Involvement of base excision repair in response to therapy targeted at thymidylate synthase

Introduction

Thymidylate synthase (TS) catalyzes the conversion of deoxyuridylate to thymidylate (TMP) for DNA biosynthesis, using N⁵,N¹⁰-methylenetetrahydrofolate as the methyl donor. Thymidylate deprivation is induced by inhibition of TS and is the therapeutic effect of several classes of antineoplastic drugs that are used to treat colorectal cancer. The folate analogue raltitrexed (RTX, ZD1694, Tomudex) and the nucleoside of FUra, 5-fluoro-2′-deoxyuridine (FdUrd), are highly selective for TS (1). TMP is unique in deoxynucleotide metabolism in that it is synthesized by a pathway not involving ribonucleotide reductase. Prolonged deprivation of TMP causes cell death in all species examined, unlike other nutritional deficiencies that generally produce cytostatic effects (2). Although the precise biological responses to thymidylate deprivation remain obscure, the following observations have been noted. Thymidylate deprivation induces DNA strand breaks and a nucleotide imbalance (3-8). The nucleotide pool imbalances seen include an accumulation of dUTP, which leads to uracil incorporation into DNA. Several studies have examined the role of dUTPase, which hydrolyzes dUTP to dUMP, in preventing uracil incorporation into DNA (9, 10). Base excision repair (BER) purges uracil from the genome, but under conditions of thymidylate deprivation, dUTP would presumably be reincorporated during repair synthesis (11). This proposed futile cycling would subsequently cause the persistence of BER intermediates, which are strand breaks.

At least four genetic loci in humans encode for uracil DNA glycosylases (UDG), which excise uracil bases from DNA to initiate BER (12). Biochemical characterization of the proteins suggests specialized roles that combat two sources of uracil introduction into the genome, deamination of cytosine and incorporation of dUMP during replication. The UDG encoded by the ung locus seems to counteract uracil misincorporation during replication (13, 14). Recent results with UDG-overexpressing cells did not suggest that UDG activity was associated with sensitivity to TS inhibitors, although the authors noted that UDG-overexpressing cells displayed sensitivity to a 24-hour (but not continuous) exposure to TS inhibitors (15). A recent study in Saccharomyces cerevisiae found that inactivating UDG induced resistance to shorter exposures to antifolates and allowed the cells to progress through S phase into G₂-M phase (16).

The enzymatic steps of the BER pathway following monofunctional DNA glycosylase activity include AP endonuclease, end trimming activity to remove the deoxyribose, DNA synthesis to replace the lost nucleotide(s), and DNA ligation. Because of the redundancy of mammalian UDGs, we chose to look downstream in the pathway to investigate the role of mammalian BER in the cellular response to TS inhibitors. DNA polymerase β (β-pol) performs two functions in BER. β-pol acts as a DNA polymerase to replace the lost nucleotide and a deoxyribose lyase to excise the 5′-deoxyribose group. XRCC1 is thought to be a scaffold protein that coordinates the final steps of BER via its interactions with β-pol, DNA ligase III, and PARP-1 (17). Although β-pol plays an essential role in the completion of short patch BER, it lacks a 3′-5′ proofreading exonuclease activity, and is significantly more error-prone than replicative polymerases (18). Interestingly, there have been reports of increased β-pol mRNA and protein levels in prostate, breast, and colon cancer tissues, compared with normal tissues and cells (19, 20), and of variant forms of β-pol in tumors (21). There is also speculation that overexpression of β-pol could contribute to genetic instability because of its poorer fidelity compared with the replicative polymerases (20). We used mouse embryonic fibroblasts (MEF) wild-type or null for β-pol to investigate the role of mammalian BER in response to TS inhibitors. The results reveal that DNA β-pol negatively influences the cellular response to TS inhibitors.

