A Truncated Human Xeroderma Pigmentosum Complementation Group A Protein Expressed from an Adenovirus Sensitizes Human Tumor Cells to Ultraviolet Light and Cisplatin¹

Individuals with the genetic disease XP³are defective in some aspect of NER, which is the primary DNA repair system responsible for repairing bulky lesions in DNA (reviewed recently in Ref. 1). There are seven complementation groups of XP (A–G), where each complementation group represents a defect in a different protein important for DNA incision at the site of damage. Of all of the different XP complementation groups, cells from group A are most severely impaired in NER and in vitrodemonstrate extreme sensitivity to UV and other types of bulky DNA damage including those induced by mitomycin C, cisplatin, melphalan,and psoralen + UVA (1, 2).

Although several reports have shown that XPA cells can be rescued by increased UV resistance or increased levels of HCR (HCR) by cell transfection with the XPA gene or the phage T4 denV gene (3, 4, 5, 6, 7, 8), to date there has been no demonstration of sensitizing tumor cells proficient in NER with a vector expressing a protein that interferes with NER as a strategy for cancer therapy. The XPA protein is a zinc-finger protein of M_r 31,000 (273 amino acids)involved in DNA damage recognition (4, 9, 10). Previous work demonstrated that the XPA protein interacts with the ERCC1 protein through amino acid residues 72–84 of XPA (11, 12, 13). Hence,we argued that by overexpressing a truncated form of the XPA protein encompassing this domain, one might be able to prevent the ERCC1/XPF nuclease complex from associating with the normal XPA protein produced by the cell and would thus interfere with the early stages of NER. By taking advantage of the efficient gene delivery and protein expression properties of adenovirus (3, 14), the XPA protein is an excellent target for gene therapeutic intervention through increased sensitization of tumor cells to chemotherapy. This strategy is particularly interesting when applied to tumor cells, allowing for more efficient therapies with a variety of drugs used for chemotherapy and to counteract increased resistance to chemotherapeutics frequently seen during the course of treatment of certain types of cancers. We demonstrate in this report that infection of human tumor cells proficient in NER with a replication-defective adenovirus expressing amino acids 59–114 of the XPA protein sensitizes the cells to UV and cisplatin.

The human non-small cell lung carcinoma cell line A549 has normal NER and p53 function (15). The GM637 and XP12RO(M1) cells have been described previously (3, 16, 17). Human embryonic kidney 293 cells (18), also normal for NER(19), were used for constructing and growing replication-defective recombinant adenovirus. Cells were routinely cultured in DMEM (Life Technologies, Inc., Gaithersburg, MD) with 10%fetal bovine serum (Irvine Scientific, Santa Ana, CA) and penicillin/streptomycin, and were subcultured twice weekly.

A truncated version of the human XPA cDNA with a 97-bp deletion between position 104 and 200 (4, 20) was cloned in the correct reading frame in pCMV-FLAG-2 (Kodak, Rochester, NY) and was expected to produce a FlagXPA fusion protein consisting of amino acids 59–273 of the human XPA protein (4). We previously demonstrated that SV40-transformed human XPA fibroblasts stably transfected with a plasmid expressing this portion of the XPA gene show normal HCR of a UV-damaged reporter gene (17). Others have also noted that amino acids 1–58 of XPA are not required for conferring UV resistance (11, 21). The FlagXPA_59–273 DNA cartridge was transferred into pZeroTG-CMV and made into an adenovirus (14). Subsequently, a 3′ deletion was made of the XPA cDNA in pZeroTG-CMVFlagXPA_59–273 by removing the DNA sequence 3′ of the unique BsaBI site (4). This plasmid is expected to produce a truncated XPA protein of 55 amino acids (59–114) tagged at its NH₂ terminus with Flag and spanning the ERCC1 binding domain (amino acids 72–84). Recombinant viruses were screened by Western blotting of extracts obtained from infected 293 cells using anti-Flag antibody. Viral clones that produced proteins of the expected sizes that reacted with the anti-Flag and anti-XPA antibodies (20) were used in subsequent experiments. Adenoviruses expressing β-galactosidase(AdLacZ), GFP (AdCMV-GFP), Flagp38 MAP kinase(AdCMV-Flagp38)⁴under control of the CMV promoter or an “empty” adenovirus(AdCMV) without any trans-gene, were constructed and used in HCR assays or as virus controls.

