A Role for Polymerase η in the Cellular Tolerance to Cisplatin-Induced Damage

Xeroderma pigmentosum (XP) is a rare autosomal recessive disorder characterized by extreme sensitivity to sunlight and a greatly increased predisposition to skin cancers (1). The molecular defects responsible for XP have now been characterized and eight distinct complementation groups have been defined, with gene products XP-A to XP-G controlling the nucleotide excision repair pathway that removes UV-induced adducts from DNA (2). The POLH gene responsible for the variant form of XP (XP-V), which is not defective in nucleotide excision repair, encodes DNA polymerase η (pol η; refs. 3, 4). This polymerase is required for the efficient and accurate translesion synthesis of intrastrand cross-linking DNA adducts caused by UV light that would otherwise impede replication (5–7).

Cisplatin (cis-diamminedichloro-platinum (II)) is one of the most commonly used and successful chemotherapeutic agents employed to date. The major cell-killing effect of cisplatin is due to DNA damage. More specifically, intrastrand cross-linking cisplatin adducts cannot be bypassed by the classic replicative polymerases α, δ, and ε, and therefore represent potential replication-blocking lesions (8). In order to complete DNA replication, a specialized DNA polymerase or polymerases are required to synthesize a few nucleotides opposite the platinated purines. Once the replication complex has bypassed the platinum adduct, the specialized translesion synthesis polymerase is replaced or switched out and DNA synthesis by the high-fidelity replicative DNA polymerases is resumed.

The identity of the translesion synthesis polymerase or polymerases responsible for cisplatin adduct bypass in vivo is unknown, although a number of in vitro studies have shown that three DNA polymerases are able to bypass cisplatin lesions efficiently by translesion synthesis, namely pol β (8, 9), pol η (7, 10), and pol μ (11). However, very few studies have shown the involvement of these polymerases in dealing with cisplatin-induced damage in cells, or the relative contribution that each might make to cellular survival following cisplatin treatment. Those studies that have been carried out have focused on the potential importance of pol β, with overexpression providing resistance to cisplatin (12, 13) and down-regulation resulting in sensitivity to cisplatin-induced damage (14, 15).

The main mechanism for the removal of cisplatin adducts from double-stranded DNA is nucleotide excision repair (16) and is thought to involve the same processes that remove UV-induced DNA adducts. Nucleotide excision repair–deficient cells are hypersensitive to cisplatin (17) and down-regulation of nucleotide excision repair components such as the DNA damage sensor protein XP-A by antisense RNA can also sensitize cells to cisplatin (18). One question that has not been previously addressed is the relative contributions of nucleotide excision repair and translesion synthesis in determining cisplatin sensitivity.

Here, we investigate the role of pol η in mediating mammalian cell survival to cisplatin and address the in vivo relevance of the ability of pol η to bypass cisplatin adducts. We also compare the cellular response to cisplatin in cells deficient in either pol η or XP-A proteins to gain an insight into the relative roles of nucleotide excision repair and translesion synthesis in dealing with cisplatin-induced DNA damage.

Cell lines and chemicals used. XP30RO/cDNA3, XP30RO/pol η cl5 and XP30RO/pol η cl6 cells were generated from SV40-transformed XP30RO xeroderma pigmentosum-variant human fibroblasts as described previously (19, 20). The XP30RO cell line (also known as GM3617) has a homozygous deletion of part of the POLH gene, leading to a truncated protein product consisting of the inactive NH₂-terminal 42 amino acids of the pol η protein (3). MRC5V1 cells are SV40-transformed normal diploid human fibroblasts (21). XP5BI is a primary fibroblast culture from XP-variant cells (22). 1BR/3 cells are nontransformed, primary human fibroblasts.

GM04312, also known as XP2OS (XP-A), and GM15876 (XP-A complemented) cells were obtained from the Coriell Cell Repository (Camden, NJ). GM04312 cells are SV40-transformed fibroblasts containing homozygous truncating mutations in the XP-A gene (23) and GM15876 cells are complemented by stable transfection of the wild-type XP-A gene (24). All experiments were done in DMEM supplemented with 10% fetal bovine serum, 2 mmol/L glutamine, and the antibiotics penicillin (100 units/mL) and streptomycin (100 μg/mL). XP30RO-derived cells were maintained in the same medium with the addition of 100 μg/mL zeocin. Note, however, that all experiments were done in the absence of zeocin. Chemotherapeutic agents were purchased from Sigma-Aldrich (St. Louis, MO), with the exception of oxaliplatin, which was purchased from LKT Laboratories, Inc. (St. Paul, MN).