Materials and Methods

Drugs and Cell Culture

RTX was generously supplied by AstraZeneca, United Kingdom. Methyl methanesulfonate, FdUrd, nocodazole, sulforhodamine B (SRB), and bromodeoxyuridine (BrdUrd) were purchased from Sigma Chemical Co (St. Louis, MO). FITC-conjugated anti-BrdUrd monoclonal antibody was purchased from Becton Dickinson (San Jose, CA). MB16tsa (wild-type) and MB19tsa (β-pol null) MEFs were obtained from American Type Culture Collection (Manassas, VA) and were maintained in DMEM (Life Technologies, Inc., Carlsbad, CA) supplemented with 10% fetal bovine serum (Hyclone, Logan, UT) and penicillin (50 units/mL)-streptomycin (50 μg/mL, Sigma) at 34°C in a 5% CO₂ humidified incubator (22). We have verified that the wild-type and β-pol null cells are uninfected with mycoplasma. MB16 (wild-type) and MB16B (stably expressing siRNA against β-pol mRNA) MEFs were a generous gift from Dr. Holly Miller, SUNY-Stony Brook. The EM9 (XRCC1-deficient) and AA8 (parental control) Chinese hamster ovary (CHO) cells were obtained from ATCC and were maintained in α-MEM (Sigma) supplemented with 10% fetal bovine serum and penicillin (50 units/mL)-streptomycin (50 μg/mL) at 37°C in a 5% CO₂ humidified incubator (23).

Cytotoxicity Assays

Short-term cell cytotoxicity was done by SRB colorimetric assay with minor modifications (24). In brief, ∼500 cells were plated on 96-well plates 24 hours before treatment. The cells were treated with various concentrations of RTX or FdUrd for 24 hours, then grown in drug-free medium for 3 days post-drug treatment. Cells were fixed with 10% trichloroacetic acid in 0.9% NaCl solution, washed, and stained with 0.4% SRB in 1% acetic acid. After washing, the bound SRB was redissolved in 10 mmol/L Tris base (pH 10.5). Absorbance was measured using a plate reader at 560 nm (UVmax Kinetic Microplate Reader, Molecular Devices, Sunnyvale, CA). The colony-forming assay was done by plating 200 to 500 cells in 100-mm culture dishes 24 hours before the addition of drug. For drug exposures in dialyzed serum, drug-containing medium was replaced with drug-free medium following a 24-hour incubation. Colonies were fixed with 100% methanol and stained with Giemsa following 9 to 10 days growth.

Cell Cycle Analysis and BrdUrd Incorporation

Exponentially growing cells were exposed to RTX at the concentrations and time points indicated in the text. In recovery experiments, drug-containing medium was removed and replaced with drug-free medium. At the appropriate times, adherent and detached cells were collected in medium, fixed in 70% ethanol, and stored at 4°C until processed for analysis. For single parameter cell cycle analysis, fixed cells were pelleted, washed with PBS, and stained with 0.5 mL RNase-PI staining solution containing propidium iodide (40 μg/mL) and RNase A (10 μg/mL). For dual parameter cell cycle analysis, cells were pulse labeled with 10 μmol/L BrdUrd for 30 minutes immediately before harvesting at the indicated treatment or recovery times, then harvested and fixed as above. The fixed, BrdUrd-labeled cells were washed with PBS and resuspended in 1 mL of 2 N HCl solution containing 0.5% Triton X-100 to denature the DNA for 30 minutes at room temperature, then neutralized with 0.1 mol/L sodium borate (pH 8.5). The cell suspensions were washed twice with PBS, once with PBS containing 0.5% Tween 20 and 0.1% bovine serum albumin, then incubated with anti-BrdUrd FITC-conjugated antibody overnight at 4°C. On the following day, cells were pelleted, washed with PBS, and resuspended in RNase-PI solution. DNA content and BrdUrd incorporation were determined using an EPICs XL-MCL flow cytometer (Beckman Coulter, Fullerton, CA), in the Instrumentation Resource Facility at the University of South Carolina School of Medicine.