For Western blot analyses, A549 or 293 cells were infected with viruses expressing XPA proteins or various control viruses and harvested 2–3 days postinfection. Samples were then briefly sonicated in loading buffer and boiled prior to SDS-PAGE and Western analysis. For affinity purification of FlagXPA proteins, anti-Flag beads (Kodak, Rochester,NY) were incubated with protein extracts isolated from cell monolayers by washing cells with PBS, then lysing the cells for 30 min on ice in buffer consisting of 50 mm Tris-HCl, 0.25 mNaCl, 0.1% NP40, 5 mm EDTA, 1 mm DTT and protease and phosphatase inhibitors (protease inhibitor mixture and phosphatase inhibitor mixture 1 and 2; Sigma Chemical Co.,St. Louis, MO) at the manufacturer’s recommended dilutions. After centrifugation, the supernatant was transferred to a separate tube, and the cell pellet was reextracted using the same lysis solution twice more for 10 min on ice. The pooled supernatants were then bound to anti-Flag affinity beads in solution for 3 h at 4°C. After centrifugation, the supernatant was discarded, and the beads were washed twice with lysis buffer. After washing, the beads were resuspended in a small volume of 10 mm TE buffer [Tris-HCl and 1 mm EDTA (pH 8.0)], and 10 μg of Flag peptide (Kodak, Rochester, NY) was added to release bound FlagXPA protein complexes from the beads. The reaction was incubated on ice for 30 min and then centrifuged. The supernatant containing the eluted proteins was transferred to a new tube and used for analysis by Western blotting. Protein samples were typically separated on a 4–20%Tris-Gly gel (Novex, San Diego, CA) at 125 V for 2 h and then transferred to a polyvinylidene difluoride membrane for 2 h at 0.6 A. After blocking in 5% milk buffer, the membranes were incubated with either mouse anti-Flag antibody (Kodak, Rochester, NY),mouse anti-ERCC1 antibody (Neomarkers, Freeman, CA), or polyclonal rabbit XPA antibody (20). Antibody binding was detected with a secondary IgG antibody conjugated with alkaline phosphatase(Santa Cruz, Santa Cruz, CA), followed by detection by chemiluminescence (Tropix, Bedford, MA). Cellular XPA appears as a band of M_r ∼40 on a gel.

A modification of our earlier procedure was followed for HCR assay(3). Briefly, a confluent cell monolayer in a 6-cm tissue culture dish was infected with adenovirus (XPA virus or control virus)at a MOI of 3–30 and after a 3 h incubation at 37°C while slowly rocking, the medium was removed and replaced with fresh culture medium. After 2 days, infected cells were trypsinized, and reseeded in triplicate in a 96-well microtiter plate as 2-fold serially dilutions with ∼1 × 10⁵ cells seeded in the first row. On the following day, AdLacZ irradiated with increasing doses of UV (0–400 J/m²) was added to the 96-well plate at a MOI of 10 for 3 h, then removed, and replaced with fresh media. After 3 days, the medium was removed from all of the wells, and the cells were fixed with a solution of 3%paraformaldehyde/0.25% glutaraldehyde for 10 min at room temperature. After discarding the fixative and washing the plate with PBS, 50 μl of the fluorescent substrate resorufin-β-d-galactose(Boehringer-Mannheim, Indianapolis, IN) at 0.15 mm in PBS supplemented with 1 mm MgCl₂ was added to each well, and the plate incubated at room temperature in the dark for 1 h. The plate was then read on a FluoroCount (Packard, Meriden,CT) fluorometer using an excitation filter of 530 nm and an emission filter of 620 nm. Fluorescence readings from cells not infected with AdLacZ were subtracted from all other readings as background. Relative HCR was determined by dividing the average readings for a given UV dose by the average readings from wells with cells infected with undamaged AdLacZ. A larger relative fluorescence indicates larger expression ofβ-galactosidase and more DNA repair. GFP fluorescence of cells infected with AdCMV-GFP is negligible at these filter settings.

