Polymerase Antimutator Allele

We have previously isolated mutants of Escherichia coli that replicate their DNA with increased fidelity. These mutants have a mutation in the dnaE gene, encoding the a subunit of DNA polymerase 111. They were isolated in a mismatch-repairdefective mutL background, in which mutations can be considered to represent uncorrected DNA replication errors. In the present study we analyze the effect of one of these alleles, dnaE911, on spontaneous mutagenesis in a mismatch-repair-proficient background. In this background, spontaneous mutations may be the sum of uncorrected replication errors and mutations resulting from other pathways. Hence, the effect of the dnaE allele may provide insights into the contribution of uncorrected DNA replication errors to spontaneous mutation. The data show that dnaE911 decreases the level of Rif, lacZ and galK mutations in this background by 1.5-2-fold. DNA sequencing of 748 forward mutants in the l a d gene reveals that this effect has a clear specificity. Transversions are decreased by -%fold, whereas transitions, frameshifts, deletions and duplications remain essentially unchanged. Among the transversions, A*T + T.A are affected most strongly (-Gfold) . In addition to this effect on transversions within the lacl gene, one previously recognized A.T + G C base-pair substitution hotspot in the lac operator is also reduced (-5-fold). The data are discussed in the light of the role of DNA replication errors in spontaneous mutation, as well as other possible explanations for the observed antimutator effects. T HE mechanisms that are responsible for the low but measurable spontaneous mutation rates in prokaryotic and eukaryotic cells are of considerable interest. To achieve low levels of mutation, cells must invest substantial resources to (i) replicate their DNA with high fidelity, (ii) minimize the occurrence of a variety of DNA damage and (iii) repair the DNA once damaged has occurred. To better understand the role of the fidelity of DNA replication in these processes, we previously undertook a search for Escherichia coli strains that replicate their DNA with increased fidelity (FIJALKOWSKA et al. 1993). This search was done in a strain deficient in postreplicative mismatch repair ( m u t L ) . In such a strain mutations may be considered to be uncorrected DNA replication errors (SCHAAPER and DUNN 1987; SCHAAPER 1993a), and suppressors of the high mutability of a mutL strain are therefore candidates for mutations that increase the accuracy of replication. We mutagenized the region of the chromosome carrying the dnuE and dm@ genes encoding, respectively, the a (= polymerase) and E (= proofreading) subunit of DNA polymerase 111, which is chiefly responsible for the replication of the chromosome (MCHENRY 1991). Mutants with lower mutation rates were identified using the gulK2 papillation assay. Seven mutants were obtained that lowered the level of papillation in the mutL background 5-30-fold and which were subsequently found to lower mutation frequencies in this background for several other markGenetics 138: 263-270 (October, 1994) ers. Genetic mapping and DNA sequencing revealed that these mutants contained mutationswithin the dnuE gene (FIJALKOWSKA and SCHAAPER 1993). These observations strongly suggest that the isolates are mutants that replicate their DNA with increased fidelity. These “antimutator” strains are useful tools for several types of studies. First, they may be used to analyze the sources of spontaneous mutation. Spontaneous mutants could represent the sum of several potential mutational pathways (such as uncorrected DNA replication errors or mutations resulting from unrepaired DNA damage, DNA repair activities, transpositions and others). Therefore, a reduction in the level of mutation due to increased mutation avoidance in a particular pathway may identify that pathway as a contributor to overall spontaneous mutagenesis and allow an estimation of its contribution. For example, a twofold reduction in the level of spontaneous mutation by a DNA polymerase conferring increased accuracy of DNA replication can be interpreted to mean that at least 50% of the mutations in a wild-type strain originate from uncorrected replication errors. [For the purpose of this study, we define replication errors as those resulting from the intrinsic inaccuracy of the DNA polymerase when copying normal undamaged DNA. Mutations resulting from replication across damaged