Cellular Responses and Repair of Single-strand Breaks Introduced by UV Damage Endonuclease in Mammalian Cells*

Although single-strand breaks (SSBs) occur frequently, the cellular responses and repair of SSB are not well understood. To address this, we established mammalian cell lines expressing Neurospora crassa UV damage endonuclease (UVDE), which introduces a SSB with a 3′-OH immediately 5′ to UV-induced cyclobutane pyrimidine dimers or 6–4 photoproducts and initiates an alternative excision repair process. Xeroderma pigmentosum group A cells expressing UVDE show UV resistance of almost the wild-type level. In these cells SSBs are produced upon UV irradiation and then efficiently repaired. The repair patch size is about seven nucleotides, and repair synthesis is decreased to 30% by aphidicolin, suggesting the involvement of a DNA polymerase δ/ε-dependent long-patch repair. Immediately after UV irradiation, cellular proteins are poly(ADP-ribosyl)ated. The UV resistance of the cells is decreased in the presence of 3-aminobenzamide, an inhibitor of poly(ADP-ribose) polymerase. Expression of UVDE in XRCC1-defective EM9, a Chinese hamster ovary cell line, greatly sensitizes the host cells to UV, and addition of 3-aminobenzamide results in almost no further sensitization of the cells to UV. Thus, we show that XRCC1 and PARP are involved in the same pathway for the repair of SSBs.

DNA single-strand breaks (SSBs) 1 are frequently produced by environmental genotoxic agents and by endogenous cellular reactions. SSBs cause double-strand breaks when replication forks encounter SSBs and, thus, result in chromosomal rearrangements and instability (1). Despite the potentially harmful effects of SSBs, however, little is known about the details of the repair mechanisms and cellular responses to SSBs in mammalian cells. This may be due to the experimental difficulty to produce SSBs alone. Genotoxic agents that produce SSBs (ionizing radiation, oxidizing agents, and alkylating agents) generate a variety of DNA lesions (2). For instance, ionizing radiation and bleomycin produce not only SSBs but also base lesions and double-strand breaks (2,3).
One of the immediate responses to SSBs in mammalian cells is thought to be the activation of poly(ADP-ribose) polymerase (PARP). PARP binds to SSBs and is activated (4). Although PARP has been considered to be involved in the repair of SSBs, especially in replicating cells (4), its precise role is still not well understood. Another player in the response to SSBs may be XRCC1, which binds to DNA ligase III and DNA polymerase ␤ (5), and is thought to be involved in a base excision repair (BER) pathway. Recently it was reported that XRCC1 is also involved in an S-phase-specific repair pathway of SSBs (6).
Since XRCC1 binds to PARP (7) and some phenotypic characteristics of XRCC1-deficient cells are similar to those of PARPdeficient cells (8), both proteins may be involved in the S phase-specific mode of SSB repair. However, the functional relationship between PARP and XRCC1 is unknown and remains to be elucidated. In addition to nucleotide excision repair (NER) for UV-induced DNA damage, the filamentous fungus, Neurospora crassa, and the fission yeast Schizosaccharomyces pombe, possess an alternative excision repair mechanism, which is initiated by an endonuclease called UV damage endonuclease (UVDE) and is referred to as UVDE-initiated excision repair (9 -14). UVDE introduces a SSB immediately 5Ј to UV-induced cyclobutane pyrimidine dimers (CPDs) and 6 -4 photoproducts, leaving 3Ј-hydroxyl and 5Ј-phosphoryl groups at the site of cleavage (9 -11). Until now UVDE has been found only in some eukaryotic microorganisms and in some bacteria including Bacilus subtilis (9) and Dienococcus radiodurans (15), but neither a similar enzymatic activity nor any homologous genes have been found in mammalian cells.