Western Blot Analysis

Cells were exposed to 50 nmol/L RTX for 24 hours, and immediately harvested by trypsinization. Cellular extracts were obtained by lysing the cells in ice-cold lysis buffer containing 1% NP40, 20 mmol/L Tris (pH 7.4), 137 mmol/L NaCl, and 10 mmol/L EDTA in the presence of protease inhibitors (5 mg/L pepstatin A and leupeptin, 25 mg/L aprotinin, 1 mmol/L pefabloc), followed by centrifugation at 10,250 × g for 10 minutes. Total protein (30 μg) was separated by 12% SDS-PAGE, and transferred to Immobilon-P membranes (Millipore, Bedford, MA) in blotting buffer (25 mmol/L Tris base, 192 mmol/L glycine, 20% methanol) overnight. After blocking for 3 hours with 5% nonfat dry milk in TBS-T buffer, the membrane was incubated with a rabbit anti-TS polyclonal antibody (1:100 dilution in TBS-T, Antibody Core, Institute for Biological Research and Technology, University of South Carolina) for 1 hour at room temperature. The membrane was then washed with TBS-T buffer and incubated with horseradish peroxidase-conjugated anti-rabbit IgG secondary antibody (1:100,000 dilution in TBS-T) for 1 hour at room temperature. The TS was visualized using the enhanced chemiluminescence advance detection kit (Amersham Biosciences, Piscataway, NJ) and autoradiography. To determine the extent of drug-TS complex formation, the cellular extracts were incubated with 5′-fluoro-2′-deoxyuridine monophosphate (15 nmol/L) and N⁵,N¹⁰-methylenetetrahydrofolate (110 μmol/L) at 30°C for 20 minutes, before separation by 12% SDS-PAGE; blotting was done as described above.

Results

Sensitivity of BER-Proficient and -Deficient Cells to TS Inhibitors

We determined the sensitivity of the wild-type and β-pol null cells to RTX. RTX is a folate analogue that binds tightly to the folate-binding site of TS. Sensitivity to RTX was determined by SRB assay as described (Materials and Methods) and the data are shown in Fig. 1A. MEFs that lack β-pol were 6-fold more resistant to RTX (IC₅₀ = 90 nmol/L) than MEFs that are wild-type (IC₅₀ = 15 nmol/L). Although both cell lines were more sensitive when the drug exposure was increased to 48 hours, β-pol null cells remained resistant compared with wild-type cells (data not shown). RTX sensitivity experiments in dialyzed serum were also done to minimize the effect of thymidine salvage. The wild-type cells remained more sensitive than β-pol null cells (IC₅₀ = 5.5 nmol/L versus 27 nmol/L, respectively), despite the relative increase in sensitivity in both cell lines (Fig. 1B). The results in dialyzed serum suggest that the difference in sensitivity is not due to a difference in the ability to salvage thymidine in regular serum. The response of the wild-type and β-pol null cells to FdUrd was also determined. FdUrd is anabolized to a nucleotide analogue of dUMP that forms a suicide complex with TS in the presence of the folate co-factor. In agreement with the results for RTX, the β-pol null cells displayed an approximately 8-fold resistance to FdUrd compared with the wild-type cells (data not shown). The β-pol null cells retained a methyl methanesulfonate–sensitive phenotype, as reported previously (data not shown; Ref. 22). Colony-forming assays were also done in dialyzed serum with the wild-type and β-pol null cells; the wild-type cells were more sensitive to 24 hours RTX exposure than the β-pol null cells, in agreement with the viability assay (data not shown). To further confirm the involvement of β-pol in the response to TS inhibitors, we used a stable β-pol “knockdown” subline of the MB16 β-pol^+/+ murine embryonic fibroblast cell line that stably expresses siRNA against β-pol mRNA. The β-pol protein levels are essentially undetectable (25). As shown in Fig. 2, the β-pol knockdown cells were >4-fold more resistant to killing induced by RTX than the wild-type cells in dialyzed serum (IC₅₀ = 18 nmol/L versus 4 nmol/L, respectively). Lastly, to determine whether the response was specific to the β-pol cell lines or a consequence of the interaction of β-pol with the remainder of BER, the toxicity of RTX in XRCC1-proficient and -deficient CHO cells was determined. The XRCC1-deficient EM9 cells were found to be approximately 1.5-fold more resistant to RTX compared with the parental AA8 cell line (Fig. 3, IC₅₀ = 32 nmol/L versus 22 nmol/L, respectively). Taken together, the data indicate that thymidylate deprivation produces a different DNA damage or damage response compared with alkylating agents in BER-deficient models.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Cell viability in wild-type and β-pol null MEFs treated with RTX in medium containing normal serum or dialyzed serum. Wild-type (▪) and β-pol null (▴) cell lines were exposed to RTX in medium containing regular serum (A) or dialyzed serum (B) 24 hours postplating on 96-well plates. After 24 hours exposure, drug-containing medium was removed. Following 3 days recovery in drug-free medium, growth inhibition was determined by SRB assay (Materials and Methods). Results are shown as the percentage compared with untreated control. Points, average of three independent experiments; bars, SD.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Cell viability in wild-type and β-pol knockdown MEFs treated with RTX in medium containing dialyzed serum. Wild-type (▪) and β-pol knockdown (♦) cell lines were exposed to RTX in medium containing dialyzed serum 24 hours postplating on 96-well plates. After 24 hours exposure, drug-containing medium was removed. Growth inhibition was determined following 3 days recovery in drug-free medium with dialyzed serum. Points, average of three independent experiments; bars, SD.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

