Comparison of the relative mutagenicities of O-alkylthymines site-specifically incorporated into phi X174 DNA.

The relative mutagenicities of O-alkylthymine-DNA adducts were analyzed in vivo by site-specific mutagenesis. Purified DNA polymerases were used to incorporate O4-methyl (Me)-, O4-ethyl (Et)-, O4-isopropyl (iPr)-, or O2-Me-dTTP onto the 3' terminus of a synthetic oligonucleotide (15-mer) hybridized to phi X174 am3 DNA. The product oligonucleotides were further extended in the presence of unmodified dNTPs to yield 21-mers containing single O-alkylthymine adducts opposite the adenine residue of the bacteriophage amber codon. Polyacrylamide gel electrophoresis and nearest-neighbor analyses confirmed the identities and nucleotide positions of the adducts. Transfection and replication of the site-specifically alkylated DNAs in ada- Escherichia coli (defective in the alkyltransferase capable of repairing O4-alkylthymine-DNA adducts) yielded mutant progeny phage with reversion frequencies of: O4-Me-dThd (19.5 X 10(-6) ) greater than O4-Et-dThd (7.5 X 10(-6) ) greater than O4-iPr-dThd (3.0 X 10(-6) ) greater than or equal to O2-Me-dThd (1.0 X 10(-6) ) approximately equal to dThd (2.0 X 10(-6) ). None of the adducts produced mutations above background following replication in ada+ E. coli. DNA sequence analyses of 40 independently isolated mutant phage derived from the O4-Me- or O4-Et-dThd-containing DNAs showed that all mutants contained guanine residues opposite the original site of the alkylthymines. These data are consistent with a mechanism of mutagenesis involving the formation of O4-alkyl-T.G base pairs during DNA replication in E. coli and suggest that the formation of A.T----G.C transition mutations is characteristic of mutagenesis by O4-Me- and O4-Et-dThds in vivo.

The relative mutagenicities of 0-alkylthymine-DNA adducts were analyzed in vivo by site-specific mutagenesis. Purified DNA polymerases were used to incorporate O'-methyl (Me)-, 04-ethyl (Et)-, 0'-isopropyl (iPr)-, or 02-Me-dTTP onto the 3' terminus of a synthetic oligonucleotide (15-mer) hybridized to 4x174 am3 DNA. The product oligonucleotides were further extended in the presence of unmodified dNTPs to yield 21-mers containing single 0-alkylthymine adducts opposite the adenine residue of the bacteriophage amber codon. Polyacrylamide gel electrophoresis and nearest-neighbor analyses confirmed the identities and nucleotide positions of the adducts. Transfection and replication of the site-specifically alkylated DNAs in a d d Escherichia coli ( N-Nitroso-alkylating agents are potent mutagens and carcinogens that mediate at least some of their biologic effects by covalent modification of DNA (see Refs. 1 and 2 for reviews). The mutagenicity of these agents is thought to be due largely to the formation of 06-alkyl-dGuo adducts which are effective miscoding lesions (3)(4)(5). However, several recent studies also strongly implicate 04-alkyl-dThd residues as causative lesions in mutagenesis (6)(7)(8)(9)(10)(11)(12) as well as carcinogenesis (13-17), and 02-alkyl-dThd adducts have been implicated as weak mutagenic lesions in DNA polymerization assays in uitro (6,9). 11 To whom correspondence should be addressed.
Studies on the molecular structure of 04-alkyl-dThds suggest that these adducts will pair with dAdo as well as dGuo residues in DNA (18)(19)(20). This hypothesis is supported by investigations in uitro, where purified DNA polymerases can incorporate either dATP or dGTP opposite 04-alkyl-dThd residues in synthetic polynucleotide templates (6-lo), and in uiuo, where the levels of 04-alkyi-dThd adducts (1) correlate with the frequency of A. T + G . C transitions produced by alkylating agents (11,12,21). Comparisons of the mutagenic (12,22) and carcinogenic ( 5 ) potencies of N-nitroso compounds in uivo also imply that ethyl adducts may be more biologically active than their methyl counterparts. However, in these studies it is difficult to determine which adduct (or adducts) is responsible for the biologic activity, since a large number of potentially mutagenic and carcinogenic adducts are formed by alkylating agents in uiuo (1).