To understand the cellular responses and repair of SSBs in mammalian cells, we have made use of UVDE. We introduced the N. crassa UVDE gene into a human and a Chinese hamster ovary (CHO) cell lines and analyzed the responses of the transfected cells to UV. Using these unique systems, we found the following. 1) Judging from the UV resistance of xeroderma pigmentosum group A (XPA) cell line expressing UVDE, UVDE-initiated alternative excision repair in human cells works almost as efficiently as NER.
2) The UVDE-initiated repair is mediated mainly by aphidicolin-sensitive DNA polymerase(s), and the repair patch size is about seven nucleotides. 3) XRCC1 and PARP cooperate and contribute to cell survival after SSBs are produced.

EXPERIMENTAL PROCEDURES
Cell Lines, Vectors, and Transfection-A human cell line derived from an XPA patient, XP12ROSV, was obtained from Dr. K. Tanaka (Osaka University) and used as the host cell for complementation of UV sensitivity by the introduced UVDE gene. The CHO cell line EM9 was purchased from the American Type Culture Collection. Plasmid pCY4B, a derivative of the one described by Niwa et al. (16), contains a chicken ␤-actin promoter (obtained from Dr J. Miyazaki, Osaka University). This plasmid was used for expression of the N. crassa UVDE gene in XPA cells and EM9 cells. pCY4B-UVDE was made by inserting the Immunoblot Analysis-Whole cell extracts were prepared from cultured cells by homogenizing cell pellets in a lysis buffer (50 mM Tris-HCl (pH 7.5), 0.3 M KCl, 0.05% Nonidet P-40, 2 mM dithiothreitol) containing protease inhibitors (protease inhibitor mixture tablets, Roche Molecular Biochemicals). After centrifugation of the homogenate at 100,000 ϫ g, supernatants were recovered and used for immunoblot analysis as well as for the incision assay described below. Western analysis was done by standard methods using whole cell extracts (20 g of protein of each), fluorotrans membrane (PALL Gelman Laboratory), and an antibody raised against N. crassa UVDE (18) (1:500 dilution). Immune complexes were detected by using an ECL plus Western blotting detection system (Amersham Pharmacia Biotech).
For detection of poly(ADP-ribose) (pADPr), cells were harvested from 35-mm culture dishes just before reaching confluence. Cells were washed with phosphate-buffered saline (PBS) and irradiated with 2.5, 7.5, and 20 J/m 2 UV. After UV irradiation, cells were incubated for various periods of time at room temperature. After incubation, cell extracts were prepared by adding 400 l of SDS-polyacrylamide gel electrophoresis sample buffer (1% SDS, 1% ␤-mercaptoethanol, 5% glycerol, 25 mM Tris-HCl (pH 6.5), and 0.05% bromphenol blue) to each dish. In another experiment, cells were preincubated in medium containing 2 mM 3-aminobenzamide (3AB, Sigma) for 2 h before UV irradiation. The cells were then washed with PBS containing 2 mM 3AB, irradiated with UV, incubated, and lysed as above. The extracts thus obtained were centrifuged at 15,000 rpm (18,000 ϫ g) for 5 min at 4°C. The supernatants were recovered after centrifugation, resolved by 8% SDS-polyacrylamide gel, and transferred to the fluorotrans membrane. To detect pADPr bound to proteins, the membrane was probed with monoclonal antibodies to pADPr (1:500 dilution; Trevigen, Inc.) and peroxidase-labeled goat antibodies to mouse IgG (Kirkegaard & Perry Laboratories, Inc.). We used actin as the control (monoclonal antibody to actin, clone C4, Roche Molecular Biochemicals).
In Vitro Incision Assay-Nicking activity of UVDE was measured as described (18). The whole cell extracts (40 g of protein of each), and the synthetic oligonucleotides containing CPD were used.
UV Survival-5 ϫ 10 3 exponentially growing XPA cells were plated per 100-mm culture dish (9.3 ϫ 10 2 cells per 60-mm dish for CHO cells) and incubated in culture medium. 9 h after plating, cells were washed with Hanks' solution (Nissui) and irradiated with UV at various doses. The cells were then cultured for 12 days (8 days for CHO cells) in either the regular medium or medium containing 2 mM 3AB.
