A Role for DNA Polymerase β in Mutagenic UV Lesion Bypass*

We report here that DNA polymerase β (pol β), the base excision repair polymerase, is highly expressed in human melanoma tissues, known to be associated with UV radiation exposure. To investigate the potential role of pol β in UV-induced genetic instability, we analyzed the cellular and molecular effects of excess pol β. We firstly demonstrated that mammalian cells overexpressing pol β are resistant and hypermutagenic after UV irradiation and that replicative extracts from these cells are able to catalyze complete translesion replication of a thymine-thymine cyclobutane pyrimidine dimer (CPD). By using in vitro primer extension reactions with purified pol β, we showed that CPD as well as, to a lesser extent, the thymine-thymine pyrimidine-pyrimidone (6-4) photoproduct, were bypassed. pol β mostly incorporates the correct dATP opposite the 3′-terminus of both CPD and the (6-4) photoproduct but can also misinsert dCTP at a frequency of 32 and 26%, respectively. In the case of CPD, efficient and error-prone extension of the correct dATP was found. These data support a biological role of pol β in UV lesion bypass and suggest that deregulated pol β may enhance UV-induced genetic instability.

ever, all lesions on the genome cannot be repaired efficiently by these processes in time for DNA replication, and some types of lesions are repaired very inefficiently. To prevent cell death through arrested DNA replication at unrepaired lesions, cells have a mechanism, referred to as translesion synthesis, that allows DNA synthesis to proceed past lesions and employs specialized DNA polymerases for promoting continued nascent strand extension.
In human cells, recent genetic and biochemical studies suggest that translesion synthesis (TLS) past a CPD-TT or a (6-4)TT lesion could be facilitated by at least four DNA polymerases, pol , , , and . In the case of pol , this process appears to be efficient and largely accurate opposite a CPD (2), whereas it could be mutagenic and limited at the 3ЈT opposite a (6-4)TT (3). Overexpression of the antisense mRNA of Rev3, one of the components of pol , leads to a dramatic drop in the extent of UV-induced mutagenesis (4), thereby implicating human pol as having a pivotal role in error-prone translesion replication in normal cells. Indeed, pol can catalyze an efficient extension of nucleotides inserted opposite the 3ЈT of both CPD and (6-4)TT lesions (3,5). Another DNA polymerase, pol , shows similar properties opposite the CPD (6). In the case of pol , the in vitro incorporation of nucleotides opposite the UV lesions and subsequent bypass can be highly error-prone, but its physiological role in TLS is still controversial (5,7,8). Presumably, all these polymerases can compete for the 3Јprimer terminus at the site of a lesion, and one would predict an effect on the quantitative and qualitative mutagenesis in UV-irradiated cells expressing these enzymes differentially. For example, mutations in the POLH (XPV/RAD30A) human gene that generate a severely truncated and inactive pol protein result in the xeroderma pigmentosum variant phenotype characterized by UV-induced hypermutability (9, 10) and a strong sunlight-induced skin cancer incidence (11)(12)(13).
The study reported here indicates that pol ␤ can now be added to the list of enzymes that can perform unassisted UV lesion bypass. pol ␤ is believed to function primarily in the repair of damaged bases in normal somatic cells (14). It is a monomeric protein of 335 amino acids (39 kDa) that lacks exonuclease activities and whose enhanced expression has been demonstrated by our laboratory to result in an increased mutation frequency (15) as well as chromosome instability and tumorigenesis (16). At the transcriptional level, pol ␤ is overexpressed in many cancer cells (17). High levels of pol ␤ have also been detected at the protein level in ovarian tumors (18) as well as in prostate, breast, and colon cancer tissues where the enzyme amount was respectively 11-, 286-, and 22-fold higher as compared with adjacent normal tissues (19). Furthermore, pol ␤ level and activity are increased by 10-fold in blood sam-ples from chronic myelogenous leukemia patients and in tumor biopsies from non-small cell lung tumors. 2 The pol ␤-dependent translesion replication that we observed here differs from that of the related DNA polymerases in the efficiency as well as in the accuracy of the reaction. These data may be relevant within the tumoral cellular context where pol ␤ is up-regulated, especially in melanomas, since several analyses showed a significant positive association between cutaneous melanoma incidence and high levels of intermittent solar exposure (20 -24).
