The role of 3-hydroxyethyldeoxyuridine in mutagenesis by ethylene oxide.

Ethylene oxide, a direct-acting mutagen and carcinogen, produces 3-hydroxyethyldeoxyuridine (3-HE-dU) after initial alkylation at N3 of dC, followed by rapid hydrolytic deamination. The significance of formation of 3-HE-dU in DNA was investigated by in vitro DNA replication of 3-HE-dU. A 55-nucleotide DNA template, containing 3-HE-dU at a single site, was constructed. DNA products, synthesized on the site-modified template, were analyzed and mutagenic bypass at 3-HE-dU estimated. The 3-HE-dU lesion blocked DNA replication by the Klenow fragment of Escherichia coli polymerase I (Kf Pol I) and bacteriophage T7 polymerase (T7 Pol) 3' to 3-HE-dU and after incorporating a nucleotide opposite 3-HE-dU. DNA synthesis past 3-HE-dU was negligible (< 3%). Substitution of Kf Pol I (exo-) and T7 Pol (exo-), polymerases lacking 3'-->5' exonuclease proofreading activity, for Kf Pol I and T7 Pol, respectively, facilitated DNA synthesis past 3-HE-dU. The bypass synthesis by Kf Pol I (exo-) was 60% and 90% by T7 Pol (exo-). These results suggest that the 3-HE-dU lesion could be bypassed, but that the extension at 3-HE-dU is rate-limiting. In the absence of proofreading, the nucleotide incorporated opposite 3-HE-dU is not excised and remains in position long enough for extension to occur. During post-lesion synthesis, both dA and dT were incorporated opposite 3-HE-dU. Since 3-HE-dU is derived from dC alkylation by ethylene oxide, incorporation of dA and dT opposite 3-HE-dU implicates this lesion in G.C-->A.T and G.C-->T.A mutagenesis.

ene oxide has been reported to induce dominant lethal mutations in mice (13) and rats (14). In monkeys, ethylene oxide exposure produced increased frequencies of sister chromatid exchange and chromosomal aberrations (15,16). In the mouse, ethylene oxide was an effective inducer of chromosomal aberrations, micronuclei, and sister chromatid exchange after intraperitoneal injection (17). Humans occupationally exposed to ethylene oxide were found to have increased frequencies of chromosomal damage and gene mutation in peripheral blood lymphocytes (18).
Ethylene oxide is a direct-acting alkylating agent. It reacts with DNA by an S,2 mechanism which favors alkylation at strongly nucleophilic endocyclic nitrogen atoms (1). Ethylene oxide binds to DNA both in vitro and in vivo (19-25) leading mainly (-90% of total alkylation) to the formation of 7 4 2 -h~-droxyethy1)deoxyguanosine (7-HE-dG1.l Other purine adducts formed are 3-(2-hydroxyethyl)deoxyadenosine (3-HE-dA) and 06-(2-hydroxyethyl)deoxyguanosine(OB-HE-dG).Hydroxyethylation at N1, N6, and N7 of the adenine moiety in DNA have also been reported (21, 23). Evidence exists that ethylene oxide can also react at DNA-phosphate groups (1, 2, 26). The purine adducts at N7 and O6 of dG and N3 of dA are not persistent in vivo and are removed from DNA through depurination and DNA repair (23,25,27,28). It has been suggested that these purine adducts may induce mutations at G C and A.T base pairs (29,30). Lack of in vivo persistence of these lesions is not consistent with this suggestion. The specific ethylene oxide-induced DNA adducts involved in mutagenesis by ethylene oxide are not established.
3-HA-dC intermediate facilitated the hydrolytic deamination and a mechanism was proposed (21, 33,35). The in vivo persistence of ethylene oxide-induced 3-(2-hydroxyethyl)deoxyuridine (3-HE-dU) is not known. The 3-HE-dU adduct is chemically stable in DNA (21). No repair activity for this lesion is known in prokaryotes and eukaryotes. 3-HE-dU may be a persistent lesion in vivo. The persistence of ethylene oxide-induced 3-HE-dU in DNA may have a significant role in ethylene oxideinduced mutagenesis.
The role of a single base modification in mutagenesis can be investigated by site-specific incorporation of the modification into DNA and analysis of the products synthesized by DNA polymerases on the site-moditled DNA template. We previously utilized such an in vitro DNA system to study the DNA replication In this paper we report that the 3-HE-dU lesion, present at a single site in the DNA template, blocked DNA synthesis 3' to 3-HE-dU and after incorporating a nucleotide opposite the lesion. Under conditions of relaxed fidelity, i.e. absence of proofreading, mutagenic bypass with dA and dT incorporation opposite 3-HE-dU occurred. Since 3-HE-dU is derived from ethylene oxide-induced 3-HE-dC, incorporation of dA and dT opposite 3-HE-dU implicates this lesion in G C + A.T and G C --j T.A mutagenesis.