Measurement of Repair Synthesis Using Autoradiography-Cells grown on glass microscope slides in culture dishes were irradiated with 7.5 J/m 2 and 15 J/m 2 UV and incubated at 37°C for 1 h in medium supplemented with 10 Ci/ml of [ 3 H]dThd (Amersham Pharmacia Biotech). The dishes were then washed with cold PBS. The cells on the microscope slides were fixed with cold 5% trichloroacetic acid, washed with ethanol, and dried at room temperature. The slides were dipped in NR-M2 emulsion (Konica, diluted 1:1 with H 2 O), dried, and exposed for 7 days at 4°C in a light-tight plastic box. The slides were then developed and counter-stained with 3% Giemsa stain (Merck). About 150 nuclei with fewer than 100 silver grains were counted for each slide. The results were shown as a percentage histogram for each slide. Repair synthesis was taken to be the difference between the percentage values of corresponding irradiated and non-irradiated samples.
Measurement of DNA Repair Synthesis by BrdUrd-induced Density Shift-DNA repair synthesis was measured essentially as described by Smith et al. (19). XPA[UVDE] cells were grown in 150-mm culture dishes in medium containing 0.4 Ci/ml [ 32 P]orthophosphate (Amersham Pharmacia Biotech) for 5 days and then subcultured in nonradioactive medium for 2 days before to UV irradiation. Then the cells were incubated for 2 h in the medium with FdUrd (1 M, Sigma), BrdUrd (10 M, Sigma), and hydroxyurea (2.5 mM, Sigma), washed twice with PBS, and irradiated with 20 J/m 2 and 40 J/m 2 UV. Cells were then incubated for 3 h in medium containing FdUrd, BrdUrd, hydroxyurea, and 5 Ci/ml [ 3 H]dThd supplemented or not with 10 g/ml aphidicolin. Cells were then washed with PBS and lysed as described by Smith et al. (19). DNA solutions thus obtained were subjected to neutral CsCl gradient sedimentation and fractionated, then assayed for radioactivity. The fractions containing unreplicated (parental-density) DNA were pooled and used for alkaline CsCl gradient sedimentation. Repair synthesis was taken to be specific incorporation ( 3 H/ 32 P) in the parentaldensity DNA of the alkaline rebanding. The ratio of repair synthesis sensitive to aphidicolin was calculated as (repair synthesis in the presence of aphidicolin/repair synthesis in the absence of aphidicolin).
Analysis of Patch Size by BrdUrd-induced Density Shift-Repair patch size was also measured as described by Smith et al. (19). XPA[U-VDE] cells were prelabeled with 32 P, incubated in medium containing FdUrd, BrdUrd, and hydroxyurea for 2 h, and irradiated with 20 J/m 2 UV as described above. Cells were then incubated in medium containing FdUrd, BrdUrd, hydroxyurea, and [ 3 H]dThd for 3 h, then lysed. The parental-density DNA was purified by two successive neutral CsCl gradient sedimentation processes. Two kinds of DNA were prepared for use as markers as follows. 32 P-Prelabeled DNA was prepared from cells immediately before irradiation with UV. Fully BrUra-substituted hybrid DNA was prepared from unirradiated cells that were incubated for 3 h in medium containing FdUrd, BrdUrd, and [ 3 H]dThd.
The isolated parental-density DNA was sonicated. The size distribution of the fragments of the sonicated DNA was determined by electrophoresis in a denaturing polyacrylamide gel as described (20). From the data, the number-average molecular size of the fragments was calculated as described (21). The size of the repair patches was measured in alkaline CsCl gradients using the parental-density DNA sonicated to a number-average molecular size of 188 nucleotides. The gradients were fractionated, and the radioactivity of each fraction was measured. The repair patch size was calculated as follows. First, the distance between the 32 P-prelabeled DNA distribution and that of the 3 H-repair label distribution was measured. This distance was then compared with the separation between the peak of the 32 P-prelabeled DNA and that of the 3 H-labeled fully BrUra-substituted DNA, which was determined from a separate analysis of DNA markers in a similar gradient. The ratio of these two distances was then multiplied by the average fragment size to give the average repair patch size.
