Methylation and Host mutHLS Gene Functions

Six different base-pair transversion mismatches are repaired with different efficiencies in an in vitro mismatch repair system. In particular, the TIT and C/C mismatches appear to be less efficiently repaired than the A/A and G/G mismatches. Four A/G and four C/T mismatches at different positions are repaired to different extents. One of the A/G mismatches is repaired equally efficiently when DNA heteroduplexes are fully methylated or hemi-methylated at the d(GATC) sequences. This type of mismatch repair appears to be unidirectional with A to C conversion by acting at A/G mispairs to restore the C/G pairs. This methylation-independent correction is not controlled by the mutH, mutL, mutS, uvrE, UWB, phr, recA, recF, and recJ gene products. The independence of the transversion mismatch repair of these genes and methylation distinguishes this from the known mismatch repair pathways. D NA base-pair mismatches may arise from spontaneous replication error and homologous genetic recombination. In Escherichia coli mismatch repair directed by dam methylation at d(GATC) sequences is believed to correct errors arising during DNA replication (CLAVERYS and LACKS 1986; RADMAN and WAGNER 1986; WAGNER and MESLSON 1976). Repair is biased to the unmethylated newly synthesized DNA strand which bears the replication errors. The d(GATC) sequence are the activating sites for mismatch repair enzymes (LAENGLE-ROUAULT, MAENHAULT-MICHEL and RADMAN 1986; LAHUE, Su and MODRICH 1987; Lu et al. 1984). The number and position of dam sites influence the repair efficiency (Lu 1987). This methyl-directed mismatch correction requires the products of genes mutH, mutL, mutS, and uvrD (or uvrE and mutU) both in vivo and in vitro (BAUER, KRAMMER and KNIPPERS 198 1; Lu, CLARK and MODRICH 1983; NEVERS and SPATZ 1975; PUKKILA et al. 1983). Mismatch repair appears to involve longpatch excision and resynthesis (WAGNER and MESELSON 1976; Lu et al. 1984). Mismatch repair in regions of heteroduplex DNA due to genetic recombination between complementary strands of two different parental molecules could be involved in gene conversion (HOLIDAY 1964). The existence of a DNA mismatch repair system has been postulated in order to account for high negative interference (NORKIN 1970; WHITE and Fox 1974) and map expansion phenomena (FINCHAM and HOLLIDAY 1970; HOLLIDAY 1974). Such repair is thought to be independent of dam methylation and may be Genetics 118: 593-600 (April, 1988). controlled by different gene products. KOLODNER and co-workers (FISHEL and KOLODNER 1983; 1984; FISHEL, SIEGEL and KOLODNER 1986) have observed two methylation-independent pathways in E . coli. One pathway involved long excision tracts and does not require mutH or mutL function, but requires the mutS and uvrD gene products. The other very weak one associated with short repair tracts requires the recF and recJ gene products. A third pathway for mismatch repair in E . coli is characterized by very short repair tracts (rarely exceed ten nucleotides in length) (LIEB 1983; LIEB, ALLEN and READ 1986). This repair results in C to T transitions at the second position within the sequence 5’CC(A/T)GG, the dcm recognition site. The repair acts to restore the G/C pair by replacing the thymine in the G/T mismatch and is apparently responsible for repairing deaminated 5methylcytosines UONES, WAGNER and RADMAN 1987b). This short patch repair requires intact mutL, mutS, and dcm genes but not mutH and mutU genes (RADMAN and WAGNER 1986; JONES, WAGNER and RADMAN 1987b). In pneumococcus, it has been shown by DNA transformation that different types of base mismatches are processed differently by the hex controlled repair system (CLAVERYS et al. 1981, 1983; LACKS, DUNN and GREENBERG 1982). Similarly, the E . coli mismatch repair system does not recognize and repair all mismatches present in M13 (KRAMER, KRAMER and FRITZ 1984) or lambda (DOHET, WAGNER and RADMAN 1985; WAGNER et al. 1984) with equal efficiency. In these experiments, heteroduplexes of 594 A.-L. Lu and D.-Y. Chang these two bacteriophages with single base-pair mismatches were used to transfect E. coli. The results with pneumococcus and E. coli are remarkably similar. In general, transition mismatches (G/T and A/C) are well repaired. Some transversion mismatches (A/ G, CIT, and T/T) appear to be poor substrates for repair. It was noted that the repair of some transversion mispairs (especially A/G and C / T ) depended on the neighboring nucleotide sequence (CLAVERYS et al. 1983; JONES, WAGNER and RADMAN 1987a; LACKS, DUNN and GREENBERG 1982). It is suggested that repair efficiency increases with increasing CIG content in the neighboring nucleotide sequences UONES, WAGNER and RADMAN 1987a). The in vitro assay for mismatch repair that has been developed (Lu, CLARK and MODRICH 1983) is based on repair of heteroduplex DNA of f l R229, which contains a base-pair mismatch within the single EcoRI site of the molecule. Transition mismatches (G/T and NC) are efficiently repaired in this in uitro system. The repair activity is dependent on the state of dam methylation of the DNA strands and on the gene €unctions of the mutH, mutL, mutS, uvrE, and ssb loci (Lu, CLARK and MODRICH 1983; Lu et al. 1984). The transversion mismatches have not been assayed in the in vitro system. In this paper, the specificity of E. coli mismatch repair mainly on the transversion mismatches was studied in the in vitro system. Similar results were obtained as in the in viuo systems, i.e., different transversion mismatches were repaired with different efficiencies. Repair of AIG and C/T mismatches was controlled by the flanking nucleotide sequences. It is the first demonstration that the repairable AIG mismatch was independent of dam methylation and did not require mutH, mutL, mutS, and uvrE gene products which are involved in methyl-directed mismatch repair. This methylationindependent correction of the A/G mismatch was quite efficient and was unidirectional with a conversion of A to C . This pathway appeared to be different from the other known methylation-independent pathways as judged by the requirement for known E. coli DNA repair gene products. MATERIALS AND METHODS Bacterial and bacteriophage strains: All E . colt strains used in this study are listed in Table 1. The bacteriophages f l used (Table 2) are derived from R229 (BOEKE 1981) which contains one EcoRI site at position 5616. Phages f l M28 and G18 containing a C to T substitution at position 562 1 and T to G substitution at position 5620, respectively, were gifts from P. MODRICH. Other fl mutants were generated in this laboratory by oligonucleotide-directed mutagenesis as described below. Other materials: Restriction enzymes BamHI, BanII, DpnI, HincII, MboI and SauSA, large fragment of DNA polymerase I and T4 DNA ligase were from Bethesda Research Laboratories. Endonuclease BsmI was from New England Biolabs. Endonucleases EcoRI and BspI and dum TABLE 1 Bacterial strains used Strain Genotype Source