Clonogenic and Mutagenic Assays-AA8 CHO cells were maintained in MEM␣ (Invitrogen) with 10% fetal calf serum, 4 mM glutamine, and antibiotics (50 units/ml penicillin and 50 g/ml streptomycin) at 37°C in a humidified 5% CO 2 atmosphere. CHO cell lines overexpressing pol ␤ were established previously after stable transfection of pUTPol␤ plasmid (15). Control cells and cells overexpressing pol ␤ were plated in 6-well plates and allowed to attach overnight. Next, they were irradiated with a 254-nm UV-C lamp at the fluence rate of 0.5 J/m 2 /s. Colonies were fixed and stained after 6 days of postincubation, and those Ͼ50 cells were scored. For the 6-thioguanine (6-TG)-resistant tests, cells were first irradiated at 20 J/m 2 and then exposed to 20 M 6-TG-containing medium (10 6 cells/14-cm plate) to determine the number of hypoxanthine guanine phosphoribosyl transferase mutants. After 1 week, plates were stained, and colonies of Ͼ50 cells were counted. Mutant frequencies were corrected for plating efficiency and for UV cytotoxicity.
Proteins, Cells, and Substrates-Rat pol ␤ was purified in Escherichia coli as described (25). One unit of rat pol ␤ corresponds to 1 pmol of dNTP incorporated into acid-insoluble materials at 37°C in 60 min by using an activated calf thymus DNA preincubated with DNase I as a substrate. Human pol ␤ was provided by Trevigen (Gaithersburg, MD) and showed a 0.68 g/l concentration and a 4 units/l activity. Calf thymus pol ␣ and HIV-1 RT were purified as described previously (26,27). AA8 CHO Sh::pol ␤ cells and AA8 CHO Sh cells were obtained after stable transfection of pUTPol␤ and empty pUT687 vectors as reported previously (15). Briefly, pol ␤ cDNA was fused in-frame with the bacterial Sh::ble gene conferring resistance to the broad-spectral zeocin xenobiotic of the phleomycin family. 30-mer UV-modified oligonucleotides and pBS-SV oriA/B vectors were prepared as described (28).
Two-step SV40 DNA Replication Assay-pBS-SvoriA(CPD) or pBS-SvoriB(CPD) plasmids were generated as described (2) Amersham Biosciences) were added to reaction mixtures and incubated for 1 h. Reactions were quenched by adding an equal volume of "stop solution" (2% SDS, 2 mg/ml proteinase K, and 50 mM EDTA), and further incubation was done for 1 h at 55°C. DNA (0.5 g of pc-DNA II, Invitrogen, containing one BamHI site and multiple DpnI sites) was added to each sample as internal purification controls. Reaction products were purified by extraction with phenol-chloroform-isoamyl alcohol followed by ethanol precipitation. The DNA was resuspended in distilled water. The samples were then treated with BamHI and DpnI (New England Biolabs), and the restriction digests were separated on a 1% agarose gel. After ethidium bromide staining of the gel, internal control DNAs were quantified. The gel was then dried, and autoradiography was performed. Quantification analysis of the resolved radioactive bands on the gel was achieved by PhosphorImager Storm-system analysis using ImageQuant software.

Overexpression of pol ␤ in Melanoma Cells as Compared with
Normal Skin Tissues-Previously, we and others found that pol ␤ was overexpressed at the protein level in many cancer tissues as compared with normal tissues (17)(18)(19). Here, we analyzed four independent melanoma protein extracts, and we compared their pol ␤ content relative to normal skin tissues (Fig. 1). More than a 10-fold increase in pol ␤ level was observed in all the melanomas tested, whereas a slight detection of the enzyme was discernible only after a long time exposure in normal skin (data not shown). In this work, we hypothesized that excess pol ␤ in skin cells exposed to UV light may predispose these cells to initiation and/or progression into tumoral melanomas by raising the UV-induced genetic instability.
