The roles of DNA polymerases alpha, beta, and gamma in DNA repair synthesis induced in hamster and human cells by different DNA damaging agents

The involvement of DNA polymerases alpha, beta, and gamma in DNA repair synthesis was investigated in subcellular preparations of cultured hamster and human cells. A variety of DNA damaging agents, including bleomycin, neocarzinostatin, UV irradiation, and alkylating agents, were utilized to induce DNA repair. The sensitivity of repair synthesis, as well as replicative synthesis and purified DNA polymerase beta activity, to inhibition by the DNA polymerase inhibitors dideoxythymidine triphosphate, aphidicolin, cytosine arabinoside triphosphate, and N-ethylmaleimide was determined. No evidence was obtained for a major role of polymerase gamma in any type of repair synthesis. In both hamster and human cells, the sensitivity of bleomycin- and neocarzinostatin-induced repair synthesis to ddTTP inhibition was essentially identical with that observed for purified polymerase beta, indicating these repair processes proceeded through a mechanism utilizing polymerase beta. Repair synthesis induced by UV irradiation and alkylating agents was not sensitive to ddTTP, indicating repair of these lesions occurred through a pathway primarily utilizing a different DNA polymerase; presumably polymerase alpha. However, replicative synthesis was much more sensitive to polymerase alpha inhibitors than was repair synthesis induced by UV irradiation or alkylating agents. Neither the amount of DNA damage nor the amount of induced repair synthesis influenced the degree to which the different DNA polymerases were involved in repair synthesis. The possibility that "patch size" or the actual type of DNA damage determines the extent to which different polymerases participate in DNA repair synthesis is discussed.

The involvement of DNA polymerases a, p, and y in DNA repair synthesis was investigated in subcellular preparations of cultured hamster and human cells. A variety of DNA damaging agents, including bleomycin, neocarzinostatin, W irradiation, and alkylating agents, were utilized to induce DNA repair. The sensitivity of repair synthesis, as well as replicative synthesis and purified DNA polymerase j3 activity, to inhibition by the DNA polymerase inhibitors dideoxythymidine triphosphate, aphidicolin, cytosine arabinoside triphosphate, and N-ethylmaleimide was determined. No evidence was obtained for a major role of polymerase y in any type of repair synthesis. In both hamster and human cells, the sensitivity of bleomycin-and neocarzinostatin-induced repair synthesis to ddlTP inhibition was essentially identical with that observed for purified polymerase j3, indicating these repair processes proceeded through a mechanism utilitizing polymerase p. Repair synthesis induced by UV irradiation and alkylating agents was not sensitive to ddTTP, indicating repair of these lesions occurred through a pathway primarily utilizing a different DNA polymerase; presumably polymerase a. However, replicative synthesis was much more sensitive to polymerase a inhibitors than was repair synthesis induced by UV irradiation or alkylating agents. Neither the amount of DNA damage nor the amount of induced repair synthesis influenced the degree to which the different DNA polymerases were involved in repair synthesis. The possibility that "patch size" or the actual type of DNA damage determines the extent to which different polymerases participate in DNA repair synthesis is discussed.
There are three DNA polymerases, a, b, and y, in mammalian cells which can be differentiated by their size, subcellular location, substrate specificities, and susceptibility to specific inhibitors (reviewed in Refs. 1 and 2 ) . The role of each polymerase in DNA metabolism has been the subject of many studies. Most studies indicate that polymerase (Y is solely responsible for nuclear DNA replication (1, 2) as well as the * This research was supported in part by Research Grants GM24677 from the National Institutes of Health, United States Public Health Service, NP-300 from the American Cancer Society, and DE-FG22-80PC30248 from the Department of Energy. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisen e n t " in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ To whom correspondence should be addressed.  (5)(6)(7)(9)(10)(11)(12), One advantage of subcellular systems is that the effect of DNA polymerase inhibitors which would not rapidly penetrate intact cells (ie. nucleotides) can be investigated. Using subcellular preparations of human cells in 0-40 m~ NaCl and UV irradiation or alkylating agents to induce DNA repair synthesis, polymerase a was reported to be responsible for repair synthesis (9)(10)(11). On the other hand, polymerase ,fl has been implicated in DNA repair synthesis using subcellular preparations of human and rodent cells in 80-120 mM NaCl and UV irradiation (5) or bleomycin (6, 7) to induce repair synthesis. We have been attempting to reconcile these conflicting reports and to clarify the roles of the different DNA polymerases in repair synthesis. The effect of the salt concentration in situ on the involvement of DNA polymerases a and , 8 in repair synthesis has been reported (13). This study reports the effect of DNA polymerase inhibitors on DNA repair synthesis in subcellular preparations of hamster and human cells, using alkylating agents, UV irradiation, neocarzinostatin, and bleomycin to damage DNA. The effect of the amount of DNA damage on the involvement of the different DNA polymerases in repair synthesis is also investigated.