RTX induced cytotoxicity in XRCC1-proficient and -deficient cell lines. The CHO AA8 (parental control, ▪) and EM9 (XRCC1-deficient, ♦) cell lines were plated on 96-well plates 24 hours before treatment in medium with dialyzed serum. Following 24 hours treatment with RTX, the drug-containing medium was removed, and growth inhibition was determined following 3 to 4 days recovery in drug-free dialyzed medium. Results are shown as the percentage compared with untreated control. Points, average of three independent experiments; bars, SD.

TS Expression Levels in Wild-Type and β-pol Null Cells before and after TS Inhibitor Treatment

Increased TS protein levels are known to inversely correlate with sensitivity to TS inhibitors such as FdUrd, RTX, and ZD9331 (26-28). Additionally, TS is induced after exposure to high concentrations of TS inhibitors and the extent of induction is dose dependent (29). We, therefore, determined TS expression levels in the β-pol^+/+ and β-pol^−/− cells and the extent of induction in TS levels after exposure to RTX. TS levels in the cell lines were analyzed in untreated cells and after exposure to 50 nmol/L RTX and corrected for the β-actin loading control (Fig. 4A). Although there was ∼1.9-fold more TS in the β-pol^−/− cells before treatment, the TS levels in the β-pol^+/+ and β-pol^−/− cells were very similar after a 50 nmol/L dose of RTX (Fig. 4A). The lack of significant induction of TS in either cell line following RTX treatment is presumably due to the relatively low dose of RTX used. Additionally, we incubated cell extracts with N⁵,N¹⁰-methylenetetrahydrofolate and 5′-fluoro-2′-deoxyuridine monophosphate to confirm that the TS present in the wild-type and β-pol null cells can be liganded to a similar extent. The resulting covalent ternary complex with TS is detectable as a more slowly migrating band than that for unliganded TS (Fig. 4B). The TS in both extracts can be shifted on ternary complex formation. Taken together, the data suggest that the difference in sensitivity does not seem to be due to differences in TS levels or inhibitor binding to TS in the wild-type and β-pol null cells.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

TS protein in wild-type and β-pol null MEFs by Western blotting. A, TS levels in untreated and treated cells. Cellular extracts were prepared from wild-type and β-pol null cells that were untreated (control) or treated with RTX (50 nmol/L) for 24 hours, and were probed with antibodies to TS or β-actin. B, extent of drug-induced complex formation of TS. Cellular extracts of wild-type and β-pol null cells were incubated in the absence (control) or presence (liganded) of FdUMP and methylene-THF to form a ternary complex of TS. Blots shown are representative experiments; experiments were repeated 3 to 5 times.

