Metal-induced infidelity of DNA synthesis.

In this paper, we consider the effects of metal ions on the accuracy of catalysis by DNA polymerases. Certain activating and nonactivating metal ions have been shown to diminish the fidelity of DNA synthesis in vitro with a variety of DNA polymerases. There is a significant correlation between the metals that decrease fidelity and those that have been reported to be mutagenic and carcinogenic. Thus, metal carcinogens are no exception to the general postulate that carcinogens can be identified by their interactions with DNA.

In this paper, we consider the effects of metal ions on the accuracy of catalysis by DNA polymerases. Certain activating and nonactivating metal ions have been shown to diminish the fidelity of DNA synthesis in vitro with a variety of DNA polymerases. There is a significant correlation between the metals that decrease fidelity and those that have been reported to be mutagenic and carcinogenic. Thus, metal carcinogens are no exception to the general postulate that carcinogens can be identified by their interactions with DNA.

The Present State of Knowledge
A number of metals have been shown to be mutagens and carcinogens (1) and to affect the accuracy of DNA replication (2). In vivo systems are too complicated to begin to unravel the mechanisms by which metals induce mutations and the effects of metal ions on the fidelity of DNA replication. Our approach to this problem has been to examine DNA synthesis in vitro, to determine the effects of different metal ions on the fidelity of this process, and then to ask whether alterations in the fidelity of DNA synthesis are related to the mutagenic and carcinogenic properties of these metals.
Prior to considering these studies in detail, it is instructive to consider the mechanism of DNA synthesis in vitro and methods for measuring fidelity of DNA synthesis.

Mechanism of DNA Polymerization
The requirements for catalysis by various DNA polymerases appear to be similar. Synthesis proceeds by a sequential addition of nucleotide monomers (deoxynucleoside monophosphates) with the concomitant release of pyrophosphates (3). DNA polymerases are part of a unique class of enzymes * The Joseph Gottstein Memorial Cancer Research Laboratory, Department of Pathology, University of Washington, Seattle, Washington 98195 August 1981 in that they primarily take direction from another molecule, a template. In cells, the template is DNA. Synthetic polydeoxynucleotides and polyribonucleotides can also serve as templates for most DNA polymerases in vitro. Synthesis is started on the 3'-hydroxy terminus of a primer-strand hybridized onto a template strand. The primer can be an oligonucleotide, one strand ofdouble-stranded DNA, a hairpin loop of single-stranded DNA, or a fragment of RNA hybridized onto DNA. Thus, DNA polymerases only elongate already existing polynucleotide chains; they fail to initiate chains de novo as do RNA polymerases. The substrates of all known DNA polymerases are deoxynucleoside triphosphates that are complementary to the template. Based on the similar requirements for activity and a spectrum of similar kinetic parameters, it is a reasonable expectation that there is a common mechanism for catalysis by DNA polymerases from different sources (4).
DNA polymerases can be classified as zinc metalloenzymes (5) in that they contain tightly bound zinc, which is required for activity (6). In addition, these enzymes require an added metal for catalysis, which in cells is presumably Mg2". In vitro, Mg2+ has been shown to coordinate the enzyme with the substrate in the form of an enzyme-metal-substrate complex. Mn2+, Co2+, Ni2?, and in certain cases, Zn2+, have been shown to substitute for Mg2`as activators (7). Analysis of E.
coli DNA polymerase I-Mn2+-substrate complexes indicates that in the absence of template, the 197 enzyme alters the conformation of the deoxynucleoside triphosphate substrate to that which it would occupy in double helical DNA (8). The function of the metal activator in the DNA polymerase reaction per se has recently been extensively reviewed (9,10).

