DNA Damage at Thymine N-3 Abolishes Base-pairing Capacity during DNA Synthesis*

3-Methylthymine was synthesized into DNA copoly- mers and deoxynucleoside triphosphate to study its effect on DNA synthesis by the Klenow fragment of Escherichia coli polymerase I and avian myeloblastosis virus reverse transcriptase. Both polymerases were greatly inhibited by template 3-methylthymine. In re- sponse to 3-methylthymine, misincorporation of dTTP increased slightly, but occurred only at low levels con- sistent with spontaneous misincorporation in vitro. Surprisingly, template 3-methylthymine resulted in a striking decrease in background misincorporation, rel- ative to normal incorporation by the Klenow fragment, of dGTP and, to a lesser extent, of dATP and dCTP. The incorporation of 3-methyl-dTTP into DNA was studied using DNA sequencing technology. The Klenow fragment failed to incorporate 3-methyl-dTTP even at 1 mm Reverse transcriptase incorporated 3-methyl- dTTP opposite adenine, cytosine, and thymine, but at only about V40,OOOth the efficiency of complementary deoxynucleoside triphosphate incorporation. Further-more, synthesis generally stalled at sites of 3-methyl- thymine incorporation. From these results, we con-clude that damage at the central hydrogen-bonding position of thymine abolishes its base-pairing capabil-ities during DNA synthesis. of the difference in abilities of 3-methyl-dTTP and dTTP to be incorporated opposite adenine by the Klenow fragment. The highest concentration (1 mM) of 3-methyl-dTTP tested showed no sign of incorporation, whereas dTTP at 250 nM gave significant replication (not shown). The results demonstrate, under our conditions, at least a 4000-fold preference for dTTP over 3-methyl-dTTP by the Klenow fragment.

publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. **Scholar of the Leukemia Society of America, Inc. erase suggest that 3-methyluracil behaves ambiguously during transcription and base-pairs with all four normal dNTPs' (10). Base pairing during transcription, however, is not a good model for base pairing during DNA synthesis (11). We have synthesized 3-methylthymidine 5'-triphosphate to high purity and used it to study the base-pairing properties of 3-methyldeoxythymine both in synthetic copolymer templates and as precursor dNTP. We report here that 3-methyldeoxythymine in template inhibited DNA replication by the Klenow fragment of E. coli DNA polymerase I and the error prone (12) avian myeloblastosis virus (AMV) reverse transcriptase. This inhibition resulted from the failure of both polymerases to incorporate efficiently any unmodified dNTP in response to template 3-methyl-dTMP. Template 3-methyldeoxythymine slightly increased misincorporation of dTTP relative to dNMP incorporated by the Klenow fragment and reverse transcriptase. However, these increased levels were consistent with levels for spontaneous misincorporation of normal dNTP during DNA replication in uitro. Surprisingly, 3-methyldeoxythymine caused a significant decrease in background dNTP misincorporation relative to correct dNTP incorporation by the Klenow fragment. This apparent increase in fidelity is discussed with respect to nonexonucleolytic proofreading mechanisms.
Consistent with these results, the Klenow fragment failed to incorporate 3-methyl-dTTP opposite any template base, and AMV reverse transcriptase incorporated 3-methyl-dTTP inefficiently (about l/40,000th the efficiency of normal dNTP incorporation) opposite only adenine, cytosine, and thymine. In addition, DNA synthesis generally stalled at sites of 3methyl-dTTP incorporation. Therefore, methylation at the central hydrogen-bonding site of thymine not only abolishes the ability of 3-methyldeoxythymine to direct dNTP incorporation, but also greatly reduces, if not eliminates, the ability of 3-methyl-dTTP to serve as a dNTP substrate during DNA replication in vitro.

