A Dominant-Negative Form of the Major Human Abasic Endonuclease Enhances Cellular Sensitivity to Laboratory and Clinical DNA-Damaging Agents

Introduction

Genetic damage can arise via a variety of means, most notably via spontaneous decay, reactions with intracellular chemicals, aberrant enzymatic processing events, and modification by environmental or clinical DNA-damaging agents (1-4). The most common intermediates or forms of DNA damage include base modifications, abasic sites, and single-strand breaks. These lesions are typically recognized and removed by the concerted effort of proteins that function in the base excision repair (BER) pathway (5). More complex lesions, such as DNA double-strand breaks, are corrected by recombinational repair responses (6). BER involves base excision by a DNA glycosylase, phosphodiester bond cleavage at the resulting abasic site by an apurinic/apyrimidinic (AP) endonuclease, gap filling and termini cleanup by a DNA polymerase/lyase, and nick ligation by a DNA ligase. Defects in participants of BER have been associated with increased sensitivity to DNA-damaging agents, genomic instability, cancer susceptibility, neurodegeneration, premature aging phenotypes, and other human diseases (7).

The major repair protein for abasic sites in mammals is AP endonuclease 1 (APE1; refs. 8, 9). This enzyme recognizes AP sites in DNA and cleaves immediately 5′ to the lesion, generating a DNA strand break with a priming 3′-hydroxyl group and a 5′-abasic residue. Prior biochemical and structural biology studies have indicated a model (known as “passing the baton”) whereby human APE1 cooperates with the next protein in BER (i.e., DNA polymerase β) to coordinately execute the next steps of the corrective response (10). In addition to AP sites, APE1 exhibits repair activity on a variety of modified (nonconventional) 3′-DNA termini, including, but not limited to, phosphates and phosphogycolyates (ref. 11 and references therein). Because abasic lesions are formed spontaneously at a frequency of roughly 10,000 times per mammalian genome per day, perhaps not surprisingly, knockouts in APE1 in mice lead to embryonic lethality (12-14), and sufficient depletion of APE1 in cells leads to cell inviability (15, 16).

Past studies using either antisense, small interfering RNA, or small molecule inhibitor strategies have revealed that APE1-deficient cells exhibit hypersensitivity to a number of DNA-damaging agents, including the laboratory agents methyl methanesulfonate (MMS), hydrogen peroxide, menadione, and paraquat, and the anticancer agents ionizing radiation, thiotepa, 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU; also known as carmustine), temozolomide, gemcitabine, and the nucleoside analogue β-l-dioxolane-cytidine (also known as troxacitabine; refs. 17-24). These findings reveal a prominent role for APE1 in the repair of specific oxidative, alkylation, and enzymatic DNA intermediates, and identify this protein as a potential target for certain therapeutic paradigms. As a novel means of inhibiting APE1 function in mammalian cells, we have used a dominant-negative strategy. Specifically, we created a mutant form of APE1, termed ED, which is catalytically inactive but possesses a higher DNA substrate binding affinity than the wild-type protein. The studies herein further identify the primary biological substrates of APE1 and reveal that ED sensitizes cells to the cytotoxic effects of the therapeutic DNA-damaging agents BCNU and dideoxycytidine, but not to bleomycin, β-d-arabinofuranosylcytosine (ara-C; also known as cytarabine), camptothecin, etoposide, mitomycin C, or cisplatin.

Results

AP Endonuclease and DNA Binding Properties of the ED Protein

Previous work had found that neutralization of a conserved (negatively charged) acidic residue within the active site of certain nucleases can produce an enzymatically inactive protein molecule with enhanced DNA-binding activity. For example, such a mutation in the flap endonuclease, FEN1 (i.e., a D to A substitution), resulted in a dominant-negative form of the human and yeast protein that bound substrate DNA with high affinity and blocked subsequent nucleic acid processing in vivo (25, 26). Prior studies have established the importance of the acidic residues E96 and D210 to the enzymatic repair activity of APE1 (Fig. 1A ; ref. 27 and references therein). We postulated that simultaneous mutation of these two amino acids would create a catalytically inactive APE1 protein with improved DNA-binding capacity.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

ED protein. A. Schematic of the APE1-AP-DNA interface. Aqua ribbon, APE1 protein. White, bound abasic substrate. Gold, abasic residue. Several key active site residues are denoted, as are the primary recognition loops α5, α8, and α11. Yellow, acidic residues mutated in ED [i.e., glutamic acid 96 (E96) and aspartic acid 210 (D210)]. B. Comparison of wild-type APE1 and the ED protein. Proteins were separated on a 12% SDS-polyacrylamide gel and stained with Coomassie blue. Arrow, position of the two proteins. Molecular weight (MW) standards from top to bottom are ∼103, 81, 48, 36, 27, and 19 kDa. C. Endonuclease activity of APE1 and ED. Proteins were assayed for AP site incision activity at 0, 0.01, 0.1, 1, and 10 ng as described in Materials and Methods. Representative incision gel. S, substrate; P, product. D. Binding activity of APE1 and ED. Electrophoretic mobility shift assays were done with radiolabeled GFA-C (100 fmol) and 0.1, 0.3, 1, or 3 ng of protein as described in Materials and Methods. Top, a representative binding gel. Bottom, points, average of six independent binding experiments; bars, SD.