Fidelity of DNA Synthesis
On the basis of spontaneous mutation rates in procaryotic and eucaryotic cells, stable misincorporation of a base during DNA replication is estimated to occur with a frequency of 10' to 10`1 per base-pair synthesized (11). This accuracy appears to be achieved by a multistep process (12). The differences between correct and incorrect Watson-Crick base-pairings involve only one or two hydrogen bonds. This difference in free energy, AG has been estimated to account for an error rate of approximately 10-2 (13). DNA polymerases also participate in base-selection, reducing the error rate to values approaching 10-5 (4). Presumably, other cellular mechanisms increase accuracy during or immediately after DNA replication. At present, studies on the effects of metals on fidelity in vitro are limited to analysis with purified DNA polymerases.

Assays of Fidelity with Polynucleotide Templates
Until very recently, all assays of the fidelity of DNA synthesis in vitro measured the ability of DNA polymerases to copy homopolymer or alternating copolymer templates. These templates contained only one or two nucleotides and the mismatched nucleotide was identified simply as one not complementary to the template nucleotides. Using this assay, one can observe the effects of both activating and non-activating metals on the fidelity of DNA synthesis. The template that we have chosen for critical measurements of fidelity is poly[d(A-T)], a synthetic polynucleotide consisting of deoxythymidine and deoxyadenosine monophosphates. Poly[d(A-T)] can be synthesized to contain less than 1 in 2 x 106 mistakes by using a de novo reaction with E. coli DNA polymerase I (14). Copied correctly, only dAMP and dTMP should be incorporated into the newly synthesized product. By using [a-32P]-dTTP, unlabeled dATP, and [3H]-dGTP or [33H]-dCTP, one can simultaneously measure the incorporation of complementary and non-complementary nucleotides (15). The incorporation of either dCTP or dGTP would represent errors. The frequency of misincorporation is obtained from the ratio of ['H] to [32p] in 198 the acid-insoluble product. Control experiments are required to show that the [3H] label in the reaction product is in the noncomplementary nucleotides and not in any radioactive contaminants. Also, it must be demonstrated that the noncomplementary nucleotides are covalently incorporated in phosphodiester linkage. Using nearest-neighbor analysis, one can determine the distribution of the noncomplementary nucleotides. Measurements of the frequency of misincorporation by DNA when copying polynucleotide templates have been summarized in a recent review (12). In general, they vary from 103 for DNA polymerases from RNA tumor viruses to 10-5 for procaryotic DNA polymerases. It should be noted that the error rates of the procaryotic DNA polymerases, those with a 3' -+5' exonuclease, are similar to those of the eucaryotic DNA polymerases, enzymes that do not have an accompanying exonuclease (16). Thus, the exonuclease in procaryotic DNA polymerases is not necessarily the major determinant of fidelity.

Metals Activators and Fidelity
The activating metal for DNA polymerase in vivo is Mg2e. In vitro, DNA polymerases from animal (16), viral (17), and bacterial sources can also use Mn2+, Co2+, or Ni2 + as the activating metal. With AMV DNA polymerase and an activated DNA template, the maximal rates of nucleotide incorporation with Mn2+, Co2+, and Ni2+ were 65%, 25%, and 7%, respectively, of that achieved with Mg2+ (18). Minimal activity has also been reported with  (17). This error rate is invariant with respect to Mg2 + concentration. At the optimal activating concentration ofMn2 + (0. 1mM), the error rate was 1 in 15,000. At greater than activating concentrations of Mn2+, there was a progressive decrease in the incorporation of the complementary nucleotide but not of the noncomplementary nucleotide, thus yielding a further increase in the frequency of misincorporation. At concentrations as great as 2mM, the error rate approached 1 in 3600, and nearest-neighbor analysis indicated that each misincorporation occurred as a single-base substitution. A similar enhancement in the error rate is observed with both activating and inhibiting con- with all templates and noncomplementary nucleotides tested. An absolute increase in the rate of incorporation of the noncomplementary nucleotide at high Mn2 + concentration can be demonstrated by simply using more DNA polymerase in the assay or prolonging the time of incubation.
The enhancement of misincorporation with an alternate metal activator appears to be a characteristic finding with DNA polymerases ( Table 1). Substitution of Mn2 + for Mg2 + results in an increase in misincorporation by E. coli DNA polymerase I (7), T4DNA polymerase (19), DNA polymerase-a (17, 20) DNA polymerase-3 (17), and avian myeloblastosis virus (AMV) DNA polymerase (15,21). The fact that Mn2 + and Co2 + alter the fidelity of the DNA polymerases that do not have an associated exonuclease (AMV, DNA polymerases-a and -I) indicates that for these DNA polymerases the metal ions do not promote misincorporation by inhibiting an error correcting exonucleolytic activity. Ni2+ can also substitute for Mg2+ as a metal activator. However, the amount of synthesis achieved with Ni2+ as the metal activator has not been sufficient to accurately measure the changes in the fidelity of DNA synthesis with any DNA polymerase except AMV DNA polymerase, in which case Ni2+ promotes misincorporation (18).
In order to relate the measurements with alternate metal activators to a situation that would be expected to occur in cells, the effects of these activators on Mg2+-activated DNA synthesis have been investigated. Co2 +, Mn2 +, and Ni2 + have been shown to enhance misincorporation by DNA polymerases in the presence of activating amounts of Mg2+ (17,21,22). Thus, these metal activators could alter the fidelity of DNA polymerases in cells.