To determine the removal of UV damage, a modification of the immune assay for pyrimidine dimers was followed (22, 23, 24). Briefly, 75% confluent dishes of A549 cells were infected with AdCMV,AdCMV-FlagXPA_59–273, or AdCMV-FlagXPA_59–114 at a MOI of 10. Two days after infection, cells were trypsinized and reseeded at 1:8 in 6-cm dishes. One dish was used for each experimental point including a dish for unirradiated control. The next morning, dishes were UV irradiated at a dose rate of 2 W/m² to a total dose of 5 or 10 J/m². The cells were then harvested at various times and cell pellets frozen on dry ice. Cell lysis buffer [0.1 m Tris (pH 7.5), 0.15 m NaCl, 12.5 mm EDTA, and 1% SDS) containing 200 μg/ml proteinase K was added to the cell pellets and sonicated followed by incubation at 42°C for 5 h. DNA was isolated by phenol:chloroform extraction followed by ethanol precipitation. The DNA was resuspended in 200 μl of TE buffer with 5 μg of RNase A and incubated at 37°C for 2 h, followed by a second phenol:chloroform extraction and ethanol precipitation. The DNA was resuspended in 10 μl of TE, and 1 μl was spotted in triplicate on duplicate GeneScreenPlus (NEN) membranes. When dry, the membranes were blocked with 5% nonfat dry milk in TBS-T [0.1 m Tris-HCl (pH 7.5), 0.1 m NaCl, and 0.005%Tween 20] overnight at 4°C. The membranes were then washed briefly in TBS-T, and incubated with a 1:3000 dilution of either anti-CBPD or anti-PP (Kamiya Biomedical, Seattle, WA) mouse monoclonal antibody in 5% BSA in TBS-T for 1 h at room temperature. After washing with TBS-T, the membranes were incubated with a 1:3000 dilution of antimouse alkaline phosphatase (Santa Cruz Biotechnology, Santa Cruz, CA) in TBS-T for 1 h and washed several times in TBS-T followed by incubation with CDP-Star (NEN), and exposure to a chemiluminescence screen. Spots were quantified using volume analysis on a Bio-Rad Molecular Imager FX using Quantity One software (Bio-Rad, Hercules,CA). To normalize the amount of DNA spotted, the membranes were hybridized to ³²P-labeled genomic DNA isolated from A549 cells. Briefly, membranes were prehybridized in Church buffer[1% BSA (w/v), 7% SDS (w/v), 0.25 m phosphate buffer (pH 7), and 1 mm EDTA] containing salmon sperm DNA (0.1 mg/ml)for blocking for several hours at 65°C. ³²P-labeled genomic DNA labeled with a“Prime-it” kit (Stratagene, La Jolla, CA) was then added to fresh prewarmed Church buffer and hybridized overnight at 65°C. Membrane were washed in Church wash buffer [1% SDS in 40 mmphosphate buffer (pH 7.0) and 0.5 mm EDTA]. Membranes were exposed to a phosphorimaging screen, scanned in Molecular Imager, and spots quantified using volume analysis. To analyze the data, the signal from the immunoassay was normalized to the DNA bound to the membrane. The signal from the spots obtained without any UV irradiation was subtracted from all of the values to adjust for background levels. Finally, to obtain the fraction of damage remaining, the normalized,background-adjusted signals obtained from the immunoassay were divided by the mean value for the 0 min time points.

