ABSTRACT
Loss of DNA mismatch repair due to mutation or diminished expression of the MLH1 gene is associated with genome instability and cancer. In this study, we used a yeast model system to examine three circumstances relevant to modulation of MLH1function. First, overexpression of wild-type MLH1 was found to cause a strong elevation of mutation rates at three different loci, similar to the mutator effect of MLH1 gene inactivation. Second, haploid yeast strains with any of six mlh1 missense mutations that mimic germ line mutations found in human cancer patients displayed a strong mutator phenotype consistent with loss of mismatch repair function. Five of these mutations affect amino acids that are homologous to residues suggested by recent crystal structure and biochemical analysis of Escherichia coli MutL to participate in ATP binding and hydrolysis. Finally, using a highly sensitive reporter gene, we detected a mutator phenotype of diploid yeast strains that are heterozygous for mlh1 mutations. Evidence suggesting that this mutator effect results not from reduced mismatch repair in the MLH1/mlh1 cells but rather from loss of the wild-type MLH1 allele in a fraction of cells is presented. Exposure to bleomycin or to UV irradiation strongly enhanced mutagenesis in the heterozygous strain but had little effect on the mutation rate in the wild-type strain. This damage-induced hypermutability may be relevant to cancer in humans with germ line mutations in only one MLH1 allele.
The stability of eukaryotic genomes depends heavily on several DNA repair processes, including correction of DNA replication errors by the DNA mismatch repair (MMR) system (reviewed in references 19, 26, and46). Mutations in genes that inactivate mismatch repair strongly elevate spontaneous mutation rates and predispose humans to cancer. Current evidence suggests that germ line humanMSH2 and MLH1 mutations account for a majority of hereditary nonpolyposis colorectal cancer (HNPCC) cases (31). Many of these result in loss of intact protein and are thus predicted to completely inactivate MMR. Others, such as theMLH1 missense mutations found in more than 30 HNPCC families (15, 25, 31), are often inferred to be pathogenic if they are nonconservative changes in evolutionary conserved amino acids, if they cosegregate with the disease, and/or if they are not observed in the normal population. However, unlike mutations that lead to protein truncation, single amino acid changes may not impair protein function or may only be partially inactivating.
To assess the functional consequences of missense mutations in humanMLH1, Shimodaira et al. (40) developed an assay in yeast, based on elimination of a dominant mutator phenotype conferred by expression of human MLH1 cDNA. They demonstrated that several human MLH1 missense mutations identified in HNPCC patients impair the function required for this dominant mutator effect. A more direct approach to address the effect of MLH1 missense mutations on MMR efficiency is based on the fact that the amino acid sequences of the human and yeast MLH1 proteins are 70% identical in the N terminus and also share significant homology in the C terminus (30). Thus, the functional consequences of certain human MLH1 missense mutations can be inferred from analysis of analogous mutations in yeast MLH1. As an example, Pang et al. (29) found that a change in yeastMLH1 gene corresponding to an HNPCC-associated missense mutation resulted in a mutator phenotype characteristic of defective MMR.
To understand the relationship between loss of MMR gene function and cancer, and with the longer-term goal of understanding the relationship between the structures of these proteins and their functions in MMR and other DNA transactions, we are also using yeast as a model of eukaryotic MMR. Here we present three sets of results in yeast that provide further understanding of the relationship between loss of MLH1 function, hypermutation, and cancer. The first is prompted by studies suggesting that, in addition to gene mutation, MMR activity can be modulated via changes in gene expression. We recently showed that MMR can be inactivated by loss of human MLH1 expression due to hypermethylation of the human MLH1 promoter (17), and we (17) and others (48) observed a correlation between promoter hypermethylation and sporadic colon cancer. In the present study, we test the opposite possibility, i.e., that overexpression of yeast MLH1 in a wild-type yeast strain might increase the mutation rate.
We also took the approach used by Pang et al. (29) to examine the effects in yeast of six missense mutations in yeastMLH1 that are homologous to those reported in HNPCC families. Our results are consistent with observations by Shimodaira et al. (40) on mutations in the human MLH1 gene and suggest complete loss of MMR activity in the mlh1 mutants. The recent description of the crystal structure of a 40-kDa N-terminal fragment of Escherichia coli MutL and the discovery of ATPase activity intrinsic to this protein (1) suggest how these missense mutations might inactivate Mlh1p function.
