The role of mutagenic metal ions in mediating in vitro mispairing by alkylpyrimidines.

A variety of alkylating mutagens and carcinogens produce pyrimidine adducts in DNA that block DNA synthesis in vitro. Since DNA synthesis past the lesion is a necessary step to produce mutations, we investigated the role of the mutagenic metal ion Mn++ in facilitating DNA synthesis past alkylpyrimidines. In the presence of the natural metal activator Mg++, N3-ethyldeoxythymidine (N3-Et-dT) and O2-ethyldeoxythymidine (O2-Et-dT), present at a single site in DNA, blocked in vitro DNA synthesis 3' to the lesion and after incorporating dA opposite each lesion. The presence of Mn++ permitted postlesion synthesis with dT misincorporated opposite N3-Et-dT and O2-Et-dT, implicating these lesions in A.T-->T.A transversion mutagenesis. The DNA synthesis block by O4-ethyldeoxythymidine (O4-Et-dT) in the presence of Mg++ was partial and was also removed by Mn++. Consistent with in vivo studies, dG was incorporated opposite O4-Et-dT during postlesion synthesis, leading to A.T-->G.C transition mutagenesis. We also have discovered a new class of DNA adducts, N3-hydroxyalkyldeoxyuridine (3-HA-dU) lesions, which are produced by mutagenic and carcinogenic aliphatic epoxides. 3-HA-dU is formed after initial alkylation at the N3 position of dC followed by a rapid hydrolytic deamination. As observed with the analogous mutagenic N3-Et-dT, the ethylene oxide-induced 3-hydroxyethyldeoxyuridine (3-HE-dU) blocked in vitro DNA synthesis, which could be by-passed in the presence of Mn++. The nucleotide incorporated opposite 3-HE-dU during postlesion synthesis is being identified. These studies suggest a role for Mn++ in mediating mutagenic and carcinogenic effects of environmentally important ethylating agents and aliphatic epoxides. ImagesFigure 2.Figure 4.Figure 5.Figure 6.


Introduction
Alkylating agents have been used extensively in studying the mechanisms of mutagenicity and carcinogenicity (1)(2)(3)(4)(5)(6)(7)(8), because of their ability to react with DNA either in vitro or in vivo. Of the best characterized alkylating agents are the N-nitroso compounds. Their occurrence is widespread in the environment, and human exposure from natural and pollutant sources is universal (9). These agents induce tumors in a wide variety of tissues of different animal species (5,(10)(11)(12)(13)(14) and probably humans (9,15).
Most N-nitroso-alkylating agents mediate their biological activities in part by interacting with genomic DNA, forming covalent adducts (4). Alkylation at thymine occurs at nucleophilic oxygen sites, such as the 04and &2positions of the base and at the N3 position (4,16). There are many factors determining the sensitivity to the toxic, mutagenic, and carcinogenic potential of N-nitroso alkylating agents. One of these factors is the capacity to repair alkylated DNA. A substantial body of experimental evidence has indicated that it is not the initial level of alkylation but the persistence (lack of repair) of premutagenic alkyl adducts in tissues which is of major importance in mutagenesis and malignancy in specific organs. This has been demonstrated for 06-alkyldeoxyguanosine (06-alkyl-dG) and 04alkyldeoxythymidine (04-alkyl-dT), where the persistence of these lesions did correlate with organotropic malignancy (17)(18)(19)(20). A wide range of independent studies has indicated that the repair of ethyldeoxythymidine (Et-dT) adducts in mammalian cells is very slow (21). These adducts are among the highly persistent DNA alkylation products in both cultured mammalian cells and animal tissues (19,22). The persistence of premutagenic Et-dT adducts would increase the probability of mutation at A-T base pairs relative to G-C. This is consistent with the prevalence of transversion and transition mutations at A-T base pairs following in vivo exposure of mice to Nethyl-N-nitrosourea (ENU) (23).
The mutational spectra of ENU has been reported in a variety of systems. In E. coli, ENU induced mainly G-C-oAT and A-T-*G-C transition mutations (24). Presumably, these mutations resulted from the unrepaired 06-Et-dG (2,25,26) and 04-Et-dT (2,5,27) lesions, respectively, as a consequence of their capacity to mispair with dT and dG, respectively, during DNA replication. Under SOS-induction, ENU generated a large fraction (46%) of transversion mutations at A-T base pairs in E. coli (28). In human cells, ENU produced a significant number (29% or more) of transversion mutations at the A-T base pairs (8,29) in addition to the same G-C->A-T and A-T-*GC transitions observed in E. coli. In vivo exposure of mice to ENU predominantly (94%) induced transversion and transition mutations at A-T base pairs (23). A study of mutational spectra in Salmonella attributed transversions at A-T base pairs to Et-dT adducts (6). Transitions at A-T base pairs can result from 04-Et-dT. The Et-dT lesions responsible for transversion mutations at A-T base pairs are not known.
