Cadmium inhibits the functions of eukaryotic MutS complexes.

Exposure of yeast cells to low concentrations of cadmium results in elevated mutation rates due to loss of mismatch repair (MMR), and cadmium inhibits MMR activity in extracts of human cells. Here we show that cadmium inhibits both Msh2-Msh6- and Msh2-Msh3-dependent human MMR activity in vitro. This inhibition, which occurs at a step or steps preceding repair DNA synthesis, is observed for repair directed by either a 3' or a 5' nick. In an attempt to identify the protein target(s) of cadmium inhibition, we show that cadmium inhibition of MMR is not reversed by addition of zinc to the repair reaction, suggesting that the target is not a zinc metalloprotein. We then show that cadmium inhibits ATP hydrolysis by yeast Msh2-Msh6 but has no effect on ATPase hydrolysis by yeast Mlh1-Pms1. Steady state kinetic analysis with wild type Msh2-Msh6, and with heterodimers containing subunit-specific Glu to Ala replacements inferred to inactivate the ATPase activity of either Msh2 or Msh6, suggest that cadmium inhibits ATP hydrolysis by Msh6 but not Msh2. Cadmium also reduces DNA binding by Msh2-Msh6 and more so for mismatched than matched duplexes. These data indicate that eukaryotic Msh2-Msh3 and Msh2-Msh6 complexes are targets for inhibition of MMR by cadmium, a human lung carcinogen that is ubiquitous in the environment.

Occupational exposure of humans to cadmium is associated with cancers of the lung, prostate, kidney, and pancreas, and experiments in rodents link cadmium exposure to lung, prostate, and pancreatic tumors (2). Thus, both the International Agency for Research on Cancer (2) and the National Toxicology Program (3) have classified cadmium as a carcinogen. Cadmium is considered a non-genotoxic carcinogen, and it has a number of biological effects (reviewed in Ref. 4). Among these, we recently reported that exposure of Saccharomyces cerevisiae to low concentrations of cadmium is highly mutagenic for base substitution and especially for insertion/deletion errors in homonucleotide runs (1). Cadmium does not increase mutation rates in strains defective in mismatch repair (MMR) 1 (1). Haploid yeast strains defective in proofreading by DNA polymerase ␦ are killed by cadmium, and proofreading defective diploid strains treated with cadmium survive and show synergistic mutagenesis with cadmium. These observations and the lack of or weak mutagenesis upon treatments with equivalent concentrations of other metals and agents causing DNA damage all suggest that cadmium-induced mutagenesis is not due to DNA damage per se but is rather due to inhibition of MMR of DNA replication errors. In further support of this hypothesis, cadmium inhibits the ability of extracts of human cells to repair a single-base insertion deletion mismatch in vitro (1), and cadmium treatment of cultured human cells suppresses MMRmediated cell cycle arrest after exposure to the alkylating agent MNNG (5).
Possible protein targets for cadmium inhibition of MMR can be considered in light of extensive knowledge of eukaryotic MMR gained over the past ten years (reviewed in Refs. 6 -15). Eukaryotic MMR is initiated when complexes of Msh2-Msh6 (MutS␣) or Msh2-Msh3 (MutS␤) bind to a mismatch. MutS␣ is primarily responsible for repairing single base-base and insertion/deletion (IDL) mismatches, MutS␤ is primarily responsible for repairing IDL mismatches containing multiple extra nucleotides in one strand, and the two complexes share responsibility for repairing IDL mismatches with one extra base. Eukaryotes also encode multiple MutL homologs that form different heterodimers. These heterodimers act as matchmakers to coordinate the various steps in MMR. MutL␣ (Mlh1-Pms1 in yeast) is involved in repairing a wide variety of mismatches, while MutL␤ (Mlh1-Mlh2) and MutL␥ (Mlh1-Mlh3) are thought to participate in repairing a subset of insertiondeletion mismatches. Several exonucleases are implicated in mismatch excision, especially including Exo1, and excision and DNA re-synthesis also require MutS␣, MutL␣, RPA, RFC, PCNA, and DNA polymerase ␦ (see Refs. 16 and 17 and references therein).
Inactivation of MMR protein functions is well known to have numerous biological consequences, including genome instability, resistance to DNA damaging agents including chemotherapeutic drugs, altered class switch recombination and somatic hypermutation of immunoglobulin genes, emergence of pathogenic bacteria, infertility, and increased susceptibility to cancer (reviewed in Refs. 6, 7, 12, 13, 15, and 18 -20). Because cadmium is ubiquitous in the environment and can accumulate in the body due to a long biological half-life (Ref. 21 and references therein), and because cadmium is a known carcinogen that inhibits MMR, we undertook the present study to identify the MMR protein target(s) and the mechanism of inhibition of MMR by cadmium. Here we present evidence that cadmium inhibits two functions of eukaryotic MutS complexes, ATP hydrolysis, and specific binding to mismatched DNA. These effects are sufficient to explain cadmium inhibition of MMR activity in vivo.