One direct approach to evaluating the mutagenicities of alkylnucleotides in vivo is to assay mutations in extrachromosomal DNA containing single, structurally defined adducts (4,(23)(24)(25). W e have recently developed an enzymatic procedure for the site-specific incorporation of purified alkyl-dNTPs into biologically active DNA (25). Coupled with a reversion assay for mutagenesis, this procedure permits a highly sensitive evaluation of the mutagenicity of alkyl-dThd adducts during DNA replication in Escherichia coli. In this report we compare the relative mutagenicities of 04-methyl (Me)'-, 04-ethyl (Et)-, 04-isopropyl (iPr)-, and 02-Me-dThds site-specifically incorporated into 4x174 am3 DNA. Our data show that 04-alkyl-dThds specifically induce A . T + G.C transitions in ada-E. coli with relative mutagenicities of: 04-Me-dThd > 04-Et-dThd > 04-iPr-dThd z 02-Me-dThd = dThd. mutagenesis in uiuo. The general strategy for the preparation of site-specifically alkylated DNA involves the enzymatic synthesis of oligonucleotides containing a single O-alkylthymine residue at a predetermined site (Fig. 1). In the first step, single-stranded 4x174 am3 DNA (+)-strand DNA is hybridized with a synthetic (-)-strand 15-mer, and the partialduplex DNA is extended by a DNA polymerase and a purified 0-alkyl-dTTP to yield a hybridized 16-mer with a single, 3'terminal 0-alkylthymine residue at nucleotide position 587. In the second step, the alkylthymine is "sealed" into the oligonucleotide by further polymerization in the presence of dATP and dCTP yielding a 21-mer with the alkylated residue located 6 nucleotides from the 3' end. The site-specifically alkylated 21-mer can then be extended with unmodified dNTPs to form complete double-stranded DNA (Fig. 1, step   3 ) . Finally, the DNA is transfected into E. coli where it is copied by the cellular DNA polymerase I11 replication complex to yield progeny phage derived from both DNA strands (30).

Enzymatic
This procedure relies on the ability of 0-alkyl-dTTPs to serve as substrates for DNA polymerases during the polymerization of 4x174 am3 DNA in uitro. The incorporated 3'terminal alkylthymines must also support further polymerization to permit complete synthesis of the extended oligonucleotides ( Fig. 1, steps 2 and 3 ) and must permit biologic expression of the nascent strands. Our laboratory previously demonstrated that DNA polymerases can site-specifically incorporate 04-Me-dTTP (25), and recent studies by Singer et al. (8,26) show that DNA polymerases also recognize 04-Etand 04-iPr-dTTPs as well as 02-Me-dTTP (6) as substrates for the polymerization of synthetic polymers and salmon sperm DNA in uitro. These data suggested that, under the appropriate conditions, polymerases might serve as general tools for the site-specific incorporation of 0-alkyl-dTTPs into biologically active DNA.
0-Alkyl-dTTPs as Substrates for DNA Polymerases-To determine the abilities of DNA polymerases to incorporate 0alkyl-dTTPs into partial-duplex DNA, single-strand 4x174 DNA was hybridized with a synthetic 15-mer and incubated with polymerase in the presence of dATP, [cx-~'P]~CTP, dGTP, and either dTTP or a single 0-alkyl-dTTP (Table I) * TABLE I 0-Alkyl-dTTPs as substrates for DNA polymerases Single-stranded 6x174 DNA was hybridized with a synthetic 15mer and incubated at 30 "C for 15 min with AMV pol or pol I in the presence of dATP, [LX-~'P]~CTP, dGTP, and either dTTP or an 0alkyl-dTTP as described under "Materials and Methods." Nucleotide incorporation was calculated from the acid-precipitable radioactivity and is expressed as the total number of nucleotides incorporated per template/unit of polymerase (mean f S.D. of four determinations). All values are corrected for the amount of acid-precipitable radioactivity in control reactions conducted in the absence of polymerase (apparent number of nucleotides incorporated per template = 0.8 f 0.9). The values in parentheses represent the average incorporation relative to unmodified dTTP. In this assay, the levels of total nucleotide incorporation should reflect the relative abilities of the 0-alkyl-dTTPs and unmodified dTTP to serve as substrates. Two polymerases were studied AMV pol, an error-prone polymerase with no detectable 3' + 5' exonuclease; and E. coli pol I, a highly accurate polymerase with a 3' + 5' exonuclease proofreading activity (30,37). Replacement of dTTP with any of the 04alkyl-dTTPs decreased the rate of DNA synthesis catalyzed by either AMV pol or pol I ( Table I). As the size of the alkyl group was increased from methyl to ethyl to isopropyl, there was a decrease in DNA polymerization. Both enzymes also utilized 02-Me-dTTP as a substrate, however, the rates of polymerization were less than those observed in the presence of the 04-Me-dTTP isomer. In all of the reactions, with the exception of those containing dTTP or 04-Me-dTTP, the error-prone polymerase AMV pol yielded greater DNA synthesis per unit of enzyme than pol I.