Confluent 6-cm dishes of A549, GM637, and XP12RO(M1) cells were infected with a MOI of 3 with either AdCMV-FlagXPA_59–114,AdCMV-FlagXPA_59–273, or control virus. For the UV toxicity assay, infected cells were incubated for 48 h, then trypsinized and UV irradiated in a small droplet of PBS at a dose rate of 1 W/m². Treated cells were plated in triplicate in 96-well plates and serially diluted. For the cisplatin toxicity assay, infected cells were seeded in 96-well plates at different dilutions at 24 h, allowed to attach overnight, and then treated with cisplatin (Aldrich, St. Louis, MO) the following day for 4 h at 37°C in complete media. After either treatment, the plates were incubated for 5 additional days. Finally, Resazuri(AlamarBlue; Molecular Probes, Eugene, OR) was added to the medium to a final concentration of 10 μm, and the plates were incubated for 3 h at 37°C and then read in the FluoroCount reader at an excitation of 530 nm and an emission of 590 nm. A larger relative fluorescence signal indicates more live cells and, thus, lower toxicity. The results were corrected and normalized as described above for the HCR assay.

To ensure that most cells in culture express the truncated XPA protein,we took advantage of the high infectivity of adenovirus and constructed a recombinant adenovirus expressing amino acids 59–114 of the XPA protein. Because cell infectivity can reach 100% in vitro,it is possible to assess the impact of this treatment by conventional cell toxicity end points (14). To facilitate detection and purification of the XPA_59–114 protein and to distinguish it from any cellular counterpart, we added a Flag epitope tag to the NH₂ terminus of the protein.

To demonstrate expression of the FlagXPA proteins from the constructed adenoviruses, A549 cells were infected with either AdCMV-FlagXPA_59–114 or an adenovirus producing functional XPA protein, AdCMV-FlagXPA_59–273, at an MOI of 30. Although, the XPA protein is M_r 31,000, it migrates as M_r 40,000 by SDS-PAGE(9). After 3 days of incubation, the cells were harvested and processed for SDS-PAGE. After protein separation on the gel,proteins were transferred to a membrane and probed with anti-Flag antibody for detection of the FlagXPA proteins (Fig. 1,A). Infection of cells with AdCMV-FlagXPA_59–273 expressing near full-length Flag-XPA_59–273 protein gave a band of M_r ∼40,000, as expected, and infection with AdCMV-FlagXPA_59–114 produced a single band of less than M_r 5,000 as would be expected for this construct. Termination of the truncated XPA protein is expected to occur in a stop codon residing in the flanking sequence. To unequivocally demonstrate that these Flag proteins were indeed of XPA origin and expressed from the viruses, we then performed parallel immunoblots using either anti-Flag or anti-XPA antibody (Fig. 1,B). We found that both of the FlagXPA proteins also reacted with the anti-XPA antibody. We occasionally detect a second XPA band by Western analysis at a position of M_r≥60,000 (Fig. 1 B, ∗), which disappears when the sample is diluted. It is presently unknown why this happens, but is possible that the FlagXPA_59–273 protein is not entirely dissociated from other proteins or else forms dimers at higher concentrations. Alternatively, the top band could be attributable to translational read-through or to different conformational properties of XPA resulting in aberrant migration by SDS-PAGE (9). In any case, this result demonstrates the efficient expression of FlagXPA proteins from adenovirus in A549 cells.

To address whether FlagXPA_59–114 protein would bind ERCC1 in vivo, which is necessary if it is meant to serve as a decoy for normal cellular XPA, A549 cells were infected with either AdCMV-FlagXPA_59–114 or a control adenovirus (AdCMV-Flagp38) expressing Flag-tagged p38 MAP kinase(25). Two days after infection, cells were harvested and lysed and the resulting protein extract incubated with anti-Flag agarose beads to purify the FlagXPA proteins. Bound protein complexes were then eluted from the beads with Flag peptide and separated by SDS-PAGE, transferred to a membrane, and probed with anti-ERCC1 antibody (Fig. 2, left). It is clear from the result shown that ERCC1 binds specifically to the FlagXPA_59–114 protein because a band of M_r ∼40,000, which is the relative molecular weight of ERCC1 (26), is detected only in extracts from cells infected with the AdCMV-FlagXPA_59–114. No such band appears in the lane with an extract from cells infected with the AdCMV-Flagp38 control virus. The same membrane was then reprobed with anti-Flag antibody to ensure the presence of Flag proteins in the extracts (Fig. 2, right). Indeed, Flag proteins of the expected sizes, i.e., M_r ∼5,000 for FlagXPA_59–114 and 38,000 for Flagp38 MAP kinase,were found in the affinity purified complexes. This result demonstrates that ERCC1 specifically interacts with FlagXPA_59–114 in vivo.