Nucleic Acids and Enzymes-Terminal deoxynucleotidyltransfer-
ase, poly(dC) CS;&,), ultrapure-grade dNTPs, and deoxyribonucleotide primers were purchased from Pharmacia P-L Biochemicals. Tritium-labeled dNTPs were purchases from ICN, and 32P-labeled and 35S-labeled dNTPs were purchased from Du Pont-New England Nuclear. The Klenow fragment was obtained from Bethesda Research Laboratories. AMV reverse transcriptase and Penicillium citrinum nuclease P1 were purchased from Boehringer Mannheim. E. coli alkaline phosphatase, snake venom 5'-nucleotidase, and yeast inorganic pyrophosphatase were obtained from Sigma.

12843
This is an Open Access article under the CC BY license.
Nucleotide Purification-All dNTPs, labeled and unlabeled, were purified by preparative HPLC using a Perkin-Elmer Series 4 liquid chromatograph, LC85B spectrophotometric detector, and data analysis software (Dynamic Solutions, Inc.). dNTPs were chromatographed on a Whatman SAX-M/9-25 anion-exchange column (9.4 mm X 25 cm) with 0.25 M potassium phosphate buffer (pH 5.95). Purified dNTPs were desalted by binding to DEAE-Sephadex equilibrated in 50 mM triethylammonium bicarbonate buffer (TEB; pH 7.8), followed by elution with a gradient of 0.2-1.0 M TEB. TEB was removed by repeated evaporations from methanol in vacuo. The purified dNTPs were stored at -70 "C and pH 7.4 in 10 mM Tris-HCl buffer.
All purified dNTPs were checked for contaminants by analytical HPLC (Whatman Partisil 10 SAX anion-exchange column (4.6 mm X 25 cm), with 0.25 M potassium phosphate buffer (pH 5.95)). Unlabeled dNTPs (0.5 pmol) were examined spectrally. The limits of spectral detection were 0.25 nmol for pyrimidine dNTPs and 0.125 nmol for purine dNTPs. The limit of detection for 3H-labeled contaminants was 7.8 X Ci, about twice the average background value. Each unlabeled dNTP was >99.95% free of contaminatingpyrimidine dNTP and >99.98% free of contaminating purine dNTP. Each 3Hlabeled dNTP was ~99.97% free of 3H-labeled contaminants.
Synthesis of 3-Methyl-dTTP-3-Methylthymidine was prepared essentially according to Farmer et al. (17), with the exception that diazomethane was generated from Diazald (N-methyl-N-nitroso-ptoluenesulfonamide, Aldrich). Following determination of purity (>95%) by reversed-phase HPLC (Varian analytical C18 column (4 mm X 30 cm), using a gradient of 20-35% methanol:H,O), the nucleoside's identity was confirmed by comparison of its UV (Amax = 267 nm at pH 7.4) and NMR spectra and HPLC elution profile with those of an authentic standard (Sigma) and with available literature values (17)(18)(19). Methylation was shown by reversed-phase HPLC to be stable to incubation at 37 "C for 18 h at pH 2.0, 7.4, and 12.0.
3-Methyl-dTMP was prepared from 3-methyldeoxythymine by modification of the phosphorus oxychloride method (20) and then reacted with tributylammonium pyrophosphate (21) to produce 3methyl-dTTP as reported elsewhere for preparation of 06-methyl-dGTP (22). 3-Methyl-dTTP (yield, 40%) was purified by DEAE-Sephadex chromatography, with elution by a gradient of 0.05-0.7 M TEB (pH 7.8), followed by preparative HPLC on the Whatman SAX-M/9-25 column. The final product was identified by anion-exchange HPLC and gave 3-methyldeoxythymine on digestion with bacterial alkaline phosphatase. 3-Methyl-dTTP was >99.95% free from contamination by the four common dNTPs (by analytical HPLC), but contained, at a very low level (el%), an unidentified by-product of the second phosphorylation reaction.
The level of unmodified dNTPs contaminating the 3-methyl-dTTP was also determined enzymatically: no incorporation of 3-methyl-dTTP (1 mM) was detected in minus reactions using the Klenow fragment, whereas 250 nM dTTP supported significant synthesis (not shown). Thus, at least for dTTP, contamination of 3-methyl-dTTP was <0.025%. Moreover, incorporation of dTTP opposite adenine resulted in uninterrupted synthesis to produce high molecular weight DNA. In contrast, stalling was consistent with incorporation of the helix-disrupting 3-methyl-dTTP, but inconsistent with incorporation of contaminating normal precursors.