MATERIALS AND METHODS
Reagents-Bleomycin and neocarzinostatin were kindly supplied by Bristol Laboratories, Syracuse, NY. MNNG' was from Aldrich, ddTTP was from P-L Biochemicals, Inc. Aphidicolin was supplied by Imperial Chemical Industries, Cheshire, England. Growth of Cells-CHO cells were grown in suspension culture and growth arrested in the GI period as previously described (12). Diploid cation of Eagles medium (Gibco, Grand Island, NY) with 15% newborn calf serum (Microbiological Associates). Growth-arrested (GI phase) cells were obtained by placing confluent cultures in medium containing 0.25% newborn serum for 48 h. In Situ DNA Synthesis-DNA replicative and DNA repair synthesis were measured in in situ preparations of CHO and HF cells, following treatment of the cells with lysolecithin tb render the cells permeable to dNTPs (8,12,14). Details for treating CHO cells with lysolecithin have been described (12,13). HF cells were removed from 100-mm culture plates with trypsin-EDTA (Gibco), followed by washing 2 times at 4 "C in solution A (35 mM 4-(2-hy~oxylethyl)-lpiperazineethanesulfonic acid, pH 7.4; 150 m M sucrose; 5 m~ potassium phosphate, pH 7.4; 5 m~ MgC12; 0.5 m~ CaCb) containing 0.2 m M phenylmethylsulfonyl fluoride, suspended in solution A (0 "C) at 8 X lo7 cells/& and permeabilized by the addition of lysolecithin (Sigma, Type I) to a final concentration of 0.5 mg/ml. DNA replicative and repair synthesis were then measured in lysolecithin-treated cells as indicated below.
Exponentially growing CHO and HF cells were permeabilized and used to measure replicative DNA synthesis, whereas growth arrested (GI) cells were used for DNA repair studies. For replicative synthesis, cells were incubated at 2 X lo7 cells/ml in solution A containing 1. repair-inducing agents, before or after permeabilization as described, Incorporation of C3H]TTP into DNA was determined as described (12).
CsCl Gradients-To characterize DNA synthesis in situ, permeable cells were incubated in the DNA synthesis solution with BrdUTP (250 ,UM) substituted for TTP; [3H]dAT? (12.5 ,UM) was the labeled nucleotide. Following incubation at 37 "C, the DNA was isolated, sheared, denatured, and centrifuged to equilibrium in CsCl gradients, as described (15). Sucrose Gradients-The amount of DNA damage induced by various agents was estimated by determining the size of DNA in alkaline sucrose gradients. Exponential HF cells were treated with 3 ,UM [3H]TdR (0.50 Ci/mmol) for 24 h, allowed to grow to confluence in r3H]TdR-free medium and growth arrested in medium containing 0.25% newborn calf serum for 48 h. The labeled cells were treated with DNA damaging agents before or after permeabilization, as indicated, and the DNA was isolated and centrifuged on 5-20% alkaline sucrose gradients in a sW41 rotor at 31,000 RPM for 10 h, 15 "C, as described (15). DNA Polymerase P-DNA polymerase p was purified to near homogeneity from hamster livers according to Kunkel et al. (16) formed in solution A containing 250 WM MTP, dCTP, dGTP, 1.25 through the hydroxyapatite step. Polymerase / 3 assays were per-p~ i3H]TTP (16 Ci/mmol), 100 ,UM CTP, GTP, UTP, 1.25 mM ATP, 5.0 mm of phosphoenolpyruvate, and 160 p g / d of DNase-activated DNA. Following incubation at 37 "C for 60 min, incorporation of r3H] TTP into DNA was determined as described (17).

RESULTS
I n Situ DNA Synthesis-DNA replication and repair synthesis have been characterized in lysolecithin-treated CHO cells (12,13). T o characterize DNA synthesis in permeable H F cells, DNA synthesized in situ was density labeled with BrdUTP and analyzed on CsCl gradients (17), as described under "Materials and Methods." In the absence of DNAdamaging agents, a small but detectable level of DNA synthesis was observed in permeable GI cells. This residual DNA synthesis was shown to be replicative in nature, due to a shift to higher than normal density on CsCl gradients (Fig. lA).
Only -35% of the DNA synthesized in situ banded with normal density DNA ( Fig. k4). The replicative synthesis observed in GI-HF and -CHO (13) cells is attributed to a small number of cells which are not growth-arrested.