The biological importance of A-T transversion mutations has been demonstrated in mammalian systems. The A*T->T-A transversion event has been proposed to account for two mutations of Environmental Health Perspectives the mouse a-and [-globin genes arising in the progeny of ENU-treated female mice (31,32). Tumors of the nervous system induced by transplacental treatment of rats with ENU contained the neu oncogene activated by an AT-4TA transversion mutation (13,14). The two activating mutations, observed in c-Ha-ras genes of liver tumors induced by treating mice with diethylnitrosamine (33), were the AT-+TA and A-T->GC that occurred at codon 61.
We also have discovered a new class of DNA adducts, 3-HA-dU, which are produced by the mutagenic and carcinogenic epoxides ethylene oxide (EO), propylene oxide (P0), glycidol, epichlorohydrin (ECH), and the epoxide of acrylonitrile (34)(35)(36)(37)(38). 3-HA-dU is formed after initial alkylation at the N3 position of dC, followed by rapid hydrolytic deamination to 3-HA-dU (34,37,38). The hydroxyalkyl group of 3-HA-dU occupies a central Watson-Crick hydrogen-bonding position and is likely to disrupt normal base pairing. 3-HA-dU is stable in DNA in vitro and may be the critical promutagenic lesion produced by aliphatic epoxides in vivo. The role of 3-HA-dU in mutagenesis by aliphatic epoxides is unknown.
To ascertain the mutagenic potential of pyrimidine alkylation in DNA, we initiated in vitro DNA replication studies on Et-dT (39-43) and 3-HA-dU lesions (44,45), site-specifically placed in the same DNA template sequence. In vitro replication studies cannot precisely mimic the actual conditions of in vivo DNA synthesis, but they have proven to be a powerful means to understand the mechanisms responsible for the fine nucleotide selection exhibited by DNA polymerases. The site-modified template used in the in vitro DNA replication studies corresponds to a portion of the bacteriophage 4X174 genome in the gene G region (46). In addition to replication studies, the same DNA sequence can be used in separate in vivo site-specific mutagenesis studies, using the 4X174-based mutagenesis system (25) to facilitate comparison between in vitro and in vivo mutagenesis studies.

Bioloqical Significance of Thymine Ethylation in DNA
The alkylating agent ENU is capable of inducing a variety of tumor types in a broad range of animal species (10)(11)(12)(13)(14) and humans (9,15). The reactivity of ENU allows it to form a diverse set of DNA adducts both in vitro and in vivo (4,6,16). The order of formation of Et-dT adducts is d-Et-dT > 04-Et-dT > N3-Et-dT. These lesions are poorly repaired in mammalian systems (21) and thus may be more biologically important in these systems. Ethylation of dT may alter its base-pairing pattern to form a miscoding lesion such as 04-Et-dT (5,61). Alternatively, ethylation may compromise the ability of the base to serve as a template during DNA replication, producing noncoding lesions such as N3-Et-dT (39,62) and CY-Et-dT (1,41). Under conditions of relaxed fidelity of DNA synthesis, the noncoding lesions may mispair to produce mutations. Transition and transversion mutations at AT base pairs form an important component of ethylating agent-induced mutagenesis in SOS-induced bacteria (6,28), human cells (8,29) and animal systems (23), suggesting that dA or dT adducts and/or a breakdown product of these adducts are responsible for A-T mutations. A*T->G*C transitions can be derived from 04-Et-dT by mispairing with dG (27). A comparison of ENU-induced mutations with base substitutions produced by other alkylating agents in bacteria and human cells has led to the suggestion that d-Et-dT may be a significant premutagenic lesion in mammalian cells capable of inducing A*T-4T*A tranversion mutations (8,29,63). Indirect support for this hypothesis was derived from mutations observed in vivo in ENU-treated mice, where mutations at A-T base pairs accounted for 94% of all mutations (23). Among the A*T mutations, 55% were A-T-4T-A transversions. To ascertain the mutagenic potential of Et-dT lesions, we studied in vitro DNA replication properties of each Et-dT lesion site-specifically incorporated into the same DNA template. The replication studies utilized the primed template system shown in Figure 1.
The replication system contains a 36nucleotide site-modified DNA template hybridized to a 32P-labeled 17-nucleotide complementary primer. The construction of site-modified templates has been described (39,42). The Et-dT adducts and their derivatives, used in the synthesis of site-modified oligomers, were fully characterized by thin-layer chromatography, high pressure liquid chromatography (HPLC), ultraviolet, mass and nuclear magnetic resonance (NMR) spectroscopy. The presence of the Et-dT moiety in the purified oligomer was demonstrated by HPLC analysis of the nucleosides released from the site-modified oligomer following diges-Environmental Health Perspectives tion with phosphodiesterase and phosphatase (39,42).