We also examined the effects of heterozygosity for mlh1mutations on the mutation rate in diploid yeast strains. We present here the first evidence of a mutator phenotype for strains heterozygous for missense mutations or deletion of one MLH1 allele. We describe a novel mechanism for this mutator effect involving loss of the wild-type allele in a small fraction of the heterozygous cells, and we then show that the mutant frequency in the heterozygous strain is strongly increased by treatment with DNA-damaging agents.
MATERIALS AND METHODS
Strains and plasmids. Saccharomyces cerevisiae haploid strain E134 (MATα ade5-1 lys2::InsE-A14trp1-289 his7-2 leu2-3,112 ura3-52) was obtained from H. Tran (45), and strain DBY747 (MAT a his3-Δ1 leu2-3,112 ura3-52 trp1-289) was provided by Y. Pavlov (St. Petersburg State University, St. Petersburg, Russia). The 2μm-based plasmids pMMR75 (13), containing the yeast MLH1gene under control of the ADH1 promoter, and pMMR74, containing the MLH1 gene with 5′ upstream and 3′ downstream regions cloned in YEplac195 (8), were provided by L. Prakash (University of Texas). Plasmid YIpMLH1 was constructed by cloning a 5.2-kb SacI fragment with the MLH1 gene from pMMR74 into SacI site of yeast integrative URA3plasmid pRS306 (41). Missense mutations in theMLH1 gene were created in YIpMLH1 by site-directed PCR mutagenesis using a QuickChange site-directed mutagenesis kit from Stratagene. To create yeast strains with mlh1 mutations, the chromosomal MLH1 gene of strain E134 was replaced with the mutant alleles. Strain E134 was transformed with the mutagenized integrative plasmids digested with NheI (which cuts inside the sequence of the mutant allele) to stimulate integration into theMLH1 locus. Ura+ transformants were plated onto 5-fluoroorotic acid-containing medium to select for excision of the plasmid with the URA3 gene. A part of the MLH1gene of the Ura− clones was sequenced by using an ABI PRISM 377 sequencer and ABI PRISM Dye Terminator Cycle Sequencing Ready Reaction kit. Isolates containing the desired missense mutation were used for further study. Oligonucleotides for site-directed mutagenesis, PCR, and sequencing were from Gibco BRL. Plasmid pmlh1Δ-LEU2, obtained from R. M. Liskay (Oregon Health Sciences University), was used to create disruption of theMLH1 gene as described elsewhere (36). S. cerevisiae E68 (MAT a ade2-1 arg4-8 leu2-3,112 thr1-4 trp1-1 ura3-52 lys2Δ cup1-1), obtained from H. Tran (National Institute of Environmental Health Sciences), and 1B-D770 (MAT a ade5-1 lys2-Tn5-13 trp1-289 his7-2 leu2-3,112 ura3-4) were used as parental strains for diploids construction.
Measurement of mutation rates and mutant frequencies.To measure the spontaneous reversion rates at thelys2::InsE-A14 or his7-2locus or the rate of forward mutation to canavanine resistance (Canr), at least nine yeast cultures were started from single colonies and grown to stationary phase in liquid YPAD medium or medium selective for the plasmids. Cells were plated after appropriate dilutions onto selective medium lacking lysine or histidine for revertant count, onto complete medium containingl-canavanine (40 mg/ml) and lacking arginine for Canr mutant count, and onto YPAD medium or medium selective for the plasmids for viable count. For strains containing plasmids, tryptophan was omitted from all media used for mutant selection. YPAD and selective dropout media were prepared as described elsewhere (38). Plates were incubated for 3 to 5 days at 30°C before counting. The frequencies of revertants and Canr mutants in each culture were calculated by dividing the revertant or Canr mutant count by the viable cell count. Mutation rates were calculated from mutant frequencies as described elsewhere (5). The 95% confidence limits for the median were determined as described previously (4). The significance of differences between mutation rates was estimated by using the Wilcoxon-Mann-Whitney nonparametric criterion (4).