Before undertaking studies assessing the effect of expressing FlagXPA_59–114 on the sensitivity of cells to DNA damage, we first used a toxicity assay to assess the relative toxicity of the recombinant adenoviruses without exposing the cells to any DNA damaging agent (Fig. 3,A). Again, A549 cells were infected with different MOI(3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30). In time course experiments, expression of FlagXPA proteins was detected on day 1, and then increased on days 2 and 3 (data not shown). On day 4, FlagXPA levels were roughly equivalent to those on day 3. We found that the AdCMV-GFP and the AdCMV-FlagXPA_59–273 viruses produced no significant increase in cell toxicity even at the highest MOI of 30. The relative growth of cells infected with AdCMV-GFP compared with uninfected cells was 0.94, and with AdCMV-FlagXPA_59–273, it was 1.03. However,infection of A549 cells with AdCMV-FlagXPA_59–114did prove toxic to cells, resulting in a relative growth of 0.56 at a MOI of 30. In parallel, samples were analyzed for the expression of FlagXPA_59–114 (Fig. 3 B). An excellent correlation between cell toxicity/growth inhibition and expression of FlagXPA_59–114 was noticed. With a MOI of 3 that produced only minor toxicity, no FlagXPA_59–114 protein was detected by Western analysis, whereas at higher MOI, both toxicity and expression were noted. This result demonstrates that FlagXPA_59–114 protein is toxic to cells when expressed at increased levels, which suggests that XPA or ERCC1 function is required even without any exogenously inflicted DNA damage to the cell. In subsequent experiments, we used a low MOI because of the relative toxicity of AdCMV-FlagXPA_59–114 in these cells.

To see whether expression of FlagXPA_59–114reduced DNA repair and to demonstrate proof of the principle of our approach, we infected A549 cells with AdCMV-FlagXPA_59–114 and, 2 days later, tested for HCR of UV-damaged AdLacZ virus (Fig. 4). HCR is an indirect assay for DNA repair that relies on the notion that damage needs to be removed in order for transcription to occur,and, thus, the greater the relative reporter gene activity, the more proficient the DNA repair (3, 7). At a dose of 400 J/m², the β-galactosidase activity in cells that had not been infected with virus prior to HCR was 0.72 relative to readings from cells infected with undamaged AdLacZ. By comparison, theβ-galactosidase activity in control cells initially infected with AdCMV-GFP and then with UV-irradiated AdLacZ was 0.94. At the same UV dose, cells infected with AdCMV-FlagXPA_59–114showed a significantly reduced fraction of β-galactosidase activity of 0.29 (P = 0.0057). This result demonstrates that infection of A549 cells with AdCMV-FlagXPA_59–114 impairs HCR and also suggests that perhaps cell survival after UV would be reduced.

Because HCR is an indirect DNA repair assay that relies on transcription of a repaired reporter gene, it was important to demonstrate that infection with AdCMV-FlagXPA_59–114 results in increased cellular sensitivity to UV exposure. Indeed, we found that infection of A549 cells with virus-expressing FlagXPA_59–114increased toxicity and inhibited growth after UV damage (Fig. 5). Growth of uninfected cells exposed to 45 J/m²of UV was reduced to 0.70 compared with cells not exposed to UV. Similarly, cells exposed to the same UV dose after infection with the control virus AdCMV reduced growth to 0.60. However, cells infected with AdCMV-FlagXPA_59–114 at a MOI of 3 before exposure to UV showed a significantly greater reduction in growth of 0.47 (P = 0.0387). Because UV irradiation of cells were carried out in a droplet, relatively high UV doses were required to see toxicity under these experimental conditions. To confirm this result using other cells and to establish the relative effect of AdCMV-FlagXPA_59–114 infection on the UV sensitivity of repair-competent cells to that of XPA cells, a second experiment was carried out with NER-deficient XPA [XP12RO(M1)] and normal NER-proficient (GM637) fibroblasts (Fig. 6). As expected, both uninfected and control-infected XP12RO(M1) cells were much more sensitive to UV than were GM637 cells, and AdCMV-FlagXPA_59–273 expressing nearly full-length XPA protein rescued XP12RO(M1) cells to GM637 levels. Most importantly, infection of GM637 with AdCMV-FlagXPA_59–114 resulted in UV sensitization that approached the XP12RO(M1) levels. These results demonstrate that infection of two different human cell lines that are normal in NER with AdCMV-FlagXPA_59–114 UV sensitizes the cells to levels approaching those of XPA.