In the DNA replication system ( Figure  1), the 3'-end of the primer is eight nucleotides away from the thymine modification (T*) present in the template. This system represents a "running start" for DNA replication in that synthesis occurs prior to the polymerase reaching the lesion. The hybridized primer is extended by the polymerase until T* is encountered. The following DNA products, reflecting the influence of template T*, are feasible. First, the progress of the polymerase is blocked 3'to T*. No nucleotide is incorporated opposite the lesion and a 25-nucleotide preincorporation blocked product accumulates. Second, DNA synthesis terminates after incorporating a nucleotide opposite T*, producing a 26-nucleotide incorporation-dependent blocked product. Finally, DNA synthesis proceeds past the lesion yielding a 36-nucleotide postlesion synthesis product. Products of DNA synthesis were analyzed by polyacrylamide gel electrophoresis. Since the 32P-end labeled primer is used to prime DNA synthesis, each product is only labeled once at the 5'end. This facilitates the quantitation of DNA synthesis products in the polymerization reaction by measuring the radioactivity associated with the individual product bands. The identity of the nucleotide incorporated opposite T* was established by DNA sequencing of the 26-nucleotide blocked and the 36-nucleotide postlesion synthesis products.

DNA Replication Properties of 02-Et-dT
The 02-position of dT does not participate in Watson-Crick base pairing. However, ethylation of the 02-position of dT interferes with normal hydrogen bonding of dT with dA, by fixing the thymine base in the enol tautomer with the loss of a hydrogen atom at the central hydrogen-bonding site (N3) of dT (42). Disruption of normal base pairing may inhibit DNA synthesis. This is consistent with our DNA replication studies where, in the presence of the natural metal activator Mg", 02-Et-dT blocked DNA replication by Kf Pol I predominantly 3' to the lesion (41). DNA synthesis past the lesion was negligible (<1%). Incorporation of dA opposite o-Et-dT occurred with increasing deoxyribonucleoside-5'-triphosphate (dNTP) concentrations (41), which was further enhanced by inhibiting the 3'-5' exonuclease proofreading activity of the polymerase with deoxyadenosine-5'-phosphate.
The postlesion synthesis remained negligible (41). The 0--Et-dT'dA base pair may occur with the formation of two hydrogen bonds between the 04 of O-Et-dT and the N6 hydrogen atom of dA, and between the N3 of (I-Et-dT and the protonated M of dA. A similar hydrogen bonding scheme has been suggested for 04-Et-dT-dA by NMR studies (64). The (I-Et-dT.dA base pair with two hydrogen bonds is expected to be thermodynamically stable. This is consistent with thermal denaturing studies where (I-Et-dT, present in the alternating poly[d(A.T)]polymers, did not alter the thermal melting profile (65).
Inhibition of DNA synthesis after incorporation of dATP opposite template 0(-Et-dT (41) or 02-Et-dT-5'-triphosphate opposite template dA (66) suggests that the geometric conformation of the o2-Et-dT-dA base pair deviates significantly from that of the normal Watson-Crick pair and may adopt a wobble conformation (64,67). In the wobble conformation, phosphodiester links (both 3' and 5' to dA) may have to be distorted to accommodate the 0 -Et-dT-dA base pair in a DNA helix. This hypothesis is consistent with molecular and computer modeling studies, indicating that the presence of (I-alkyl-dT in DNA may cause distortion in the DNA structure (1). Additional support for this hypothesis is derived from 31P NMR studies of DNA duplexes, containing the wobble base pairs 06-Et-dG dC and 04-Et-dT-dA (64). Distortion of the phosphodiester links 3' and 5' to dT and dG, respectively, was observed. The conformational changes associated with phosphodiester bonds during the formation of the (I-Et-dT.dA base pair are expected to adversely effect the catalysis of phosphodiester links on both the 3' and 5' sides of the incoming dA opposite 02-Et-dT during DNA replication. This would suggest that incorporation of dA opposite O-Et-dT and extension of the resulting 02-Et-dT-dA base pair will be inefficient. This is supported by in vitro DNA replication studies in the presence of Mg", where DNA synthesis was terminated predominantly (94%) 3' to 0-Et-dT, and postlesion synthesis did not occur (41). Similar results were obtained during DNA replication using bacteriophage T7 DNA polymerase (T7 Pol) (42). The block by the 02-Et-dT.dA base pair appears to be an inherent property of the spatial conformation of this base pair, since it was observed repeatedly during synthesis by Kf Pol l (41) or T7 Pol (42) in the presence of Mg++ or Mn++ from running (41,42) or standing (68) starts. Under normal cellular conditions, extension of the (I-Et-dT.dA base pair may either not occur or may occur with low efficiency. Our DNA replication studies in the presence of Mg++ (41) suggest that (I-Et-dT may contribute in part to the cytotoxicity of ethylating agents. Since DNA replication past the lesion is a necessary step in the production of mutations, we investigated the role of the mutagenic metal ion Mn++ in mediating DNA synthesis past the (I-Et-dT adduct. When Mn++ was substituted for Mg++ in the polymerization reaction, incorporation of a nucleotide opposite 02-Et-dT and subsequent postlesion synthesis were enhanced (41). Increasing the dNTP concentration and inhibiting the proofreading activity of Kf Pol I increased postlesion synthesis (which reached 66% at 200 pM dNTP) (41). DNA sequencing of the blocked and postlesion synthesis products revealed that while dA was present opposite 02-Et-dT in the blocked product, both dA and dT were present opposite the lesion in the postlesion synthesis product (41). The presence of dA opposite (I-Et-dT in both the blocked and postlesion synthesis products indicates that, in the presence of Mn+, the 0-Et-dT.dA base pair at the 3'-end of the growing chain can be extended but inefficiently. This is in contrast to DNA replication in the presence of Mg", where the 02-Et-dT-dA base pair was not extended. ciently extended. Other mis4pairs, including 06-Et-dG-dT and 0 -Et-dT-dG, which contain one hydrogen bond and retain normal Watson-Crick alignment, have been shown to be efficiently extended in vitro (61,69). These studies suggest that correct alignment of the backbone is crucial in DNA replication (64,67). The strength of hydrogen bonding is of secondary importance. Molecular and computer models together with physicochemical studies on the 02-Et-dT-dT base pair will provide insight into this hypothesis.