Colony formation assays. For transformed fibroblasts, either 750 (XP30RO lines, MRC5) or 1,000 (XP-A lines) cells were seeded in each well of a six-well plate and allowed to attach overnight. Drugs were added to wells in triplicate to give final concentrations as indicated. Cells were exposed to drugs for 24 hours before replacement with normal media. Cells were allowed to grow for a further 6 to 10 days in order to form colonies, after which time, medium was removed and replaced with modified Giemsa stain for 1 hour to fix and stain the colonies. Colonies were counted automatically using a colony counter (Oxford Optronics, Oxford, United Kingdom). Survival curves and cytotoxic concentration (CC₅₀) values were calculated using the four-parameter sigmoidal analysis of the Xlfit program (IDBS, Emeryville, CA). Data shown are calculated from the mean values of three independent experiments.

Western blotting. Whole cell lysates were immunoblotted and probed with the following antibodies directed to pol η (1:1,000 dilution; ref. 25) or XP-A (1:2,500, clone 12F5; BD PharMingen, San Jose, CA). After probing with these antibodies, blots were stripped at 50°C in 62.5 mmol/L Tris-HCl (pH 6.8), 2% SDS, 0.2% 2-mercaptoethanol, and reprobed with antibodies to β-actin (1:10,000, Sigma-Aldrich) or pol β (1:2,500, a kind gift from Sam Wilson, NIEHS, NIH, Research Triangle Park, NC).

Flow cytometry. Cells (5 × 10⁵) were plated in 10 cm dishes and allowed to attach overnight. Cisplatin was added to final concentrations as indicated and cells were exposed for 1 hour, after which time, the medium was changed. Twenty-four hours later, bromodeoxyuridine (BrdUrd) was added to a final concentration of 10 μmol/L for 2 hours. Cells were subsequently harvested using trypsin-EDTA and fixed in 70% ethanol overnight. Cells were stained for incorporation of BrdUrd using an anti-BrdUrd FITC-conjugated antibody (BD PharMingen) and conditions recommended by the supplier. Cells were subsequently counterstained for DNA content by incubating in the presence of 50 μg/mL propidium iodide/RNase solution at 37°C for 1 hour; for each treatment, 10,000 to 20,000 cells were analyzed using a Partec CyFlow flow cytometer. Cell cycle profiles were analyzed using FlowMax software (Partec, Munster, Germany) and were gated between one and two DNA contents prior to measurement of percentage of cells in each of G₁, S, or G₂-M phases of the cell cycle.

Analysis of proliferating cell nuclear antigen-ubiquitination. Cells were grown to 60% confluence then either mock-treated, incubated with 50 μg/mL cisplatin in serum-free medium for 1 hour or UV-irradiated at 20 J/m². After treatment, the cells were placed in fresh serum-containing medium for 24 or 6 hours, respectively. Cells were scraped in Laemmli buffer, sonicated, and boiled. Ten micrograms of each extract was analyzed by probing Western blots, after 10% SDS-PAGE, with the PC10 anti–proliferating cell nuclear antigen (PCNA) monoclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA) at 1:5,000 dilution in 5% milk/PBT overnight and subsequently detected using enhanced chemiluminescence.

Subcellular localization studies. MRC5 cells were plated onto glass coverslips at 60% confluence. 16 hours later they were transfected with “Living Colours” vectors (Clontech, Mountain View, CA) expressing peCFP-pol η, or cotransfected with peCFP-pol β and peYFP-pol η using 6 μL Fugene 6 reagent (Roche, Indianapolis, IN), 2 μg each plasmid and 200 μL serum-free medium for each coverslip. Twelve hours later, the cells were either mock-treated, UV-irradiated (7 J/m²), or incubated with cisplatin (9 μg/mL for 1 hour), then allowed to recover in fresh medium for 16 or 24 hours, respectively. The coverslips were washed in PBS, and either fixed in 2% paraformaldehyde in PBS, or washed in CSK buffer (10 mmol/L PIPES, pH 7.0; 100 mmol/L NaCl, 300 mmol/L sucrose, 3 mmol/L MgCl₂) and treated with 0.2% Triton X-100 in CSK for 5 minutes to remove soluble proteins prior to fixation and mounting. For visualization of endogenous PCNA, fixed cells were treated with methanol for 20 minutes at −20°C, blocked in 5% bovine serum albumin in PBS plus 0.5% Tween 20, incubated with PC10 antibody at a dilution of 1:1,000 in block buffer for 1 hour, washed, then incubated in Alexa fluor 555–conjugated goat anti-mouse secondary antibody (Invitrogen Molecular Probes, Eugene, OR) for 45 minutes in block buffer before washing and mounting in Vectashield (Vector Laboratories, Burlingame, CA). Fluorescent images in yellow and cyan channels were captured using a Zeiss LSM 510META inverted confocal microscope. Mid-plane confocal sections are presented after contrast and brightness adjustment and psuedo-coloring using Adobe Photoshop software.