Polymer Synthesis-Deoxyribonucleotide polymers were produced using terminal deoxynucleotidyltransferase and purified dNTPs essentially as described elsewhere (23). Approximate yields of polymers were determined from their UV absorbance at pH 7.4. Extinction coefficients were corrected for the presence of modified base. Molar composition of each polymer was determined by reversed-phase HPLC of the free nucleosides resulting from complete digestion of polymers with nuclease P1 (pH 5.3) and bacterial alkaline phosphatase (pH 8.  transcriptase (as indicated in the figure legends) and incubation at In all assays, "complementary" dNTP is complementary to the unmodified template residue. All other dNTPs are "noncomplementary." Incorporation rate is the ratio of noncomplementary to total dNTP incorporated. Turnover rate is the ratio of noncomplementary dNTP turned over to the total of complementary dNTP incorporated plus noncomplementary dNTP turned over. Incorporation and turnover assays were performed as described elsewhere (26).
For minus reactions catalyzed by the Klenow fragment, lyophilized DNA was resuspended in 32-36 p1 of 0.1 M Tris-HCI buffer (pH 7.4 at 0 "C), 50 mM MgCl,, and 50 mM P-mercaptoethanol. Five-pl reactions contained 2 pl of resuspended DNA, 2 pl of the appropriate minus mixture (25 p~ concentration of each of three dNTPs plus 3methyl-dTTP at various concentrations), and 0.8 unit of the Klenow fragment in 1 pl of dilution buffer. Reactions were incubated at 0 "C for 30 min and then stopped by addition of 5 p1 of stop mixture (0.025% bromphenol blue, 0.025% xylene cyanol, and 10 mM EDTA in formamide). All stopped reaction mixtures were heated to 100 "C, cooled, and then electrophoresed as described (22). After electrophoresis, the gel was soaked for 15 min in 10% acetic acid, 10% methanol, dried at 80 'C under vacuum onto Whatman No. 3MM chromatography paper, and autoradiographed.

RESULTS
Polydeoxyribonucleotide Templates-The chemical nature of 3-methylthymine suggested that it would be relatively free of spontaneous chemical rearrangements. Therefore, we synthesized 3-methyldeoxythymine and confirmed its stability by reversed-phase HPLC after incubation at 37 "C for 18 h at pH 2.0, 7.4, and 12.0 (not shown). Polymers were synthesized using terminal deoxynucleotidyltransferase, 3-methyl-dTTP, and purified dNTPs (Table I). It is interesting to note that terminal deoxynucleotidyltransferase was biased for or against 3-methyl-dTTP depending on the unmodified dNTP in the reaction (Table I).
Inhibition of Replication by 3-Methyldeoxythymine in Template-3-Methylthymine in poly(dT,3-mdT) (Fig. 1) and in poly(dA,3-mdT) and poly(dC,3-mdT) (Fig. 2) copolymer templates inhibited DNA synthesis by the Klenow fragment. If 3-methyldeoxythymine acts as a strong block to strand elon-   gation independent of neighboring sequence, one might expect the percentage inhibition of DNA synthesis on each copolymer to reflect only the mole fraction of 3-methyldeoxythymine. Fig. 2 suggests that this is the situation and t h a t maximum inhibition of synthesis is reached at 17.3% 3methyldeoxythymine. Replication by AMV reverse transcriptase also was inhibited by 3-methyldeoxythymine in template Template-primer complexes were annealed at 4 "C under concentration and buffer conditions identical to those used in all assays. The UV absorbance was monitored as the temperature was increased at 1 "C/min as described under "Materials and Methods." Neither of the two single-strand polymers nor oligo(dA)8 exhibited increased absorbance over the temperature range tested. (Fig. 2), b u t t o a lesser extent than replication by the Klenow fragment.