Synthesis and site-specific incorporation of 3-HE-dU into the oligodeoxynucleotide are described below. Reverse-phase HPLC was performed on a waters Associates system consisting of two pumps (models 6000A and 590), a variable wavelength detector (model 490) and a System Interface Module. The system was controlled by a Digital Equipment Corporation (Professional 350) computer operating a Waters 840 data and chromatography control station. Proton homonuclear magnetic resonance (lH NMR) spectra were recorded on a Bruker AM360 (360 MHz) instrument, and all chemical s h i h are assigned relative to tetramethylsilane (0.0). Mass spectra (MS) were obtained on a VG-70SE high resolution mass spectrometer interfaced to a VG 1l/250 data system N G Manchester, United Kingdom).
TLC plates were developed in the following solvent systems: I, chloroform/ methanol (9:l); 11, dichloromethandmethanoUtriethylamine (45:l:l); 111, dichloromethandtriethylamindethyl acetate (18:1:2). Fig. 1)-A 50-ml roundbottom flask was charged with 2.00 g (8.76 mmol) of 2'-deoxyuridine and 2.42 g (17.52 mmol) of potassium carbonate in 5 ml of dimethyl sulfoxide. The mixture was stirred a t room temperature for 30 min and then treated with 1.60 g (9.58 mmol) of 1-bromo-2-acetoxyethane. The reaction mixture was stirred at room temperature for 24 h. TLC analysis (solvent system I) revealed the presence of two UV-absorbing spots (R,= 0.27 and 0.60) with the more polar spot corresponding to unreacted dU. The reaction mixture was diluted with 20 ml of water and neutralized with dilute acetic acid. The resulting mixture was concentrated under diminished pressure at 40 "C, treated with 50 ml of methanol, 10 g of silica, and again concentrated. The residue was coevaporated with methanol (2 x 50 ml) and dried in uucuo overnight. The dried, impregnated silica was applied to a silica gel column (3.8 cm x 16.5 cm) and eluted with 7% methanol in chloroform. Fractions of 10 ml each were collected and analyzed by TLC (solvent system I). The fractions con- . These assignments are based on one-and two-dimensional proton homonuclear correlation spectra. The absence of a signal due to the proton at N3 of dU confirmed that 2-acetoxyethylation had occurred at this position ( Fig. 1).

3-(2-Hydroxyethyl)-2'-deoxyuridine
(3-HE-dU; Fig. 11-3-AE-dU (100 mg; 0.32 mmol) was deacetylated by heating with concentrated ammonium hydroxide (2 ml) at 70 "C for 30 min. These conditions were used for deprotection of the synthesized oligomer. To confirm that 3-HE-dU is exclusively formed and is stable following deacetylation of 3-AE-dU, the following experiments were performed. After cooling, excess of ammonia was removed by bubbling nitrogen gas through the solution. The deacetylated product, 3-HE-dU (Fig. 11, was recovered by lyophilization and fully characterized. The product moved as a single spot on TLC (solvent system I; R -0.31) and coeluted as a single sharp peak (retention time, 26.3 mln) with authentic synthetic marker, previously characterized by UV, MS, and NMR (21), during reversed-phase HPLC on a Beckman Ultrasphere ODS semipreparative column (5-pm particle size; 4.6 mm x 25 cm) utilizing a flow rate of 2.4 mUmin and monitoring the UV absorption a t 254 nm. Chromatography was performed using water as the mobile phase with 10% acetonitrile ramped in over 30 min initially (curve 61, followed by a n increase to 25% at 32 min (curve 6) and ultimately to 75% a t 38 min (curve 6). CI-MS analysis of 3-HE-dU formed by deacetylation of 3-AE-dU had the identical mass spectra to authentic 3-HE-dU with the expected molecular ion MH at m l z 273, establishing the molecular weight as 272. The presence of the base ions bH, at m l z 157 and bH,-H,O at m / z 139 established the deacetylation of 3-AE-dU to yield 3-HE-dU. These results demonstrate that concentrated ammonium hydroxide a t 70 "C eficiently deacetylated 3-AE-dU to 3-HE-dU with no detectable side products. The pyridine was removed under high vacuum, and the flask was again charged with 17 ml of anhydrous pyridine and 46 mg (0.38 mmol) of 4-dimethylaminopyridine. The mixture was cooled in an ice-bath and treated, all at once, with 3.42 g (10.1 mmol) of 4,4'-dimethoxytrityl chloride. The mixture was stirred in a n ice-bath for 1 h followed by 2 h f .-Ethylene Oxide-induced 3-Hydroxyethyl Uracil at room temperature. The reaction mixture was diluted with 150 ml of dichloromethane and washed successively with saturated sodium bicarbonate (1 x 200 ml) and brine (1 x 200 ml). The organics were dried over anhydrous magnesium sulfate and the solvent removed under reduced pressure at 25 "C. The resulting syrup was dried by coevaporation with 200 ml of anhydrous toluene under reduced pressure at 45 "C. TLC analysis (solvent system 11) showed a major product spot (Rf = 0.34).