EXPERIMENTAL PROCEDURES
Cell-free Extracts, Mismatch Repair, and in Vitro Replication Assays-Procedures for extract and heteroduplex substrate preparation and for measuring repair activity were as described (22). The substrates used are described in the legend to Fig. 1. SV40 origin-dependent DNA replication assays were performed as described (23). Where indicated, the reaction mixture was preincubated with 50 M CdCl 2 for 10 min at 0°C prior to adding substrate.
Purification of Heterodimers-yMlh1-yPms1 was purified as described previously (24). yMsh2-yMsh6 was overexpressed in selection * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. medium containing 2% galactose and purified as previously described with the following modifications (25). The heterodimer was step eluted from a nickel resin column with 100 mM imidazole, 20 mM Tris, pH 8, 200 mM NaCl and 5 mM ␤-mercaptoethanol. After eluting from a 1-ml heparin column, the fractions were applied to a 1 ml HiTrap Q column. Eluted fractions were dialyzed against 20 mM Tris, pH 8, 200 mM NaCl, 1 mM EDTA, and 1 mM dithiothreitol and stored at Ϫ70°C.

RESULTS AND DISCUSSION
Cadmium Inhibition of in Vitro Human Mismatch Repair-We previously showed that low concentrations of cadmium inhibit MMR activity in vitro by extracts of human cells using a substrate containing a 1-base IDL mismatch and a nick to direct strand-specific repair that is located 5Ј to the mismatch (see Ref. 1 and left most entry in Fig. 1A). To determine whether cadmium inhibits both MutS␣-dependent and MutS␤dependent MMR, and to determine whether cadmium inhibits MMR involving excision initiated at both 3Ј as well as 5Ј nicks, we began this study by examining the ability of cadmium to inhibit MMR using four additional substrates. The MutS␣-dependent repair of a G⅐G mismatch with a nick either 3Ј or 5Ј to the mismatch was inhibited by 50 M CdCl 2 (Fig. 1A). Additionally, the repair of a 2-base IDL mismatch that was shown previously to depend on MutS␤ (28) was also inhibited by 50 M CdCl 2 . These results in human cell extracts are consistent with genetic evidence in yeast (1) indicating that cadmium inhibits both MSH2-MSH3-and MSH2-MSH6-dependent repair of base substitution and IDL mismatches. They further show that cadmium inhibition occurs when the nick that serves as the strand discrimination signal is located either 3Ј or 5Ј to the mismatch.
Effect of Other Divalent Metals on MMR Activity-Two proteins involved in MMR, the large subunit of RPA (29) and DNA polymerase ␦ (30), are known to contain zinc fingers. Like zinc, cadmium has a high affinity for sulfhydryl groups in proteins and can compete with zinc for binding to zinc metalloproteins. We therefore tested whether zinc can prevent cadmium inhibition of MMR activity. In the absence of zinc, 15 M CdCl 2 moderately inhibited MMR (Fig. 1B). Addition of an equal amount of ZnSO 4 did not reverse this inhibition (Fig. 1B). This is interesting because zinc is reported to protect against cadmium-induced carcinogenesis (see Ref. 31 and references therein). Thus, the protection awarded by zinc in vivo may occur through another mechanism. Rather than protection, here we observe that zinc alone inhibited MMR activity (Fig. 1, B and C), and the inhibitory effects of cadmium and zinc were additive (Fig. 1B). These data suggest that cadmium is not inhibiting mismatch repair by replacing zinc in a zinc metalloprotein. Instead, they are consistent with the possibility that both metals may inactivate a common MMR protein target. Note that zinc inhibition of MMR is not observed in yeast (1), suggesting that other factor(s) affect the accessibility of metal ions to MutS␣ in vivo. In contrast to inhibition of MMR activity by cadmium and zinc, neither CoCl 2 , MnCl 2 , or NiSO 4 inhibited MMR activity (Fig. 2C and Ref. 1).