Decreased Sensitivity to UV Radiation and Enhanced Induced Mutagenesis in CHO-pol␤ Cells-To investigate whether high levels of pol ␤ can affect genetic stability after UV irradiation in mammalian cells, we examined UV sensitivity as well as UV-induced mutagenesis in two independent transfected CHO cell lines that overproduce the enzyme by 3.2-and 2.4-fold (AA8 pol ␤2::Sh cells and AA8 pol ␤3::Sh cells) (16). Firstly, we conducted clonogenic experiments after treatment with increasing doses of UV-C irradiation concomitantly with the isogenic control AA8 Sh cells. In at least three separate experiments performed in duplicate, we demonstrated a significant 1.5-2-fold resistance of cells up-regulating the enzyme as compared with the control cells ( Fig. 2A). To compare the mutation frequency in the surviving irradiated cells, we used the con- ventional methodology testing the appearance of mutational events leading to a resistance phenotype at the locus encoding the purine salvage enzyme hypoxanthine guanine phosphoribosyl transferase. After irradiation, cells were allowed to grow for 1 week before plating in 6-thioguanine-supplemented medium and then grown for one additional week, and 6-TG R mutant colonies were counted. A 2.6 -50-fold increase in mutagenesis for the pol ␤::Sh cells relative to the Sh cells was observed in three independent experiments after a 20 J/m 2 UV dose (Fig. 2B). The lack of correlation between pol ␤ expression level and UV resistance as well as hypermutability may be due to the mutator phenotype induced by pol ␤ overexpression (15). It is possible that the higher expression of pol ␤ may cause deleterious side effects and may affect other genes that would interfere with cell viability after UV treatment.
To investigate the molecular bases for these in vivo phenotypes, we tested a potential translesion ability of UV lesions of the replicative extracts from these cell lines. We performed in vitro primer extension reactions with a 30-mer template containing a CPD adduct, annealed to a 5Ј-32 P-labeled 16-mer primer (Fig. 3A, upper part). The primer was localized at a position so that the two first nucleotides were always incorporated opposite the lesion. In the presence of replicative extracts prepared from the control cells and one pol ␤::Sh cell line, we found that the pol ␤::Sh cell extracts could replicate past the CPD more efficiently as compared with the control extracts (Fig. 2C), demonstrating that excess pol ␤ facilitated the bypass process. Addition of purified pol ␤ to the control extracts also increased a bypass synthesis capability of the CPD lesion (data not shown). In contrast, we did not observe any TLS of the heavy distorting (6-4)TT lesion with either cell extract (data not shown). These results suggest that bypass synthesis of CPD damage by excess pol ␤ may contribute to the in vivo resistance and hypermutagenesis toward UV irradiation in the cells overexpressing pol ␤.
