The crystal structure of Escherichia coli MutS bound to an oligonucleotide with a G-T mismatch has been determined using a derivative of the MutS protein, MutSΔ800, which lacks the C-terminal 53 amino acids (10). The MutSΔ800 mutant crystallizes as a dimer and retains the ability to bind DNA and ATP, just as full-length MutS (853 amino acids) does. The atomic structure of a truncated MutS from Thermus aquaticus has also been determined and is very similar to that of E. coli MutS (14). The physiological effects of the mutSΔ800 mutation have so far been studied only in multicopy (2, 4, 10), and we show below that, in single copy, it imparts a MutS null phenotype.

The procedure used to construct the mutSΔ800 chromosomal mutation is outlined in Fig. ​Fig.1.1. The sequence around and including the bla gene (ampicillin resistance) and its promoter was amplified by PCR (Fig. ​(Fig.1A)1A) from strain TP879 to produce a product (Fig. ​(Fig.1B)1B) bearing the bla region flanked by 50-bp regions. The 5′ flanking region has the DNA sequence immediately upstream of residue 800 of mutS plus a termination codon (Fig. ​(Fig.1B),1B), and the 3′ flanking region has the downstream sequence immediately following the termination codon of mutS. The PCR product was electroporated (13) into strain TP798, which constitutively expresses the products of the exo (exonuclease) and bet (beta protein) recombination genes of bacteriophage lambda (15). Recombination between the homologous regions of the PCR product and the mutS gene and its flanking sequence (Fig. ​(Fig.1C)1C) produces a recombinant sequence in which the mutS gene is truncated at residue 800 and has an adjacent bla gene (Fig. ​(Fig.1D).1D). By changing the upstream PCR primer sequence, we also constructed mutSΔ2, in which all but the first two and last codons of mutS were deleted, and the control mutS⁺ construct with the flanking bla gene.

We measured the levels of native and mutant MutS, by Western blotting (5, 7), in strains with the chromosomal constructs as well as multicopy plasmids which were in a mutS null host (Fig. ​(Fig.22 and Table ​Table1).1). The levels in GM8311 (mutS⁺) were the same as those in AB1157 (mutS⁺) and were increased fourfold in GM7451, which harbors pMQ372 (mutS⁺), but no MutS was detected in GM8313 (mutSΔ2). Strain GM8315 (mutSΔ800) contained 2.5-fold less MutS than that contained by GM8311 (mutS⁺), but in multicopy (GM7453), the level was the same as that for the wild-type strain.

The strains bearing the mutSΔ2, mutSΔ800, and wild-type alleles were tested for reversions of the argE3 marker and for mutations to rifampin resistance as described elsewhere (4). The results in Table ​Table22 show that with the wild-type strain, a low rate of resistance or reversion was observed. However, for the strains with mutSΔ2 and mutSΔ800, the number of rifampin-resistant mutants increased about 60-fold for both, and the increases for Arg⁺ revertants were about 60- and 40-fold, respectively.

Each of the chromosomal mutS mutants was used as a recipient in conjugal crosses with E. coli (homologous) or Salmonella enterica serovar Typhimurium (homeologous) donors as described previously (4). For the homologous crosses with the E. coli donor, recombinants are formed at the same frequency, indicating that there is no effect on homologous recombination stemming from the construction of the chromosomal mutS mutations (Table ​(Table3).3). With the Salmonella donor and a wild-type mutS recipient, no recombinants were detected. In contrast, the mutSΔ2 and mutSΔ800 recipients increased the ability to form recombinants 11,000- and 27,000-fold, respectively, with the same Salmonella donor (Table ​(Table3).3). These increases were abrogated in crosses with recA deletion derivatives of the mutSΔ2 and mutSΔ800 mutant recipients (data not shown), indicating that the mutS mutations reduce antirecombination function in a recA-dependent manner.

E. coli dam mutS⁺ mutants are more sensitive to exposure to MNNG (N-methyl-N′-nitro-N-nitrosoguanidine) (Fig. ​(Fig.3A)3A) and cisplatin (Fig. ​(Fig.3B)3B) than the wild type (6, 8, 9) is, but the dam mutSΔ2 deletion strain is as resistant to both treatments as the wild type is (Fig. ​(Fig.3),3), based on determinations using the protocol described previously (4). Figure ​Figure33 also shows that dam mutSΔ800 bacteria are as resistant to MNNG and cisplatin as the dam mutSΔ2 cells are.

We conclude that the mutSΔ800 mutation in a single copy on the chromosome confers a mutS null phenotype to a cell to the same degree as the mutSΔ2 deletion mutation does for mutation avoidance, antirecombination, and resistance to cytotoxic agents. On a multicopy plasmid, the mutSΔ800 mutation in a dam mutS host confers a “split” phenotype, where mutation avoidance (2, 4, 10) and MNNG sensitivity are at the wild-type levels but antirecombination and resistance to cisplatin are severely diminished (4).

The MutS null phenotype of the mutSΔ800 strain is due in part to the decreased cellular level of MutSΔ800 compared to that of MutS (Fig. ​(Fig.2),2), indicating that the C-terminal 53 amino acids impart stability to the protein. Even when corrected so that the levels of the proteins are similar, as in strains with the multicopy plasmid mutSΔ800 and single-copy mutS, there is still the deficiency of antirecombination and resistance to cisplatin (4). Furthermore, purified MutSΔ800 protein has a lower affinity than MutS does for certain oligonucleotides with base pair mismatches and, in the presence of other MMR components, reduces the efficiency of MutH-induced incision at hemimethylated GATC sequences in vitro (3). The lower cellular amount of MutSΔ800 protein, therefore, cannot be the sole explanation for the phenotypic differences between wild-type and mutSΔ800 strains.