To be sure that 3-methyldeoxythymine inhibition was not the result of reduced annealing of primer with a 3-methyldeoxythymine-containing copolymer, DNA melting curves of primer-template were measured (Fig. 3), and DNA synthesis product lengths were monitored. The melting curves measured and hypochromicity values calculated (15.7% for the annealed homopolymer-primer uersus 14.2% for the annealed copolymer-primer) demonstrated that both complexes were maximally annealed at 4 "C. The lower T,,, and broad melting range of annealed copolymer-primer relative to annealed homopolymer-primer are expected for a helix with random unpaired bases (29). In addition, electrophoresis of the polymerization reaction products (on a DNA sequencing gel after incorporation of radiolabeled nucleotide) revealed all high molecular weight product in the absence of 3-methyldeoxythymine, but primarily low molecular weight product in the presence of 3methyldeoxythymine (not shown).

3-Methyldeoxythymine-directed
Incorporation by the Klenow Fragment-To determine if inhibition of replication by 3-methyldeoxythymine in template resulted from lack of incorporation opposite 3-methyldeoxythymine, incorporation of dATP and dTTP was studied during replication of poly(dC, 3-mdT). Similarly, incorporation of dCTP and dGTP was studied during replication of either poly(dA,3-mdT) or poly(dT,3-mdT). Incorporation of dNTPs complementary to the primer dNTP could not be determined due to possible nucleotide addition to the template terminus at overlapping ends.

3-Methyldeoxythymine-directed
Incorporation by AMV Reverse Transcriptase-Since the Klenow fragment apparently only incorporated a small amount of d T T P opposite 3-methyldeoxythymine, we tested the error-prone AMV reverse transcriptase. Incorporation of noncomplementary dNTP was tested opposite poly(dC) and poly(dC,24.0% 3-mdT) since only poly(dC) is used efficiently as template by AMV reverse transcriptase (30). The nucleotide pool was biased 20-fold in favor of the noncomplementary dNTP; and, as with the Klenow fragment, only stimulation of dTTP incorporation (10-fold) was detected (Table 111). A small increase in dCTP incorporation was noted, but it was within experimental error.    Turnover of Noncomplementary d N T P by the Klenow Fragment-Noncomplementary dNTPs could have been incorporated by the Klenow fragment, but then removed as dNMPs (turnover) by the polymerase's proofreading 3'+5' exonuclease. Therefore, turnover was measured concomitant with incorporation. Only dGTP opposite poly(dA) gave measurable levels of turnover; 3.1 & 0.02 bases turned over per 1000 bases incorporated (not shown). This result agrees with results from incorporation studies (Table 11), in which the highest level of misincorporation was seen with dGTP during replication of poly(dA). Turnover of dGTP was decreased to undetectable levels by 3-methyldeoxythymine in template, which suggests that the dramatic decrease in dGTP misincorporation in response to template 3-methyldeoxythymine resulted from selectivity in the incorporation step rather than in an exonucleolytic proofreading step.
Incorporation and Turnover of Complementary d N T P by the Klenow Fragment-The Klenow fragment did not turn over any noncomplementary dNTP in response to 3-methyldeoxythymine in template. However, "idling" of polymerase at a 3-methyldeoxythymine block to elongation would result in increased turnover of the complementary dNTP. Therefore, turnover and incorporation of complementary dNTP were determined during replication of copolymers containing 3-methyldeoxythymine (Table IV).
Turnover of each complementary dNTP increased in response to 3-methyldeoxythymine in template (Table IV), whereas incorporation of each complementary dNTP decreased, in agreement with results in Figs. l and 2. Turnover of dATP and dGTP was low relative to that of dTTP, which probably reflects the relative stacking interactions on the nascent strand at the growing point (26), coupled with weaker A . T base pairing. The increase of complementary dNTP turnover in response to 3-methyldeoxythymine is further evidence that 3-methyldeoxythymine is affecting the polymerase rather than primer-template annealing.