5'-0-~4,4'-dimethozytrityl-3-(2-acetozyethyl)~2'-deoxyuridine (DMlh
The product was purified by silica gel column (3.8 cm x 25 cm) chromatography using solvent system I1 as the eluant. Fractions of 10 ml each were collected and analyzed by TLC (solvent system 11). The fractions containing the pure desired product (single spot; Rf = 0.34) were pooled and concentrated in uucuo to yield 3.04 g (60%) of DMTr-(3-AE)dU (Fig. 1). CI-MS established the expected molecular weight of 617. The structure was confirmed by 'H NMR in deuterated chloroform (CDCI,). The chemical shift assignments are: 6 7.80 (d, lH, J = 8.1 Hz, H6), 7.37-7.16 (m, 9H, phenyl ring and meta to -OCH, in anisole rings), 6.80 (d, 4H, J = 8.8 Hz, ortho to -OCH, in anisole rings), 6.28 (t, lH, J = 6.1 HZ, Hl'), 5. The mixture was brought to 0 "C and treated, dropwise over 5 min, with 156 mg (0.67 mmol) of N,N-diisopropylamino-(2-cyanoethoxy)-chlorophosphine. The mixture was stirred at 0 "C for 1.5 h, allowed to warm to room temperature and stirred an additional 1 h. The reaction was diluted with 15 ml of anhydrous ethyl acetate and transferred to a separatory funnel containing 20 ml of ethyl acetate and 50 ml of saturated brine. The organics were separated, washed with saturated brine (3 x 50 ml), and dried over anhydrous magnesium sulfate. The solvents were removed under reduced pressure at 25 "C. The resulting gum was purified by silica gel column (4 cm x 23 cm) chromatography using solvent system I11 as an eluent. Fractions of 10 ml each were collected and analyzed on TLC Site-modified Oligonucleotides-Seventeen-nucleotide long oligomers, 5'-TAAAAGTCU*AAAACATG (U* = 3-HE-dU and dU as a control), were assembled by Midland Certified Reagent Company (Midland, T X ) on an Applied Biosystems model 381Asynthesizer using phosphite triester chemistry (48). Release of the oligomer from the solid support and complete removal of all protecting groups was accomplished in a sealed tube with concentrated ammonium hydroxide at 70 "C for 30 min. The oligomers were purified by anion-exchange HPLC and subsequently desalted utilizing gel-filtration (Midland Certified Reagent Company). The purified oligomers were phosphorylated at the 5'-OH group with T4 polynucleotide kinase and [y-32PlATP and fully characterized. Homogeneity of the oligomers were checked by electrophoresis on a 20% polyacrylamide, 8 M urea sequencing gel. To quantitate the nucleoside content of the oligomers, approximately 1 A,, unit of the oligonucleotide was incubated with 3 units of phosphodiesterase in 100 m M Tris-HC1 buffer, pH 8.2, for 10 min at 37 "C (total reaction volume = 50 J) followed by treatment with 2 units of alkaline phosphatase for 10 min under the same conditions. Following the 20-min total reaction time, the proteins were denatured by heating at 80 "C for 10 min, and the solution was analyzed by reversed-phase HPLC on a Beckman Ultrasphere ODS semipreparative column (5-pm particle size; 4.6 mm x 25 cm) utilizing a flow rate of 2.4 mVmin and monitoring the W absorbance at 254 nm. Elution was performed as described for 3-HE-dU (above). Under those conditions authentic standards exhibited good separation with retention times of dC, 15.8 min; dU, 18.4 min; deoxyinosine (dl), 22.3 min; dG, 23.2 min; dT, 25.6 min; 3-HE-dU, 26.3 min; dA, 30.8 min. Response factors of the deoxynucleosides (at 254 nm) were established by integrating the peak areas when equimolar amounts were injected. Deoxythymidine was used as the reference and all others were normalized with respect to it.