After the introduction of E96Q and D210N substitutions into the APE1 coding sequence, we expressed and purified the resulting double mutant protein, which we have termed ED, and biochemically characterized the product (Fig. 1B). As shown in Fig. 1C, the ED protein exhibited no AP site incision activity at a concentration 10-fold higher (i.e., at 10 ng) than required to incise all DNA substrate by the wild-type enzyme (i.e., at 1 ng). Based on prior knowledge of the effect of independent substitutions at E96 or D210 (28), we estimate that the ED protein is >56,000,000-fold reduced in nuclease capacity. Furthermore, using a previously established electrophoretic mobility shift assay (see Materials and Methods), we found that ED displayed an ∼13-fold higher affinity for AP-DNA compared with wild-type APE1 protein (Fig. 1D); this enhanced affinity is presumably due to the loss of the repulsion between the negatively charged acidic residues and the phosphate backbone of DNA. In studies not shown, ED did not bind undamaged DNA with increased stability in electrophoretic mobility shift assays, indicating that the improved affinity was specific for substrate DNA.

ED Interferes with Wild-type APE1 Repair Activity In vitro

As an initial means of assessing the potential of ED to act as a dominant-negative protein, we examined the effect of ED on wild-type APE1 repair capacity using in vitro reconstitution assays. We define dominant-negative here as a protein that binds cognate DNA substrates but prevents subsequent repair processing (in this case, AP site incision). In the first set of in vitro experiments, an increasing amount of ED protein was preincubated with 200 fmol of AP-DNA. After 10 min on ice, wild-type APE1 protein was added and cleavage efficiency was immediately measured via a 2-min incubation at 37°C. As shown in Fig. 2A , prebinding of ED to AP-DNA inhibited wild-type APE1 incision capacity, with nearly 100% inhibition seen at a 1:1 ratio of ED (6.7 ng or 200 fmol) to substrate DNA. In the second set of experiments, ED and APE1 were preincubated at varying ratios with abasic DNA, and then MgCl₂ was added and the cleavage efficiency was measured as above. As reported in Fig. 2B, a 4:1 ED to APE1 ratio resulted in roughly 80% inhibition of wild-type incision activity. The results, in total, indicate that ED can bind AP-DNA with high affinity and interfere with normal APE1 endonuclease activity in vitro.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

In vitro competition assays with APE1 and ED. A. Competition incision assay. ED (at 0, 0.67, 2, 6.7, or 20 ng) was preincubated with radiolabeled GFA-C substrate (200 fmol) on ice. APE1 (0.3 ng or 9 fmol) was then added, and incision was allowed to proceed at 37°C for 2 min. Following incubation, conversion of substrate to product was determined on a denaturing polyacrylamide gel (shown above) as described in Materials and Methods. Columns, relative incision efficiency at the indicated amount of ED. B. Competition ratio assay. APE1 and ED (at the specified ratio; at 100 fmol total) were incubated with the GFA-C substrate (50 fmol) for 10 min on ice. MgCl₂ was then added, and incision was allowed to proceed at 37°C for 2 min. Relative incision efficiency was determined as in A. Columns (A and B), average of six independent incision experiments; bars, SD. NE, no enzyme.