Nonactivating Metal Ions and Fidelity
Beryllium, a known carcinogen, has been shown to decrease the fidelity of catalysis with M. luteus DNA polymerase (23) and AMV DNA polymerase (24). Be2+ is unable to substitute for Mg2+ as a metal activator. However, as a nonactivating cation,

Screening for Metals that Alter Fidelity
To date, over 40 metal compounds have been tested in graded concentrations for their effects on the fidelity of DNA synthesis. The method of analysis and the results are summarized in Figure  2. In the initial study, 22 of these metal salts were tested by using a triple-blind protocol in which the assays, computations, and designation of each unknown compound with respect to fidelity were carried out independently (2). Compounds which increased infidelity by greater than 30% at two or more concentrations were scored as positive. Metals were designated as carcinogens or mutagens by an evaluation of the literature prior to assessment of their effects on fidelity. An enhancement in the infidelity of DNA synthesis was observed with all of the known mutagens and/or carcinogens tested at that time (Ag, Be, Cd, Co, Cr, Mn, Ni, Pb). The evidence in the literature on the mutagenicity or carcinogenicity of three of the metal ions was considered equivocal. Of these, Cu2+ increased misincorporation; Fe2+ and Zn2+ did not alterfidelity.
All other metal salts that were tested were considered to be neither carcinogenic nor mutagenic, and they did not increase misincorporation. Only a few of the metal salts that did not alter fidelity are listed in Figure 2.
With only a few exceptions, these results have been confirmed by Miyaki et al. (22) and Sirover et al. (7), using E. coli DNA polymerase I, and by Seal et al., using DNA polymerases-a and -,B from human placenta (17). Most  of these two metal ions. We have also examined the effect of different ionic species of chromium and have found that both Cr (III) and Cr (VI) alter the fidelity of E. coli DNA polymerase I (26). Chain initiation by RNA polymerases can be stimulated to Pb2+, Cd2+, Co2 , Cu2+ and Mn2+, whereas Zn2+, Mg2+, Li+, Na+ and K+ are inhibitory (28). The similarity between the effects caused by particular metal ions on fidelity with DNA polymerases and on chain initiation with RNA polymerase could point to metal interactions with the DNA template as the common underlying mechanism for these two phenomena.
Fidelity of DNA Synthesis with Natural DNA Templates All of the aforementioned studies on the effects of metal ions on the fidelity of DNA synthesis depended on measuring the incorporation of noncomplementary nucleotides by using synthetic polynucleotide templates of limited composition. It has been assumed that the results with such a model system are similar to those that would be obtained copying natural DNA containing all four bases. It is known, however, that slippage of the primer relative to the template can occur when primed templates of a repeating nucleotide se-Environmental Health Perspectives quence are copied. Thus, metal-mediated changes in the fidelity of DNA synthesis could result from such slippage of the primer on the template, an event that presumably does not occur during copying of natural DNA templates. Also unique to homopolymers or repeating heteropolymers is the fact that a single noncomplementary nucleotide can occupy a looped-out structure without changing the reading frame of subsequent codons. Thus, metals could enhance misincorporation by increasing the frequency of such looped-out structures. To circumvent these limitations, a system has been recently developed (29) to monitor the fidelity of in vitro DNA synthesis using a natural DNA template, DNA from the bacteriophage XX174 carrying a suppressible nonsense mutation, amber 3 (am3) (Fig. 3). Certain nucleotide substitutions within the am3 locus that occur during in vitro replications of this DNA will cause a reversion to the wild type phenotype. Such reversions are detected by the bioassay method diagrammed in Figure 3. Thus, measurement of the reversion frequency of the progeny phage indicates the accuracy with which the DNA in the region of this mutation was copied. This assay system has been used with AMV DNA polymerase (30) and E. coli DNA polymerase I (31). With homogeneous AMV DNA polymerase Mg2", and equal concentration ofnucleotides, the in vitro mutation rate is approximately 1 in 1000. With E. coli DNA polymerase I, variations in the divalent metal ion activator used in the copying reaction markedly affects the reversion frequency of copied 4X174 am3 DNA. Thus, the calculated error rate observed with 5mM Mg2 + can be increased severalfold by the substitution of Mn2 + or Co2 + for Mg2 +. The error rate can also be increased by copying in the presence of inhibiting concentrations of Mg2 + or by the presence of Cr3+ or Cr6+ in Mg2+ -activated reactions (32). These limited studies suggest the metal mutagens and carcinogens also diminish accuracy with natural DNA templates. By determining the sequence of the DNA synthesized in the presence of mutagenic metals using the XX174 DNA template, it should be possible to define the specificity of interactions of these metals with the template nucleotides.