Effects of Cadmium on DNA Replication-Our initial study (1) demonstrated that MMR was inhibited in vivo by concentrations of cadmium that had little effect on yeast cell growth. This implies that cadmium does not strongly inhibit DNA replication. To test this in the human system, we examined the ability of CdCl 2 to inhibit SV40 origin-dependent DNA replication catalyzed by the same extracts that were used to measure MMR activity. Addition of 50 M CdCl 2 , which strongly inhibits MMR activity (Fig. 1A and Ref. 1), had a negligible effect on replication activity or on the distribution of replication products (Fig. 1D). The fact that cadmium does not inhibit yeast cell growth and has little effect on SV40 origin-dependent DNA replication in human cell extracts suggests that those proteins common to replication and MMR, i.e. RPA, RFC, PCNA, and DNA polymerase ␦, are unlikely targets for cadmium inhibition of MMR activity. In addition, the assay used to detect MMR activity in our initial study (1) and in Fig. 1A does not require repair DNA synthesis (32). Therefore, cadmium is most likely inhibiting human MMR activity in vitro at a step or steps prior to repair DNA synthesis. Moreover, Lü tzen et al. (5) have recently reported that cadmium does not inhibit the 5Ј-exonuclease activity of hEXO1 or formation of hEXO1 protein complexes. Collectively, all these facts focused our attention on examining whether cadmium alters the functions of those proteins required for the earliest steps of MMR, i.e. MutS␣ and MutL␣.
Cadmium Inhibition of ATP Hydrolysis by Yeast MutS␣ but Not MutL␣-Eukaryotic MutS␣ and MutL␣ both bind and hydrolyze ATP, and these abilities are essential for MMR activity. To determine whether cadmium could inhibit MMR activity by inhibiting ATP hydrolysis, the ATPase activities of yeast MutS␣ (Msh2-Msh6) and MutL␣ (Mlh1-Pms1) were assayed in the absence and presence of cadmium. While cadmium had little effect on ATP hydrolysis by MutL␣ ( Fig. 2A), 50 M CdCl 2 reduced ATP hydrolysis by MutS␣ to only 20% of that observed in the absence of cadmium ( Fig. 2A). As a negative control, 50 M MnCl 2 did not inhibit the ATPase activity of MutS␣ ( Fig.  2A). After incubation of yMsh2 His -yMsh6 in the presence of 50 M CdCl 2 , both proteins were pulled down from the mixture by nickel agarose beads (data not shown). Because cadmium does not eliminate heterodimerization of the two proteins, the inhibition of MutS␣ ATPase function by cadmium is not likely to be due to disruption of the Msh2-Msh6 complex. Inhibition of ATP hydrolysis by MutS␣ increased with increasing CdCl 2 concentration, and the dose response for this inhibition was remarkably similar to that observed for inhibition of MMR activity in extracts (see relative values in Fig. 2B). The effects of cadmium on ATP hydrolysis by yeast MutL␣ and MutS␣ were further examined by steady state kinetic analysis. The K m , K cat , and K cat /K m values for MutL␣ were similar in the presence and absence of cadmium (Table I), whereas for wild type MutS␣ (left panel in Fig. 2C), the presence of 50 M CdCl 2 increased the K m and decreased the K cat resulting in a 6-fold decrease in catalytic efficiency (Table I). These results demonstrate that cadmium inhibits the ATPase activity of yeast MutS␣ but not MutL␣.
Evidence That Cadmium Inhibits ATP Hydrolysis by Msh6 but Not Msh2-MutS␣ belongs to the ABC transporter family of ATPase proteins, and it has two composite ATPase active sites, each comprised of five conserved motifs contributed by one subunit and a sixth motif contributed in trans by the other subunit. Among these motifs, the Walker B motif of Msh2 and Msh6 contains a conserved glutamate residue that is essential for ATP hydrolysis (25). To determine whether cadmium inhibits ATP hydrolysis by Msh2, Msh6, or both, we performed steady state kinetic analysis of ATP hydrolysis by purified mutant heterodimers containing an inactivating glutamate to alanine substitution in the Walker B motif of either Msh2 (E768A) or yMsh6 (E1062A). Cadmium reduced the efficiency of ATP hydrolysis by the Msh2 E768A -Msh6 heterodimer by 20fold but had no effect on the efficiency of ATP hydrolysis by the Msh2-Msh6 E1062A heterodimer (Table I and Fig. 2C). These results indicate that cadmium inhibits ATP hydrolysis by yMsh6, but it does not inhibit ATP hydrolysis by yMsh2. This could be due to selective interaction of cadmium with Msh6 protein. Alternatively, because residues from Msh2 contribute to the composite ATPase active site in Msh6, ATP hydrolysis by Msh6 could be inhibited by cadmium interactions with Msh2. Experiments are under way to identify the target residues within Msh2 and/or Msh6 that are responsible for cadmium inhibition.