Establishment of ED-Expressing Mammalian Cell Lines

Given that ED was able to inhibit wild-type APE1 nuclease activity in the biochemical assays above, we next determined if ED expression would affect APE1 repair capacity in vivo. Toward this end, we created T-REx Chinese hamster ovary (CHO)-K1 cell lines that expressed ED protein in a tetracycline-inducible (tet-on) manner (see Materials and Methods). Specifically, after selection and isolation of clonal integrates that stably harbor the recombinant pcDNA4/TO-ED plasmid, we identified four cell lines that differentially expressed ED in a tightly regulated, tet-on manner: ED8 > ED11 > ED5 >> ED6 (from highest to lowest; Fig. 3A ). Because ED11 displayed intermediate expression similar in nature to ED8 and ED5, we omitted this line from most subsequent analysis. We note that, in our hands, none of the cell lines exhibited a reliable tet concentration–dependent response (data not shown); that is, ED expression was either on (at tet concentrations of 0.1-10 μg/mL) or off (in the absence of tet), with no clear titratable expression being observed. Moreover, there was no obvious departure in growth characteristics (such as doubling time or survival), adhesion properties, or cellular morphologies of the ED-expressing clones upon tet induction (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Establishment of CHO cell lines stably expressing ED. A. CHO clonal isolates expressing ED in a tet-dependent manner. A subset of clones (ED8, ED11, ED5, and ED6) expressing the ED protein were identified by Western blot analysis. The experiment was carried out without (−) or with (+) tet addition. APE1, Western blot control protein. MW, lane with protein standards. Note that no cross-reactivity of the antibody was observed to the endogenous CHO APE1 protein. B. Total AP incision activity of CHO ED-expressing cells. Incision assays were done on whole-cell extracts produced from the indicated ED-expressing lines, which had either not been exposed to tet (tet minus) or were exposed to tet (tet plus). T-REx is the nontransfected control cell line. Columns, averages of at least five data points for each test variable; bars, SD.

As a first means of determining whether ED expression might affect in vivo AP site repair, we generated whole-cell extracts from noninduced and tet-induced ED5, ED6, and ED8 cells and measured total AP site incision capacity. We reasoned that if ED was expressed at sufficient levels and was indeed inhibitory of wild-type APE1 function, extracts containing ED protein would exhibit reduced incision efficiency. Consistent with this hypothesis, we found that both ED5 and ED8 (the high-expressing clones; Fig. 3A) displayed a tet/ED-dependent reduction (∼75%) in total AP site incision capacity (compare tet minus with tet plus), whereas ED6 (the low-expressing clone) and the T-REx control extracts did not (Fig. 3B). Given these results, we postulated that sufficient ED expression would enhance cellular sensitivity to DNA-damaging agents that generate APE1 (or BER) DNA substrates by inhibiting normal repair processing.

ED Expression Enhances Sensitivity to Laboratory DNA-Damaging Agents

To evaluate whether ED potentiates the cytotoxic effects of DNA-damaging agents, we first examined the sensitivity of the stably integrated CHO cell lines to the monofunctional alkylating agent MMS, a classic BER substrate–generating compound, which produces mainly alkylated base damage and AP sites (29). Initial colony formation assays were done with and without tet induction, and included the various ED-expressing lines (ED5, ED6, ED8, and ED11) and the T-REx parental cells. As shown in Supplementary Fig. S1 (and Fig. 4A ), the presence and level of ED expression corresponded qualitatively with the degree of MMS sensitivity. Specifically, the high-expressing ED5, ED8, and ED11 clones displayed decreased cell survival after MMS challenges, whereas the low-expressing ED6 cell line and the T-REx control exhibited “normal” sensitivity. Moreover, the impaired cell survival seen with ED5, ED8, and ED11 was not detected in the absence of tet (denoted by “−”), indicating that ED expression was required for MMS-induced cell death; the ED6 and T-REx cells exhibited wild-type resistance independent of tet exposure (Supplementary Fig. S1). A more quantitative analysis of the MMS response documented an ∼5- to 6-fold increased sensitivity (as determined by the relative LD₃₇) of the high-expressing ED clones (ED5, LD₃₇ 0.20 mmol/L; ED8, LD₃₇ 0.26 mmol/L), compared with the low-expressing ED6 line (LD₃₇ 1.15 mmol/L) and the T-REx control (1.25 mmol/L; Fig. 4A).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Cell survival analysis and AP site damage. A. MMS cytotoxicity. After tet induction, T-REx, ED5, ED6, and ED8 were exposed to MMS at various concentrations (0, 0.1, 0.2, 0.4, 0.8, and 1.6 mmol/L) for 1 h, then grown and stained for colony formation (see Materials and Methods). B. H₂O₂ cytotoxicity. Procedure was done as in A, except for using 0, 20, 40, 80, 120, and 160 μmol/L H₂O₂ as the challenge. Points (A and B), mean of at least four independent data points; bars, SD. C. Steady-state AP site levels. After 24 h tet exposure, T-REx, ED5, ED6, and ED8 cells were exposed to 0.4 mmol/L MMS for 1 h (or not treated), then washed and collected. After genomic DNA was isolated, the number of AP sites (expressed as AP sites per 1 × 10⁶ bp) was quantified using an aldehyde reactive probe–based colorimetric assay (see Materials and Methods). Columns, average of at least five independent data points; bars, SD.