Mechanism of Genetic Miscoding by Metals
The exact mechanism by which certain divalent metal ions decrease the fidelity of DNA synthesis in vitro is not known. On the basis of the available data, three alternatives can be unambiguously eliminated, while three others may still be considered viable mechanisms and will require further investigation.
The following three possibilities by which metal MUTATION  It can be argued that the observed increase in error frequency at high metal concentration represents the selective acid precipitation of metal ion complexes containing unincorporated noncomplementary nucleotides. However, physical and enzymatic studies of the products synthesized with AMV DNA polymerase (33), E. coli DNA polymerase (7), and DNA polymerases-a and -,B (17), indicate that the noncomplementary nucleotides are incorporated into a polynucleotide chain, predominantly as single-base substitutions.
Metal-Substrate Interactions. Metal-induced infidelity does not appear to result from selective interactions between particular metals and particular nucleotides. For example, it could be argued that Co2 + selectively interacts with the complementary nucleotide and reduces its effective concentration in the reaction mixture. However, at a high concentration of Co2+ (5mM), the incorporation of dGTP as the complementary nucleotide with a poly (C) template is markedly inhibited, whereas the incorporation of dGTP as the noncomplementary nucleotide with poly [d(A-T)] as the template is undiminished (18). Similar results have been obtained with Mn2+ using various DNA polymerases with different template combinations.
Inhibition of "Proof-reading" Exonuclease by Metal Ions. The possibility that decreases in fidelity with divalent metal ions are mediated by inhibition of 3' --5' exonucleolytic activity is also unlikely. Eucaryotic DNA polymerases and DNA polymerases from RNA tumor viruses are devoid of such an activity (16), yet mutagenic metal ions decrease the fidelity of these enzymes. Detailed studies on the effect ofMn2" on fidelity, exonucleolytic activity, and monophosphate generation have been carried out with E. coli DNA polymerase I. Under conditions in which Mn2+ diminishes fidelity, there is no diminution of the 3' -* 5' exonucleolytic activity (7).
The decrease in fidelity of metal ions during in vitro DNA synthesis can be explained most directly by any one or more of the following types of interactions (Fig. 4).
Altered Substrate Conformation. The ability ofMn2 + Co2+, Ni2+, and possibly Zn2+ to substitute for Mg'+ as a metal activator focuses on the possibility that the mechanism of change in fidelity by these metals occurs by a substitution at the substrate binding site on the polymerase.  Mg2 . Magnetic resonance studies indicate that the interaction of the metal activator involves an enzyme-metal-substrate bridge complex involving the -y-phosphoryl group of the substrate (35). Studies with E. coli DNA polymerase I in the absence of template indicate that the bound metal changes the conformation ofthe substrate to that ofthe nucleotidyl Environmental Health Perspectives unit in double-helical DNA. Sloan et al. (8) noted that this conformation could reduce the frequency of misincorporation. Thus, it could be argued that differences in conformation of the bound substrate with different metal activators might account for differences in the fidelity of DNA synthesis, particularly at activating concentrations of these cations. However, the current data are not sufficient to eliminate the possibility that differences in fidelity reflect interactions of metal ions with the template and not the enzyme even when the metal serves as an activator. Thus, the parallel incorporation of complementary and noncomplementary nucleotides at activating metal concentrations could simply indicate that polymerization is a rate-limiting event and the metal-mediated change in fidelity could be at a site other than the substrate site on the enzyme. Altered Enzyme Conformation. The decrease in fidelity observed at inhibiting concentrations of metal activators suggests binding of metals at sites in addition to the catalytically active site. Ancillary binding sites for Mn2+ were detected on E. coli DNA polymerase I by nuclear magnetic resonance studies (35). The demonstrations that nonactivating metal ions alter the fidelity of other DNA polymerases is compatible with this concept. Also, evidence has been presented that Be2", a nonactivating cation, binds to AMV DNA polymerase directly and diminishes the fidelity of DNA synthesis in vitro (24). Thus, interactions of metals or metal-nucleotide complexes at distant sites could change the conformation of the polymerase so as to promote misincorporation. To date, attempts in our laboratory to generate an altered DNA polymerase with diminished fidelity by treatment with denaturing agents and heat have not been successful.
Altered Template-Base Specificity. The direct interaction of metal ions with phosphates and bases on polynucleotides have been measured by a number of physical techniques (36). Studies on the interaction of Mn2+ with activated DNA template by paramagnetic resonance (35) indicate 5 + 2 very tight sites and 52 weaker sites having an invariant association constant of 68 FM. The largest decreases in fidelity with Mn2 + were observed at much higher concentrations (2-5mM). Weak Mn2 + binding sites on E. coli DNA polymerase I have been reported (35). However, it is also possible that very weak binding sites on polynucleotides are responsible for diminished fidelity, and these would not be observed in the magnetic resonance experiments. Eichhorn and collaborators initially observed that metal ions can cause enhanced mispairing upon renaturation of polynucleotides (36). Conceivably, the metal ions can directly interfere with comple-August 1981 mentary base-pairing or cause a shift in the ketoenol equilibria of the nucleotide. Recent studies by Murray and Flessel (37) indicate that Mn2+ and Cd2+ promote mispairing during hybridization of the synthetic templates. Moreover, the mispairing with Mn2+ can be demonstrated to occur at millimolar concentrations.

