13), a frequent oxidation product of DNA (4). This removal allows repair polymerases to restore a C across from the 8-oxoG (25), allowing the MutM protein to remove the 8-oxoG (11, 13; for reviews, see references 12 and 15). Subsequent repair synthesis restores the original G · C base pair. MutY also removes the A from A · G mispairs (2), from A · 8-oxoA mispairs (13), and, to a lesser extent, from A · C mispairs (13, 21). Because the MutY protein does not discriminate between old and new strands, the ability to remove A from A · C mispairs may potentially immortalize mutations stemming from A · C mispairs in which A is the correct base, even though these mispairs may be substrates for correction by the MutSHL-dependent mismatch repair (MMR) system (for a review, see reference 17). In these cases, the original A · T base pair would be converted to a G · C base pair. In this sense, the MutY protein competes with the MMR system for the processing of A · C mispairs. Figure 1 portrays the different outcomes of A · C mispairs arising from replication errors at an A · T base pair.

FIG. 1.

Competition between the MutY protein and the MutHLS MMR system. (A) Misreplication at an A · T base pair leads to an A · C mispair that is corrected by the MutHLS MMR system but converted to a G · C pair by MutY. (B) Misreplication at a G · C base pair leads to an A · C mispair that is converted back to a G · C pair by both MutY and the MMR system.

The E. coli rpoB/Rif^r system.

We decided to study the contribution of the MutY protein to A · T→G · C mutations that can arise via an A · C mispair by using the previously described E. coli rpoB/Rif^r system to monitor mutations (7). We have extended the work of others (8, 19, 22-24) to generate a system that can analyze 69 different base substitutions in the rpoB gene (7). Recent work (20; E. Wolff, M. Kim, and J. H. Miller, unpublished data) has added several sites to this collection, so that as many as 73 different base substitutions can be monitored by analyzing E. coli Rif^r mutants at 37°C (Table 1). Of these 73 mutations, one particular A · T→G · C mutation, at bp 1547, is a hot spot in a wild-type background and a very strong hot spot in a mutS background (7, 16). Although it is not clear what proportion of the A · T→G · C transition mutations at this site result from A · C rather than T · G mispairs, comparing the rates of mutations at this hot spot in wild-type and mutY backgrounds seems to be a straightforward way to look for effects of the presence or absence of MutY on transitions. However, G · C→T · A mutations are elevated in a mutY background, significantly increasing the Rif^r mutant frequencies (17) (Table 2). Only a small percentage of the Rif^r mutants would be caused by other mutations. Thus, 13 of 15 mutations in rpoB obtained in a mutY strain result from G · C→T · A transversions (Table 1). Therefore, we sought to increase the level of A · C mispairs by employing a strain with a defect in the ndk gene (16), which encodes the enzyme nucleotide diphosphate kinase. Nucleotide diphosphate kinase is involved in maintaining nucleotide triphosphate levels, and ndk mutant strains have 20-fold-higher levels of dCTP and 7-fold-higher levels of dGTP, as well as higher levels of spontaneous mutations (10), than do strains without a defect in the ndk gene. We have analyzed the mutator effect of ndk strains and shown that certain base substitutions and frameshifts are considerably elevated in ndk mutS double mutants, indicating that replication errors are involved (16). Among mutations in rpoB leading to Rif^r, the mutations in both ndk and ndk mutS strains predominate at the A · T→G · C hot spot at bp 1547 (16). Because the level of mutations at this hot spot in ndk strains is high enough, we can determine whether the MutY protein is involved in generating A · T→G · C mutations at the bp 1547 hot spot by comparing the levels of mutations at this site in ndk and ndk mutY strains (for methods, see references 7 and 14).

TABLE 1.