In addition to MMS, we also examined the cellular sensitivity of the ED5, ED6, and ED8 lines to the laboratory chemical H₂O₂. This agent leads to the formation of predominantly oxidized bases, AP sites, and single-strand breaks with refractory 3′-termini, primarily 3′-phosphate ends (30). Although not as pronounced as seen with MMS (Fig. 4A), a similar trend of increased sensitivity was observed in the high-expressing ED clones (ED5 and ED8), whereas no differential survival was seen between the low-expressing ED6 clone and the T-REx control after H₂O₂ treatment (Fig. 4B). We note that at the highest H₂O₂ concentration (160 μmol/L), there were no colonies formed with the ED5 and ED8 lines, unlike ED6 and T-REx. The comparative LD₃₇ values for H₂O₂ were as follows: T-REx control, 100 μmol/L; ED5, 73 μmol/L; ED6, 100 μmol/L; and ED8, 65 μmol/L.

To develop a more precise picture of the effect of ED on DNA repair, we measured the steady-state level of AP sites using an established aldehyde-reactive probe–based method (see Materials and Methods). In these experiments, we found that upon tet induction (i.e., ED expression), all of the cell lines exhibited identical levels of endogenous AP sites (Fig. 4C); this finding is consistent with the fact that the various cell lines display comparable cell growth and viability characteristics after ED production. Conversely, after MMS treatment, the high ED–expressing clones (ED5 and ED8) accumulated roughly twice as many AP lesions (over background) than the low-expressing ED6 clone or the T-REx control. Consistent with this observation, we also found (using an alkaline Comet assay) that the ED5 and ED8 cells exhibited a 38% to 47% increase in alkaline-labile sites and single-strand breaks after a 0.4 mmol/L MMS treatment, compared with the 20% to 30% increase over background seen with the ED6 strain and T-REx control (data not shown). These data indicate that accumulating DNA damage is likely responsible for the enhanced cell death seen in the MMS-challenged, high ED–expressing cells.

ED Expression Sensitizes Cells to Clinical DNA-Damaging Agents

Because an emphasis of our effort was to generate a resource that would enhance cellular sensitivity to anticancer agents in future targeted gene therapy paradigms, we determined the comparative cell survival of the ED-expressing clones ED5, ED6, and ED8, as well as the T-REx control, to a variety of chemotherapeutic agents. In particular, we tested the following compounds: the alkylating agent BCNU, which is commonly used for treatment of brain tumors; the radiomimetic bleomycin, an antibiotic with potent antitumor activity against a range of lymphomas, head and neck cancers, and germ cell tumors; the chain terminating nucleoside analogues, dideoxycytidine, which is used mainly to block the infectivity of the HIV; and ara-C, which is used in the treatment of leukemias; the topoisomerase I inhibitor camptothecin, shown to have significant antitumor activity against lung, ovarian, breast, pancreas, and stomach cancers; the topoisomerase II inhibitor etoposide, used to treat a wide spectrum of human cancers; and the DNA cross-linking agents mitomycin C and cisplatin, both of which are used in the systemic treatment of a range of malignancies. Because these anticancer agents elicit their cell killing effects via the generation of largely distinct cytotoxic DNA intermediates, our cell survival analysis also provided a means of identifying primary DNA substrates of APE1 in vivo. As shown in Supplementary Fig. S2, no obvious increased sensitivity to bleomycin, ara-C, camptothecin, etoposide, mitomycin C, or cisplatin was observed for any of the ED-expressing clones. Conversely, colony formation assays did reveal that ED expression enhanced the cell killing effects of BCNU and dideoxycytidine by up to 1.4- to 2.2-fold and 2.1- to 2.8-fold, respectively (Supplementary Fig. S2; Fig. 5 ).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Colony formation analysis with specific chemotherapeutics. A. BCNU cytotoxicity. T-REx, ED5, ED6, and ED8 were exposed to BCNU at various concentrations (as in Fig. 4A, except for exposure times; concentrations 0, 30, 100, 150, 200, 250, and 300 μmol/L). B. Dideoxycytidine (ddC) cytotoxicity. T-REx, ED5, ED6, and ED8 were exposed to dideoxycytidine at various concentrations (as in Fig. 4A, except for exposure times; concentrations 0, 1.5, 3, 6, 9, 10, and 12 mmol/L). Points, mean of at least four data points taken from two independent experimental runs; bars, SD.