To measure the Lys+ revertant frequencies in yeast cultures grown from bleomycin-treated cells, nine independent cultures were grown from single colonies to saturation in liquid YPAD medium. Cells from each culture were collected by centrifugation and resuspended in sterile water at a density of 108 cells per ml. Freshly dissolved bleomycin sulfate (Sigma) was added to a final concentration of 5 μg/ml. Cells were treated with aeration for 1 h at 30°C. Approximately 105 bleomycin-treated cells were immediately transferred into 2.5 ml of liquid YPDA medium, and the cultures were grown to stationary phase at 30°C with aeration. The frequency of Lys+ revertants in these cultures was determined as described above for spontaneous Lys+ reversion.
Qualitative mutagenesis assays.For spontaneous mutant frequency estimation, single yeast colonies were patched onto YPAD plates, incubated for 2 days at 30°C, and replica plated onto selective media. For bleomycin-induced mutagenesis, yeast strains were grown to stationary phase in liquid YPAD medium, collected by centrifugation, and resuspended in sterile water at a density of 108 cells per ml. Freshly dissolved bleomycin sulfate was then added to a final concentration of 5 μg/ml. Cells were treated with aeration for 1 h at 30°C, washed, and resuspended in water at a density of 107 cells per ml. Approximately 0.2 ml of this suspension was plated with a 150-pin replicator onto YPAD plates, incubated for 2 days at 30°C, and replica plated onto selective medium lacking lysine. The selective plates were incubated for 4 to 5 days at 30°C to allow visualization of Lys+ colonies.
Isolation of independent Lys+ revertants and PCR genotyping.To isolate independent Lys+ revertants, yeast strains were grown overnight in liquid YPAD medium and the cultures were diluted in water to a density of 106 cells per ml. This suspension was plated with a 150-pin replicator onto YPAD plates (approximately 0.2 ml per plate) and selective plates lacking lysine to confirm that the suspension plated did not contain Lys+ revertants. Selective plates were incubated for 5 days at 30°C. YPAD plates were incubated for 2 days at 30°C, and cells were replica plated onto selective medium lacking lysine. One Lys+ colony was picked from the replica of each patch and streaked for single colonies on selective medium. The MLH1allele status was determined by PCR using primers designed to specifically amplify a portion of the wild-type MLH1 gene and/or the mlh1Δ::LEU2 mutant allele. Primers M11 (5′ GCTATCGTTGTAGGGTCC 3′), M3 (5′ CTCCTCGGAATCCATACG 3′), and LEU2 (5′ ACAGTACCACGGAAGTCG 3′) were from Gibco BRL. M11 is complementary to nucleotides −291 through −274 in the 5′ upstream region of the MLH1 gene. M3 is complementary to nucleotides 276 through 294 of the MLH1coding region. LEU2 is complementary to nucleotides 1037 through 1054 of the LEU2 coding region. PCR mixtures of 100 μl contained 100 pmol of primer M3, 100 pmol of primer LEU2, 200 pmol of primer M11, 200 nM deoxynucleoside triphosphates, 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl2, and 0.001% gelatin. Single yeast colonies were resuspended in the PCR mix, the suspension was boiled for 5 min, 2.5 U of Taq DNA polymerase (Perkin-Elmer) was added, and PCR was performed for 30 cycles at 95°C (1 min), 51°C (1 min), and 68°C (4 min). PCR products were analyzed by electrophoresis in a 2% agarose gel.
RESULTS
Mutagenic effects of MLH1 overexpression.To determine if increased expression of the MLH1 gene affects the mutation rate, the MMR-proficient haploid strain E134 was transformed with high-copy-number 2μm plasmids which express theMLH1 gene from the natural MLH1 (pMMR74) or theADH1 (pMMR75) promoter (13). Three genetic markers were used to examine mutation rates in the transformants: reversion of the lys2::InsE-A14 andhis7-2 alleles and forward mutation to Canr. Thelys2::InsE-A14 mutation reverts via loss of a single base pair in a run of 14 A · T base pairs (45). The his7-2 allele was sequenced and found to contain a single base pair deletion in a run of eight A · T base pairs at positions 472 to 479 in the HIS7 gene. Sequencing of the HIS7 gene of a limited number of His+ revertants has shown that the his7-2mutation reverts spontaneously via +1 insertions and −2 deletions in a stretch of 43 base pairs including the A7 run. Single-base addition frameshifts in the A7 run are almost exclusively scored in the mlh1 background (data not shown). The Canr forward mutation assay scores a wide variety of base substitutions, frameshifts, and complex mutations in theCAN1 gene (3, 24, 43).