Primed Site-modified Template-The primed template was assembled by annealing a 17-nucleotide complementary primer to a 55nucleotide long 3-HE-dU-containing template. The template was constructed by ligating a 5'-phosphorylated 17-nucleotide oligomer (5 nmol) to the 3'-end and a 21-nucleotide oligomer (5 nmol) to the 5'-end of the 5"phosphorylated 3-HE-dU-containing 17-nucleotide oligomer (5 nmol). The three oligomers were held together in the desired orientation for ligation by two 18-nucleotide complementary oligomers (5 nmol each). The five oligomers in 100 pl of 25 m M Tris-HC1, pH 7.5, 10 m M MgCl,, 2 m M DTT, and 100 m M NaCl were heated at 65 "C for 5 min, followed by slow cooling to 4 "C (4 h). The annealed mixture was made 1 m M ATP and 5 m M DTT. Ligation was initiated by adding T4 DNA ligase (100 units as reported by the supplier). After incubation at 16 "C for 12 h, the reaction was terminated by adding EDTA at a final concentration of 15 m~. The reaction volume was reduced to about 30 p1 using a Speed Vac concentrator (Savant), mixed with 50 p I of formamide dye mix (49), and the mixture heated at 100 "C for 3 min before separation on a 16% polyacrylamide, 8 M urea gel (1 mm thick). The product bands were visualized by placing the gel on a TLC plate containing a fluorescent indicator. The major band, corresponding to the 55-nucleotide template, was excised, crushed, and extracted with methanol (4 x 5 mlj at 37 "C to remove urea. The product was eluted in 10 m M NaCl (700 pl). The yield of the 3-HE-dU-containing 55-mer template was 65%. The template was characterized by DNA sequencing (data not shown). The site-modified template contained a single 3-HE-dU adduct at position 26 from the 3'-end.
A 2-fold molar excess of the complementary 5'-3ZP-labeled (3000 Ci/ mmol) 17-nucleotide primer was annealed to the site-modified 55-nucleotide template as described previously (36). The yield of the resulting primed template, as estimated by electrophoresis on a non-denaturing 12% polyacrylamide gel at 4 "C (cold room), was >95%.
DNA Polymerase Reaction-The reaction mixture (7 111) contained 10 m M HEPES buffer, pH 7.5, 5 m M MgCl,, 2 m M DTT, 0.1 pmol of primed template, and 0.2 unit (as reported by the supplier) of the DNA polymerase. The concentrations of dNTP were varied and are included in the figure legends. The incubation was at 37 "C for 30 min. The reaction was terminated by adding 1 1. 11 of 100 m M EDTA and 1.5 volumes of the formamide dye mix. After heating at 100 "C for 3 min, DNA synthesis products were separated on a 16% polyacrylamide, 8 M urea sequencing gel. DNAsynthesis products were analyzed as described previously (37).
Identification of the Nucleotide Incorporated Opposite 3-HE-dU-The nucleotides incorporated opposite 3-HE-dU during in vitro DNA replication were identified by sequencing the DNA synthesis products. The products, sufficient for DNA sequencing, were obtained by DNA polymerase reaction using 10 pmol of the site-modified primed template. The products were isolated as described previously (37). DNA sequencing of the blocked products used the modified Maxam-Gilbert (50) method, while sequencing of the bypassed products was performed by the Sanger et al. (51) dideoxy method.

Chemical Synthesis
of Site-modified OEigodeoxynucEeotide-A 17-nucleotide long oligomer, containing a single 3-HE-dU adduct, was synthesized by the solid-phase phosphite triester method (48). During synthesis, 3-HE-dU was site specifically incorporated into the oligomer by the use of DMTr-(3-AE)dU-phosphoramidite (Fig. 1). The phosphoramidite was synthesized by reacting 1-bromo-2-acetoxyethane with dU in a bimolecular (S,2) manner under mildly basic conditions, followed by sequential protection of the 5'-OH group of the resulting 3-AE-dU adduct with 4,4'-dimethoxytrityl (DMTr) group and conversion of the 3"OH into the 3'-phosphoramidite. The overall yield of the phosphoramide was >40%. A key feature in the synthesis of DMTI"(3-AE)dU-phosphoramidite (Fig. 1) was the utilization of an acetyl (Ac) group for protection of the N3 side chain hydroxyl moiety. The acetyl group is stable under the acidic conditions employed during oligonucleotide aynthesis and afforded the added advantage of being cleaved in the final step used for deprotection of the synthetic oligomer.
Before being used in the next synthetic step, the desired intermediate product was purified to homogeneity as judged by a sharp single spot on TLC and absence of the starting material, and fully characterized by UV, MS, and NMR spectroscopy.
The U V spectrum of 3-AE-dU at pH 6 (H,O) displayed a A , , , at 262 nm, which is consistent with alkylation at N3 of dU and excludes the addition of a 2-acetoxyethyl group a t O4 and O2 of dU. The CI mass spectrum of 3-AE-dU displayed an MH ion at mlz 315 confirming the expected molecular weight of 314. The presence of the bH, ion at mlz 199 established that modification had occurred on the base moiety. The 3-AE-dU was further characterized by NMR. All NMR assignments, based on oneand two-dimensional homonuclear correlation spectra, were consistent with the structure 3-AE-dU (Fig. 1). Assignment of the side chain residue in 3-AE-dU was confirmed by the presence of two correlated resonances (6 4.04 and 4.18) corresponding to the methylene protons of the side chain. Based on the NMR spectrum of 3-(2-acetoxypropyl)dU,2 the up field resonance (6 4.04) was tentatively assigned to the methylene protons adjacent to N3 of dU and the down field one (6 4.18) to the protons adjacent to the acetate group. As demonstrated by MS and co-chromatography with an authentic marker (21) 3-HE-dU was exclusively produced when 3-AE-dU was deblocked by concentrated ammonium hydroxide at 70 "C. The stability of 3-HE-dU in hot aqueous ammonia was important to establish since these conditions were employed to deprotect the 3-HE-dU-containing oligomer.