Incorporation of Precursor 3-Methyl-dTTP by the Klenow Fragment-Besides the base-pairing capacity of template 3methyldeoxythymine, the base-pairing capacity of precursor 3-methyl-dTTP was also determined during DNA synthesis. DNA sequencing technology enables characterization of modified precursor incorporation by DNA polymerases over a wide range of sequence (28). Using this technology, we determined that 3-methyl-dTTP (0.25-1 mM) was not incorporated   A minimum estimate was made of the difference in abilities of 3-methyl-dTTP and dTTP to be incorporated opposite adenine by the Klenow fragment. The highest concentration (1 mM) of 3-methyl-dTTP tested showed no sign of incorporation, whereas dTTP at 250 nM gave significant replication (not shown). The results demonstrate, under our conditions, at least a 4000-fold preference for dTTP over 3-methyl-dTTP by the Klenow fragment. Incorporation of 3-Methyl-dTTP by AMV Reverse Transcriptase-Similar to the lack of 3-methyl-dTTP incorporation by the Klenow fragment, AMV reverse transcriptase incorporated 3-methyl-dTTP, but only inefficiently even at 1 mM. Incorporation occurred opposite adenine, cytosine, and thymine template residues, but not opposite guanosine residues. Examples of 3-methyl-dTTP incorporation, demonstrated by a decrease in band intensity relative to control, can be seen in minus adenine reactions at positions 3669 and 3698 (Fig. 4a), in minus guanosine reactions at positions 3672 and 3700 (Fig. 4b), and in minus thymine reactions at positions 3664 and 3674 (Fig. 4c), but not in minus cytosine reactions (Fig. 4d).
To determine the relative abilities of 3-methyl-dTTP and dTTP to be incorporated opposite adenine by AMV reverse transcriptase, concentrations of the two dNTPs required for similar extension of DNA strands past blocked thymine sites were compared. One mM 3-methyl-dTTP (Fig. 4c) was required to produce effects similar to those seen with 25 nM dTTP (Fig. 5). This suggested that the AMV reverse transcriptase incorporation rate for 3-methyl-dTTP was about 1/40,00Oth that of dTTP under our conditions.
3-Methyl-dTTP Incorporation Stalls DNA Synthesis-3-Methyl-dTTP was incorporated by AMV reverse transcriptase in place of three of the four normal dNTPs, although with low efficiency. However, strand elongation frequently did not continue beyond the incorporation site, as seen in autoradiograms where intensity lost from one band due to 3methyl-dTTP incorporation accumulated in the following band (Fig. 4). 3-Methyl-dTTP incorporated primarily either at single sites (i.e. at nonrepetitive sequences) or at the first position in a repeat of identical bases. AMV reverse transcriptase failed to incorporate past doublet AA at positions 3676-3677 (Fig. 4a), GG at 3662-3663, and 3678-3679 (Fig. 4b), TT at 3664-3665, 3684-3685, and 3701-3702 (Fig. 4c). At two adenine doublets (positions 3669-3670 and 3707-3708, Fig. 4a), 3-methyl-dTTP incorporated sequentially, although at low levels. Lack of incorporation may be due to steric effects or the instability of two sequential base pairs containing 3methyl-dTTP. Alternatively, stalling after the incorporation of a single 3-methyl-dTTP may occur regardless of the dNTP that is to be incorporated next since stalling occurred at almost all single sites in which 3-methyl-dTTP was incorporated.

DISCUSSION
DNA replication in vitro was used to study the response of the Klenow fragment of E. coli DNA polymerase I and AMV reverse transcriptase to deoxyribonucleotide templates containing 3-methyldeoxythymine. DNA replication by both polymerases was inhibited by these templates. For the Klenow fragment, the extent of inhibition appeared to be dependent upon the amount of template 3-methyldeoxythymine, but not upon the identity of the unmodified template residue. This relationship was not examined for AMV reverse transcriptase. Replication by these two polymerases is also inhibited by noncoding apurinic sites, with replication by AMV reverse transcriptase less inhibited than replication by the Klenow fragment (31).