The pMMR75 vector, which expresses Mlh1p at readily detectable levels (13), produces a 4,400-fold increase in Lys+reversion rate, as well as substantial increase in His+reversion and Canr mutation rate (Table 1). The mutation rate approaches that observed in a strain in which MLH1 gene has been deleted. A 2μm plasmid containing the MLH1 gene under the control of its natural promoter (pMMR74) was also mutagenic, but to a lesser extent (Table 1).
Mutator effect of MLH1 overexpression
Such strong mutator effects were unanticipated, since Pang et al. (29) did not observe any increase in reversion rate ofhom3-10 frameshift mutation when the MLH1 gene was expressed from the ADH1 promoter in wild-type cells. This is particularly surprising in view of the fact that thehom3-10 allele reverts to wild type via single base deletions in a run of seven T · A pairs (24), a specificity consistent with thelys2::InsE-A14 reversion spectrum. To eliminate the possibility that the plasmids we used contain mutations in the MLH1 gene, we sequenced the MLH1 coding sequences in pMMR74 and pMMR75 and confirmed that they were indeed wild type. To examine if the dominant mutator effect was unique to the strain we used, we also introduced plasmid pMMR75 into a nonrelated MMR-proficient yeast strain, DBY747. This led to an increased rate of forward mutations to α-aminoadipic acid resistance, with the rate elevated almost to the same extent as observed for disruption of theMLH1 gene (data not shown). The differences in results obtained by Pang et al. (29) and by us (Table 1) could reflect differences in the strain genetic background, the expression vectors, or the reporter genes used. Nevertheless, our data indicate that MLH1 overexpression destabilizes the yeast genome at several different loci and in two different strains, even when wild-type copies of all the MMR genes are present in the cells. This mutator effect is in marked contrast to observations with yeastMSH2 gene, which does not increase the rate oflys2::InsE-A14 reversion or Canr mutation when introduced into wild-type yeast strain on plasmids expressing the gene at low level from its natural promoter or at high levels from the GAL10 promoter (6).
Mutator phenotypes of yeast mlh1 missense mutants.Six missense mutations were introduced into the yeast MLH1gene to mimic germ line human MLH1 mutations found in HNPCC families, P28L (49), M35R (42), G67R (42), I68N (42), T117M (22), and G244D (32) (Fig. 1). The strong dependence of mutation rate on MLH1 expression level described above suggests that natural expression conditions should be maintained in order to assess the impact of mlh1 mutations on MMR function. Therefore, to analyze the effects of the missense mutations, we replaced the wild-type chromosomal copy of theMLH1 gene in strain E134 with the mutant alleles. The impact of these mutations in haploid cells was determined by comparing the spontaneous mutation rates in the mlh1 mutants and the isogenic wild-type strain. Rates of Lys+ and His+ reversion and Canr mutation in all sixmlh1 mutant strains were much higher than in the wild-type yeast strain and similar to the mutation rate in an mlh1Δmutant (Table 2). These strong mutator effects suggest complete or near-complete inactivation of DNA mismatch repair of replication errors. (The Lys+ and His+ reversion rates in the P25L mutant strain were twofold lower than the rates observed with the other mutants. The differences were calculated to be statistically significant [α ≤ 0.05 for His+ reversion and α ≤ 0.01 for Lys+reversion], leaving open the possibility that this mutant retains partial Mlh1p function.)
Missense mutations in MLH1. Alignment of the amino acid sequences of MutL homologs is shown. hMLH1, human MLH1; yMLH1, S. cerevisiae Mlh1p; rMLH1, Rattus norvegicus MLH1; EcMutL, E. coli MutL; StMutL,Salmonella typhimurium MutL; HexB, Streptococcus pneumoniae HexB. Sequences are from SWISS-PROT or GenBank databases. Amino acids that are identical for at least five proteins are in bold. Boxes indicate positions where amino acid changes were found in HNPCC patients and made in yeast MLH1 in this study. The amino acid substitutions are shown below the alignments.