Inhibition of yMsh2-Msh6 Mismatch Recognition by Cadmium-The structure of bacterial MutS proteins (33)(34)(35)(36)(37) and extensive biochemical evidence (reviewed in Refs. 6 -15) have revealed that MMR requires close coordination between adenine nucleotide binding and ATP hydrolysis by the ATPase active sites in MutS␣ with its binding affinity for mismatched DNA. We therefore used an electrophoretic mobility shift assay to examine the effects of cadmium on MutS␣ binding to 41 base A, the effect of cadmium on MutL␣ and MutS␣ ATPase activity was assayed using 30 nM yMsh2-yMsh6 or 1 M yMlh1-yPms1 and 10 M ATP. Where indicated, reaction mixtures were preincubated on ice with CdCl 2 or MnCl 2 prior to initiating reaction with ATP. Numbers at the bottom of the lanes indicate fraction of ATP hydrolysis, normalized to the control. B, the relationship of CdCl 2 concentration and inhibition of yMsh2-yMsh6 ATPase activity. The CdCl 2 concentration in the reaction is indicated by the numbers above the lanes. The fraction of ATPase activity relative to control is displayed below the lanes. Below the ATPase activity is the fraction of repair activity relative to control observed in MMR reactions with TK6 extracts and identical CdCl 2 concentrations (1). C, plots of ATPase reactions measuring the effect of cadmium on yMsh2 and yMsh6 ATPases. The effects of 50 M CdCl 2 on yMsh2 and yMsh6 ATPases were measured using 220 nM yMsh2-Msh6 E1062A and 50 nM yMsh2 E768A -yMsh6, respectively. Wild type yMsh2-yMsh6 ATPase concentration was 30 nM. Reactions lacking CdCl 2 are represented by Ⅺ. Reactions containing CdCl 2 are represented by f. pair duplex DNAs that were either completely matched or contained a single G⅐T mismatch. The results show that in the absence of cadmium, and as expected based on earlier studies (38), yeast MutS␣ binds to mismatched DNA with 34-fold higher affinity than it binds to matched DNA (Table II). However, in the presence of 50 M CdCl 2 , MutS␣ binding affinity for mismatched DNA was reduced by 18-fold, to the K d value only slightly higher than that for binding to matched DNA (Table  II). Cadmium also reduced binding to matched DNA, but the K d only increased by 1.6-fold, from 34 to 55 nM (Table II).These results demonstrate that cadmium reduces the DNA binding affinity of MutS␣ as well as its ability to discriminate between matched and mismatched DNA. Summary and Implications-The results presented here are consistent with cadmium inhibition of MutS␣ and MutS␤-dependent MMR activity through interactions that inhibit ATP hydrolysis and reduce DNA binding affinity and mismatch recognition specificity by MutS␣ and, by extrapolation, perhaps MutS␤. Alone or in combination, these biochemical effects are sufficient to explain the inhibition of human MMR activity observed in vitro and of yeast MMR in cells (1) and the altered response of human cells to alkylation damage (5). However, the results presented here do not completely exclude that cadmium might also interfere with other proteins or steps in MMR.
The interactions of cadmium with Msh2 and/or Msh6 that are responsible for inhibition are unknown. Cadmium could perhaps inhibit ATP hydrolysis by competing with magnesium binding at the ATPase active site that is required for hydrolysis. Although this possibility cannot yet be excluded, it seems unlikely for three reasons. First, cadmium cannot replace magnesium for activation of the ATPase activity of MutS␣ (data not shown), and the effects we report are observed using cadmium concentrations (e.g. 50 M) that are much lower than the magnesium concentration (2 mM) used in the ATPase assays. Second, magnesium is required for ATP hydrolysis by both Msh2 and Msh6, yet the results in Fig. 2C and Table I suggest that cadmium preferentially inhibits ATP hydrolysis by Msh6 but not by Msh2. Third, if cadmium was to compete with magnesium at the ATPase active site, it would be expected to also inhibit ATP hydrolysis by bacterial MutS, yet in an initial experiment, 50 M CdCl 2 failed to inhibit ATP hydrolysis by Taq MutS (data not shown). An alterative possibility is that cadmium binds to Msh2 and/or Msh6 at a different location, e.g. through coordination with cysteines that contain sulfhydryl groups and/or histidine, glutamate, and aspartate residues in tetrahedral geometries (21,39). Such a cadmium coordination site could be in Msh2 or in Msh3 and Msh6 or some combination of these. This site should be highly specific to account for cadmium inhibition of MMR but not DNA replication (Ref. 1 and Fig. 1D) or other types of DNA repair (1) and to account for the asymmetric reduction in ATP hydrolysis (Table I) and the reduced DNA binding discrimination between mismatched and matched DNA. Theoretically, one cadmium binding site may be sufficient to influence both ATP hydrolysis and DNA binding specificity, e.g. if cadmium binding interferes with conformational changes in this large and conformationally active het-erodimer. Experiments are under way in an attempt to identify a cadmium binding site.