Only dTTP incorporation was detected in response to template 3-methyldeoxythymine. To evaluate the significance of this incorporation, the amount of incorporation possible opposite 3-methyldeoxythymine encountered by the Klenow fragment and AMV reverse transcriptase while replicating poly(dC,24.0% 3-mdT) was determined. 3-Methyldeoxythymine was randomly distributed in the copolymers (32), so the number of 3-methyldeoxythymine residues encountered by polymerase given the percent of DNA replicated was calculated using a hypergeometric distribution (33). The result was 2,500 3-methyldeoxythymine residues encountered by AMV polymerase and 2,537 3-methyldeoxythymine residues encountered by the Klenow fragment per 10,000 bases replicated. These expected incorporation rates exceeded the observed dTTP incorporation rates opposite 3-methyldeoxythymine using AMV reverse transcriptase (Table 111) by 2,155fold and using the Klenow fragment (Table 11)  Thus, the overall incorporation of dNTP in response to 3methyldeoxythymine in template was similar to that usually reported for background polymerase infidelity in vitro; about one misincorporation/~03-~04 template residues tested (see the background levels of misincorporation reported in this study; and for a review, see Table I in Ref. 34).
Turnover studies confirmed that the Klenow fragment did not efficiently incorporate any dNTP opposite template 3methyldeoxythymine. Whereas no efficient incorporation or turnover of noncomplementary dNTPs occurred in response to 3-methyldeoxythymine residues, increased turnover of complementary dNTPs did occur. Thus, the polymerase was blocked by its inability to replicate past 3-methyldeoxythymine template residues. Consequently, the Klenow fragment idled, as previously reported for E. coli DNA polymerase I in response to pyrimidine dimers in template (35).
Surprisingly, introduction of 3-methyldeoxythymine into template apparently increased the overall fidelity of the Klenow polymerase without a concomitant increase in turnover of noncomplementary dNTP. In fact, both noncomplementary dNTP turnover and incorporation decreased in response to 3-methyldeoxythymine relative to complementary dNTP incorporation, indicating that the effect was in the polymerization step of DNA synthesis. This response was most evident for dGTP misincorporation, which was the highest of any noncomplementary dNTP. This apparent increase in fidelity would appear to lend support to the notion that polymerase proofreading of errors can also occur in the polymerization step, independent of the proofreading exonuclease, through release of dNTP (36, 37). It is not clear to us why a stuttering (progress impeded by 3-methyldeoxythymine) polymerase should be more accurate than a processive polymerase. Possibly, the stuttering polymerase can take greater advantage of the difference in off rates between complementary and noncomplementary dNTPs (36,37) to enhance its incorporation accuracy.
DNA sequencing technology demonstrated that methylation a t N-3 abolishes dTTP's base-pairing ability during DNA synthesis. Whereas the Klenow fragment did not substitute 3-methyl-dTTP, even at 1 mM, for any of the four normal dNTPs, unmodified dNTPs significantly incorporated at 250 nM (this study and Footnote 2), demonstrating a greater than 4,000-fold difference between the incorporation abilities of normal dNTPs and 3-methyl-dTTP. AMV reverse transcriptase substituted 3-methyl-dTTP only at high concentration for dATP, dGTP, and dTTP, but not for dCTP and not at every site. Comparison of 3-methyl-dTTP incorporation with the efficient levels of incorporation of the complementary unmodified dNTPs indicated a 40,000-fold difference in incorporation efficiency.   Fig. 4 for all other conditions. Note the shift to higher molecular weight DNA products with increasing dTTP concentration.
during DNA synthesis because of their favorable incorporation properties during DNA synthesis; their apparent K , values are only about 10-fold higher than for unmodified, complementary dNTP. Indeed, 04-methylthymine, supplied in culture medium, is incorporated in amounts rivaling amounts produced by direct alkylation (45). Our results, however, indicate that 3-methyl-dTTP and other dNTPs damaged at the central hydrogen-bonding position are unlikely to contribute to mutagenesis by this route, based on an apparent insertion rate for 3-methyl-dTTP that is greater than 4,000-40,000-fold below that of dTTP.