Ability of Purified pol ␤ to Bypass in Vitro CPD and (6-4)TT Adducts-To investigate in more depth the specific ability of pol ␤ to bypass UV photoproducts, we performed a kinetic study on the 30-mer template containing either CPD or (6-4)TT adduct, annealed to a 5Ј-32 P-labeled 16-mer primer (Fig. 3A). We used purified human and rat pol ␤, and we compared their behavior to pol ␣, which was reported previously as unable to incorporate nucleotides opposite the CPD or the (6-4)TT (2). As can be seen in Fig. 3B, by using amounts of enzymes allowing efficient and complete primer extension on undamaged template (Fig. 3B, right part), pol ␤ was able to incorporate nucleotides opposite both the CPD and the (6-4)TT lesions (17-and 18-mer products) as well as to perform extension beyond the adducts (products with a size larger than 18-mer) in a time-dependent manner, whereas pol ␣ is not, as expected. Some discrete radioactive fragments were also observed as 24-mer products; the mechanism involved in the generation of these products will be addressed in more depth later in the manuscript when describing Fig. 4. We also reported here that the HIV-1 RT, which shares structural and inaccuracy features with pol ␤, catalyzed efficient translesion synthesis of UV photoproducts (Fig. 3B). To better visualize and quantify the pol ␤-dependent bypass process, the 30-mer template was annealed in the presence of the 16-mer-labeled primer at a ratio of 1:1 and an excess of 11-mer oligonucleotide complementary to the 3Ј-end of the template to generate a 3-nucleotide gapped DNA, a preferential substrate for pol ␤ that offers the possibility to analyze the incorporation opposite the lesion and further extension of one nucleotide (Fig. 3A). Primer extension reactions were performed in the presence of 0.05 or 0.5 units of human pol ␤, leading to a 1:1 or 10:1 molar ratio, respectively, as compared with the primer-template. We found a more efficient pol ␤-mediated bypass in both a time-and dose-dependent manner on this gapped DNA as compared with the non-gapped template (Fig. 3C). The higher efficiency for nucleotide incorporation opposite the lesions may be favored by the ability of the pol ␤ 8-kDa domain, which binds to the downstream 5Јterminus, to promote processive extension of misinserted nucleotides on undamaged gapped DNA (32). In the presence of 0.05 and 0.5 units of pol ␤, the efficiency of the bypass of CPD into the 3-nucleotide gap represented 20 and 75% extension, respectively, of the primer for 60 min of incubation time (Fig.  3C). A minor bypass product also showed full-size synthesis in the presence of higher polymerase concentration, probably resulting from the previously reported in vitro strand displacement activity by pol ␤ of the 11-mer oligonucleotide (18). In the case of the (6-4)TT adduct, the presence of the 11-mer oligonucleotide allowed a 3-fold increase of nucleotide incorporation opposite the 3ЈT of the adduct by 0.05 units of pol ␤ (Fig. 3C). A complete 3-nucleotide gap-filling reaction was achieved in the presence of 0.5 units of human pol ␤, and bypass products represented more than 55% of the extended oligonucleotides after 20 min of incubation (Fig. 3C). Taken together, these results demonstrated that purified pol ␤ can bypass CPD and (6-4)TT adducts during in vitro primer extension.
Specificity of pol ␤-dependent Incorporation Opposite the CPD and (6-4)TT-A steady-state "single hit" gel kinetic assay (33) was performed using primed unmodified or UV-modified 30-mer DNA templates to quantitatively determine the specificity of nucleotide incorporation opposite the 3ЈT of CPD and (6-4)TT. For damaged templates, the concentration of incoming dNTP varied from 5 to 1000 M, and incubation time was 1 h in the presence of 0.5 units of pol ␤. Regarding the undamaged templates, the concentration of incoming dNTP varied from 1 to 500 M, and incubation time was 15 min in the presence of 0.001 units of pol ␤ when using dATP and 30 min with 0.5 units of pol ␤ when using dCTP or dGTP. All the data we obtained are summarized in Table I, and these data revealed that the dATP represented 55 and 71% of the inserted nucleotides opposite the CPD and the (6-4)TT lesions, respectively, leading to an error-free insertion. Insertion of dCTP opposite the 3ЈT of the lesion represented the major error-prone insertion with 32 and 26% of the inserted nucleotides opposite the CPD and the (6-4)TT lesions, respectively. Some dGTP residues can be inserted opposite the 5ЈT but at a lesser extent. Interestingly, when comparing kinetic parameters, we found that the ability of pol ␤ to insert the incorrect dGTP nucleotide opposite the 3ЈT of an undamaged template was only 4 -16-fold higher as compared with the pol ␤ efficiency to misinsert dATP or dCTP opposite the 3ЈT of the CPD or the (6-4)TT, signifying the high capacity of pol ␤ to incorporate nucleotide opposite distorting lesions. Finally, pol ␤ inserted dATP opposite the 3ЈT of the CPD or the (6-4)TT lesion with an efficiency 3500 -8000 less as compared with the insertion of dATP opposite the 3ЈT of the undamaged template.