The kinetic mechanisms by which o Et-dT impedes DNA synthesis and miscodes were studied (70). The kinetic parameters, K. and Vmax, for dA and dT insertion opposite and extension past o-Et-dT by Kf Pol I in the presence of Mg++ and Mg", were determined using a polyacrylamide gel assay (71)(72)(73). Insertion and extension frequencies of 02_-Et-dT.dA and d-Et-dT.dT base pairs, relative to the right base pair (dT.dA), were estimated from V1mnIK ratios. The prema m liminary results revealed (70) that, in the presence of Mn+, 02-Et-dT inhibited insertion and extension of dA and dT at this lesion with efficiencies of 10-4 or lower. As compared to Mg++, Mn++ increased insertion frequencies of dA and dT opposite 02-Et-dT by 10-fold or more. The insertion was enhanced primarily through Km discrimination. The extension frequency at the 02-Et-dT.dT base pair was enhanced 6-fold when Mn++ was substituted for Mg++ in the polymerization reaction. Mn++ had little effect on extension of the 02-Et-dT-dA base air. As compared to d-Et-dT-dA, the 0-Et-dT.dT mispair was extended 30 times more efficiently (70). A higher extension frequency of the 02-Et-dT-dT mispair was primarily achieved through an increase in V.max The results suggest that, as compared to Mg++, Mn++ may increase the residence time of the 02-Et-dT-dT mispair at the catalytic site of the polymerase. These kinetic studies are consistent with in vitro DNA replication studies of 0 -Et-dT discussed above (41,42). They suggest that the 02-Et-  (27). The alkyl group of 04-alkyl-dT is located within the Watson-Crick base pairing region and may interfere with normal hydrogen bonding of dT with dA. The 04-alkyl-dT lesion may behave as a miscoding and/or noncoding lesion during DNA replication. The noncoding lesions usually inhibit DNA replication by distorting the DNA structure. Miscoding lesions alter the precision of base pairing during DNA synthesis leading to mutation. The role of 04-alkyl-dT in blocking DNA synthesis and the reaction conditions allowing DNA synthesis past the lesion were studied through in vitro DNA replication studies of 0 -Et-dT, site-specifically incorporated into a DNA template. The replication studies utilized the primed-template shown in Figure 1.
In the presence of the natural metal activator Mg", 04-Et-dT presented a partial (54%) block to DNA replication by Kf Pol I, with replication mainly (48%) interrupted 3' to 04-Et-dT. DNA synthesis past the 04-Et-dT lesion was 46%. Accumulation of the blocked product, obtained after incorporation of a nucleotide opposite 04-Et-dT, was low (6%) and remained constant over a wide range of dNTP concentrations (10-200 pM). The results suggest that, during DNA replication past 04-Et-dT in the presence of Mg", insertion of a nucleotide opposite this lesion may be the rate-limiting step. The replication block by 04-Et-dT was removed when Mn++ was substituted for Mg++ and a postlesion synthesis product was obtained in high yield (>90%). DNA sequencing revealed that predominantly dG was incorporated opposite 04-Et-dT ( Figure 2) during DNA synthesis past the lesion. The results implicate Mn++ in facilitating the insertion of dG opposite 0 -Et-dT and extension of the resulting base pair. Kinetics of insertion opposite template 0 -Me-dT have been described (61). (nucleotide insertion) rather than the effect of proofreading because the efficiencies using Kf Pol I were similar to those using the Drosophila melanogaster polymerase a-primase complex, which does not contain detectable 3'-*5' exonuclease proofreading activity (75). In the absence of any dNTP, the "weak" 3'-5' exonuclease activity of Kf Pol I did not excise the 04-Me-dT'dG base pair (5). This is consistent with our DNA replication studies where the 04-Et-dT'dG base pair can be easily extended.