Sensitivity of xeroderma pigmentosum-variant cell lines to cisplatin. In order to investigate the role of pol η in response to cisplatin in a cell-based system, a set of isogenically matched cell lines were used that were derived from SV40-transformed fibroblasts obtained from a patient with XP-V. These XP30RO cells express a truncated nonfunctional pol η protein and stable transfectants were generated using a cDNA expressing recombinant wild-type pol η to produce the independent clones XP30RO/pol η cl5 and XP30RO/pol η cl6, or with the vector alone to generate the control line XP30RO/cDNA3, referred to here simply as XP30RO (Fig. 1A; refs. 19, 20). XP30RO cells have been previously shown to be sensitive to UV radiation and this sensitivity is complemented by pol η in the XP30RO/pol η cl5 clone (20).

We examined the response of these cells to cisplatin in clonogenic colony formation assays, and as an additional reference, used an unmatched SV40-transformed normal fibroblast line, MRC5 (Fig. 1B). The XP30RO cells were found to be significantly more sensitive to cisplatin compared with the pol η-complemented clones XP30RO/pol η cl5 or cl6 or the MRC5 cells, demonstrating an ∼4.5-fold decrease in the cisplatin dose resulting in 50% survival (CC₅₀; Table 1). The similarity of results obtained with the two independently isolated XP30RO/pol η clones suggests that this response is a true reflection of functional pol η and not an indirect effect caused, for example, by an insertion of the pol η construct into an unrelated site. Moreover, the similarity in the cisplatin response of the two XP30RO/pol η clones compared with the MRC5 cells, suggests that it is the pol η-deficient XP30RO cells that are sensitive to cisplatin, rather than the XP30RO/pol η clones demonstrating an atypical resistance to the platinum compound.

In order to investigate whether this differential cisplatin response was due to a peculiarity of the XP30RO cell line or transformation with SV40, primary XP-V fibroblast cells (XP5BI) were examined for their response to cisplatin in survival assays (Fig. 1C). XP5BI cells were also found to be more sensitive to cisplatin (CC₅₀ 119 ng/mL) than 1BR/3 normal primary fibroblast cells (CC₅₀ 274 ng/mL) and the MRC5 cells (CC₅₀ 422 ng/mL), and showed a similar, although less pronounced sensitivity to cisplatin as the XP30RO cells (CC₅₀ 71 ng/mL). Taken together, these data provide evidence that the differential response to cisplatin is due to the expression of pol η alone.

It is worth noting that the cell survival and cell cycle responses of XP30RO cells exposed to UV radiation are generally apparent only in the presence of low levels of caffeine (20, 26). However, no significant difference in response to cisplatin was observed in the presence or absence of 75 μg/mL caffeine (Fig. 1D), a concentration previously shown to enhance UVC-dependent toxicity of XP-V cells (22).

Sensitivity of xeroderma pigmentosum-variant cells to other chemotherapeutic agents. The response of the XP30RO cells to other platinum anticancer agents, which have a similar mechanism of action to cisplatin, was also examined. The XP30RO cells exhibited pol η-dependent sensitivity to both carboplatin and oxaliplatin (Fig. 2A), indicating that a pol η-dependent response is common in these additional platinum agents. In contrast, the response to transplatin, a stereoisomer of cisplatin which preferentially forms interstrand, rather than intrastrand cross-links, displayed no apparent difference between the XP30RO lines (data not shown).

The XP30RO cells were also examined for their response to other chemotherapeutic agents. No difference was observed in either the response to the microtubule-depolarizing agent paclitaxel (Fig. 2B), or the antimetabolite 5-fluorouracil (data not shown), indicating that the differential response to cisplatin is not due to any general fragility of the XP30RO cells. The response to an interstrand DNA cross-linking agent, mitomycin C, was also examined to assess any potential intrastrand versus interstrand cross-linking effects. However, no significant sensitization with mitomycin C was observed (Fig. 2B). Together, these results (summarized in Table 1) suggest that the different susceptibilities of pol η-deficient cells to cisplatin are likely to involve a very specific DNA damage response.