DNA are considered a separate class of mutational origin.] Second, antimutator polymerase alleles might be used to investigate the role of the DNA 264 A. R. Oller and R. M. Schaaper polymerase in mutagenesis occurring at sites of DNA damage. This could be done by combining defined deficiencies in DNA repair activities that result in a mutator phenotype with our DNA polymerase alleles. Any reduction in the mutator activity will reveal the importance of DNA polymerase fidelity in this type of mutagenesis. Third, the antimutator DNA polymerases may facilitate in vitro and in vivo studies directed at the precise mechanisms by which DNA polymerases achieve their high accuracy of polymerization. In the present study, we use the E. coli antimutators to analyze the contribution of unrepaired DNA replication errors to spontaneous mutations. We examine the antimutator effect of one of the previously isolated dnuE antimutator alleles, dnuE911, in an otherwise wild-type background ( i. e., mismatch-repair proficient). We also characterize in detail its mutational specificity using the ladfornard mutation system. Analysis of specificity is important since spontaneous mutagenesis has been shown to consist of a number of different mutational classes (SCHAAPER and DUNN 1991) and DNA polymerase mutants may affect certain classes but not others (see below). dnaE911 was chosen since it is one of the better characterized dnuE antimutator alleles (FIJALKOWSKA et al. 1993; SCHAAPER 1993b). Some information on the specificity of dnaE911 in backgrounds other than the wild-type has already been reported. A study of the specificity in a mutL background, in which transitions (A-T + G C and G C + A*T) are the predominant base substitutions (SCHAAPER and DUNN 1987a), showed that dnaE911 reduces both transitions by approximately 3-5-fold (SCHAAPER 1993b). Transversions were only decreased -20% based on the limited number of transversion mutations observed in this background. dnaE911 was also capable of reducing mutagenesis in a mutT mutator strain, where mutations are exclusively A.T -+ C.G transversions (SCHAAPER and DUNN 1987b), indicating that this type of transversions could also be decreased (FIJALKOWSKA et al. 1993). However, no effect was observed in a recA730 strain (FIJALKOWSKA et al. 1993). This strain is a spontaneous mutator due to the constitutive induction of the SOS system (WITKIN et al. 1982; WALKER 1984), and the majority of mutations are transversions (MILLER and LOW 1984; OLLER et ul. 1993; YATACAI et aL 1991). It thus appears that the effect of the DNA antimutator alleles depends on (i) the type of mutation being considered and (ii) the genetic background in which the mutation is analyzed. Other antimutator polymerases, particularly those of bacteriophage T4 (DRAKE 1993), also display a defined specificity. The results presented here describe a total of 748 sequenced l a d mutants in wild-type and dnaE911 strains. Based on these results we discuss the possible contribution of unrepaired replication errors to spontaneous mutations as well as other possible roles of DNA polymerase I11 in preventing spontaneous mutagenesis. MATERIALS AND METHODS Strains and media: The strains used for ladspectrum studies are derivatives of KA796 ( m a , thi, Aprolac) (SCHAAPER et al. 1985). NR10171 is KA796, but dnaE911, zae::Tn 1 Od-Cam. The dnaE911 allele was introduced along with 2ae::Tnl Od-Cam by P1 transduction as described (FIJALKOWSKA et al. 1993; FIJALKOWSKA and SCHAAPER 1993). F'prolac (F'128-27) was introduced into these strains by conjugation, yielding, respectively, NR10179 (dnaE+) and NR10175 (dnaE911). F'128-27 provides the target to score forward lac1 mutants (SCHAAPER and DUNN 1991). NR10179 (dnaE') is identical to NR9102, whose mutational specificity has been previously reported (SCHAAPER and DUNN 1991). GalK2 revertant frequencies were measured in NR9355 (ara-9, fhuA1, lacy1 or lacZ4, tsx-3 , supE44, galK2, h-,hisG4(0c), @Dl?, trp-3(0c), rpsL8 or rpsL9, malAl (A"), metE46, mtl-1, thi-1) (FIJALKOWSKA et al. 1993) and NR9901, its dnaE911 zae::TnlOd-Cam counterpart (FIJALKOWSKA and SCHAAPER 1993). LB broth and minimal medium (MM) were as described (SCHAAPER et al. 1985). LBRif plates contained 100 pg/ml of rifampicin. "Gal plates contained 0.2% galactose as sole carbon source. Phenyl-k-lactopyranoside (Pgal) plates contained 750 pg/ml Pgal as sole carbon source. Mutant frequency determination: To determine the frequency of