In another experiment, [ 3 H]BrdUrd (Moravek) was used as the isotopic label in place of [ 3 H]dThd. Hydroxyurea was not used. The number-average molecular size of the sonicated parental-density DNA was 211 nucleotides in this case, and the experiment was performed in the same way as above.
Alkaline Gel Analysis-The relative number of SSBs in the genomic DNA from cells collected at various periods of time after UV irradiation was determined by using alkaline gel analysis. Briefly, cells were prelabeled with [ 32 P]orthophosphate, washed with Hanks' solution, and irradiated or unirradiated with 20 J/m 2 UV. Cells were incubated in medium for various periods at 37°C. In a separate experimental series, measurements over a smaller time scale from 0.5 min to 40 min were done. In this case, after UV irradiation, cells were incubated in Hanks' solution at room temperature. At appropriate periods of time, cells were lysed by incubating in 0.5% SDS, 100 g/ml proteinase K (Wako), 10 mM Tris, 1 mM EDTA (pH 8) at 37°C overnight. Genomic DNA was isolated by phenol/chloroform extraction and ethanol precipitation. Each 5-g DNA sample was mixed with alkaline loading buffer (22) and electrophoresed in a 3.5% alkaline-agarose (Agarose H, Wako) gel. A set of 5Ј-end 32 P-labeled DNA fragments (Marker 8GT; Nippon Gene) was prepared by using T4 polynucleotide kinase (Takara) and [␥-32 P]ATP (NEN™ Life Science Products, Inc.). These size markers were electrophoresed in an alkaline gel containing the DNA sample. The gel was dried and analyzed using FLA-2000 (Fujifilm). As a measure of the relative amount SSBs, we took the following ratio: the radioactivity from the 5.6-to14.3-kilobase area/the total radioactivity.  (Fig. 1A). The nicking activity of the extract prepared from XPA[UVDE] cells to UV damage is shown in Fig. 1C. The extract introduced an incision immediately 5Ј to the CPD, as judged by the decrease of a 49-mer band and concomitant strong appearance of the 20-mer band. This incision activity is the same as previously reported for recombinant UVDE (18). These data indicated that XPA[UVDE] cells express UVDE and retain its nicking activity.

Establishment of XPA Cell Line
Survival of XPA Transfectants-The colony-forming ability of the XPA transfectants after UV irradiation was assessed. XPA[Vector] cells were extremely sensitive to UV irradiation, whereas XPA[cXPA] cells exhibited UV resistance (Fig. 2). XPA[UVDE] cells showed almost the same level of UV resistance as XPA[cXPA] cells at low UV doses (Fig. 2). As the UV dose increased, XPA[cXPA] cells became more UV-resistant than XPA[UVDE] cells (Fig. 2). Thus, the alternative excision repair found in eukaryotic microorganisms provided NER-deficient human cells with UV resistance of almost wild-type level. Now the question is how UV-induced DNA damage is repaired in these cells.