Mutator phenotypes of haploid mlh1 mutants
Mutator phenotype of MLH1/mlh1 heterozygous diploids.To examine if these missense mutations exert a dominant mutator effect, we constructed diploid yeast strains heterozygous for each of the six mlh1 mutations. The haploid strain E134 and the mlh1 mutants were crossed to theMLH1 + strain E68, and Lys+ reversion rates were measured in the resulting diploids. A portion of theLYS2 gene including the site of thelys2::InsE-A14 mutation is deleted in strain E68. Thus, Lys+ revertants can arise in diploids only by lys2::InsE-A14 reversion, which occurs predominantly via frameshift mutations within the homopolymeric run. The reversion rates in strains heterozygous for themlh1 missense mutations were 2.7- to 6.8-fold higher than in the homozygous wild-type diploid strain (Table3). A similar result was observed in aMLH1/mlh1Δ heterozygote, indicating that the mutator phenotype of heterozygous strains is not due to an inhibition of the wild-type Mlh1p function in the presence of mutant Mlh1p.
Mutator phenotypes of diploids heterozygous formlh1 mutations
The mutator phenotype of the MLH1/mlh1 strain results from loss of heterozygosity.The mutator phenotype of the heterozygous strain could reflect slightly reduced MMR activity in allMLH1/mlh1 cells or loss of the wild-type allele in a small population of cells. In the latter case, Lys+ revertants would display the strong mutator phenotype characteristic of a null mutant. To examine this, we constructed diploid strains to first monitor reversion of thelys2::InsE-A14 mutation and then examine the Lys+ revertants for a mutator phenotype by using a second reporter locus, his7-2. Strain E134 and itsmlh1Δ derivative were crossed to strain 1B-D770, which carries the lys2-Tn5-13 and his7-2mutations. Thus, the diploids carry two different mutant alleles ofLYS2 (lys2::InsE-A14 andlys2-Tn5-13) and are homozygous for thehis7-2 mutation. The lys2-Tn5-13allele is an insertion of bacterial transposon Tn5 in theLYS2 gene that reverts at frequencies less than 10−8 in both wild-type (12) and MMR-deficient strains (11, 39). Thelys2::InsE-A14 andlys2-Tn5-13 mutations are located in the same site of the LYS2 gene (44, 45), such that the heteroallelic diploids can not revert to Lys+ phenotype via intragenic recombination. Thus, Lys+ revertants of diploid strains are expected to result from single base pair deletions within the A14 run.
Lys+ reversion rates in the 1B-D770-derived diploids (not shown) were indistinguishable from those shown in Table 3, with theMLH1/mlh1Δ heterozygous strain having a rate 5.2-fold higher than that of the wild-type diploid. We isolated 104 independent Lys+ revertants from the wild-type diploid and 102 independent revertants from the MLH1/mlh1 strain and then qualitatively analyzed His+ reversion in these clones as described in Materials and Methods. The 104 Lys+ clones derived from the MLH1 + strain appeared to revert to His+ at a frequency similar to that of an MMR-proficient strain. However, 67 of 102 Lys+ clones derived from the heterozygous strain showed a mutator phenotype, similar to the phenotype of the mlh1Δ haploid strain (data not shown).
We then examined the status of the MLH1 locus in Lys+ revertants by using a PCR approach (Fig.2A) whereby the wild-type MLH1allele yields a 586-bp fragment and the mutantmlh1Δ::LEU2 allele yields a 810-bp fragment. Among 15 Lys+ revertants of the wild-type strain analyzed, we observed only the fragment characteristic of the wild-typeMLH1 gene (seven examples are shown in Fig. 2B). All of the 15 revertants of the heterozygous strain that did not show a strong mutator phenotype yielded both the 586- and 810-bp fragments (Fig. 2B), indicating retention of both alleles. In contrast, we observed only the 810-bp fragment characteristic of the mutant mlh1Δ allele in 15 revertants of the heterozygous strain that had a mutator phenotype (Fig. 2B). Thus, Lys+ revertants of the heterozygote that have a mutator phenotype lost the wild-typeMLH1 allele.