Effects of 0-Alkyl-dThds on 6x1 74 Suruiual-To determine the effects of 0-alkyl-dThds on the biologic expression of 4x174, 0-alkyl-dTTPs were incorporated into the wild-type strand of a 23-mer/4X174 am3 heteroduplex molecule; the wild-type progeny phage were then titered following transfection and replication of the 0-alkyl-dThd-containing DNAs in ada-E. coli (Table 11). Direct comparison of the wild-type titers showed that all of the 0-alkyl-dThds suppressed expression of the (-)-strand when compared to unmodified dThd. The levels of suppression ranged from 40 to 60% and were less than that of dUrd (80%) which is known to be lethal in ung+ E. coli (38). In this assay, the expression of the wildtype (-)-strand should be a function of both the lethality of each adduct and the number of adducts incorporated. When the wild-type titers are normalized for the extent of adduct incorporation in the (-)-strands, the relative survivals were: (Table 11).
Site-specific Incorporation of 0-Alkyl-dTTPs onto the 3'-Termini of Synthetic Oligonucleotides-The data in Tables I  and I1 demonstrate that 0-alkyl-dTTPs can replace dTTP during 6x174 DNA polymerization in uitro to produce biologically active molecules. To determine the abilities of DNA polymerases to site-specifically incorporate 0-alkyl-dTTPs, [5'-32P]15-mer/4X174 am3 partial-duplex DNA was incubated with polymerase in the presence of single dTTP analogues ( Fig. 1, step l ), and the extended oligonucleotides were analyzed by polyacrylamide gel electrophoresis (PAGE). A

TABLE I1
Effects of 0-alkyl-dThds on 6x1 74 survival dThd analogues were incorporated into the wild-type (-)-strand of wild-type 23-mer/&X174 am 3 heteroduplex DNA by incubation with AMV pol in the presence of dATP, [cu-"P]dCTP, dGTP, and a single dTTP analogue. The number of dThd analogues incorporated per (-)-strand was estimated from the levels of [a-"P]dCTP incorporation, and the relative biologic expression of the analogue-containing (-)-strands was determined from the titers of the progeny phage on am Sup' and Sup-indicator bacteria following transfection and replication in ada-E. coli ( I % Expression, wild-type titer/total titer X 100. The values separated by commas represent the results of duplicate titers, and the values in parentheses represent the average expression relative to the dThd-containing control. The average total titer from the dThdcontaining DNA was 1.6 X 10' plaque-forming units/80 ng of transfected DNA.
*Average of duplicate determinations with a range 518%. The values in parentheses represent the number of analogues relative to unmodified dThd.
The values are expressed relative to dThd and are corrected for the content of the dThd analogues in the 6x174 (-)-strand DNA as follows: relative survival = (relative (-)-strand expression) X (relative number of analogues per (-)-strand). comparison of four different polymerases showed that the yields of extended oligonucleotides were dependent both on the source of the polymerase and on the structure of the alkyl adduct (Figs. 2 and 3).