Sensitivity of xeroderma pigmentosum-A cell lines to cisplatin. Cisplatin adducts are removed from DNA by the nucleotide excision repair pathway and the importance of nucleotide excision repair components in cisplatin survival is well-documented (17, 18, 27). We compared the effects on survival to cisplatin treatment of the loss of pol η with those caused by the loss of the XP-A protein, a zinc-finger DNA-binding protein that is a key component of nucleotide excision repair previously shown to be involved in cisplatin sensitivity (17, 18). GM04312 cells are derived from an XP patient with homozygous truncating mutations in the XP-A gene (23), whereas those cells stably transfected with a wild-type XP-A gene give rise to the isogenic line, GM15876, that has been used previously in nucleotide excision repair studies (Fig. 3A; ref. 24). These matched cells were examined for their survival to cisplatin treatment under identical conditions used for the XP30RO experiments. GM04312 cells were found to be sensitive to cisplatin compared with normal unmatched MRC5, and complementation with the wild-type XP-A protein partially restored resistance (Fig. 3B). The level of cisplatin sensitization caused by the loss of XP-A protein expression was ∼2-fold (CC₅₀ GM04312, 107 ± 19 ng/mL and GM15876, 183 ± 32 ng/mL), less than that of the pol η-deficient lines XP30RO, indicating from these experiments that translesion synthesis by pol η may be at least as important as functional nucleotide excision repair in determining cellular survival following cisplatin treatment.

Cell cycle analysis of xeroderma pigmentosum-variant and xeroderma pigmentosum-A cell lines in response to cisplatin. Pol η is required for replication after UV-induced damage (28). Loss of pol η expression results in a prolonged delay in S phase after UV irradiation in the presence of caffeine and this effect is rescued in the pol η-complemented lines (20). This has been attributed to a requirement for translesion synthesis past the UV-induced DNA adducts before replication can be completed. In a similar manner, we predict that pol η would be required for efficient completion of S phase after cisplatin treatment.

Asynchronous populations of MRC5, XP30RO, and XP30RO/pol η cl5 cells were treated with cisplatin for 1 hour before being allowed to recover for 24 hours. The percentage of cells in G₁, S, and G₂M cell cycle phases was then examined by BrdUrd incorporation (Fig. 4). Cisplatin treatment of MRC5 cells led to a dose-dependent increase in the percentage of cells in S phase, consistent with the replication-blocking effect of cisplatin DNA adducts (Fig. 4A). The XP30RO cells lacking pol η expression were extremely sensitive to the S phase block induced by cisplatin, with a dramatic increase in the number of cells in S phase and an ∼6-fold increase in S/G₁ ratios (Fig. 4B and C). Complementation of pol η expression in the XP30RO/pol η cl5 cells led to a restoration of S phase progression to a level comparable with the MRC5 cells. These data therefore show a clear requirement for pol η in overcoming the S phase block induced by cisplatin, which is consistent with a role in translesion synthesis past cisplatin-induced DNA adducts.

The cell cycle response of XP-A and complemented cells to cisplatin was examined under the same conditions (Fig. 4D). The presence or absence of the XP-A protein did not affect the cell cycle response to cisplatin, with S/G₁ ratios of the GM04312 and GM15876 cells similar to those observed for the control MRC5 cells (Fig. 4D). These data provide a clear distinction between the roles of translesion synthesis and nucleotide excision repair in the cellular response to cisplatin.