NMR studies on DNA duplexes containing 04-Me-dT have indicated that the 04-Me-dT-dA base pair has a wobble conformation with the alkylated base moved towards the major groove of the helix (64). A wobble alignment for 04-Me-dT-dG was ruled out and it was suggested that this base pair retains the normal Watson-Crick alignment (64). Due to steric hindrance by the 04-methyl group, the normal alignment may have only one hydrogen-bond between the 2amino of dG and the o2 of 04-Me-dT. The 04-Me-dT-dG mispair having less hydrogen bonding than 04-Me-dT-dA is consistent with the optical melting profiles of duplexes where the duplex containing 0 -Me-dT'dA pairs melted at higher temperatures than the one containing 04-Me-dT-dG pairs (76).
An important factor in miscoding by 04-Me-dT is that the 04-Me-dT.dG mispair retains the Watson-Crick alignment, with no distortion of phosphodiester links 3' and 5' to the dG, as revealed by NMR studies (64). This is consistent with the greater efficiency of dG incorporation opposite 04-Me-dT as compared to dA (61). Stimulation of dG incorporation opposite 04-Et-dT in the presence of Mn+ in our studies suggests a role for Mn++ in stabilizing the 04-Et-dT'dG mispair in a normal Watson-Crick alignment to facilitate the formation of phosphodiester bonds. Based on NMR studies on nucleotide binding to E. coli DNA polymerase I (77), it has been suggested that the polymerase-Mn++ complex may be less selective of the sugar ring conformation of the nucleotide than the enzyme DNA Relication Properties of N3-E -d N3-alkyl-dT is formed, both in vitro and in vivo, but in relatively small amounts (4,6,16). Among the alkylating agents, ENU demonstrates a significant amount of binding to O-alkyl and N3-alkyl pyrimidines. N3-alkyl-dT is stable in DNA in vitro. No DNA repair activity has been reported for N3-alkyl-dT. This DNA lesion may be persistent in vivo and exerts its biological consequences long after exposure has occurred. The alkyl group of N3alkyl-dT occupies the central Watson-Crick hydrogen bonding site (N3) of thymine and is likely to interfere with normal hydrogen bonding of dT, probably leading to mispairing and/or inhibition of DNA synthesis. N3-alkyl-dT is likely to be a potentially cytotoxic and mutagenic lesion produced by alkylating agents. The biological significance of the N3-alkyl-dT lesion was ascertained through in vitro DNA replication studies of N3-Et-dT (39,40,43) present at a single site in the DNA template shown in Figure 1.
In the presence of the natural metal activator Mg++ and a low dNTP concentration (10 11M), N3-Et-dT blocked DNA synthesis by Kf Pol I predominantly 3' to N3-Et-dT. DNA synthesis past the lesion was not observed (39). Incorporation of dA opposite N3-Et-dT occurred with increasing dNTP concentrations, but no postlesion synthesis was obtained (39). Similar results were obtained during DNA replication with T7 Pol, where incorporation of dA opposite N3-Et-dT blocked DNA synthesis in the presence of Mg++ (43). These studies implicate N3-Et-dT as a potentially cytotoxic lesion produced by ethylating agents.
No postlesion synthesis suggests that the N3-Et-dTdA base pair, formed at the replication fork, did not retain the Watson-Crick alignment and may cause distortion in the DNA structure. Due to steric hindrance by the lesion, translocation of the polymerase to the nucleotide past the lesion may be extremely slow. Polymerization of the nucleotide past N3-Et-dT may also be very slow, owing to the need to extend the distorted terminus formed by the wobble N3-Et-dT-dA base pair. During a pause at the lesion, the polymerase may dissociate from the lesionblocked primer-template complex and cease elongation. Once dissociated, rebinding of the polymerase to the primer-tem-plate complex may lead to formation of a defective initiation complex of lower stability. This complex may be stable enough to allow the relatively easy first polymerization step of inserting dA opposite N3-Et-dT, but not the next very slow step of extending the N3-Et-dT*dA base pair. This notion is consistent with DNA replication studies (39), where in the presence of Mg++ the blocked product, obtained after incorporation of dA opposite N3-Et-dT, was formed but not extended.
When Mn++ was substituted for Mg", incorporation of dA opposite N3-Et-dT was increased from 4% in the presence of Mg++ to 60% in the presence of Mn++ at 10 ,uM dNTP (40). At this dNTP concentration, DNA synthesis past the lesion was not obtained. Postlesion synthesis occurred at higher dNTP concentrations and reached 68% at 200 pM. During postlesion synthesis, dT was incorporated opposite N3-Et-dT (40), implicating this lesion in transversion mutagenesis at the A-T base pair by ethylating agents. The results suggest a role for Mn++ in mediating an AT--TA transversion mutation by the N3-Et-dT lesion.