In polymerizations with the error-prone polymerases AMV pol ( Fig. 2 A ) or MLV pol (Fig. 2B, left), the "correct" nucleotide d T T P efficiently extended the oligonucleotide to give about 25% or 90% yields of X-mer, respectively. Both enzymes also produced 17-mers and 18-mers, presumably via the misincorporation of dTMP opposite dThd and dGuo residues on the template strand. The ability of AMV pol to insert mismatches during polymerization (37) was demonstrated by partial extension (about 25% yield) of the 15-mer in the presence of the "incorrect" nucleotide dATP (Fig. 2 A ) . Like dTTP, the 04-alkyl-dTTPs and 02-Me-dTTP supported oligonucleotide extension to produce 16-, 17-, and 18-mers in relatively high yields. The total yields of extended oligonucleotides were greatest with MLV pol where virtually none of the original 15-mers remained unextended. Under these conditions, the size of the alkyl group did not appear to affect the yields of extended oligonucleotides.

A. LARGE FRAGMENT POL I
In contrast to the error-prone RNA-dependent DNA polymerases, extension of the oligonucleotides by the highly accurate polymerase pol I was limited to the insertion of single nucleotides opposite the template dAdo residue to yield exclusively 16-mers (Fig. 3A, left, and 3B). Under these conditions, all of the dTTP analogues produced 16-mers in nearly quantitative yields. Similar results were observed with either the large fragment of pol I (Fig. 3A, left) or intact pol I (Fig.   3 B ) . However, the absolute yields with intact pol I appeared less than with the large enzyme fragment; this was presumably due to removal of the 5'-"P label by the 5' + 3' exonuclease associated with the intact enzyme (30).
Evidence for the Presence of 0-Alkyl-dThds in the Extended Oligonucleotides-Examination of the electrophoresis patterns from the reactions containing 04-alkyl-dTTPs and either AMV pol (Fig. 2 A ) or MLV pol (Fig. 2B, left) shows that the mobilities of the extended oligonucleotides are altered in comparison with oligonucleotides extended by unmodified dTTP. As the size of the alkyl group increased from methyl to ethyl to isopropyl, there was a concomitant decrease in the electrophoretic mobilities of the oligonucleotides extended by more than 1 nucleotide. This was most clearly seen in the 17and 18-mers where the incorporated 04-alkyl-dTMPs would represent 12% and 17%, respectively, of the total nucleotides in the extended polymers. No changes in mobility were observed in the 16-mer products where the newly incorporated nucleotides would comprise only 6% of the total oligonucleotide length. The products from reactions with 02-Me-dTTP were unchanged in their electrophoretic mobilities when compared to oligonucleotides extended with dTTP. Thus, the

Mutagenesis by O-Alkylthymines
relative mobilities of the oligonucleotide products were dependent on both the size and isomeric position of the alkyl group on dTTP.
To directly assay for the presence of incorporated 04-alkyl-dTMPs, the extended oligonucleotides prepared from AMV pol (Fig. 2 A ) or intact pol I (Fig. 3B) were subjected to nearest-neighbor analyses (39) following further polymerization in the presence of [(u-~'P]~ATP. This assay will permit the detection of 3"terminal O-alkyl-dTMPs in the 16-mer and 18-mer products where the next correct base on the nascent strand is an adenine (Fig. 1). The resultant [3'-32P] dNMPs were analyzed by thin layer chromatography and compared to O-alkyl-dTMP standards.
The relatively large amounts of [3'-32P]dTMP detected in these analyses most likely arose via hydrolysis of the 04-alkyl-dThd residues during the incubations with excess nuclease and phosphodiesterase, since PAGE analyses of the 04-alkyl-dThd-containing oligonucleotides prior to nearest-neighbor analysis revealed only faint bands ( 4 0 % of the total) that co-migrated with unmodified oligonucleotides (Fig. 2 A ) . These trace amounts of unmodified oligonucleotides were not observed in all 04-alkyl-dThd-oligonucleotide preparations (compare Fig. 2, A and B, left) and presumably arose by slow chemical dealkylation of the 04-alkyl-dThd-oligonucleotides or 04-alkyl-dTTP substrates during storage (40).