SSBs in Genomic DNA of XPA[UVDE] Cell after UV Irradiation-To examine whether SSBs are actually produced after UV irradiation in XPA[UVDE] cells, we conducted alkaline gel electrophoresis analysis of the genomic DNA of UV-irradiated XPA[UVDE] cells. After 20 J/m 2 UV irradiation, cells were incubated in buffer for various periods of time before genomic DNA was isolated and electrophoresed on an alkaline-agarose gel. Unirradiated DNA migrated as a discrete band near the origin, whereas the DNA isolated after UV irradiation showed a broad smeared band on the gel (Fig. 3A). These results indicate that the SSBs are actually produced by UVDE in intact XPA[UVDE] cells immediately after UV irradiation. To quantify the SSBs, we measured the amount of DNA between 5.6-and 14.3-kilobase DNA. The amount of smeared DNA gradually increased up to 40 min after UV irradiation and reached a plateau level (Fig. 3A). Two hours after UV irradiation, the amount of the SSBs was significantly decreased (Fig.  3B), indicating repair of SSBs. Under the conditions used, no significant difference in the extent of the smear was observed in XPA[Vector] and XPA[cXPA] cells (Fig. 3B). These data suggest that in XPA[UVDE], SSBs are produced by UVDE, which initiates an alternative excision repair in human cells.
Repair  (Fig. 4). The unscheduled DNA synthesis in XPA[UVDE] was slightly less than unscheduled DNA synthesis determined in HeLa cells and increased with UV doses (Fig. 4). We next examined the sensitivity of the repair synthesis to aphidicolin, a specific inhibitor of DNA polymerase ␣, ␦, and ⑀. We measured repair synthesis using the BrdUrd density shift technique (see "Experimental Procedures"). Exponentially growing cells were exposed to 20 J/m 2 and 40 J/m 2 UV. Hydroxyurea was added to the growth medium to reduce the level of semi-conservative DNA synthesis. At both doses used, most of UV-induced repair synthesis was aphidicolin-sensitive (Table I). These data suggest that repair synthesis is mediated mainly by aphidicolin-sensitive DNA polymerase(s): presumably by DNA polymerase ␦ and/or ⑀ (see "Discussion").

Estimation of Repair Patch Size in XPA[UVDE]
Cells-The measurement of repair patch size is an extension of the method used to measure the repair synthesis. 32 P-Prelabeled XPA[U-VDE] cells were irradiated with 20 J/m 2 UV and incubated in medium containing [ 3 H]dThd, hydroxyurea, and BrdUrd for 3 h. Parental-density DNA was isolated by two successive processes of neutral CsCl gradient sedimentation. This DNA was sonicated to an average size of 188 nucleotides and then centrifuged to equilibrium in alkaline CsCl gradients. Under these conditions, the increase in density of DNA fragments that contain repair patches (synthesized in the presence of BrdUrd) is large enough to be measured and can be compared with the increase in density of DNA completely substituted with BrUra. Gradients were fractionated, and the radioactivity profiles of 3 H and 32 P were determined (Fig. 5). The density of the DNA molecules containing repair patches (shown by the profile of 3 H) was clearly larger than that of bulk genomic DNA (shown by the profile of 32 P) (Fig. 5). Based on the shift between the profiles of 3 H and 32 P and referring to the position of fully BrUra-substituted DNA, the patch size was determined as 8 Ϯ 2 nucleotides. In the second experiment we did not add hydroxyurea, and [ 3 H]BrdUrd was used as the isotopic label. In this case, the patch size was determined as 7 Ϯ 2 nucleotides (radioactivity profiles were not shown).
Involvement of PARP Activation in the UV Resistance of XPA[UVDE]-We further investigated whether PARP and XRCC1 are involved in the repair process. In the presence of 3AB, a widely used inhibitor of PARP, enhanced lethality after UV irradiation was observed in XPA[UVDE] cells (Fig. 2). By contrast, in XPA[Vector] cells and XPA[cXPA] cells, no significant increase in sensitivity to UV was observed in the presence of 3AB (Fig. 2). These results demonstrate the involvement of PARP in the repair of the SSBs introduced by UVDE. By immunoblot analysis with monoclonal antibody to pADPr, we examined whether the activation of PARP occurs in XPA[U-VDE] in response to UVDE-induced SSBs. Thirty seconds after 20 J/m 2 UV irradiation in XPA[UVDE] cells, a significant amount of pADPr was synthesized (Fig. 6). This is consistent with the result that the SSBs had already been introduced in

FIG. 3. SSB frequencies of various XPA transfectants irradiated with UV and incubated for various time intervals.