Public Health Implications
The results in this working paper suggest that mutagenic metal ions alter the fidelity of DNA synthesis. This has been demonstrated with purified DNA polymerases using both synthetic and natural DNA templates. We argue that in studying fidelity of DNA synthesis by DNA polymerases, one is studying mutagenesis in vitro. Correlations observed between alterations in fidelity in vitro and mutagenicity or carcinogenicity in vivo are in accord with the hypothesis that infidelity during DNA synthesis may cause mutations. However, we recognize that metal ions have many other effects in vivo. Considerable evidence will be required to document whether or not alterations in the fidelity of DNA synthesis are causally associated with mutations and malignancy. Irrespective of a defined mechanism, the correlation between alterations in fidelity and mutagenicity and/or carcinogenicity indicates the practicality of using fidelity assays as a screen for evaluating possible mutagens and carcinogens. Since these assays are carried out in vitro in defined homogeneous systems, it is possible to design experiments to understand how metals alter the fidelity of DNA synthesis.
With respect to metals, diminished fidelity and somatic mutations, the following hypothesis can be generated. Metal-induced mutations may occur by the interaction of metal ions with the DNA template or with the DNA polymerase. In the latter case, a normal polymerase could be exposed to an abnormal concentration of physiologically required metals, or to exogenous metals that are usually not present during cellular metabolism. Alternatively, metal ions that are normally not used for DNA replication could serve as activators for DNA polymerases that have been previously altered. In either case, an abnormal polymerase-metal combination might decrease the fidelity with which the DNA is replicated, and thus lead to the synthesis of DNA containing mutations. This newly synthesized DNA may contain certain critical errors (e.g., genes which code for altered polymerases). Furthermore, continued replication of the DNA by an altered polymerase or in the presence of mutagenic metals could also lead to an accumulation of additional errors during subsequent rounds of replica-tion. Such critical errors and/or cascading errors caused by an accumulation of mutations may account for the progressive change in cellular properties during tumor progression.