MLH1 allele status in Lys+revertants of the MLH1/mlh1Δ strain. (A) Locations of primers used for PCR amplification of the wild-type andmlh1Δ::LEU2 alleles. Open box, MLH1open reading frame; solid box, the LEU2 gene replacing 230 bp of upstream and 300 bp of MLH1 coding region (36). Arrows indicate locations of primers. (B) PCR analysis of the MLH1 locus in Lys+ revertants. Lane 1,MLH1/MLH1 diploid; lane 2, MLH1/mlh1Δ diploid; lanes 3 to 9, Lys+ revertants obtained in the wild-typeMLH1/MLH1 strain; lanes 10 to 16, Lys+revertants with normal mutability obtained in theMLH1/mlh1Δ strain; lanes 17 to 23, Lys+revertants with a mutator phenotype obtained in theMLH1/mlh1Δ strain. Sizes of the amplified fragments (in base pairs) are shown on the right.
Damage-induced loss of heterozygosity.We then determined if the mutator effect in the MLH1/mlh1Δ strain is enhanced by treatment with bleomycin, a radiomimetic chemical used in cancer chemotherapy. Bleomycin induces mitotic recombination in yeast (16, 27), which could lead to loss of heterozygosity at theMLH1 locus, thus selectively enhancing the mutator phenotype in the MLH1/mlh1Δ but not the MLH1/MLH1 strain. When Lys+ reversion was examined in a qualitative test, the mutator phenotype of the untreated heterozygous strain relative to the untreated wild-type strain was readily apparent (Fig.3, plates on the left). Treatment with bleomycin (5 μg/ml) had little apparent effect on the reversion frequency in the wild-type strain (Fig. 3, top right) but induced a strong increase in reversion of the MLH1/mlh1Δ strain (Fig. 3, bottom right). This increase was quantified by analysis of revertant frequencies (see Materials and Methods), which demonstrated a 56-fold-higher bleomycin-induced revertant frequency in the heterozygous strain than in the wild-type strain (Table4). Similar results were observed upon irradiation of the two strains with 260-nm UV light (30 J/m2); the revertant frequency was selectively increased in the heterozygous strain but not in the wild-type strain (data not shown).
Effect of bleomycin treatment on Lys+reversion in diploid strains. The analysis was performed as described in Materials and Methods.
Effect of bleomycin treatment on Lys+reversion frequency in diploid strains
DISCUSSION
It is well established that mutations in MLH1, or loss of MLH1 expression due to promoter hypermethylation, can inactivate MMR, leading to genome instability. Here we show that the opposite is also true, i.e., that too much Mlh1p can destabilize the yeast genome. MLH1 expression from the ADH1promoter yields a strong mutator effect, while a lesser but significant effect is observed when the MLH1 gene on a multicopy plasmid is expressed from its natural promoter. This instability is in marked contrast to results obtained with the yeast MSH2 gene, which does not confer a mutator phenotype when overexpressed in the wild-type cells (6). The mutator effects of excess Mlh1p are reminiscent of the elevated mutation rates of methotrexate-resistant, cultured mammalian cells overexpressing human MSH3 (7, 23). That mutator effect was suggested to result from binding of excess human MSH3 to MSH2, leading to a reduced level of human MSH2-MSH6 heterodimer, with subsequent inability to repair base-base mismatches. By analogy, the overproduced Mlh1p may also sequester other proteins and prevent their participation in mismatch repair and/or other DNA transactions important for controlling the mutation rate. Mlh1p is known to interact with Pms1p (human PMS2) to form a heterodimer which functions in mismatch repair (20, 37) and with other proteins, including MutS homologues and PCNA (13, 14, 47). The mutator phenotype caused by elevation of Mlh1p level suggests thatMLH1 expression may be strictly regulated to ensure a low mutation rate.
All six yeast mlh1 missense mutations resulted in strong mutator phenotypes in haploid strains. This finding suggests that they strongly inactivate mismatch repair and, since the corresponding mutations in humans cosegregate with HNPCC (22, 32, 42, 49), that they are important for cancer predisposition. Similar interpretations resulted from a recent study (40) of four homologous human MLH1 missense mutations, M35R (M32R in yeast), G67R (G64R), I68N (I65N), and T117M (T114M). Each of the four human MLH1 mutations eliminated the dominant-negative mutator effect characteristic of expression of wild-type humanMLH1 cDNA in yeast. A mutator phenotype was also demonstrated (29) for a different yeast mlh1missense mutation that mimics an HNPCC-associated mutation (A41F). Taken together, these studies demonstrate the value of using homologous yeast genes to examine the functional effects of missense mutations found in human cancer patients.