Bleomycin and neocarzinostatin, which release bases and break DNA strands (18)(19)(20), induced DNA synthesis when incubated directly with permeable cells. Maximum DNA synthesis was obtained with 2 p g / d of bleomycin or 50 pg/ml of neocarzinostatin. Dithiothreitol ( 5 m~) was required for neo-carzinostatin activity, in agreement with other reports (19,20); however, the presence of a reducing agent did not enhance DNA synthesis induced by bleomycin (not shown). c s c l gradient analysis of BrdUTP-labeled DNA synthesized in permeable HF cells in response to bleomycin (Fig. 1B) or neocarzinostatin (not shown) demonstrated a lack in shift from normal density, which is'typical of repair synthesis (17).
MNNG and NMU, which methylate specific bases (21,22), and UV irradiation, which creates pyrimidine dimers (2, 23), induced DNA synthesis when growth-arrested H F cells were exposed to these agents prior to permeabilization. Maximum DNA synthesis was obtained by treating GI cells with 50 ~L M MNNG or 10 mM NMU in growth medium for 1 or 2 h, respectively, at 37 "C. For UV irradiation, H F cells were rinsed with phosphate-buffered saline and exposed to UV light at 200 microwatts/cm2 for 1 min. Following treatment with the alkylating agents or UV irradiation, cells were immediately collected, permeabilized, and assayed for DNA  in situ was conducted with BrdUTP in place of TTP and E3H]dATP was the radioactive deoxynucleotide. Following incubation at 37 "C for 30 min, permeable HF cells were lysed and DNA was isolated, sheared, denatured, and centrifuged to equilibrium on CsCl gradients as described (15). Centrifugation was in a Ty 75 rotor at 45,000 rpm for 48 h, 25 "C. Fractions were collected from the top and acid precipitated on GF/A filters  Table I. Repair synthesis induced by NMU and MNNG in CHO and HF cells is inhibited to the same extent by DNA polymerase inhibitors (13).3 To simplify Figs. 2-5, the data obtained with NMU is presented in CHO cells, and data obtained with MNNG is presented in HF cells. In CHO cells, repair synthesis induced by bleomycin was inhibited 50% by -15 ~L M ddTTP, whereas -900 p~ ddTTP was required to produce 50% inhibition of repair synthesis induced by methylating agents and of replicative synthesis ( Fig. 2A and Table I). Inhibition of DNA polymerase p and of bleomycin repair synthesis in CHO cells exhibited very similar sensitivities to inhibition by ddTTP. In HF cells, repair synthesis induced by bleomycin and neocarzinostatin were inhibited 50% by -16 and 8 p~ ddTTP, respectively; however, repair synthesis induced by methylating agents and UV irradiation required much higher concentrations of ddTTP (-900 p~) to attain 50% inhibition. These studies demonstrate that DNA repair synthesis induced by bleomycin or neocarzinostaiin is much more sensitive to ddTTP inhibition than is repair synthesis induced by NMU, MNNG, or UV irradiation.
The effect of aphidicolin on DNA synthesis in situ is shown in Fig. 3 and Table I. In both CHO and HF cells, replication was much more sensitive to aphidicolin than was repair synthesis induced by any agent. At a l l concentrations of aphidi-Colin tested (1-20 ,ug/ml), purified DNA polymerase , ! ? was unaffected.
As observed with aphidicolin (Fig. 3), DNA replication in both CHO and HF cells was inhibited by much lower concentrations of araCTP (50% inhibition at 235 p~ araCTP) than was repair synthesis induced by any of the DNA-damaging a M. R. Miller and D. N. Chinault, unpublished information.

TABLE I
Concentration of DNA polymerase inhibitors which decrease DNA synthesis 50% Using the data presented in Figs. 2-5, the concentrations of ddTTP, aphidicolin, araCTP, and NEM which decrease replication or repair synthesis induced by different agents 50% in CHO and HF cells are indicated below. The concentration of inhibitors which reduce the activity of purified hamster liver DNA polymerase p 50% is also indicated. N.E. indicates no effect, and * indicates the highest concentration of inhibitor tested reduced DNA synthesis 30%. DNA  Hamster Human agents tested (Fig. 4 and Table I). Purified DNA polymerase ,!? was significantly inhibited by high concentrations of araCTP; 1.1 mM araCTP inhibited polymerase p ~3 0 % (Fig.   a).