pol ␤-dependent Efficiency of Extending Primers with One Base Opposite the CPD and the (6-4)TT-As it could be of biological significance to determine whether the incorporated nucleotide can be extended, 17-mer primers in which each of the four bases was paired to the 3ЈT of each adduct were annealed to damaged templates. Extension of these primers was assayed after a 1-h reaction in the presence of 0.5 units of pol ␤ and either all four dNTPs (200 M) (Fig. 4A) or a unique dNTP (Fig. 4, B and C). The most efficient extension to the full-size product of a primer annealed to the CPD-containing template occurred with the correctly paired dATP in the presence of all 4 dNTPs (Fig. 4A). We analyzed the 5ЈT incorporation specificity at the AT primer and found that, although dATP was mostly incorporated, all the other dNTPs could be also incorporated with a slightly lower efficiency (Fig. 4B). Indeed, there was a significant misincorporation of dTTP, dCTP, and dGTP opposite the 5ЈT of the CPD, and after dGTP incorporation, an incorporation opposite the adjacent undamaged dCTP in the template occurred, leading to a complete lesion bypass. With the GT primer in the presence of all 4 dNTPs, the obtaining of full-size product was as efficient as compared with the correctly paired AT primer (Fig. 4A), probably initiated by dATP incorporation (Fig. 4B). In contrast, pol␤-dependent extension of CT and TT mispairs was inefficient since it was aborted after incorporation of one nucleotide   (Fig. 4A). Extension reactions of primers annealed to the (6-4)TT-containing template revealed 1-nucleotide incorporation but no further extension, as shown in Fig. 4, A and C. Extension of AT, CT, and GT mispairs occurred only in the presence of dTTP, rendering this weak process highly mutagenic (Fig.  4C). A discrete radioactive fragment product migrating as a 24-mer product was observed when the CT mispair was extended on both the CPD and the (6-4)TT templates (Fig. 4A). This seems likely to correspond to a 6-nucleotide synthesis resulting from an annealing event of the microsequence ATGC at the 3Ј-terminus of the 17-mer primer that can pair to a homologous sequence TACG at positions 20 -23 of the template, generating a template loop. Such a misalignment incorporation mechanism facilitated by pol ␤ has already been described during TLS of an abasic site (34,35), an 8-oxodeoxyguanosine (36), and propano-deoxyguanosine lesions (37). This specific ability to catalyze template misalignment by searching microhomology sequence is shared by pol , another member of the DNA polymerase X family (38). pol ␤-dependent Extension of Primers with Two Bases Opposite the CPD or the (6-4)TT-To investigate whether pol ␤-dependent incorporated nucleotides opposite the two damaged bases could be extended, we used 18-mer primers whose termini were located directly opposite the CPD or (6-4)TT (Fig. 5). We focused on a set of 11 primers, 8 primers representing the best incorporations opposite the dimers (primers ending with AA, AG, AT, AC, GA, GT, CA, CT for the CPD; primers ending with AT, CT, GT for the (6-4)TT), and 3 primers randomly chosen (primers ending with GG, CG, GC). pol ␤ was able to extend efficiently a primer with two dA residues opposite the CPD, generating a full-size product. Interestingly, efficient extension was also observed with the AG, GA, AC, and CA primers with a decreasing efficiency (AGϾACϾGAϾCA). None of the primers randomly chosen were extended by pol ␤ opposite the CPD. Taken together, this shows that the best efficiencies were obtained with the nucleotides specifically incorporated opposite the CPD by pol ␤. Discrete radioactive fragments were also observed as 21-and 24-mer products when we used the AC and CT primers opposite either lesion, probably reflecting a misalignment incorporation mechanism facilitated by pol ␤ between TGAC or ATGCT at the 3-terminus of the 18-mer primers and the homologous sequence ACTG (position 24 -27) or TACGA (position 20 -24) of the 30-mer template, respectively. In the case of the (6-4)TT-containing template, we did not detect any significant primer extension with all the primers tested (Fig. 5, right part). Taken together, these results suggest that pol ␤ is able to extend efficiently mutagenic as well as correct nucleotides incorporated opposite the CPD.