Two of the yeast missense mutations that we studied, G64R and I65N, change conserved amino acids in a proposed ATP-binding motif (2,28) shared by bacterial MutL, its homologs, and several other proteins, including DNA gyrase, Hsp90, type II topoisomerases, and bacterial histidine kinase. More recent studies (1) reveal that E. coli MutL does bind ATP and has intrinsic ATPase activity and that an N-terminal fragment of MutL has a crystal structure homologous to that of a two-domain ATPase-containing fragment of NgyrB, a DNA gyrase. When the locations of the mlh1missense mutations that we studied are mapped on the structure of the homologous MutL, five are found in or near four conserved structural motifs (1), designated I, II, III, and IV in Fig.4 (shown in red). In NgyrB, these motifs contain residues that directly interact with ADPnP, an ATP analog (see Fig. 4b in reference 1). The sixth MLH1missense mutation that we examined (G243D) maps to a second MutL domain (lower domain in Fig. 4). By analogy to other family members, this domain may move significantly to stabilize ATP binding, since it has been shown that certain structural elements in the ATP-binding domains of homologous proteins undergo substantial conformational changes in apo versus ligand-bound forms (34, 35). Collectively, the available data suggest that the six missense mutations that we studied may inactivate MMR by altering the capacity of Mlh1p to bind ATP, to change conformation, and/or to hydrolyze ATP.
Putative locations of MLH1 missense mutations based on the structure of E. coli MutL. In the ribbon diagram of the 40-kDa N-terminal fragment of E. coli MutL protein (1), the four conserved ATP-binding motifs are shown in red, and the amino acids homologous to residues in yeast Mlh1p where the missense mutations were made are marked with black balls. Amino acid residue numbers for E. coli MutL are shown first, with the numbers for homologous, wild-type yeast residues given in parentheses.
By using the highly sensitivelys2::InsE-A14 reporter system, this study provides the first evidence of a mutator phenotype in cells heterozygous for missense mutations or deletion of one MLH1allele. The data indicate that the mutator phenotype does not reflect a small increase in mutation rate in all cells but rather reflects loss of the wild-type MLH1 allele with a strong increase in mutation rate in a small fraction of the cells. Loss of the wild-type allele may result from any of several mechanisms, including mitotic recombination, deletion, or chromosome loss. In yeast, the predominant mechanism is reciprocal mitotic recombination between the gene and the centromere (reviewed in reference 33), which can be induced by a variety of physical and chemical agents. We demonstrate here that treatment of yeast cells with the radiomimetic antibiotic bleomycin or with UV irradiation results in hypermutability of the heterozygous MLH1/mlh1Δ strain but not the MMR-proficient strain. If a similar situation exists in human cells, then exposure of heterozygous individuals to environmental mutagens, including those that induce mitotic recombination (9, 18, 21, 50), might be a decisive factor in tumor development. The approach used here with yeast strains heterozygous for an MMR gene mutation suggests several experiments that can be performed to test this hypothesis with heterozygous human and mouse cell lines. For example, a human cell line in which MLH1 deficiency was corrected by chromosome transfer has anhprt mutation rate that is slightly higher than that of normal cells (10). Experiments are in progress to determine if hprt mutant clones isolated from this line are mutators at a second locus and have lost the wild-type human MLH1allele and if a selective mutator effect can be induced in this cell line by DNA damage.
ACKNOWLEDGMENTS
We are grateful to Hiep Tran and Youri Pavlov for providing yeast strains, to Louise Prakash for providing yeast plasmids pMMR74 and pMMR75, and to R. Michael Liskay for providing yeast plasmid pmlh1Δ-LEU2. We also thank Wei Yang for generously providing information prior to publication of her work on E. coli MutL, including the coordinates for the N-terminal fragment of MutL that were needed to prepare Fig. 4. We thank Dmitry A. Gordenin for many helpful discussions and Karin Drotschmann and Leroy Worth for critical comments on the manuscript.
FOOTNOTES
- Received 16 October 1998.
- Returned for modification 23 November 1998.
- Accepted 14 December 1998.
- Copyright © 1999 American Society for Microbiology