The absence of dT opposite N3-Et-dT in the blocked product and its presence only in the postlesion synthesis product (40) suggest that the N3-Et-dT-dT mispair formed at the replication fork is not inhibitory to DNA synthesis. This mispair is efficiently extended, leading to an AT-*TA mutation. These results are similar to those observed in the case of (I-Me-dG-dT (64), 04-Me-dT-dG (64) and (Y-Et-dT-dT (41,42) mispairs, which were efficiently extended and led to mutations.
As shown for d-Me-dG.dT and 04-Me-dT-dT mispairs (64), the N3-Et-dT-dT mispair may also retain the normal Watson-Crick alignment, facilitating extension of this mispair. This is consistent with postlesion synthesis in high yield (68%) observed in our DNA replication studies (40).
Since the N3-Et-dT'dT mispair is efficiently extended, incorporation of dT opposite N3-Et-dT appears to be the ratelimiting step during posdesion synthesis. In contrast to the low dNTP concentration (10 pM), higher dNTP concentrations facilitated insertion of dT opposite N3-Et-dT (40). Polymerization of the next correct nucleotide following the N3-Et-dT-dT mispair protected the mispair from excision by the proofreading activity of the polymerase. Since the polymerase has difficulty extending from the N3-Et-dT-dA base pair, this base pair becomes susceptible to Volume 102, Supplement  proofreading. Excision of dA from N3-Et-dT-dA makes the template N3-Et-dT lesion available for insertion of dA and dT opposite the lesion with dT insertion efficiently extended. The net result of this cycling is accumulation of the postlesion synthesis product, containing dT opposite N3-Et-dT in high yield (68%) (40).
In contrast to Kf Pol I, T7 Pol was highly specific in incorporating only dT opposite N3-Et-dT in the presence of Mn++ during postlesion synthesis (>95%) even at 10 pM dNTP (43). The greater ability of T7 Pol to incorporate dT opposite N3-Et-dT may reflect the large difference in processivity between T7 Pol and Kf Pol I (67). T7 Pol incorporates thousands of nucleotides prior to dissociating from the template (78). The high affinity of T7 Pol for the DNA template may increase the likelihood of insertion and extension of dT opposite N3-Et-dT by the polymerase. The high yield of postlesion synthesis probably was achieved through protection of the N3-Et-dTdT mispair by extension and excision of the N3-Et-dTdA base pair by the highly potent 3'-*5' exonuclease proofreading activity associated with T7 Pol.

Biological Significance of Aliphatic Epoxide-induced 3-HA-dU Lesions
Aliphatic epoxides include a number of very reactive reagents. They contain threemembered rings, which are highly strained. The least substituted carbon, which is sterically more accessible, is the site of attack of most epoxides (SN2 mechanism) under neutral conditions. These epoxides are a class of direct-acting alkylating agents, which are effective mutagens and animal carcinogens (79)(80)(81). Because of their usefulness as chemical reagents, the simple volatile epoxides, especially EO, PO, and ECH, are used extensively in industry and result in potential human exposure in the work place. An association between exposure to EO and cancer has led to the suggestion that EO may be involved in the etiology of human cancer. PO and ECH have been shown to be rodent carcinogens.
Alkylation of DNA, followed by the persistence of premutagenic adducts during the period of DNA replication and cell division, seem to be necessary but not sufficient requirements for initiation of carcinogenesis with many alkylating agents. We discovered a new class of DNA adducts (34-37), 3-HA-dU, which is produced by mutagenic and carcinogenic epoxides. The 3-HA-dU lesion is stable in DNA in vitro. This lesion is formed in DNA after rapid hydrolytic deamination of a 3-hydroxyalkyldeoxycytidine (3-HA-dC) adduct (34,38). The initial reaction of SN2 epoxides occurs at N3 of dC. The charge on the protonated 3-HA-dC intermediate (pK=9) is delocalized over the C4 carbon, which is then attacked by the side chain OH group, resulting in the loss of ammonia (deamination) with retention of the charge at C4. Hydroxyl attack at this carbon results in ring opening and formation of 3-HA-dU (Solomon, unpublished). The halflives of hydrolytic deamination of 3-HE-dU and 3-HP-dU at pH 7.4 and 37°C were 10 hr and 6 hr respectively (34,37). The 3-hydroxyalkyl group of 3-HA-dU occupies a central Watson-Crick hydrogen bonding position and is likely to disrupt normal hydrogen bonding, leading to mispairing and/or inhibition of DNA replication. 3-HA-dU may be the critical premutagenic lesion produced by SN2 epoxides in vivo. In order to understand the role of 3-HA-dU in mutagenesis and cancer induction, we initiated in vitro DNA replication studies of 3-HE-dU and 3-HP-dU, produced by EO and PO respectively.