Recruitment of Excess pol ␤ during in Vitro SV40 Replication to Bypass the CPD-To investigate whether excess pol ␤ could interfere with the replicative machinery during replication of UV-damaged duplex DNA, we performed a two-step in vitro SV40 replication assay. This assay can be used to observe CPD bypass as demonstrated for pol in HeLa cell extracts (2). We used two covalently closed circular templates containing the SV40 origin of DNA replication with a single CPD located on each side of the SV40 origin (Fig. 6A). Replication forks encounter the lesion during lagging strand synthesis in the case of pBS-SvoriA(CPD) and in the course of the leading strand synthesis in the case of pBS-SvoriB(CPD). These plasmids were first incubated for 4 h with 400 g of Hela extracts in the reaction buffer, then [␣-32 P]dATP and purified pol ␤ were added for an additional hour. During the first incubation period, DNA replication machinery stalled at the lesion on the damaged strand, and during the shorter period of the second incubation in the presence of radioactive dATP and purified pol ␤, radioactivity will be incorporated preferentially into products of damage bypass replication. Then, DNA was purified, linearized by BamH1 and DpnI, and subjected to electrophoresis onto a 1% agarose gel. Ethidium bromide staining and autoradiography of the gel are shown in Fig. 6B. DpnI digestion was done to visualize only the DNA population that was replicated once. Additionally, we verified that addition of up to 0.024 units of pol ␤ in reaction mixtures replicating undamaged FIG. 6. Excess pol ␤-mediated translesion synthesis of CPD during SV40 replication. A, possible implications of excess pol ␤ in the bypass of the CPD adduct during the two-step SV40 replication assay. SV40-DNA constructs are shown here during bidirectional semiconservative SV40 replication that began at the origin (Ori A or Ori B for pBS-SV(CPD) oriA or oriB, respectively). In B, 100 ng pBS-SV oriA and oriB were replicated by 400 g of cell-free extracts from human Hela cells in a T-antigen-dependent manner in the presence or absence of 0.012 units of pol ␤, and then they were linearized by BamHI (one unique site) or digested by BamHI and DpnI (multiple sites). The two-step SV40 DNA replication and analysis of the products are described under "Experimental Procedures." DNA did not result in an increase of the radioactive replication signal (39). As observed in Fig. 6B, radioactivity incorporation during DNA replication is lower with the pBS-SVoriA DNA as compared with the pBS-SVoriB DNA. As suggested in a previous report, SV40 replication of a UV lesion-containing plasmid could be synchronous between the two parental strands in the case of pBS-SVoriA(CPD) and asynchroneous in the case of pBS-SVoriB(CPD) (2); during lagging strand synthesis, the replication fork moves past the lesion, and reinitiation occurs at the next Okasaki fragment, leaving a small single-stranded gap; during the leading strand replication, the progression of the fork is inhibited, and uncoupling of leading and lagging strand occurs; the replication machinery continues to synthesize the lagging strand (40,41).