A, top, SSB frequencies at short periods. XPA[UVDE] cells prelabeled with 32 P were either irradiated with 20 J/m 2 UV or unirradiated. They were incubated for various periods of time in buffer at room temperature and then lysed for collection of the genomic DNA. The samples (5 g of each) of purified DNA were electrophoresed through a 3.5% alkaline agarose gel. The gel was dried, and the radioactivity of 32 P was visualized as described under "Experimental Procedures." C stands for the unirradiated control. Bottom, the ratio (radioactivity from 5.6-to 14.3-kilobase (kb) area/total radioactivity) is shown on the histogram as a measure of the relative amount of SSBs.  cells is dependent on UV dose, and 7.5 J/m 2 irradiation was necessary to detect pADPr synthesis in our assay (Fig. 6B).
Cell Survival of EM9 Transfectants-Since XRCC1 is thought to be involved in the processing of SSBs, we investigated whether XRCC1 is actually necessary for repair of SSBs introduced by UVDE. The Neurospora UVDE gene was introduced into the CHO cell line EM9, which is mutated in the XRCC1 gene (23). The obtained transfectant was designated as EM9 [UVDE]. Anti-Neurospora UVDE antibody detected UVDE expression in EM9[UVDE] cells (data not shown). The colony-forming ability of the CHO transfectants after UV irradiation was assessed. EM9[UVDE] cells were much more sensitive to UV than EM9[Vector] cells (Fig. 7). This indicates the involvement of XRCC1 in the repair of UVDE-introduced SSBs.
In the presence of 3AB, only a very slight increase in sensitivity to UV was observed in EM9[UVDE] cells (Fig. 7). Thus, the inhibition of PARP activation does not influence the survival of EM9 cells, which lack active XRCC1 protein. DISCUSSION We established mammalian cell lines, which enabled us to examine the repair characteristics and cellular responses of SSBs produced by a foreign UV endonuclease, UVDE. UVDE introduces a nick immediately 5Ј to UV-induced CPDs and 6 -4 photoproducts and initiates an alternative excision repair in several eukaryotic and prokaryotic microorganisms. The first interesting question was how much of the UV sensitivity of NER-deficient human cells is complemented by the expressed UVDE. We previously found that UVDE-initiated repair is a rapid global genome repair and is effective for most UV-induced DNA damage before NER occurs in S. pombe cells (24). As shown in Fig. 2, NER-deficient human host cells acquired the UV resistance of the wild-type level. This is the first example of an extensive complementation of UV sensitivity in NERdeficient human cells by a foreign repair protein.
As previously reported, pyrimidine dimer DNA glycosylase only partially complements the UV sensitivity of NER-deficient human cells (25), far less than UVDE shown here. This different efficiency of complementation between UVDE and pyrimidine dimer DNA glycosylase is explained by the different structures of SSBs produced by the endonucleases. UVDE produces 3Ј-OH ends, which are common intermediates during DNA replication, repair, and recombination processes and constitute appropriate primer terminus for DNA polymerases. However, 3Ј-unsaturated aldehyde termini produced by pyrimidine dimer DNA glycosylases have to be removed before repair synthesis. Another explanation is that pyrimidine dimer DNA glycosylase repairs only CPDs, whereas UVDE can repair both CPDs and 6 -4 photoproducts (14). Only at higher UV doses were XPA[cXPA] cells more UV-resistant than XPA [UVDE]. This difference may be due to the lack of transcription coupling in UVDE-initiated repair or due to the increase of UV-induced DNA lesions other than CPDs or 6 -4 photoproducts, which are not recognized by UVDE but repaired by NER. This may also be due to the incomplete processing of the SSBs excessively produced by UVDE during the short time span. We showed that a large number of SSBs are introduced in XPA[UVDE] after irradiation with a high dose of UV within half a minute, and considerable SSBs remain to be repaired even 2 h after irradiation (Fig. 3).