Extension of Oligonucleotides Containing 3 "Terminal 0-Alkyl-dTMPs-The nearest-neighbor analyses of the oligonucleotides extended with 04-Me-or 04-Et-dTTP and the PAGE analyses of the oligonucleotides extended with 04-Me-, 04-Et-, 04-iPr-, or 02-Me-dTTP provide strong evidence that the product oligonucleotides contained the predicted adducts at the 3"termini. To produce oligonucleotides with aIkyl adducts on internal nucleotides, it was necessary to further extend the 3"alkylated oligonucleotides in a second step of polymerization with unmodified dNTPs (Fig. 1, step  2). In the presence of dATP and dCTP (i.e. the next 5 correct nucleotides), the oligonucleotides prepared from O-alkyl-dTTPs and either MLV pol or the large fragment of pol I were extended to yield primarily the predicted 21-mers (Figs. 2B, center and 3A, right). The 16-mers and 17-mers derived from 02-Me-, 04-Me-, or 04-Et-dTTPs and the 16-mers from 04-iPr-dTTP were extended nearly quantitatively. In contrast, the 17-mers derived from 04-iPr-dTTP were poor substrates for further oligonucleotide extension (Fig. 2B, center); approximately 50% of the 04-iPr-dTTP-derived 17-mers still remained unextended after incubation for 60 min in the presence of dATP and dCTP (data not shown). Both polym-erases also formed varying amounts of 25-, 26-, and 28-mers, presumably due to the presence of trace amounts of dGTP in the commercial polymerase preparation^.^ Limited digestion of the extended alkyl oligonucleotides from Fig. 2B (center) with the 3' "-* 5' exonuclease of T4 DNA polymerase yielded a series of products ranging from 15 to 28 nucleotides in length (Fig. 2B, right). The stable presence of the 04-alkyl nucleotides was evidenced by the reappearance of the step-like electrophoretic pattern corresponding to the slowed mobilities of the alkylthymine-containing 17-mers (Fig. 2B, left).
Mutagenicity of O-Alkylthymines in E. coli-To compare the relative mutagenicities of the O-alkylthymines in uiuo, the site-specifically alkylated 16-mer/dX174 am3 DNAs prepared with pol I (Fig. 3B) were further extended in the presence of unmodified dNTPs and transfected into and allowed to replicate in E. coli spheroplasts. The mutagenicities of the alkylthymines were determined from the yields of wild-type progeny phage growing on Sup-indicator bacteria as previously described (25,28). In this amber codon reversion assay, alkylthymines that base pair with dGuo, dCyd, or dThd residues in vivo will revert the amber codon to yield phage with a wildtype phenotype (41). Therefore, all possible mutagenic base pairings at nucleotide position 587 are detected with high sensitivity.
None of the O-alkyl-dThd-containing DNAs produced mutations on transfection into wild-type (ada+) E. coli (data not shown). However, in ada-E. coli the 04-Me-dThd-and 04-Et-dThd-containing DNAs induced about 10-and 4-fold increases in reversion frequency, respectively, compared to DNA prepared from unmodified dTTP (Table 111). The 04-iPr-dThd-containing DNA induced only a slight increase in reversion frequency (2.5-to 3-fold) in two of the three experiments, and DNA prepared by site-specific polymerization in the absence of dNTPs or in the presence of dCTP, dUTP, or 02-Me-dTTP failed to induce mutations above background.