Analysis of DMTr-(3-AE)dU by CI-MS revealed the expected molecular weight of 617, with alkylation occurring on the base and tritylation occurring on the sugar moiety. The one-dimensional NMR spectrum of DMTr-(3-AE)dU exhibited resonances which are compatible with its structure (Fig. 1)  Reduced volatility of DMTI"(3-AE)dU-phosphoramidite prevented CI-MS analysis of this compound. However, the phosphoramidite was fully characterized by NMR. The differences in NMR spectra, of the phosphoramidite and the precursor DMTr-(3-AE)dU, highlighted by the absence of the resonance arising from the 3'-OH proton (6 4.60 in DMTrPr-(3-AE)dU) and the concomitant presence of signals arising from the diisopropyl (6 1.20) and cyanoethoxy (6 2.45) residues, established the structure of the phosporamidite (Fig. 1).
The pure and fully characterized phosphoramidite was used to insert the 3-HE-dU adduct into the oligomer during oligodeoxynucleotide synthesis. Since 3-HE-dU is stable to alkaline conditions, all protecting groups from the synthesized oligomer were fully removed by treatment with concentrated ammonia at 70 "C. The completely deprotected oligomer was purified by anion-exchange HPLC followed by electrophoresis on a 16% polyacrylamide, 8 M urea gel. The 5'-32P-labeled purified oligomer moved as a sharp band on a 20% polyacrylamide, 8 M urea sequencing gel and had the same mobility as a standard 17-mer marker ( Fig. 2A, inset). The 3-HE-dU adduct in the oligomer was quantitated by HPLC analysis (Fig. 2 A ) of the nucleosides released by enzymatic hydrolysis of the site-modified oligomer. The 3-HE-dU peak was unambiguously identified by coelution with an authentic synthetic marker (21) and by UV absorption characteristics. No additional peak, except dI peak (retention time, 22.3 min), was observed (Fig. 2.4) peak was present in the hydrolysates of both site-modified and control oligomers (Fig. 2). The dI was derived from deamination of dA by the contaminating 2'-deoxyadenosinedeaminase in the phosphodiesterase preparation used to hydrolyze the oligomers. This was evident by the increase in the dI peak with the corresponding decrease in the dA peak (data not shown) with longer incubation periods during enzymatic hydrolysis. For quantitation, the amount of dI formed was added to the amount of dA obtained. Analysis of the nucleosides released from the site-modified oligomer revealed a ratio of 2.1:1.9:3.0:9.2:1.2 between dC/dG/dT/dA/3-HE-dU, as compared with the predicted ratio of 2:2:3:9:1. The results demonstrate that 3-HE-dU is present in all oligomer molecules. This is consistent with the absence of the dU peak (retention time, 18.4 min) in the analysis of the site-modified oligomer ( Fig. 2A) as compared to its presence in the control dU-containing oligomer (Fig. 2B).
In Vitro DNA Replication System-The in vitro DNA replication system (Fig. 3) was used to study the DNA replication properties of 3-HE-dU, present at a single site in the DNA template. The system utilizes a primed template, similar to the system used in our laboratory for DNA replication studies of the ethylating agent-induced 3-Et-dT (36,37,53) and 02-Et-dT (47,52,53), except that the length of the site-modified template was increased from 36 to 55 nucleotides without changing the location of the lesion in the system. The increase in the length of the site-modified template was designed to facilitate DNA sequencing of the lesion-bypass products by the Sanger dideoxy method. The 3-HE-dU lesion was present at template position 26 from the 3'-end and was eight nucleotides away from the 3' terminus of the hybridized primer. This system represents a "running start" for 3-HE-dU in DNA replication in that synthesis occurs prior to the polymerase reaching the lesion. When the polymerase encounters 3-HE-dU, the lesion may block DNA replication producing a 25-nucleotide "preblocked" product, terminating 3' to 3-HE-dU, without incorporation of a nucleotide opposite the lesion, and a 26-nucleotide "blocked" product, terminating after incorporating a nucleotide opposite 3-HE-dU. The 3-HE-dU lesion may allow DNA synthesis past the lesion producing a 55-nucleotide "bypassed" product. The representative DNA products produced during DNA synthesis past 3-HE-dU are shown in Fig. 4. During DNA replication using a 3' + 5' exonuclease-deficient polymerase, a 56-nucleotide bypassed product, due to blunt-end addition (54) at the synthesized 55-nucleotide duplex, may also be obtained. In that case, the 55-and 56-nucleotide products were combined for analysis. DNA Replication Block by 3-HE-dU-The 3-HE-dU lesion, present a t a single site in a DNA template, blocked DNA synthesis by Kf Pol I. In the control, containing dU in place of 3-HE-dU, DNA synthesis was not interrupted and proceeded to the 5'-end of the template yielding a full-length (55-nucleotide) product (data not shown). The results indicated that in the site-modified template, 3-HE-dU was responsible for blocking DNAreplication. In the presence of 5 mM Mg2' and 10 p~ dNTP, the major DNA synthesis product was a 25-nucleotide preblocked product (72%) obtained by DNA replication inter- rupted 3' to 3-HE-dU (Fig. 5A). KfPol I was able to incorporate a nucleotide opposite 3-HE-dU, but DNA synthesis was blocked after incorporation resulting in the accumulation of a 26-nucleotide blocked product (25%). Since DNA synthesis past 3-HE-dU was not observed, the lesion represented a complete block to DNA replication by Kf Pol I.