The replication system for 3-HA-dU ( Figure 3) is similar to the one used for Et-dT adducts (Figure 1), except that the primer used was a 12-mer and the sitemodified DNA template was a 50-mer. Use of the 50-nucleotide template was necessitated by the need to facilitate sequencing of the postlesion synthesis product by Sanger's dideoxy-sequencing protocol, and to incorporate the site-modified 50-mer into a 50-nucleotide gap iñ X174 replicative form DNA for in vivo site-specific mutagenesis studies. To facilitate comparison of the results between Et-dT and 3-HA-dU adducts, U* was located in the same DNA sequence from 4X174.
The site-modified oligomers were fully characterized for their purity, expected DNA sequence and presence of U* in essentially all oligomer molecules. The details of the synthesis of site-modified oligomers and their characterization will be described elsewhere. In this replication system, the expected DNA synthesis products are a 20-nucleotide preincorporationblocked, a 21-nucleotide incorporationblocked and a 50-nucleotide postlesion synthesis product. DNA Replication Properties of 3-HE-dU As observed with the analogous mutagenic lesion, N3-Et-dT, 3-HE-dU blocked DNA synthesis by Kf Pol I (Figure 4). In the presence of the natural metal activator Mg++ and at 10 pM dNTP, the majority (70%) of the block was 3' to 3-HE-dU and the remainder (27%) was blocked after incorporating a nucleotide opposite 3-HE-dU (44). Synthesis past the lesion was negligible (<3%). Incorporation opposite 3-HE-dU increased with increasing dNTP concentrations, reaching 60% at 200 pM. Postlesion synthesis remained negligible (<3%). DNA sequencing of the 21nucleotide blocked product revealed that w w .  dA is incorporated opposite 3-HE-dU ( Figure 5). Since postlesion synthesis is negligible, the results suggest that the 3-HE-dU-dA present at the growing replication fork is inhibitory to DNA synthesis. The results are similar to the analogous N3-Et-dT lesion, where the N3-Et-dT-dA base pair was not extended in the presence of Mg++ (39). Our DNA replication studies implicate 3-HE-dU as a potentially-cytotoxic lesion produced by EO. During in vitro DNA replication, the 3-HE-dU-dA base pair behaved in a manner similar to N3-Et-dT-dA, 02-Et-dT-dA, 04-Et-dT-dA, and d5-Me-dG.dC The Mn++-mediated synthesis past 3-HE-dU suggests that the role of Mn++ in modifying the fidelity of Kf Pol I in DNA replication may be comparable to the SOSinduced functions in bacteria. In bacteria, SOS-induced proteins may alter the fidelity of the DNA replication complex, facilitating the incorporation and subsequent extension at 3-HE-dU. This hypothesis suggests that mutagenesis by aliphatic epoxide-induced 3-HA-dU requires induction of the SOS system in bacteria. This is supported by the production of SOSdependent mutagenesis by PO at template cytosines (45). Involvement of the SOSlike system in mammalian cells is not known. Inside the mammalian cell, DNA polymerase-accessory proteins may facilitate incorporation and subsequent extension at 3-HA-dU. Since 3-HA-dU is derived from deamination of aliphatic epoxide-induced 3-HA-dC, extension of all base pairs (except 3-HA-dU.dG) at 3-HA-dU will produce mutations. Our in vitro DNA replication studies have demonstrated formation of a 3-HE-dU-dA base pair. In vivo extension of this pair will produce a GC->A-T transition mutation and implicate the 3-HE-dU lesion in G-C-4AT transition mutagenesis by EO. Support for this hypothesis is derived from our mutagenesis studies with PO, where G-C-A-T transitions represent an important component of PO-induced mutational spectra (45). DNA Replication Properties of 3-HPdU PO-induced 3-HP-dU behaved in a similar manner as 3-HE-dU (44) and the analo-gous N3-Et-dT lesion during in vitro DNA replication (39,40). 3-HP-dU blocked DNA synthesis by Kf Pol I in the presence of Mg++ 3' to 3-HP-dU and after incorporating a nucleotide opposite the lesion (Figure 4). Postlesion synthesis was negligible. Substitution of Mn++ for Mg++ mediated DNA synthesis past 3-HP-dU. The specificity of the nucleotide incorporated opposite 3-HP-dU in the blocked and postlesion synthesis products is being investigated.

Mispairing
Manganese is known to be highly mutagenic in vivo and to reduce the fidelity of DNA synthesis in vitro (82). It also shows a strong co-mutagenic effect with UV (60). The mechanism of Mn++-induced mutagenesis has been studied using E. coli DNA polymerase I (50). The role of a free Mn++ concentration on Mn++-induced mutagene- sis in vitro was determined by an analysis of the polymerase error rate and a comparison with dissociation constants of Mn++ from the enzyme, template, and dNTP. These studies suggest that, in the presence of physiologically-relevant free Mn++ concentrations (10-100 pM), Mn++ induced misincorporations by interacting with the DNA template rather than with the polymerase (50). At higher Mn++ concentrations (0.5-1.5 mM), Mn++-induced mutagenesis is probably due to Mn++ association either with single-stranded regions of template DNA or with weak sites in the polymerase. The results with T4 DNA polymerase suggest that the mutagenic action of Mn++ can be attributed primarily to a significant differential increase in binding of mispaired relative to correctly paired nucleotides to the polymerase-template complex (49). The resulting increase in residence times for mispaired nucleotides on the complex results in their increased frequency of misinsertion. A smaller contributing factor to Mn++induced mutagenesis was the loss of proofreading specificity (49). These conclusions are supported by other kinetic studies of incorrect vs. correct incorporation during DNA synthesis (51,83).