Ubiquitination of proliferating cell nuclear antigen and subcellular localization of pol η and β after cisplatin treatment. UV irradiation of human fibroblasts results in an accumulation of cells blocked in S phase with replication foci containing both PCNA and pol η (19). We previously showed that a band of reduced mobility detected with anti-PCNA antibody on Western blots of cell extracts from UV-irradiated cells resulted from the monoubiquitination of PCNA (25). Figure 5A shows that a band of identical mobility can also be detected after cisplatin treatment of either normal or XP-V cells. After UV treatment, this monoubiquitinated PCNA specifically interacts with pol η to facilitate translesion synthesis, and is likely to assist in targeting the pol η to replication sites containing DNA damage (25). Cisplatin treatment of cells also resulted in a dramatic increase in the number of cells with nuclear pol η foci (Fig. 5B and C). As in the case of UV irradiation (19), the pol η foci formed after cisplatin treatment colocalized with endogenous PCNA and were resistant to extraction with detergent (Fig. 5B). In the absence of any treatment, ∼5% of the cells in an untreated population show pol η foci, whereas in the remainder, pol η is represented by a diffuse nuclear pattern. After 24 hours of treatment with 25 μg/mL cisplatin, this proportion increased to 90% of cells with pol η foci, which is even higher than the 72% of cells seen 16 hours after UV (7 J/m²). In contrast to pol η, pol β did not accumulate in replication foci (Fig. 5C), irrespective of cell treatment with UV or cisplatin. Similar results were obtained whether the pol β was tagged with eYFP or with HA (data not shown). Moreover, the nuclear staining of pol β was completely removed from the cells after detergent extraction prior to fixation (data not shown). These data provide evidence that pol η, but not pol β, is associated with replication foci formation following cisplatin treatment, an observation reminiscent of the pol η response to UV.

Cisplatin is a widely used chemotherapeutic drug and the prototypical member of a family of platinum-based DNA damaging agents. The major limitations to its use are a high degree of toxicity and the development of acquired drug resistance in the tumor being treated. The toxicity and concomitant narrow therapeutic window means that only small increases in resistance are required to render treatments ineffective. Multiple mechanisms of cellular resistance to cisplatin have been proposed (29), although there is no clear consensus for which, if any, has the most clinical relevance. However, it is now clear that DNA repair processes have a critical impact in both acquired and intrinsic cellular resistance to this agent (30).

Nucleotide excision repair is believed to be the main process responsible for removing platinum adducts from DNA, and it has been suggested that impaired nucleotide excision repair status in metastatic testicular cancer cells may be responsible for the excellent clinical response of this disease to cisplatin combination therapy (31, 32). However, nucleotide excision repair is not the only DNA repair process that can influence the response to cisplatin. Enhanced replicative bypass or translesion synthesis may be another mechanism providing cellular tolerance to cisplatin (33, 34). The relative roles of nucleotide excision repair and translesion synthesis in determining cisplatin survival have not been previously established.

Here, we have presented data which indicate that both processes are integral to surviving cisplatin-induced damage. Our findings are consistent with a cellular response to cisplatin involving both tolerance (mediated by translesion synthesis) and repair (mediated by nucleotide excision repair). In this model, a cisplatin adduct resulting in a stalled replication complex (due to the inability of the replicative DNA polymerases to accommodate the distorted DNA structure in its active site) will trigger a polymerase switch involving pol η (and/or pol β). This will result in error-prone bypass of the lesion by translesion synthesis so that replication can resume. The nucleotide excision repair pathway has the capacity to recognize cisplatin adducts in double-stranded DNA and then excise them from the genome to restore undamaged DNA at these sites. This model would therefore predict that cells with functional translesion synthesis but deficient in nucleotide excision repair genes should accumulate mutations when challenged with cisplatin and such an observation has indeed been made in cells deficient in XP-A (35).

The data presented here indicate that the loss of pol η results in a striking sensitization to cisplatin and other platinum agents using well-characterized mammalian cells derived from XP-V patients. Moreover, the loss of pol η results in a pronounced S phase block upon cisplatin treatment that is consistent with a role for pol η in translesion synthesis during S phase. Our findings are also consistent with previous reports demonstrating that pol η can bypass cisplatin adducts in DNA by translesion synthesis in vitro (36). Intriguingly, we find that XP-V cells show substantial sensitivity to the lethal effects of cisplatin, which is not increased by postincubation in the presence of caffeine. This is in contrast with UV irradiation, to which XP-V cells are only very mildly sensitive in the absence of caffeine and only with postirradiation incubation in the presence of caffeine are they substantially sensitized to killing by UV (26). This may imply that for cell survival, pol η-dependent translesion synthesis is more important for the tolerance of cells to cisplatin than to UV. Consistent with this idea, the level of sensitization to cisplatin that we observe in the absence of functional pol η is greater than that observed as a consequence of XP-A deficiency, suggesting that pol η is at least as important as nucleotide excision repair in mediating cisplatin survival in our assays. The smaller effect of loss of XP-A on cisplatin survival that we observed is comparable to that found in a previous study of down-regulation of XP-A by antisense RNA (18) and fits with a model in which cells are able to tolerate cisplatin damage in the absence of nucleotide excision repair by translesion synthesis past cisplatin lesions. In contrast, XP-A cells are much more sensitive to UV irradiation than XP-V cells. The crucial role of pol η-dependent translesion synthesis in response to cisplatin is also attested to by the high proportion of cells that contain ubiquitinated PCNA and pol η-containing replication foci after such treatment (Fig. 5). This suggests that pol η is mediating translesion synthesis at stalled forks in response to this agent. These observations are further supported by a study which showed that cisplatin treatment of XP-V cells delayed replication and caused an accumulation of shorter replication products, indicative of deficient translesion synthesis of cisplatin lesions (37).