The effect of NEM on DNA synthesis in permeable CHO and HF cells is shown in Fig. 5 and Table I. In both cell lines, bleomycin-induced repair synthesis was least sensitive to NEM; 50% inhibition occurred at -600 p~ NEM. Replication in both cells was inhibited 50% by -290 NEM. In CHO cells, NMU repair synthesis was inhibited 50% at -90 p~ NEM, and in HF cells, UV irradiation and MNNG-induced repair synthesis were inhibited 50% at =80 and 170 p~ NEM, respectively. The ability of neocarzinostatin to damage DNA was dependent on reducing agents and was strongly inhibited by NEM; therefore, the sensitivity of neocarzinostatin-induced repair synthesis to NEM was not reported. Amount of DNA Damage-The possibility that the amount of DNA damage influenced the extent to which the different DNA polymerases participated in repair synthesis was investigated in HF cells. HI? cells were treated with different concentrations of MNNG and bleomycin and the amount of DNA damage was estimated. Bleomycin directly releases bases and breaks DNA strands (18,24) and the amount of DNA damage generated by bleomycin can be quantitated by determining the average size of DNA on alkaline sucrose gradients. MNNG methylates the following DNA bases; guanine at N-7, adenine at N-3, and guanine at 0-6, in the relative proportions 801423, respectively (21,22). The N-methylated DNA bases are heat labile and can be selectively and quantitatively released by heating DNA at 45 "C for 18 h, resulting in apurinic sites (22). Subsequent treatment of the DNA with alkali results in specifically cleaving the DNA at apurinic sites. Following MNNG treatment, the amount of DNA damage can, therefore, be estimated by heating the DNA at 45 "C for 18 h, then determining the average size of the DNA on alkaline sucrose gradients. HF cells, prelabeled with [3H]TdR, were treated with 0.05 and 0.25 rn MNNG or with 0.75 and 5 p g / d of bleomycin. The DNA from control, MNNG-, and bleomycin-treated cells was isolated as described (17), which involved heating at 45 "C, for 18 h, and analyzed on alkaline sucrose gradients.    Table I1 shows the effect of DNA polymerase inhibitors on repair synthesis induced by the different concentrations of bleomycin and MNNG. The range in amount of DNA repair synthesis induced by different concentrations of DNA-damaging agents in different experiments is also presented in Table 11. Although different concentrations of bleomycin and MNNG altered the amount of DNA damage (Fig. 6), the ability of DNA polymerase inhibitors to decrease repair synthesis was not significantly affected by the amount of DNA damage (Table TI). The values for per cent inhibition of repair synthesis by polymerase inhibitors at 0.75 and 5 pg/ml of bleomycin and at 0.05 and 0.25 nm MNNG (Table 11)

DISCUSSION
Although the DNA polymerase inhibitors used in this study are not absolutely specific, the differential effect of the inhibitors on DNA repair synthesis induced by different agents helps clarify the role of polymerases a, b, and y in DNA repair synthesis. A primary role of DNA polymerase y in repair synthesis would be indicated by inhibition of repair synthesis by relatively low concentrations of ddTTP as well as NEM; however, such inhibition was not observed in repair synthesis induced by any of the agents studied (Table I). Polymerase y, therefore, appears to play a minor, if any, role in DNA repair synthesis in mammalian cells. Repair synthesis induced by MNNG, NMU, or UV irradiation was relatively insensitive to ddTTP, whereas repair induced by bleomycin or neocarzinostatin was inhibited by much lower concentrations of ddTTP (Fig. 2). This observation suggests that different DNA polymerases are involved in repairing different types of DNA damage. The similarity of ddTTP inhibition of neocarzinostatin and bleomycin repair to the ddTTP inhibition of purified polymerase / 3 (Fig. 2) indicated that polymerase /3 was primarily responsible for repair synthesis induced by these two agents. Although very high concentrations of araCTP inhibited repair synthesis induced by bleomycin and neocarzinostatin, similar conentrations of araCTP also inhibited polymerase /3 (Fig. 4). It is difficult to explain why repair synthesis induced by neocarzinostatin and bleomycin was inhibited by high levels of aphidicolin, but purified polymerase p was not inhibited by aphidicolin (Fig. 3). This observation may indicate that: 1) polymerase /3 is required for one step in bleomycin or neocarzinostatin repair synthesis while polymerase a is required for a different step, 2) polymerase p alone resynthesizes most of the DNA areas damaged by bleomycin and neocarzinostatin, and polymerase a alone resynthesizes other areas or 3) in addition to polymerase a, aphidicolin interacts with another cellular protein(s) in such a manner to reduce polymerase p activity in situ. Indeed, decreased sensitivity of polymerase a to aphidicolin during purifkation has been interpreted as loss of a polymerase a accessory protein which interacts with aphidicolin (25,26).