ED Expression Enhances the Killing of MMS, BCNU, and Dideoxycytidine in NCI-H1299 Cells

To assess the potential value of ED in the treatment of human cancer, we determined the effect of ED expression on MMS, BCNU, and dideoxycytidine sensitivity in the non–small cell lung cancer line NCI-H1299 (31). This cell line was selected largely due to its ability to be transfected efficiently (≥96%), a feature that allowed us to evaluate the effect of ED without a significant number of background (nontransfected) cells. Thus, after transient transfection of either the pcDNA4/TO control vector or the pcDNA4/TO-ED expression construct, we plated, challenged with either MMS, BCNU, or dideoxycytidine, and permitted colony formation of the NCI-H1299 cells (see Materials and Methods). As with the CHO line, ED expression enhanced cellular sensitivity to MMS, BCNU, and dideoxycytidine relative to the controls, with the increase in sensitivity for the human cancer line being ∼2.6-, 3.2-, and 1.6-fold, respectively (Fig. 6 ). A similar effect of ED was observed on cellular sensitivity to MMS and BCNU (dideoxycytidine was not tested) in the human osteosarcoma cell line U-2 OS (Supplementary Fig. S3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Effect of ED expression on sensitivity of NCI-H1299 cells to MMS, BCNU, and dideoxycytidine. A. MMS cytotoxicity. NCI-H1299 cell line was transfected with pcDNA4/TO (Vector) or pcDNA4/TO-ED (ED), and subsequently exposed to MMS at various concentrations (0, 0.5, 1, 2, and 3 mmol/L) for 1 h, then grown and stained for colony formation (see Materials and Methods). B. BCNU cytotoxicity analysis. Procedure was done as in A at 0, 25, 50, 100, and 150 μmol/L. C. Dideoxycytidine cytotoxicity analysis. Procedure was done as in A, except exposure was for 24 h, at 0, 1.5, 3, 6, and 9 mmol/L. Points (A-C), mean percentage survival values of at least four data points.

Discussion

Prior laboratory studies have documented the value of active site nuclease mutants specifically and dominant-negative DNA binding proteins in general. For instance, Resnick and colleagues (25, 26) have used a FEN1 mutant that harbors a D to A substitution at a residue found to be critical for DNA processing to determine the in vivo substrates of the flap endonuclease. Using either genetically engineered yeast strains or exogenously expressing yeast systems, they show that the FEN1 mutant causes genetic instability and increased sensitivity to MMS, likely by preventing processing of DNA replication intermediates and promoting the formation of recombinogenic DNA double-strand breaks. A DNA binding fragment (lacking the catalytic domain) of the critical strand break response protein, poly(ADP-ribose) polymerase 1, has also been shown to act as dominant-negative by blocking subsequent repair steps (32, 33). Upon tissue-specific expression, the poly(ADP-ribose) polymerase 1 fragment (its DNA-binding domain) was found to enhance the sensitivity of prostate carcinoma cells to ionizing radiation and etoposide, highlighting its potential usefulness in certain gene therapy–based anticancer treatment protocols (34). We report herein the design of a dominant-negative APE1 protein, which we have termed ED.

ED harbors two active site mutations at acidic residues (E96 and D210) previously shown to be essential for efficient catalytic activity (ref. 27 and references therein). Notably, this mutant protein exhibits a >56,000,000-fold reduced nuclease capacity, while displaying a better than wild-type AP-DNA binding affinity (Fig. 1). In vitro reconstituted assays also revealed that ED interferes with (or inhibits) wild-type APE1 incision activity (Fig. 2). Given these biochemical results, we did a series of experiments to determine the effects of ED expression in mammalian cells, postulating that ED (by binding DNA with high affinity and blocking subsequent wild-type APE1 repair activity) would (a) facilitate the identification of the primary biological substrates of APE1 and (b) serve as a resource in gene-therapy paradigms to selectively sensitive cancer cells to certain chemotherapeutics.

We have shown within that ED production leads to a markedly increased sensitivity to MMS and a concomitant accumulation of AP site damage (Fig. 4). Because abasic sites are prominent DNA intermediates of MMS exposure (29), our results underscore the critical importance of APE1 in the repair of AP lesions (and potentially, certain DNA single-strand breaks). The less dramatic hypersensitivity of ED-expressing cells to H₂O₂ implies a lesser role for APE1 in the repair of oxidative 3′ blocking lesions, such as 3′-phosphate groups, yet still indicates a significant contribution of APE1 to the efficient processing of oxidative DNA damage. This sensitivity profile of the ED-expressing cells is consistent with APE1 exhibiting a robust AP endonuclease activity in vitro (even for oxidized AP sites ref. 35), yet a comparably weak 3′-phosphodiesterase function (36), and with the polynucleotide kinase/phosphatase protein operating as the main human DNA 3′-phosphatase (37). The data also suggest that AP sites are effective cytotoxic lesions, a conclusion that is consistent with the observation that sufficient APE1 depletion in mammalian cells leads to an accumulation of AP sites and an accompanying cell inviability (15). We propose that the T-REx CHO system described within could be a useful asset in characterizing the functional activities of other human APE1 mutants and/or population variants (38).