The next interesting question is how the UVDE-initiated repair proceeds in human cells. The repair synthesis was shown to be mostly dependent on aphidicolin-sensitive DNA polymerase, and the determined patch size of the repair was about seven nucleotides. Since the repair patch size of the proliferating cell nuclear antigen-dependent pathway of BER has been reported as between 7 and 14 nucleotides (26), or less than 10 nucleotides in length (27,28), the patch size for UVDEinitiated excision repair fits in reasonably well with that of the BER pathway. It has also been reported that the repair synthesis of the proliferating cell nuclear antigen-dependent pathway is not catalyzed by DNA polymerase ␣ (29) and is catalyzed by DNA polymerase ␦ or ⑀ (27,28,29). By in vitro assays with purified recombinant proteins, we and other groups (30,31) show that the SSBs produced by UVDE became substrates for cleavage by FEN1 (flap endonuclease 1), which has already been shown to be a factor involved in the proliferating cell nuclear antigen-dependent BER pathway (27,28,32,33). Thus, the SSBs produced by UVDE in XPA[UVDE] cells may be processed by DNA polymerase ␦ and/or ⑀ and components that are common with long patch repair pathway of BER.
The third interesting question about UVDE-initiated repair of UV damage concerns the cellular responses to the induced SSBs in mammalian cells. Western blot analysis showed that immediately after irradiation with a high dose of UV, cellular proteins were poly(ADP-ribosyl)ated in XPA[UVDE] cells (Fig.  6). This is in contrast to the response in XPA[cXPA] cells, which showed no significant synthesis of pADPr (Fig. 6A). By adding 3AB, a competitive inhibitor of PARP, to XPA[UVDE] cells, pADPr synthesis was suppressed (Fig. 6A). These results give additional clear evidence for PARP activation by SSBs in human cells. 10 min after UV irradiation, pADPr was no longer observed (Fig. 6A). This is explained by reports that, after excessive activation of PARP, pADPr has a short half-life close to 1 min (34), and the levels of NAD, which is a substrate of PARP, are depleted (4).
We have shown here that 3AB enhanced the UV lethality of XPA[UVDE], whereas 3AB did not make any significant differ- ence to survival in XPA[cXPA] cells, indicating the involvement of PARP in the repair of the SSBs. These results are consistent with reports that cells treated with alkylating agents and xrays, which are known to produce SSBs in cells, are sensitive to 3AB (4). The lethal effect of 3AB on cells treated with these agents is known to be maximal in S phase (35,36), suggesting that PARP is a survival factor playing an essential role during recovery from SSBs in S phase. PARP is known to interact directly with XRCC1 (7). It has been reported that S-phasespecific repair of SSBs mediated by XRCC1 is indispensable for resistance to alkylating agents in CHO cells (6). Therefore, to link the effect of 3AB on XRCC1, we introduced the UVDE gene into a CHO cell line, EM9, that is defective in the XRCC1 gene. We showed that EM9 cells expressing UVDE (EM9[UVDE]) are extremely sensitive to UV, indicating the involvement of XRCC1 in repair of the SSBs (Fig. 7). The addition of 3AB results in almost no increase in the UV sensitivity of EM9[UVDE] (Fig. 7). This result suggests that PARP and XRCC1 play essential roles in the same pathway, probably in the same S-phase-specific recovery pathway for SSBs. It has been shown that nuclear foci of XRCC1 co-localize with Rad51 (6). Thus, PARP and XRCC1 may function in concert with a homologous recombination pathway in the processing of SSBs as well as double-strand breaks, which are produced from SSBs during the replication process. Recent molecular and genetic analyses of repair-deficient strains from various organisms suggest that SSBs are one of the major risk factors for genome instability induced by oxidative DNA damage. We consider that UVDE-expressing cell lines offer a unique experimental system for the analysis of the cellular response to SSBs in mammalian cells.