Rif, lac1 and galKmutants, 12 to 20 individual colonies from each strain were toothpicked into 1 ml LB. After overnight growth at 37", appropriate dilutions were plated on LB or MM plates to determine the number of viable cells, and undiluted cultures were plated on the selective plates (LBRif, Pgal or "Gal) to determine the number of mutants. To calculate mutation frequencies, we divided the median number of mutants/ml by the average number of viable cells/ml. Each experiment was repeated two or three times and the average of the medians is reported. Collection of l a d mutants and DNA sequencing: l a d mutants were collected from two separate experiments. In experiment 1, three independent isolates for each strain (NR10179 and NR10175) were grown overnight in LB to saturation. Each culture was diluted in fresh LB to -10,000 cells/ml and distributed in 200pl aliquots into the wells of 9Gwell microtiter dishes. Ten dishes were started per strain (960 independent cultures). After overnight growth at 37" on a shaking platform, 10 pl from each well were spread on a quarter section of a Pgal plate; on these plates only mutants expressing the lac operon constitutively are capable of forming colonies (MILLER 1978). At this time, lac1 mutant frequencies were measured by taking samples from 20 wells/strain. The Pgal plates were incubated for 60 hr at 37". Four mutants per quarter section were then picked randomly and placed in a gridded pattern (100grid) on Pgal plates (-3800 mutants per strain). After overnight growth these plates were replicamated (F' transfer) into strain CSH52 to identify the dominant mutants as described (SCHAAPER and DUNN 1987a, 1991). About 500 independent dominant mutants were obtained for each strain and regridded on Pgal plates (only one dominant mutant per quarter Pgal plate was taken, and if more than one existed per culture, the first one was selected). Dominant mutants can have mutations in l a c 0 or lacl. To detect two frequent classes of lacO mutants (SCHAAPER et al. 1986), the dominant colonies were also toothpicked onto NEN nylon filters and probed as described below. The dominant mutants that were negative with the lacO probes (-200 mutants per strain) were regridded on Pgal plates and replica-mated (F' transfer) into strain S9OC (SCHAAPER et al. 1985). The resulting conjugants were infected with phage mRS81 for recombinational transfer of the lac1 gene, as described (SCHAAPER et al. 1985, 1986). Recombinant phage were purified and their DNAs DNA Polymerase Antimutator 265 sequenced as described (SCHAAPER et al. 1985,1986). Base pairs 30 through 250 of lacl were sequenced. Experiment 2 was essentially as experiment 1, except that more independent cultures (1056 per strain) were started and more cells (25 pl) were plated per Pgal quarter plate. These modifications resulted in the recovery of -4200 total mutants and -700 independent dominant mutants per strain. Colony hybridization: l ac0 mutants (operator-constitutive mutants), which also display a dominant phenotype in the complementation test with strain CSH52 (SCHAAPER et al. 1986), have been found to be of mainly two types: deletions in the operator region and an A*T -+ G-C base pair-substitution hotspot at position +6 of lac2 (SCHAAPER et al. 1986). In a wild-type strain these two mutations can comprise a significant fraction (-50% or more) of spontaneous dominant mutants recovered on Pgal plates. We have found it advantageous to identify these mutants prior to DNA sequencing by using oligonucleotide probing as described by HALLIDAY et al. (1990). In brief, dominant mutant colonies were gridded on duplicate nylon membranes (NEN Research Products, Boston, Massachusetts) placed on LB plates. After overnight growth at 37", the membranes were autoclaved for 2 min to fix the DNA. To detect mutations that had deletions of the lac operator we used a 20-mer oligonucleotide corresponding to lacO base pairs izations were performed at 56". To detect mutants that had an A-T -j G.C transition at +6 lacO, we used a 17-mer spanning base pairs -3 to +I4 (5'-TGGAATTGCGAGCC'), but containing the mutation at +6 (underlined). Because the +6 lacO position is in a region of the operator that forms a strong palindrome (SCHAAPER et al. 1986), we used the unlabeled 20mer oligonucleotide to help increase the hybridization signal as suggested by KUPCHELLA and CEBULA (1991). Hybridizations were performed at 52". 10 to -29 (5'-TITATGCITCCGCCTCGTAT-3'). Hybrid-