Emerging work has documented the potential of regulating specific DNA damage responses as a means of promoting selective cell survival or cell death in the clinic (3, 4, 39). This concept has particular usefulness in anticancer (and in some instances antiviral) treatment paradigms, where genotoxic agents are often used to eradicate cancerous (or rapidly dividing) tissue. As noted throughout, a goal of our effort was to create a protein source with potential therapeutic value. We therefore examined whether ED expression enhanced the cell killing of a variety of chemotherapeutic agents, initially in the various CHO cell lines. We found that the most pronounced effect on cell survival upon ED induction was seen with BCNU and dideoxycytidine (Fig. 5). This increased cell killing was subsequently observed in the human cancer cell line NCI-H1299 when combining transient ED expression with MMS, BCNU, or dideoxycytidine treatment.

BCNU is an alkylating agent that reacts mainly with the N-7 or O-6 position of guanine, and the N-3 position of cytosine (40). Although the O-6 alkylation product is a major cytotoxic intermediate (primarily upon the formation of DNA interstrand cross-links), many of the other base modifications likely give rise to increased abasic sites via either the enhanced destabilization of the N-glycosylic linkage or removal by DNA glycosylases (41, 42). This amplified level of AP sites and accompanying inhibition of repair by ED is likely responsible for the enhanced cell killing observed with BCNU (Fig. 5). It is thus tempting to speculate that simultaneous inhibition of the major repair protein for O-6 alkylation damage (i.e., the O-6 methylguanine DNA methyltransferase; ref. 43), and a central BER component such as APE1, would elicit a more pronounced cytotoxic response to BCNU than single protein–targeted paradigms. Notably, the results of others support the notion that the APE1 repair nuclease capacity is a determinant in BCNU resistance, particularly in brain tumors (21, 44).

The enhanced cell killing exhibited by the ED-expressing cells after treatment with dideoxycytidine, but not ara-C, is consistent with the known 3′ to 5′ exonuclease preferences of APE1 for these chain-terminating d-configured nucleoside analogues. In particular, APE1 exhibits a ∼3-fold more robust excision activity for dideoxycytidine than for ara-C (45). Because APE1 harbors an even more proficient removal activity for l-configured nucleoside analogues, such as β-l-dioxolane-cytidine (excised roughly 8-fold faster than dideoxycytidine; ref. 45), future studies will need to determine the effect of ED on the cytotoxic potential of these (not readily available) agents. Prior investigations have indeed revealed that APE1 repair capacity correlates with sensitivity to certain l-configured nucleoside analogues (22, 46).

The lack of enhanced sensitivity to bleomycin was surprising given that APE1 was found to be the primary activity for excising 3′-phosphoglycolates (a major product of this antibiotic) from single-strand breaks in DNA (47). This observation may suggest that the main toxic intermediates formed by bleomycin are those related to clustered DNA damage, including juxtaposed AP sites and/or 3′ damage–containing single-strand breaks, or complex DNA double-strand breaks (48). Although APE1 has been shown to excise a synthetic mimic of a trapped 3′-topoisomerase I protein (i.e., a 3′-DNA tyrosyl residue; ref. 11), the fact that ED expression does not enhance sensitivity to camptothecin exposure suggests that APE1 is not a major player in the removal of this form of DNA damage. Indeed, evidence indicates that tyrosyl DNA phosphodiesterase serves as the primary repair enzyme for such 3′-blocking lesions in mammalian cells (49, 50). The observation that ED induction does not alter cellular sensitivity to etoposide, mitomycin C, and cisplatin is consistent with APE1 not playing a known role in the repair of 5′-trapped topoisomerase-DNA intermediates or interstrand/intrastrand DNA cross-links. In closing, the data herein not only shed important insights into the in vivo nucleic acid substrates of APE1 but also highlight the fact that APE1 is a rational therapeutic target (depending on the agent used and damage generated) and that ED is a potential resource for future gene therapy paradigms.

Materials and Methods

Reagents

DNA oligonucleotides were obtained from Midland Certified Reagent Company (Midland, TX). [γ-³²P]ATP (3,000 Ci/mmol) was purchased from Roche Pharmaceuticals (Indianapolis, IN), and T4 polynucleotide kinase was from New England BioLabs (Beverly, MA). All chemicals were purchased from Sigma-Aldrich (St. Louis, MO) unless otherwise specified. Denaturing polyacrylamide gel reagents were acquired from National Diagnostics (Atlanta, GA). Acrylamide and the low-range prestained protein standards were obtained from Bio-Rad (Hercules, CA). GetpureDNA kit and the DNA Damage Quantification kit were purchased from Dojindo Molecular Technologies, Inc. (Gaithersburg, MD). T-REx CHO-K1, pcDNA4/TO, and DMEM were obtained from Invitrogen Corporation (Carlsbad, CA). The human cancer cell lines NCI-H1299 and U-2 OS were purchased from American Type Culture Collection (Manassas, VA).