Repair synthesis induced by MNNG, NMU, and UV irradiation were inhibited only by high concentrations of ddTTP, which also inhibits polymerase a (Fig. l). 4 The lack of sensitivity of repair synthesis induced by methylating agents and UV irradiation to ddTTP indicated that neither polymerase p nor (Y participated in these repair synthesis processes. However, MNNG-, NMU-, and UV-induced repair were much less sensitive to both aphidicolin and araCTP than was replication. A strong correlation between sensitivity of polymerase a and sensitivity of repair synthesis induced by MNNG, NMU, and UV irradiation to polymerase a inhibitors has not been established, Thus, while polymerase p appears to be involved in repair of bleomycin or neocarzinostatin damage, we can only conclude that a ddTTP-insensitive polymerase is responsible for repairing MNNG, NMU, and UV irradiation damage. Our results may indicate that different forms of DNA polymerase a, (28,29), with different sensitivities to aphidicolin and araCTP, participate in DNA replication and repair synthesis or that a novel polymerase is involved in MNNG-, NMU-, and UV-induced repair synthesis.
Interpretation of NEM inhibition of DNA synthesis in situ is difticult since, in addition to polymerases a and y, any other proteins requiring a sulfhydryl group for activity would be inactivated. Nonetheless, repair synthesis induced by MNNG, NMU, and UV irradiation were much more sensitive to NEM inhibition than was repair synthesis induced by bleomycin (Fig. 5).
Our results explain conflicting reports in the literature (5)(6)(7)(8)(9)(10)(11). The similar sensitivity of repair synthesis in rodent (CHO) and human (HF) cells to DNA polymerase inhibitors (Figs. [2][3][4][5] indicates that differences in cell species was not the origin of the conflicting reports. Differences in salt concentrations employed in situ has also been shown not to alter the extent to which polymerase inhibitors reduce repair synthesis (13). The extent of involvement of different polymerases in DNA repair synthesis appears to be related to the agent used to damage DNA. The only exception to our results, of which we are aware, is a report by Hubscher et al. ( 5 ) , indicating I. Goldberg, unpublished results. polymerase / 3 was responsible for UV-induced repair synthesis in human neuronal nuclei. These cells did not contain polymerase (Y and may be viewed as atypical or special cells. At least three factors may be involved in determining which polymerase participates in repair synthesis 1) the amount of DNA damage, 2) the "patch size" of repaired DNA and 3) the actual type of damage being repaired. Data in this report ( Fig.  6 and Table 11) indicate that the amount of DNA damage does not significantly alter the involvement of polymerase , 8 or the ddTTP-insensitive polymerase. The size of DNA resynthesized after damage by some agents, such as X irradiation, is relatively small (1-5 nucleotides) and is termed "short patch," whereas longer areas of DNA (-100 nucleotides) are synthesized in response to agents such as UV irradiation (23) and alkylating agents (22). Because DNA polymerase a requires a gap ranging from -25-50 nucleotides to initiate synthesis, while polymerase /3 is active on DNA with smaller gaps, Grossman (30) and Cleaver (31) have hypothesized that the patch size may influence which polymerase participates in repair synthesis. We offer an alternative explanation of our findings. During repair of many types of DNA damage, incision processes create 3' and 5' termini on damaged DNA strands which are susceptible to typical repair nucleases. The repair nucleases create gaps ("short" or "long") which are acted on by a ddTTP-insensitive polymerase. On the other hand, bleomycin and neocarzinostatin create breaks in DNA which may not be susceptible to the same repair nucleases. After bleomycin treatment, the 3' end of the broken strand is "blocked" by a CH2-CH-COOH group (26), and following neocarzinostatin treatment, the 5' end appears to be "blocked" by a residual sugar moiety (32). Such blocked termini may not be substrates for typical repair nucleases. Repair of damaged DNA containing blocked termini may require excision of blocked termini by a polymerase / 3 associated nuclease, followed by resynthesis primarily, but not necessarily exclusively, with polymerase p. An example of such a nuclease is DNase V, described by Mosbaugh and Meyer (33). Studies are in progress to test this hypothesis.
Another factor which may contribute to our results is that bleomycin-and neocarzinostatin-induced repair synthesis were initiated in permeable cells, whereas MNNG, NMU, and UV repair synthesis were initiated in intact cells. Preparation of in situ cell systems may alter initiation of normal repair processes in some fashion, causing DNA polymerases to participate differently in repair synthesis. It will be imperative to investigate this possibility to determine whether in situ SYS-tems are valid models for studying DNA repair synthesis.