Distribution of mutation in rpoB

Mutation	Site (bp)	Amino-acid change	No. of strains with rpoB mutation
			WT	mutS mutant	mutS mutY double mutant	mutY mutant	ndk mutant	ndk mutY double mutant
AT → GC	443	Q148R	5	0	0	0	1	0
	1522	S508P	0	0	0	0	0	0
	1532	L511P	2	7	7	0	2	0
	1534	S512P	9	11	7	0	2	3
	1538	Q513R	0	3	0	0	1	0
	1547	D516G	23	48	17	1	70	2
	1552	N518D	1	3	4	0	0	0
	1577	H526R	0	0	0	0	0	0
	1598	L533P	0	0	0	0	0	0
	1703a	N568S	0	0	0	0	0	0
	1715	I572T	0	0	0	0	0	0
GC → AT	1520	G507D	0	0	0	0	0	0
	1535	S512F	4	0	0	0	0	0
	1546	D516N	3	9	1	0	0	0
	1565	S522F	3	0	0	0	0	0
	1576	H526Y	7	0	0	1	0	0
	1585	R529C	0	0	0	0	0	0
	1586	R529H	2	1	0	0	0	0
	1592	S531F	5	1	0	0	0	0
	1595	A532V	0	0	0	0	0	0
	1600	G534S	0	3	0	0	0	0
	1601	G534D	0	0	0	0	0	0
	1610a	G537D	0	0	0	0	0	0
	1691	P564L	4	0	0	0	0	0
	1721	S574F	4	0	0	0	0	0
	2060b	R687H	0	0	0	0	0	0
AT → TA	443	Q148L	18	0	0	0	6	18
	1532	L511Q	1	0	0	0	3	0
	1538	Q513L	7	0	0	0	0	0
	1547	D516V	1	0	0	0	0	0
	1568	E523V	0	0	0	0	0	0
	1577	H526L	0	0	0	0	0	0
	1598	L533H	2	0	0	0	0	0
	1714	I572F	4	0	0	0	0	0
	1715	I572N	1	0	0	0	0	0
AT → CG	437	V146G	0	0	0	0	0	0
	443	Q148P	0	0	0	0	0	0
	1525	S509R	1	0	0	0	0	0
	1532	L511R	3	0	0	0	1	0
	1534	S512A	0	0	0	0	0	0
	1538	Q513P	0	0	0	0	0	2
	1547	D516A	0	0	0	0	0	1
	1577	H526P	1	0	0	0	0	1
	1598	L533R	0	0	0	0	0	0
	1687	T563P	0	0	0	0	0	0
	1714	I572L	14	0	0	0	1	0
	1715	I572S	4	0	0	0	2	0
GC → TA	436	V146F	0	0	1	0	0	0
	442	Q148K	0	0	0	1	0	0
	444	Q148H	1	0	0	2	0	0
	1527	S509R	2	0	0	2	0	0
	1535	S512Y	0	0	0	0	0	0
	1537	Q513K	0	0	0	0	0	0
	1546	D516Y	2	0	0	0	0	0
	1565	S522Y	0	0	0	1	0	0
	1576	H526N	2	0	0	0	0	0
	1578	H526Q	0	0	0	0	0	0
	1586	R529L	0	0	0	0	0	0
	1592	S531Y	0	0	0	2	0	0
	1595	A532E	0	0	0	0	0	0
	1600	G534C	3	0	0	2	0	0
	1601	G534V	1	0	0	1	0	0
	1708	G570C	0	0	0	0	0	0
	1721	S574Y	2	0	0	2	0	0
	1585b	R529S	0	0	0	0	0	0/PICK>
GC → CG	444	Q148H	0	0	0	0	0	0
	1527	S509R	1	0	0	0	0	0
	1574	T525R	1	0	0	0	0	0
	1576	H526D	3	0	0	0	0	0
	1578	H526Q	0	0	0	0	0	0
	1585	R529G	0	0	0	0	0	0
	1601	G534A	1	0	0	0	0	0
	1691	P564R	0	0	0	0	0	0
	1709	G570A	0	0	0	0	0	0
	1716	I572M	0	0	0	0	0	0
None found			4	0	0	0	0	0
Totalc			152	86	37	15	90	27

^aSites that have been detected in a related study (J. H. Miller, E. Wolff, and M. Kim, unpublished data).