A number of in vitro lesion bypass studies have shown the ability of certain specialized DNA polymerases to bypass intrastrand cisplatin cross-links with greater or lesser efficiency (8, 36, 38). However, most in vitro studies have been done under different artificial conditions, and the relevance to the situation in cells has not been clearly addressed. Indeed, it is still unclear which polymerase(s) is supplying the translesion synthesis function in mammalian cells. The most compelling evidence that has been obtained thus far at the cellular level suggests that pol β plays an important role in cisplatin survival and that down-regulation of pol β by antisense oligonucleotides (14) or small interfering RNA (15) sensitizes cells to cisplatin, whereas overexpression of pol β provides resistance to cisplatin (12, 34). This is likely to be a clinically relevant observation because pol β overexpression seems to be a frequent event in tumors (15) and has been correlated with cisplatin-resistant cancer cells (34, 39). In a notable contradiction to these data, pol β-deficient mouse embryonic fibroblasts (stem cells) are not sensitive to cisplatin (40), although the reason for this is not clear.

In the study presented here using human cell lines, alterations in pol η levels alone result in substantial differences in sensitivity to cisplatin, even though pol β levels are comparable in all of the XP-V-derived and MRC5 cells that were used (Fig. 1A). Similarly, cisplatin sensitivity mediated by small interfering RNA directed against pol β was independent of pol η levels (ref. 15; data not shown). This suggests that although both pol η and pol β seem to be able to bypass cisplatin adducts in cells, inhibition of either alone is sufficient to sensitize cells to cisplatin, suggesting that translesion synthesis of cisplatin-induced DNA adducts is a complex affair and that pol η and pol β are not obviously redundant. Due to the stability of the pol η protein, we have thus far been unable to knockdown pol η and pol β levels simultaneously in order to determine whether this effect will be epistatic, although work is ongoing to address this question.

Very recently, the cisplatin response of an alternative matched system of pol η-complemented XP-V-derived lines was examined (38). In contrast to the findings presented here, no apparent sensitivity to cisplatin was observed in this system, although a requirement for pol η in error-free bypass of cisplatin lesions was proposed. The major difference between the studies is the experimental cells used; more specifically, their study employs hTERT-transfected primary cells and the transfection of a very high level (50-fold) of pol η overexpression, whereas this study uses controlled SV40-transformed lines in which the level of complemented pol η is approximately the same as in the wild-type pol η controls in cl5 or slightly higher in cl6 (Fig. 1A). Another experimental difference between the two studies is the time of exposure to cisplatin, which was 24 hours in our experiments and only 1 hour in the work of Bassett et al. (38), although we also found enhanced sensitivity of XP-V cells with 1-hour exposures to cisplatin (data not shown). Whereas we cannot rule out that the cisplatin sensitivity seen here is due to the combination of SV40 transformation and the loss of pol η in the XP30RO line, this would seem unlikely because we also see cisplatin sensitivity in our primary diploid XP-V fibroblasts in comparison with normal diploid fibroblasts.

Irrespective of the explanation for this discrepancy, the data presented here clearly show that pol η plays an important role in determining cellular survival to the commonly used class of platinum chemotherapeutic agents. Moreover, there is an obvious differentiation between the cisplatin-dependent cellular response to defects in translesion synthesis and nucleotide excision repair. In the same way that nucleotide excision repair status has been proposed as an explanation for the sensitivity of testicular cancer to cisplatin (31), we suggest that the translesion synthesis status of cancer cells could also have a profound impact on their response to this important class of chemotherapeutic agents, which in turn, could influence patient stratification or suggest novel therapeutic approaches to improving platinum-based cancer treatment.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We are grateful to Limei Ju for technical assistance; Alan Lau for advice and help on flow cytometry and comments on manuscript; and to Sam Wilson for the generous gift of the pol β antibody.