APE1 Wild-type and Mutant Protein

The bacterial expression system for producing wild-type recombinant human APE1 protein (i.e., pETApe) was generated as described (51). Overlapping PCR mutagenesis was used (see refs. 28, 52) to sequentially introduce two codon changes (i.e., the E96Q and D210N amino acid substitutions) into the pETApe construct to create the double mutant APE1 protein, ED. ED was expressed and purified from the recombinant pET-ED plasmid as detailed for the wild-type APE1 protein (51).

AP Endonuclease and DNA Binding Assays

AP endonuclease activity was measured using a duplex substrate termed GFA-C essentially as described (53). In brief, an F (tetrahydrofuran)–containing 26-mer oligonucleotide [GFA; 5′- AATTCACCGGTACG(F)ACTAGAATTCG-3′] was 5′-³²P-end labeled and annealed to an unlabeled complementary strand (GFA-C comp; 5′-CGAATTCTAGTCCGTACCGGTGAATT-3′). Subsequent incision reactions consisted of 25 mmol/L MOPS (pH 7.2), 100 mmol/L KCl, 1 mmol/L MgCl₂, 1 mmol/L DTT, and 50 μg/mL bovine serum albumin; either APE1 or ED protein at 0, 0.01, 0.1, 1, or 10 ng; and 1 pmol of GFA-C in a final volume of 10 μL. Reactions were carried out for 10 min at 37°C, stopped by the addition of 10 μL formamide buffer, and analyzed by electrophoresis in an 18% denaturing polyacrylamide gel. AP endonuclease activity is the amount of radiolabeled F-DNA converted to the shorter incised DNA product per minute. Percentage conversion was determined on a Molecular Dynamics Typhoon phosphorimager using ImageQuant software (version 5.2, Amersham Biosciences Corp., Piscataway, NJ).

The ED competition incision assay was conducted using the above AP endonuclease reaction conditions with a few modifications. ED (in the amount of 0, 0.67, 2, 6.7, or 20 ng) was first preincubated with 200 fmol GFA-C DNA at 0°C for 10 min. Subsequently, 0.3 ng (or 9 fmol) of wild-type APE1 was added, and the reaction was placed at 37°C for 2 min. The reaction was stopped and the percentage of substrate converted to product was determined as above.

The competition ratio assay was conducted using the AP endonuclease reaction conditions above with a few exceptions. Specifically, a defined ratio of APE1 to ED was first incubated [at amounts of 0%, 20%, 50%, 80%, or 100% wild-type APE1 (totaling 100 fmol of combined protein)] with 50 fmol of radiolabeled GFA-C DNA in the presence of 0.5 mmol/L EDTA at 0°C for 10 min. Subsequently, MgCl₂ was added to a final concentration of 10 mmol/L, and the reaction immediately transferred to 37°C for 2 min. As above, the reaction was stopped and the percentage of substrate converted to product was determined.

For the DNA binding electrophoretic mobility shift assay, protein (either APE1 or ED at 0.1, 0.3, 1, or 3 ng) was incubated with 5′-³²P-end labeled duplex GFA-C DNA (100 fmol) for 10 min at 0°C in 1× electrophoretic mobility shift assay buffer [25 mmol/L MOPS (pH 7.2), 100 mmol/L KCl, 1 mmol/L DTT, 50 μg/mL BSA, 20% glycerol, and 4 mmol/L EDTA]. The binding reactions were subsequently analyzed on an 8% nondenaturing gel [which contained 8% acrylamide (37.5:1), 2.5% glycerol, 20 mmol/L Tris-HCl (pH 7.2), 10 mmol/L sodium hydroxide, and 0.5 mmol/L EDTA] at 4°C (54). The percentage of labeled DNA bound by protein was determined by phosphorimager analysis as above.

ED-Expressing CHO Cell Lines

The ED cDNA was amplified from the pET-ED expression construct described above and subcloned into the BamHI and EcoRI restriction sites of pcDNA4/TO. Accuracy of the nucleotide sequence and the presence of the appropriate nucleotide substitutions for ED were verified. To create putative ED-expressing CHO cell lines, the pcDNA4/TO-ED construct was transfected into the T-REx CHO cell line using the Amaxa Nucleofactor System (Amaxa Biosystems, Gaithersburg, MD). The T-REx CHO line harbors an endogenous copy of pcDNA6/TR, which permits tet-regulatable (tet-on) gene expression from pcDNA4/TO. Stable integrates were selected by growth in DMEM [supplemented with 1% penicillin, streptomycin and glutamate, and 10% fetal bovine serum (tet minus)] containing zeocin (pcDNA6/TR) and blasticidin (pcDNA4/TO). ED expression was subsequently evaluated by standard Western blot analysis using whole-cell extracts (100 μg) and polyclonal antibodies to the human APE1 protein (Trevigen, Gaithersburg, MD), after exposure of several putative T-REx-ED isolates to 1 μg/mL tet (see more below). Detection was carried out using goat anti-rabbit horseradish peroxidase–conjugated secondary antibody (Pierce Biotechnology, Rockford, IL).