The effect of dNTP concentration on nucleotide incorporation opposite 3-HE-dU was significant. Fig. 5A presents the effect of dNTP concentration on the relative formation of various DNA products synthesized by Kf Pol I. When dNTP concentration was increased from 10 to 200 PM, the incorporation opposite 3-HE-dU was increased from 25 to 60%. This increase was accompanied by a corresponding decrease in the preblocked product. The accumulation of the 26-nucleotide blocked product represents an equilibrium between nucleotide insertion opposite 3-HE-dU by the polymerase activity and removal of the inserted nucleotide by the 3' + 5' exonuclease proofreading activity of Kf Pol I. Since DNA synthesis past 3-HE-dU was not observed (Fig. 5A), even at high dNTP concentration (200 p d , the results suggest that Kf Pol I could incorporate a nucleotide opposite 3-HE-dU, but the resulting base pair was not a suitable substrate for extension. These results are consistent with the analogous 3-Et-dT lesion, where 3-Et-dT presented a complete block to DNA replication by Kf Pol I in the presence of M e (36).
Kf Pol I, a"repair" polymerase used in the above experiments has low processivity (dissociating before the incorporation of an average 10 nucleotides) and a weak 3' + 5' exonuclease proofreading activity (55). In order to investigate the effects of polymerase processivity and efficient proofreading on bypass replication of 3-HE-dU, DNA replication was catalyzed by a "replicative" polymerase, T7 Pol. The in vitro DNA replication system used in these studies was the same as used for Kf Pol I. T7 Pol is a highly processive enzyme (incorporating thousands of nucleotides on the same template before dissociating) and 3-HE-dU blocked DNA synthesis (92%) by T7 Pol 3' to 3-HE-dU (preblocked product), and 6% after incorporating a nucleotide opposite the lesion (blocked product). DNA synthesis past 3-HE-dU was negligible (<2%; Fig. 6A). These results are similar to those obtained with Kf Pol I except that the blocked product terminating after incorporation opposite 3-HE-dU was higher (25%) with Kf Pol I. The results suggest that the nucleotide incorporated opposite 3-HE-dU was efficiently removed by the potent exonuclease activity of T7 Pol. As shown in Fig.  6 A , nucleotide incorporation opposite 3-HE-dU increased with increasing dNTP concentration, reaching 39% at 200 p~, as compared to 60% with Kf Pol I (Fig. 5A). Absence of post-lesion synthesis, even at higher dNTP concentration (200 p~) , suggests that 3-HE-dU presents a strong block to DNA replication by T7 Pol. Synthesis of the 26-nucleotide blocked products (Figs. 5A and 6 A ) indicated that T7 Pol and Kf Pol I could incorporate a nucleotide opposite 3-HE-dU, but that the resulting base pair, present at the growing 3'-end of the primer, was difficult to elongate.