It was of interest to determine whether infecting cells with AdCMV-FlagXPA_59–114 could also be applied to increase cell toxicity to a clinically important drug such as cisplatin, which causes DNA damage that is repaired by NER. We found that infection with virus expressing FlagXPA_59–114 also inhibited the growth of A549 cells after exposure to cisplatin (Fig. 7). Uninfected cells given a dose of 20 μm cisplatin for 4 h showed reduced growth compared with untreated cells of 0.49,and cells infected with AdCMV control virus before the same cisplatin treatment were slightly more sensitive with a fraction of growth relative to untreated infected cells of 0.32. Cells infected with AdCMV-FlagXPA_59–114 before challenge with cisplatin were much more sensitive, showing only 0.15 of the growth of cells infected with the same virus but not exposed to cisplatin(P = 0.0026). These results demonstrate that infection with AdCMV-FlagXPA_59–114 renders cells more sensitive to cisplatin as well as to UV, which suggests that this approach could potentially be used to increase the sensitivity of tumor cells to cisplatin and other clinically important drugs that produce bulky DNA damage.

To investigate the mechanism by which AdCMV-FlagXPA_59–114 sensitizes normal cells to UV, we examined the removal of pyrimidine dimers from the DNA of irradiated cells using a immune assay for CBPDs and PPs (22, 24). First, we examined the rate of CBPD and PP removal in A549 cells exposed to 10 J/m² (Fig. 8,A). As expected, we found that PP lesions were removed much faster from the DNA than were CBPDs with half lives of <1 h and 15–24 h, respectively, in agreement with previous reports (22, 24, 27). We found no difference in the removal of CBPDs between uninfected cells, cells infected with AdCMV (control), and cells infected with AdCMV-FlagXPA_59–114 (Fig. 8,B), which suggests that expression of FlagXPA_59–114 did not interfere with the removal of CBPDs. However, when the removal of PPs were examined, we found dramatic differences between cells infected with AdCMV-FlagXPA_59–114 compared with uninfected and AdCMV-infected cells (Fig. 8 C). To our surprise, cells infected with AdCMV-FlagXPA_59–114 appeared to generate more PPs over time compared with controls. Because it is inconceivable that more lesions were generated with time after UV irradiation, the most reasonable interpretation of this result is that the PP epitope is better exposed to the antibody, perhaps resulting from partial DNA repair of the PPs. In any case, the results suggest that the removal of PPs but not CBPDs is affected by the expression of FlagXPA_59–114.

Drugs such as cisplatin, mitomycin C, psoralen in combination with UVA (PUVA), and melphalan are commonly used for the treatment of ovarian, testicular, prostate, and leukemic cancers. The DNA damage induced by these drugs is to a large extent repaired by NER (2, 28). One major problem with current modalities for the treatment of cancer using drugs that cause DNA damage is the relatively small therapeutic indices that can be achieved. In addition, during the course of chemotherapy, tumor drug resistance occasionally develops,and doses may need to be escalated to control tumor growth(29). It would be highly desirable to develop approaches that would increase therapeutic ratios and lower the doses administered to the patient. Attempts are now made in many laboratories to target chemotherapy specifically to tumor cells by using efficient gene delivery tools, such as adenovirus (29, 30, 31), and to attack cellular functions that influence drug resistance, such as apoptosis,thereby improving cancer therapy. Thus far, little has been done to specifically target NER to improve chemotherapy (32).