^bSites that show temperature effects (8) between 30 and 42°C and that may not yield Rif^r colonies at 37°C.

^cThe results for 73 sites were considered for the total, but the number increases to 75 with the inclusion of the sites indicated by footnote b. Mutations in the wild-type, and the mutS mutant were described previously (7), as was the distribution of a smaller subset of the ndk mutations (16).

TABLE 2.

rpoB frequencies and notes of mutations^a

Strain	Mean value (95% CL)b
	f (10−8)	μ (10−8)
WTc	7.6 (5.2-8.8)	1.5 (1.1-1.7)
mutY mutant	79 (68-90)	11 (7.4-12)
ndk mutant	570 (530-630)	64 (60-70)
mutS mutant	935 (848-1,110)	120 (110-140)
ndk mutY mutant	206 (126-280)	26 (17-34)

^aThe rpoB mutation frequencies (f) per cell were calculated by dividing the median number of mutants by the average number of cells in a series of cultures, and the mutation rates (μ) per replication were calculated from these values by the method of Drake (5).

^bCL, confidence limits (3).

^cWT, wild type.

Distribution of mutations in rpoB.

Table 1 shows the distribution of rpoB mutations in wild-type, mutS, mutY, and ndk strains and in the ndk mutY and mutS mutY double mutants. Table 2 shows some comparative mutation frequencies and mutation rates per replication. ndk strains have 43-fold-higher rates of rpoB mutations that lead to Rif^r colonies than the wild type, and mutS strains have approximately a 80-fold-higher mutation rate (Table 2) than the wild type, although most of the mutations in the ndk and mutS strains are at one site (position 1547; A · T→G · C) (Table 1). The defect in mutY has no detectable effect on the mutation rate in mutS strains (data not shown) and no effect on the mutS spectrum (Table 1; note the different sample sizes). This lack of effect occurs because in the absence of MMR, which is the consequence of being a mutS mutant, there is no way to repair A · C mismatches, derived from A · T base pairs (Fig. 1A), regardless of the presence or absence of the MutY protein. However, in an ndk strain that is MMR proficient, the defect in the mutY gene lowers the frequency of mutations in rpoB 2.5- to 3-fold (Table 2) and, most importantly, virtually eliminates the position 1547 A · T → G · C hot spot from the mutational spectrum (Table 1). In the absence of this hot spot, the next most frequent mutation, A · T→T · A at position 443, now becomes the most frequent mutation and appears as a new hot spot. Figure 2 incorporates the mutation frequencies into the comparison of the ndk and ndk mutY spectra and shows that MutY-deficient derivatives of ndk strains have 30-fold-lower levels of A · T→G · C mutations at position 1547 in rpoB. (The apparent severalfold increase of A · T→T · A mutations at position 443 may not be significant, because of the small sample size of these mutations at position 443 in the distribution for the ndk mutants.) This finding demonstrates an in vivo effect of MutY on increasing transition mutations at certain A · T base pairs due to its ability to remove A from A · C mispairs and represents an example of a repair enzyme actually being involved in creating mutations under certain conditions. The effect of MutY on lowering A · T→G · C transversions in a mutT background has been described previously (6, 26), as has a 4.6-fold decrease in A · T→G · C mutations at one site in trpA (6).

FIG. 2.

Comparative mutation frequencies in ndk and ndk mutY strains. The frequencies of mutations in the rpoB gene at five different sites are shown for both ndk (white bars) and ndk mutY (black bars) backgrounds (see also Tables 1 and 2). The five sites (left to right) are as follows: 1, A · T→T · A at bp 443; 2, A · T→T · A at bp 1532; 3, A · →G · C at bp 1532; 4, A · T→G · C at bp 1534; 5, A · T→G · C at bp 1547.