Cell Cycle Analysis of Wild-Type and β-pol Null Cells following RTX Treatment

We determined the effects of RTX on cell cycle distribution during treatment with and recovery from RTX to investigate the cellular responses of the wild-type and β-pol null cells. As shown in Fig. 5, the wild-type and β-pol null cells display a similar response to a 12-hour or 24-hour exposure to RTX. There was an induction of an early S-phase arrest, accompanied by a progressive loss of the G₂-M cell population (Fig. 5). The arrest is interpreted as early S, not G₁-S boundary, because the peak in the treated population has shifted to the right of the control G₁ population (Fig. 5). To determine S-phase progression, we measured the extent of BrdUrd incorporation into DNA at 24 hours RTX treatment and during recovery at time points of 3, 6, and 24 hours. The cells were pulse labeled with BrdUrd for 30 minutes before the end of the RTX treatment or recovery time point. In Fig. 6, the left-hand panel shows DNA content as measured by PI staining and the right-hand panel shows a density plot for cells with DNA content on the horizontal axis and BrdUrd incorporation on the vertical axis. At a dose of 50 nmol/L RTX, the profiles were very similar at 24 hours treatment. In particular, the number of cells scored as BrdUrd positive increases from 50% to 60% up to >90% in both wild-type and null cells (Fig. 6, right-hand panel, 24 h). BrdUrd is itself a thymidine analogue rapidly incorporated during S phase; we, therefore, conclude that the RTX treatment arrested the cells in early S phase. Although the initial arrest in response to RTX seems to be very similar for the wild-type and β-pol null cells, differences become apparent following recovery. Specifically, the β-pol null cells, which were more resistant, appeared to resume progression through the cell cycle (Fig. 6, Post 3 h, Post 6 h). The wild-type cells, however, resume cell cycle progression slower (Fig. 6, Post 3 h, Post 6 h). We confirmed this observation by including nocodazole, an agent that arrests cells at G₂-M (Fig. 5, bottom). The cells were treated for 24 hours with 50 nmol/L RTX; the RTX-containing medium was removed and replaced with medium containing nocodazole for 12 hours. A significant accumulation of G₂-M cells occurred in the β-pol null cells, whereas no accumulation of G₂-M cells was observed in the wild-type cells. By 24 hours, a substantial proportion of cells possess an intermediate DNA content between G₁ and G₂, but did not stain BrdUrd positive (Fig. 6). The results suggest that these BrdUrd-negative cells might have resumed replication following removal of RTX but then replication stalled and was not active at the time of BrdUrd labeling. In summary, although the initial cell cycle arrest seems to be similar for wild-type and β-pol null cells, the β-pol null cells recovered more rapidly and progressed through the cell cycle.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Single-parameter cell cycle profiles in wild-type and β-pol null cells treated with RTX. Exponentially growing cells were exposed to 0 nmol/L (Control) or 50 nmol/L RTX for 24 hours in medium containing normal serum. Cells were harvested at indicated time points, stained with propidium iodide, and processed by fluorescence-activated cell sorting. Control, untreated cells; 12h and 24h time points, cells treated for 12 hours and 24 hours, respectively, and immediately harvested. Post, cells treated with 50 nmol/L RTX for 24 hours followed by 12 hours incubation in either drug-free or nocodazole-containing medium. Nocodazole Only 6h, cells that were incubated with nocodazole alone for 6 hours. Histograms shown are representative experiments; experiments were repeated at least 3 times.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Dual parameter cell cycle analysis of wild-type and β-pol null cells treated with RTX. Wild-type and β-pol null MEFs were untreated (Control) or treated with RTX for 24 hours. Cells were pulse labeled with BrdUrd for 30 minutes immediately before harvesting at the indicated time points during and after treatments with RTX. Cells were then fixed, incubated with conjugated BrdUrd-FITC antibody, stained with propidium iodide, and dual parameter cell cycle analyses were done by FACS. Post 3 h, 6 h, and 24 h, recovery in drug-free medium for 3, 6, or 24 hours following 24 hours exposure to RTX. Histograms shown are representative experiments; experiments were repeated 3 times.