The alkyl group of alkylated pyrimidines studied by us, whether located at a Watson-Crick base-pairing position (i.e., N3-Et-dT, 3-HE-dU, 3-HP-dU and 04-Et-dT) or not (i.e., d-Et-dT), interferes with normal hydrogen bonding of the alkylpyrimidines, leading to mispairing or blocking of DNA synthesis. Since normal hydrogen bonding is disrupted, a nucleotide inserted opposite alkylpyrimidine represents a mismatch and may be recognized by the 3'->5'-exonuclease proofreading activity of the polymerase. Mispairs that retain the Watson-Crick alignment are efficiently extended, leadin,g to misincorporation. Mispairs, such as (I-Me-dG-dT and 04-Me-dT-dG, shown to retain Watson-Crick conformation, are efficiently extended (64). Mispairs that adopt a wobble conformation are either not extended or extended inefficiently, owing to the distortion caused by the mispair in the DNA structure. We postulate that the role of Mn++ in mediating mispairing by alkylpyrimidines is to increase the probability of inserting a nucleotide opposite the lesion in a conformation that retains the Watson-Crick alignment. In our DNA replication studies, a higher Mn++ concentration (500 jiM) was used. At this concentration, Mn++ binds to DNA polymerase, template DNA, and dNTP substrates (50). The normal Watson-Crick alignment of the mispairs formed by the 0O-,0 and N3-Et-dT lesions may have been achieved through interactions of Mn++ with the polymerase-template-dNTP complex. NMR studies of nucleotide binding to E. coli DNA polymerase I have shown that binding of the dNTP substrate to the active site of the polymerase in the presence of Mn++ may occur in conformations favorable for Watson-Crick base pairing in DNA B form. These types of conformations may not be favored in the presence of Mg++ (77). This hypothesis is consistent with our DNA replication studies, where 02-Et-dT-dT, O4-Et-dT-dG and N3-Et-dT*dT mispairs are formed and efficiendy extended in the presence of Mn++ but not Mg". Conclusion Alkylation of thymine in DNA is toxic, mutagenic, and carcinogenic. Et-dT adducts block in vitro DNA synthesis, often after incorporating dA opposite the lesions. In vivo extension of the Et-dT-dA base pair is nonmutagenic. Failure to extend the Et-dT'dA base pair implicates Et-dT lesions in cytotoxicity by ethylating agents. Mn++-mediated mispairing and bypass of Et-dT adducts suggest a comutagenic role for Mn++ in the mutagenicity of ethylating agents and implicates Et-dT adducts in A T-4T-A and A-T-*G*C mutations.
The implication, that ENU-induced Et-dT lesions can produce A-ToTA and AT->GC mutations, and the observation of these mutations at the activating site of oncogenes isolated from ENU-induced tumors, emphasize the existence of an important, but unproved, relationship among the formation of N3-Et-dT, d-Et-dT and 04-Et-dT lesions, the induction of transversion and transition mutations at A*T base pairs, and the subsequent development of cancer by N-nitrosoethylating agents. The prevalence of N-nitroso compounds in the environment and their formation in vivo have led to the suggestion that they may be involved in the etiology of human cancers (9). The demonstration of N-nitroso alkylating agent-induced mutations in activated oncogenes (11,13,14,32 suggests that cellular protooncogenes represent important targets for these agents. Although Et-dT adducts are rapidly repaired in bacteria (30), their repair in mammalian systems is not efficient (21). The Et-dT adducts are among the highlypersistent DNA ethylation products in mammalian systems (19,22. Our in vitro DNA replication studies suggest that ethylation of thymine in DNA is biologically significant and may exert cytotoxic, mutagenic and carcinogenic activity long after the exposure has occurred. Preliminary DNA replication studies of 3-HE-dU and 3-HP-dU suggest a dual role for epoxide-induced 3-HA-dU lesions. They block DNA replication in vitro and may terminate DNA synthesis in vivo, contributing to the cytotoxicity of aliphatic epoxides. Under relaxed polymerase fidelity (Mn++), DNA synthesis past 3-HE-dU and 3-HP-dU occurs, implicating these lesions in mutagenesis at GC base pairs by epoxides. The studies provide a basis for understanding molecular mechanisms by which environmentally important aliphatic epoxides produce mutations and contribute to the process of carcinogenesis.