Whole-Cell Extract Assays

CHO cell lines were cultured in DMEM (as above). Once cells reached 80% confluence, they were treated with tet (final concentration 1 μg/mL) for 24 h. For extract preparation, cells were trypsinized, washed with PBS, and harvested by centrifugation. Cell pellets were frozen at −80°C for 1 h before extract preparation. Cells were resuspended in 1 mL lysis buffer [50 mmol/L Tris (pH 7.4), 1 mmol/L EDTA, 1 mmol/L DTT, 10% glycerol, 0.5 mmol/L phenylmethylsulfonyl fluoride] and then sonicated using a Misonix Microson sonicator (Farmingdale, NY). Following centrifugation, a Bradford assay (Bio-Rad) was run to determine the protein concentration of the supernatant (whole-cell extract). Total AP endonuclease activity was measured using radiolabeled 34F double-stranded DNA (55). AP endonuclease reactions consisted of 50 mmol/L HEPES (pH 7.5), 100 mmol/L KCl, 1 mmol/L MgCl₂, 10 pmol of the oligonucleotide duplex, and 1 μg whole-cell extract in a final volume of 10 μL. Reactions were done for 10 min at 37°C, stopped, and analyzed as above.

AP Site Measurements

Identified ED-expressing CHO cell lines (e.g., ED5, ED6, and ED8) and the T-REx control were grown to 80% confluence and treated with 1 μg/mL tetracycline as above. Cells were then challenged with 0.4 mmol/L MMS for 1 h at 37°C or received no treatment. Cells were harvested and processed as outlined by McNeill et al. (56). In brief, abasic site quantification consisted of isolating the chromosomal DNA from MMS-treated or untreated cells using the GetpureDNA kit according to the manufacturer's specifications. AP sites were then measured using the DNA Damage Quantification kit from Dojindo Molecular Technologies.

Colony Formation Survival Assays

Specified ED-expressing CHO cell lines and the T-REx control were grown to confluence, then trypsinized and counted. One hundred fifty cells of each cell line were subsequently transferred to each well of a six-well plate. Cells were allowed to adhere for 2 h before being treated with 1 μg/mL tet (see above). At the end of the 24-h incubation, cells were treated at the indicated concentrations with one of the following (see figure legends): MMS (for 1 h), H₂O₂ (for 1 h), BCNU (for 24 h), bleomycin (for 1 h), dideoxycytidine (for 24 h), ara-C (for 4 h), camptothecin (for 24 h), etoposide (for 4 h), mitomycin C (for 1 h), and cisplatin (for 4 h). The cells were then washed twice with 1× PBS and incubated for 10 days with fresh medium to allow colonies to form. At this time, colonies were stained with methylene blue and counted. The percentage survival was calculated by taking the number of colonies in the treatment well and dividing by the number of colonies in the mock-treated group.

For assessing NCI-H1299 sensitivity to DNA-damaging agents, cells were transiently transfected with either pcDNA4/TO or pcDNA4/TO-ED using the Amaxa Nucleofactor kit C (Amaxa Biosystems). After transfection, cells were grown to confluence over 2 days, then trypsinized and counted. One hundred fifty cells from each transfection were subsequently transferred to each well of a six-well plate, and allowed to adhere for 2 h before treatment with MMS (for 1 h), BCNU (for 1 h), or dideoxycytidine (for 24 h) at the indicated concentration. The cells were then washed twice with 1× PBS and incubated for 10 days in fresh RPMI 1640 (supplemented as above) to allow colonies to form. At this time, colonies were stained with methylene blue and counted, and the percentage survival was calculated as above. To determine transfection efficiency, NCI-H1299 cells were transfected with the pEYPF-C1 plasmid (Clontech, Mountain View, CA) and grown on a six-well chamber slide for 2 days to ∼80% confluence. Cells were then washed twice with 1× PBS and analyzed for YFP expression using an epifluorescence Zeiss (Berlin, Germany) microscope. At least 50 cells were analyzed for 4′,6-diamidino-2-phenylindole and FITC staining to determine transfection efficiency. Cell survival experiments with U-2 OS were done essentially as above, except that Amaxa Nucleofactor kit V was used for transfection and cells were cultured in DMEM (supplemented as above). The transfection efficiency of U-2 OS was ≥94%.