We found that, in the presence of 0.012 units of rat pol ␤, DpnI-resistant products increased by 4-fold with pBS-SVori-A(CPD) and by 2-fold with pBS-SVoriB(CPD) as compared with the control reactions without pol ␤ (Fig. 6B, right tracks). For the global replication products (without digestion by DpnI; Fig.  6B, left tracks), a pol ␤-dependent increase was also detected. When pol ␤ and the radioactive nucleotide were added at the beginning of the reaction (one-step reaction), a 2-, 3.5-, and 5-fold signal increase was observed with pBS-SVoriB(CPD) in the presence of 0.0048, 0.012, and 0.024 units of rat pol ␤, respectively (data not shown). Taken together, these results suggest that when DNA synthesis during replication of duplex DNA is stopped by a CPD, excess pol ␤ can be notably recruited to overcome the lesion. DISCUSSION We showed here that pol ␤, an enzyme required in somatic cells for the base excision repair pathway (14), can facilitate translesion replication of a CPD as well as, to a lesser extent, a thymine-thymine pyrimidine-pyrimidone (6-4) photoproduct ((6-4)PP). Such a result was obtained by using the well calibrated primer extension assay using site-specific UV-modified oligonucleotides as well as the SV40 replication assay, which reconstitutes the mammalian DNA replication fork, using CPD-modified duplex DNA. pol ␤ mostly incorporated the correct dATP opposite the 3ЈT of the CPD and the (6-4)PP but could also misinsert dCTP. For the CPD, we found that the 5ЈT incorporation specificity by pol ␤ at the AT and CT primers was highly mutagenic. Whether the nucleotides were correctly or incorrectly inserted opposite the CPD, some of them were efficiently extended by pol ␤, and this extension is highly errorprone, supporting the possibility that pol ␤ could compete with pol (5) or pol (6) to extend nucleotides incorporated opposite the 3ЈT of the CPD adduct. Opposite the (6-4)TT lesion, the incorporation by pol ␤ opposite the 3ЈT, essentially the dATPlike pol , is poorly efficient and is most of the time aborted, probably because of the strong distortion of DNA. This low extension capability of pol ␤ is shared with pol (3) and pol (5,7). It has been proposed that pol is responsible for the subsequent extension of the nucleotide incorporated opposite the 3ЈT of the (6-4)TT damage (3,5). This suggests that in vivo, an efficient, mostly accurate, but potentially error-prone TLS of the (6-4)TT lesion may result from the combined activities of pol ␤ and pol .
To date, among the 12 eukaryotic DNA polymerases that have been identified, only pol , pol , pol , and pol have been shown to exhibit such potential involvement in CPD and (6-4)TT photoproducts bypass activity (2,(5)(6)(7). In normal somatic cells, the majority of translesion replication is normally pol -dependent since in xeroderma pigmentosum variant cell extracts, in which pol is inactive, only 10% of the lesion bypass activity of normal cell extracts is observed (2,42). So what could be the biological significance of such a pol ␤-dependent bypass? Analysis of the mutagenic spectra observed after exposing human cells to UV light suggests that most mutations are, in fact, targeted to the 3Ј-site of a di-pyrimidine containing a dC (at CC and TC) (1 ,43). However, some minor mutations T 3 A and T 3 C targeted to the 5Ј-site of TT can also be observed (10,44), and these match to the pol ␤-dependent mutations that we observed here in vitro, suggesting a role of pol ␤ in some of the UV-induced mutations in somatic cells. The frequency of this kind of mutation increases strongly up to 45% in xeroderma pigmentosum variant cells (1 ,10, 45), supporting that pol ␤, like pol , may be involved in the TLS process at the TT sites in the absence of pol .
Moreover, situations in which the imbalance of pol ␤ expression in cells occurs may be of interest in such translesion process of UV lesions. Interestingly, we observed in this work that high levels of pol ␤ can be found in various melanomas tumors, which are known to be associated with UV radiation exposure. We recently showed that pol ␤ can interfere in vitro with duplex DNA replication when up-represented, rendering the process inaccurate (39). The data presented here suggest strongly that interference of excess pol ␤ at the replication forks not only can affect the accuracy of the process but can also modulate the genotoxicity of UV lesions when present on the genomic DNA. Although the mutagenic translesion replication experiments reported here were performed entirely in vitro, we believe that they shed light on the mutagenic process in vivo in melanoma cells and that excess pol ␤ may enhance CPD translesion in a mutagenic manner by competing with pol . By using isogenic CHO cells, we found that the sole pol ␤ overexpression event resulted in a resistant phenotype toward UV treatment and can dramatically enhance the induced mutagenesis. Both phenotypes may result from the TLS catalyzed by pol ␤ during the elongation of the replication forks. Overexpression of pol ␤ could be therefore identified as a host risk factor that may potentiate the genetic instability in cells exposed to UV and may consequently affect melanoma risk.