Nucleotide Sequence Changes Induced by O-Alkylthymines-DNA sequence analysis of 20 independent revertant phage derived from the dThd-containing DNA (Table 111, Experiment 1) showed that all of these "spontaneous" mutants contained dGuo residues at nucleotide position 587 of the (+)-strand. Similarly, 20 out of 20 revertants induced by either 04-Me-dThd or 04-Et-dThd contained dGuo residues opposite the original site of the O-alkyl-dThd. In these experiments (Table 111, Experiment l), the incorporation of 04-Me-dTTP or 04-Et-dTTP resulted in a 25-or 6-fold increase in mutation frequency, respectively, when compared to the incorporation of unmodified dTTP. Thus, in a sampling of independently arising mutant plaques, 24 out of 25 and 5 out of 6 plaques should be derived from the 04-Me-dThd-and 04-Et-dThd-containing DNAs, respectively. In control experiments where dATP was forced opposite the dAdo residue at nucleotide 587 by AMV pol (Fig. 2 A ) , the proportion of induced mutations (dThd residues at position 587 of the (+)strand) and spontaneous mutations (dGuo residues at position 587) corresponded with the increase in mutation frequency (data not shown).  Mutagenicity of 0-alkylthymines in E. coli 6x174 am3 DNA was hybridized with a (-)-strand 15-mer and site-specifically polymerized with intact pol I in the presence of a single dNTP (Fig. 1, step 1; Fig. 3B). The product partial-duplex molecules were further extended with unmodified dNTPs (Fig. 1, step 3 ) and allowed to replicate in vivo following transfection into GW5352 (&a-) E. coli spheroplasts. Progeny phage were titered on am Sup+ and Sup-indicator bacteria, and the reversion frequencies were calculated as the ratio of Sup-:Sup+ titers. See "Materials and Methods" for details. The average Sup-and Sup' titers from the DNA site-specifically polymerized with dTTP were 8.2 X 10' and 4.5 X 10' progeny phage/100 ng of transfected DNA, respectively. The data in the table represent the results of three separate experiments from three different preparations of each site-specifically polymerized oligonucleotide. All DNA sequences were determined from mutants produced in Experiment 1. The values in parentheses represent the number of independent revertant phage that were sequenced. ND, not determined.
for incorporation opposite a dAdo residue in 4x174 am3 DNA in uitro. Investigations on the polymerization of synthetic polymers i n uitro show that 0-alkyl-dTTPs can substitute for dTTP and incorporate opposite template adenines (6,26). Our data show that polymerases in high concentrations can also copy long stretches of natural, heteropolymeric DNA where 0-alkyl-dTTPs are the only source of thymidine nucleotide substrate (Table I). As predicted from simple steric factors, the rates of polymerization decreased as the size of the alkyl group increased. However, the size of the alkyl group alone does not determine substrate usage, since 02-Me-dTTP was poorer than its 04-Me-dTTP isomer and approximately the same as 04-Et-dTTP in supportingpolymerization. PAGE analysis of oligonucleotides extended by single dTTP analogues (Figs. 2 and 3 ) directly demonstrated that polymerases can site-specifically incorporate 02-alkyl-and 04-alkyl-dTTPs opposite dAdo residues in natural DNA. With the error-prone polymerases, AMV pol and MLV pol, which lack 3' + 5' exonucleolytic activity (37), extension by 04-alkyl-dTTPs occurred beyond a single nucleotide to yield oligonucleotides with 3"terminal adducts opposite dThd and dGuo as well as dAdo residues in the template strand. In contrast, extension by 02-Me-dTTP or unmodified d T T P with MLV pol did not proceed significantly beyond the first dAdo residue. These data show that, in addition to forming base pairs with dAdo and dGuo residues (6-lo), 04-alkyl-dThds can be forced opposite dThds. With the highly accurate polymerase pol I, however, only single 02-alkyl-and 04-alkyl-dThds were added opposite the template adenine. This suggests that pol I recognizes 0-alkyl-dThds opposite dThds as noncomplementary and avoids their formation either by stringent nucleotide selection during polymerization or by subsequent 3' + 5' exonucleolytic removal (37).
Characterization of the site-specifically extended oligonucleotides by PAGE (Figs. 2 and 3) and by nearest-neighbor analysis provided direct evidence that the 0-alkyl-dThd adducts were incorporated at the desired nucleotide position. Moreover, 3' .--, 5' exonuclease digestion of the alkyl oligonucleotides showed that the incorporated 04-alkyl-dThds are relatively stable during the procedures of enzymatic synthesis (Fig. 2B, right). The PAGE and nearest-neighbor data also demonstrate that oligonucleotides with 3'-terminal adducts can be further extended by DNA polymerases in the presence of unmodified dNTPs. The lack of extension of oligonucleotides containing multiple 04-iPr-dThds (Fig. 2B, center) probably reflects a steric hindrance imparted by the relatively large branched isopropyl group. Apparently this is not significant with the smaller alkyl groups or with oligonucleotides containing single isopropyl-dThds.