We previously reported on a virus similar to the one described here that expressed the same truncated XPA protein without the Flag epitope that sensitized cells to UV (33). However, we were unable to clearly show expression of this truncated protein in infected cells. Detecting the presence of the Flag epitope on the truncated XPA protein made by the virus described in the present study unequivocally demonstrates that this protein is made by the virus and not by the cell. Results from the present study confirm earlier findings that ERCC1 binds to XPA (12, 13). However, the in vivo binding between full-length ERCC1 and XPA proteins was previously demonstrated only with GAL4-fusion proteins in a yeast two-hybrid screen (11, 13). The result from our study clearly demonstrates that interaction between the 59–114-amino-acid domain of XPA and ERCC1 occurs in human cells in vivo. We are currently investigating whether other proteins, including XPF(12), are also present in this protein complex. The presence of an epitope tag combined with the efficient expression from a highly infectious adenovirus should be very useful in isolating large quantities of XPA protein complexes and for determining the composition and changes in posttranslational modifications in these complexes that occur in vivo in response to DNA damage.

It was demonstrated previously that a mutant XPA (ΔG) with 4 amino acids (GGGF) deleted in the G-domain was unable to bind ERCC1 but was able to compete with normal XPA protein and to reduce DNA repair in vitro (11). It was suggested that ΔG reduced repair because its DNA binding ability was still retained but was not loading ERCC1 protein on to the damaged DNA site(11). Our results suggest that the reciprocal event is also possible, i.e., the FlagXPA_59–114 protein encompassing the G-domain is able to bind ERCC1, and perhaps the resulting complex competes with the formation of normal XPA/ERRC1/(XPF?) complexes. However, at this point, we have no direct evidence that this occurs. Very recently, it was shown that the interaction between XPA and ERCC1 is modulated indirectly by a protein kinase (34), which would provide another means by which NER may be inhibited to increase the sensitization of tumor cells to chemotherapy.

The results from this study demonstrate that overexpressing truncated XPA protein is toxic to cells even without treating the infected cells with any DNA-damaging agent, which suggests that either NER is needed at all times or that the function of some other protein needed for normal cellular growth is inhibited by the truncated XPA protein. However, we were able to establish the virus dose range within which a sensitizing effect on treatment with UV or cisplatin could be clearly distinguished from the toxicity observed from the virus itself. Perhaps by controlling the expression of the FlagXPA_59–114 protein by using an inducible promoter, one could alleviate the toxicity of the protein when cells are not treated with a drug. Along the same line, future virus vectors may be designed with a DNA damage-inducible promoter to reduce the expression of truncated XPA in the absence of DNA damage and, thus,limit toxicity in the absence of drug treatment. Use of such vector would be expected to improve the therapeutic ratio.

How might the expression of Flag-XPA_59–114affect NER? From these studies, we can conclude that the most likely mechanism for the increased cell toxicity of AdCMV-FlagXPA_59–114 is that expression of FlagXPA_59–114 interferes with the removal of PP lesions. This effect is specific for PPs because the removal of CBPDs was not affected. Of interest is the result from the immune assay that suggests that the PPs in the DNA isolated from cells infected with AdCMV-FlagXPA_59–114 is exposed more efficiently to the antibody specific for the PP epitope. The most reasonable explanation for this result is that the PP has been altered in some fashion but is still attached to the DNA. Conceivably, this alteration could be attributable to the incision on one side of the lesion,presumably the 3′ side (35), resulting in a ‘flap’ that is more exposed to the antibody and thus would result in an increased signal. However, at the present time, we can only speculate as to what the exact structure of this lesion might be.

In summary, our results suggest that using an adenoviral vector expressing a truncated XPA protein to inhibit NER in tumor cells could be an effective combined modality for the treatment of various cancers with drugs known to cause bulky lesions repaired by NER. With this approach, the targeting of tumor cells should be more specific, and perhaps the doses of chemotherapy could be lowered, resulting in more effective treatment with fewer side effects to the patient.