To identify the nucleotide incorporated opposite 3-HE-dU during the replication block, the 26-nucleotide blocked products, synthesized by Kf Pol I and T7 Pol, were sequenced by the modified Maxam-Gilbert procedure. The sequencing results, obtained from the 26-nucleotide blocked product synthesized by Kf Pol I, are shown in Fig. 7. An identical sequencing pattern (not shown) was obtained from the blocked product synthesized by T7 Pol. The results, indicate that both Kf Pol I and T7 Pol preferentially incorporated dA opposite 3-HE-dU. Incorporation of dT or other nucleotides at low levels may not be detected by this sequencing procedure. The results suggest that although specificity of nucleotide incorporation opposite 3-HE-.$ 100 60 40t. DNA Synthesis Past 3-HE-dU-In general, DNA lesion bypass is strongly inhibited in the presence of 3' + 5' exonuclease proofreading activity of the polymerase (57). Since DNA synthesis past 3-HE-dU is a necessary step to produce mutation, the effect of the polymerase proofreading activity in mediating DNA synthesis past 3-HE-dU was investigated. When Kf Pol I (exo-), a polymerase lacking 3' + 5' exonuclease activity was substituted for Kf Pol I, DNA synthesis past 3-HE-dU occurred (Fig, 4A). At 10 PM dNTP, the bypass synthesis was 32% which increased with increasing dNTP concentration reaching 58% at 200 VM (Fig. 5B). This is in contrast to post-lesion synthesis by Kf Pol I, where DNA synthesis past 3-HE-dU was negligible (<3%; Fig. 5 A ) . At 200 J~M dNTP, only 4% of DNA synthesis by Kf Pol I (exo-) was blocked 3' to 3-HE-dU indicating that incorporation of a nucleotide opposite 3-HE-dU, to produce a 26-nucleotide blocked product (38%) and a bypassed product (58%), was 96%. In the absence of proofreading the nucleotide incorporated opposite 3-HE-dU in the blocked product was not cleaved, allowing the nucleotide incorporated opposite 3-HE-dU to remain in position long enough for elongation past 3-HE-dU to occur (58%). Accumulation (38%), even at high dNTP concentration (200 PM), of the 26-nucleotide blocked product indicates that the 3-HE-dU-containing base pair could be extended, but with low efficiency.
In vitro DNA replication studies of 3-HE-dU with Kf Pol I (exo-) and T7 Pol (exo-) indicate that, in the absence of proofreading, the 3-HE-dU lesion could be bypassed. T7 Pol (exo-) was more efficient in elongation at 3-HE-dU than Kf Pol I (exo-). At 200 p~ dNTP, while bypass synthesis by Kf Pol I (exo-) was <60% (Fig. 5B), the extension by T7 Pol (exo-) was >90% (Fig. 6B ). Efficient extension by T7 Pol (exo-) is consistent with the high processivity of T7 polymerase as compared to the low processibity of Kf Pol I. In general, processive polymerases translocate along DNA faster than they dissociate from primer template termini (57). were incorporated opposite 04-Et-dT (53), which is consistent with published reports (58,59). Since 3-HE-dU is induced a t a dC moiety, incorporation of dA and dT opposite 3-HE-dU implicates this lesion in transversion and transition mutagenesis a t G-C base pairs.

DISCUSSION
The in vitro DNA replication studies described here demonstrate that under conditions of relaxed fidelity, i.e. absence of proofreading, dA and dT are incorporated opposite 3-HE-dU.
Since 3-HE-dU is derived from ethylene oxide-induced 3-HE-dC, incorporation of dA and dT opposite 3-HE-dU implicates this lesion in G-C + A-T and G C + T.A mutagenesis. Our studies suggest that 3-HE-dU may be a critical premutagenic lesion produced by the mutagenic and carcinogenic ethylene oxide in vivo and epoxide-induced 3-hydroxyalkyldeoxyuridine (3-HA-dU) lesions may contribute to mutagenesis and initiation of cancer by other environmentally important epoxides.
DNA synthesis past a lesion is a necessary step to produce mutation by the lesion. In general, post-lesion synthesis results from three independent reactions (60): nucleotide insertion opposite the lesion, excision by proofreading of the inserted nucleotide before extension, and extension past the lesion. Proofreading by exonuclease becomes more proficient when the extension step is slow. In studies reported here significant accumulation of the 26-nucleotide blocked product (Figs. 5A and 6A), terminating after incorporation of a nucleotide opposite 3-HE-dU during polymerization by Kf Pol I and T7 Pol, suggests that addition opposite 3-HE-dU is a relatively easy step. Negligible post lesion synthesis (<3%) indicates that extension at 3-HE-dU is rate-limiting. The 3' + 5' exonuclease proofreading activity, associated with the polymerases, made the extension a t 3-HE-dU impossible by excising the incorporated nucleotide opposite 3-HE-dU before extension could occur. The potent exonuclease activity of T7 Pol (56), as compared to the weak activity of Kf Pol I (551, is manifested in the accumulation of the 26-nucleotide blocked product in lower yields during DNA replication by T7 Pol (Fig. 6 A ) compared to Kf Pol I (Fig.  5A). Use of Kf Pol I (exo-) and T7 Pol (exo-), polymerases lacking proofreading activity, facilitated DNA synthesis past 3-HE-dU (Figs. 5B and 6B). The results are consistent with the suggestion that the elongation step at 3-HE-dU is rate-limiting. The absence of proofreading facilitated bypass at 3-HE-dU by allowing the nucleotide incorporated opposite 3-HE-dU to remain in position long enough for elongation to occur.