Required Research
To date, only the initial studies have been conducted on genetic miscoding by metals. The main gap in our knowledge concerns the relationship between the studies with metals and purified polynucleotides in vitro, as well as those involving infidelity by DNA polymerases, to the realities of chromosomal replication in cells. The central question is whether or not the infidelity of DNA synthesis by metals is causally associated with mutagenesis and perhaps carcinogenesis. The approach to the problem requires an analysis of the effects of metals on the accuracy of DNA synthesis at three different levels of cellular organization.

At the Level of DNA Polymerases
Current evidence suggests that metal ions can alter the accuracy of catalysis by DNA polymerases from a variety of cells. The fact that infidelity of DNA synthesis correlates with mutagenic or carcinogenic properties of these metals may be only circumstantial when considering the multiple effects of metals in cells. The mechanism by which metals alter the fidelity of DNA synthesis has not been detailed. Careful physical and biochemical measurements are required on the binding of metals to DNA templates and polymerases before mechanistic details can evolve. Of interest is the possibility that particular metals interact preferentially with different nucleotides on DNA templates. An analysis of changes in sequence using natural DNA templates, such as 4X174 DNA copied in the presence of different metal ions, should provide more definitive results concerning specificity of metal-nucleotide interactions.

At the Level of DNA Replicating Complexes
The most faithful DNA polymerase copies polynucleotide templates in vitro with an error rate approximating 1iO This rate is considerably greater than the spontaneous mutation rates per nucleotide synthesized observed in cells (10-8-1O1). Clearly, other proteins in cells function in accuracy of DNA synthesis. From studies with E. coli DNA polymerase I andT4 DNA polymerase, it is apparent that DNA replication proceeds by the concerted action of multiple proteins (3). Further studies are required to determine the effect of each of these proteins on the replication process. With the establishment of enzyme/protein complexes capable of replicating DNA faithfully in vitro, it will be necessary to study the effects of metals on these complexes. An analysis of the effects of metals on the accuracy of catalysis by DNA replicating proteins is perhaps somewhat closer to the complexity involved in chromosomal replication in human cells.

At the Cellular Level
The main problem in studying detrimental effects of metals at the cellular level is that metals are required for many cellular processes. Nevertheless, a number of studies on cells in tissue culture demonstrate the mutagenic potential of metals. Thus, a study of the effects of metals on the fidelity of DNA synthesis in vivo is important since replication of DNA is required for the establishment of mutational alterations. It would be desirable to extend these studies using various eukaryotic cells that are defective in different pathways of DNA repair, since these cells would be more likely to exhibit metal-induced mutational alterations. There is also a large gap in knowledge of metal content and localization in cells. An analysis of metal content of eukaryotic cells, particularly in the nucleus and chromosomes, is required to better define the relationship between studies on fidelity in vitro and mutation rates in cells.