Transfection of the site-specifically alkylated DNAs (prepared with intact pol I) into E. coli and analysis of the progeny phage permitted a comparison of the relative mutagenicities of the 0-alkyl-dThds during DNA replication in uiuo. In &a-E. coli, which are defective in the alkyltransferase capable of repairing 04-alkylthymine-DNA adducts (29, 42-44), the apparent mutagenicities were: 04-Me-dThd > 04-Et-dThd > 04-iPr-dThd 2 02-Me-dThd = dThd. The absence of 04alkyl-dThd-induced mutations in ada+ E. coli provides strong evidence that these mutations were produced specifically by the alkyl adducts and shows that the ada gene product plays a key role in protecting against mutagenesis by 04-Me-and 04-Et-dThds in uiuo. The absence of 02-Me-dThd-induced mutations in either ada+ or &a-E. coli may reflect the inherent low mutagenic potential of 02-Me-dThd (6,9) or, more likely, the repair of this adduct by the E. coli alkA gene product (43, 44). DNA sequence analyses of the 04-Me-and 04-Et-dThd-induced mutants showed that all of the mutant phage contained dGuos in place of the dAdo residues originally present opposite the adducts. These data are consistent with a mechanism of mutagenesis involving the formation of 04alkyl-T. G base pairs during DNA replication in the E. coli and suggest that the production of A . T .--, G.C transition mutations is characteristic of mutagenesis by 04-alkyl-dThds both in uivo (this report; Ref. 25) and in vitro (6-10).
Comparison of the mutation frequencies induced by the 04alkyl-dThds in ada-E. coli suggests that the mutagenicities of these adducts decreases as the size of the alkyl group increases (Table 111). Since the yields (Fig. 3B) and lethalities ( Table 11) of all of the O4-a1kyl-dThd-containing DNAs were very similar, it is likely that the lower mutagenicities of the larger alkyl adducts is due to either a difference in their mispairing efficiencies during DNA replication or a difference in the repair of these adducts in E. coli. Studies comparing the mutagenicities of N-alkyl-N-nitrosoureas in E. coli show that as the size of the alkyl group increases, the protective effects of the ada repair system are reduced while those of the uur excision repair system are increased (23, 45). The preferential repair of small alkyl groups by the adu gene product (an 04-alkylthymine-and 06-alkylguanine-DNA alkyltransferase; Refs. 43 and 44) has been confirmed in vitro (46-48), and recent studies on the repair of 04-alkylthymines and OQalkylguanines in repair-deficient E. coli provide direct evidence that both the ada and the uur gene products contribute to the repair of these adducts i n Since the uur repair system characteristically recognizes "bulky" lesions (49), it is likely that the larger ethyl and isopropyl adducts are preferred substrates relative to methyl adducts. Thus, the lower mutagenicities of the ethyl and isopropyl derivatives in our assay may be due to partial repair of these adducts by uur nucleases.
Our procedures for the enzymatic synthesis of site-specifically alkylated oligonucleotides offers several advantages over procedures involving the chemical synthesis of oligonucleotides and could provide a rapid assay for screening the mutagenicities of a variety of nucleotide adducts in vivo. In principle, our technique is limited by the abilities of polymerases to accept dNTP analogues as substrates for DNA polymerization ( Fig. 1, step 1 ) and to extend the product oligonucleotides containing 3"terminal analogues (Fig. 1, steps 2 and 3 ) . However, the mild conditions of enzymatic oligonucleotide synthesis (30-37 "C, pH 7-8) are well-suited for mutagenesis studies on labile adducts such as the 04-alkyl-dThds that are unstable (40) at the extreme pH values and temperatures frequently employed in the chemical synthesis of oligonucleotides (50). Moreover, the mild conditions inherent to polymerase reactions should minimize the formation of unwanted (and often mutagenic) side products that may arise during oligonucleotide synthesis by chemical methods (51-56). Our data illustrate the applicability of this procedure for studying the mutagenicities of 0-alkyl-dThd adducts and suggest that other analogues might be similarly studied. Thus, by conducting sequential polymerizations in the presence of select dNTP adducts, it should now be possible to rapidly and simply prepare small amounts of oligonucleotides of virtually any length containing single or multiple alkyl adducts at predetermined sites.