Base substitution mutations constitute an important component of the mutational spectra induced by ethylene oxide in vivo at the hypoxanthine-guanine phosphoribosyltransferase-(hprt) locus in different systems. In T-lymphocytes of ethylene oxide exposed B6C3F1 mice (29), A.T + T.A, A.T + G C , G C + T.A and G C "-f A T mutations were induced with almost equal frequency. In human diploid fibroblasts (301, ethylene oxide induced both G C --j T.A transversions (30%) and G C -A.T transitions (30%) at G.C base pairs, but only A.T + T.A transversions (40%) were produced at A,T base pairs. In peripheral lymphocytes of workers exposed to ethylene oxide (181, a hot spot for G C "-f AT transition mutation was observed at the hprt locus. Mutagenicity studies using S . typhimurium test strains have emphasized the importance of G C "-f T,A, G.C + A.T and A.T + T.A mutations in mutagenesis by aliphatic epoxides (61). The DNA adducts involved in mutagenesis induced by ethylene oxide are not known.
Ethylene oxide may induce base substitution mutations at A.T base pairs through adduction at dA or dT moieties. Transversion mutations at A-T base pairs may result from insertion of dA opposite an ethylene oxide-induced 3-HE-dA adduct or through incorporation of dA opposite an apurinidapyrimidinic ( A P ) site, created by the spontaneous loss or glycosylase-mediated removal of 3-HE-dA. Since 3-HE-dA is rapidly removed from DNA both in vitro and in vivo (23,251, the involvement of this lesion in mutagenesis by ethylene oxide a t A.T base pairs is uncertain. A substantial body of experimental evidence has indicated that it is not the initial level of DNA adduction but the in vivo persistence (lack of repair) and mutagenic bypass of premutagenic DNA adducts which is of major importance in mutagenesis and carcinogenesis (62)(63)(64)(65). Other dA adducts that can contribute to A.T + T.A mutagenesis either by mispairing with dA or through AP site formation include IP-, N1-, and N7-hydroxyethyldeoxyadenosine.
We have recently reported the in vitro formation of 3-(2-hydroxyethy1)deoxythymidine (3-HE-dT) in DNA after reaction with ethylene oxide (21). This lesion, which is formed in a relatively small amount (0.5%), is chemically stable in DNA (21). In the absence of a known DNA repair activity for 3-HE-dT, this lesion may be persistent in vivo. The 3-HE-dT lesion may contribute to A.T + T.A transversion mutagenesis by mispairing with dT during DNA replication. Support for this hypothesis is derived from our in vitro DNA replication studies of the analogous 3-Et-dT lesion (37) where, during post-lesion synthesis, dT was incorporated opposite 3-Et-dT leading to A.T 4 T.A transversion mutagenesis.
The ethylene oxide-induced modifications of dG and dC, responsible for transversion and transition mutations at G C base pairs, are not known. 7-HE-dG is the major adduct of dG produced by ethylene oxide both in vivo and in vitro (21,24,25). 7-HE-dG may not be premutagenic, but the lesion may contribute to mutagenesis through depurination or imidazole ring opening. AF' sites have been shown to induce mainly G.C + T.A and G C + A.T mutations (66)(67)(68). AP sites produced in vivo are usually repaired rapidly and without error (66). A small amount of 06-HE-dG is also induced in DNA by ethylene oxide (24,25). 06-Alkyl-dG is of major importance in mutagenesis by alkylating agents (69) giving rise to G.C + A.T transition mutations (70,71). Unrepaired 06-HE-dG may contribute to G.C + A-T mutagenesis by ethylene oxide. However, 06-HE-dG is efficiently removed from DNA of most tissues, except brain, presumably by 06-alkylguanine-DNA alkyltransferase and excision repair (27,28). Persistence studies have indicated that the in vivo half-life of 7-HE-dG is significantly greater than 06-HE-dG (25). Based on the formation of 7-HE-dG as the major ethylene oxide-induced DNA adduct and induction of the same G C + A T and G C + T-A mutations by ethylene oxide and AP sites, depurination of 7-HE-dG has been suggested to contribute to ethylene oxide mutagenesis at G.C base pairs (29, 30). However, lack of in vivo persistence of the lesion (25) is not consistent with this suggestion. We have shown that ethylene oxide reacts with dC at the N3 position and that the 3-HE-dC adduct undergoes rapid hydrolytic deamination to produce a potentially mutagenic lesion, 3-HE-dU (21, 34). This lesion is chemically stable in DNA (211, and no repair activity has been reported in prokaryotes or eukaryotes. In vivo persistence of 3-HE-dU may contribute to mutagenesis at G C base pairs by ethylene oxide. Our DNA replication studies have implicated 3-HE-dU in G C + A.T and G-C + T.A mutagenesis. The studies described here suggest that 3-HE-dU may be a critical mutagenic lesion produced by ethylene oxide in vivo and that 3-HA-dU may contribute to transition and transversion mutagenesis at G C base pairs by other epoxides.