Discussion

Thymidylate deprivation has been investigated because of the clinical use of TS inhibitors in chemotherapy. Furthermore, folate deprivation resulting from dietary insufficiency or induced by antifolate therapy and some anticonvulsant agents leads to thymidylate deprivation. Folate deprivation has been shown to induce nucleotide pool imbalance, uracil misincorporation, and strand breaks (30, 31). Previous studies in Escherichia coli and S. cerevisiae provided reasonable evidence that uracil misincorporation and removal results in DNA strand breaks (32-36). Interestingly, resistance to thymidylate deprivation has previously been noted in S. cerevisiae strains deficient in post-replicative repair (35). The involvement of BER in response to thymidylate deprivation has been investigated because BER plays a crucial role in removing uracil from the genome (11). Studies in mammalian cells undergoing thymidylate deprivation have also revealed nucleotide imbalance, which includes increases in dUTP and/or dATP, and strand breaks (3-5). Cell cycle studies showed an early S-phase arrest in response to TS inhibitors (37-39). Matsui et al. (39) concluded that cells were entering S phase before arresting, rather than arresting at the G₁-S boundary. Our results agree; the wild-type and β-pol null cells seem to arrest after entering S phase during a 24-hour exposure to RTX. We, therefore, feel that the difference in sensitivity for the wild-type and β-pol null cells is not due to the initial response to TS inhibition, but that the difference lies in the recovery from TS inhibition.

The role of uracil incorporation and removal in the toxicity of TS inhibitors has been investigated through several approaches. Studies using dUTPase to suppress levels of dUTP available for misincorporation indicated that dUMP incorporation contributes to the cytotoxicity of TS inhibitors, although the effect of uracil misincorporation and removal appeared to be both time and cell line dependent (10, 38). A recent study isolated cells that overexpressed UDG and determined sensitivity to TS inhibitors (15). Although a statistically significant effect was observed on cell viability following a 24-hour treatment (UDG-overexpressing clones were more sensitive), UDG overexpression did not affect sensitivity following longer drug incubation times or clonogenicity (15). In a recent study in S. cerevisiae, inactivation of UDG resulted in resistance to shorter exposures to antifolates and allowed the cells to progress through S phase into G₂-M phase (16). Our results agree in that the β-pol null cells were resistant to TS inhibitors and appeared to progress through S phase following recovery from RTX treatment. The above studies suggest a role for uracil incorporation and removal in the response to TS inhibitors, but also imply that the effect of uracil misincorporation and/or its removal by UDG on survival is transient and sensitive to the duration of treatment and recovery. Prolonged/severe thymidylate deprivation has other effects that might mask the problems associated with uracil incorporation and removal. However, β-pol null cells were relatively resistant to RTX exposures of 24 and 48 hours in viability assays and 24 hours in the colony-forming assay. Inactivation of XRCC1, which interacts with β-pol, DNA ligase III, and PARP-1, also led to a modest resistance to RTX. Taken together, the studies suggest that uracil misincorporation followed by the action of BER to remove uracil from DNA contributes to the cytotoxicity of TS inhibitors. The results presented here suggest that β-pol plays an important role.

It is not clear whether the negative influence of β-pol during thymidylate deprivation results from its polymerase activity, deoxyribose lyase activity, functional association with other BER components, or other interactions outside the known BER pathway. It is difficult to envision how loss of the deoxyribose lyase activity might contribute to resistance, because the deoxyribose group seems to be a toxic BER intermediate (40-42). Overexpression of β-pol results in a mutator phenotype (43); the lower fidelity of β-pol might, therefore, contribute to genomic instability under conditions of thymidylate deprivation and nucleotide imbalance. Proliferating cell nuclear antigen and β-pol were recently shown to interact (44), although the significance of this interaction is not yet known; proliferating cell nuclear antigen is required for long-patch but not short-patch BER (45, 46). Long patch-BER, which seems to be preferred at replication forks (14, 47), would presumably act in the absence of β-pol. Alternatively, uncoupling the cross-talk between BER and apoptotic processes might lead to resistance under conditions of thymidylate deprivation. BER might communicate with cell death machinery (48, 49); interactions between p53 and β-pol have also been reported (50, 51). BER-mediated communication with cell death might also occur through poly(ADP)-ribose polymerase 1 (PARP-1) even though its precise role in BER remains debated (52-56). Lastly, BER intermediates seem to be targets for recombination in a process that is independent of p53 and MMR (57). The influence of BER on recombination during thymidylate deprivation has not yet been examined to our knowledge. We are currently exploring the above possibilities to better define how BER contributes to the cellular response to chemotherapeutics that inhibit TS. Regardless of the precise mechanism, it seems that β-pol can modulate the toxic effects of TS inhibitors. The reported alterations in β-pol levels in different tumor types including colon suggest that these observations might have clinical relevance with TS-directed therapies for colon cancer.