Sensitization of Human Carcinoma Cells to Alkylating Agents by Small Interfering RNA Suppression of 3-Alkyladenine-DNA Glycosylase

Research Article

Johanna Paik, Tod Duncan, Tomas Lindahl, and Barbara Sedgwick

Clare Hall Laboratories, Cancer Research UK London Research Institute, South Mimms, Hertfordshire, United Kingdom

Abstract cancers has been linked to the use of high-dose chemotherapeutic One of the major cytotoxic lesions generated by alkylating regimens. To reduce the risk of secondary cancers by decreasing agents is DNA 3-alkyladenine, which can be excised by treatment doses, cells must be sensitized to the cytotoxic but not to 3-alkyladenine DNA glycosylase (AAG). Inhibition of AAG the mutagenic effects of alkylating agents. One possible approach to may therefore result in increased cellular sensitivity to achieving this is to inhibit the repair of lethal DNA lesions that are chemotherapeutic alkylating agents. To investigate this not highly mutagenic. O6-Methylguanine DNA methyltransferase (MGMT) directly possibility, we have examined the role of AAG in protecting 6 human tumor cells against such agents. Plasmids that express demethylates DNA O -methylguanine. This lesion is highly small interfering RNAs targeted to two different regions of mutagenic due to its ability to mispair with thymine (reviewed in AAG mRNA were transfected into HeLa cervical carcinoma ref. 6). Inhibition of MGMT activity may not only increase the chemotherapeutic effects of alkylating agents but also enhance the cells and A2780-SCA ovarian carcinoma cells. Stable deriva- 6 tives of both cell types with low AAG protein levels were risk of secondary tumors caused by accumulation of mutagenic O - sensitized to alkylating agents. Two HeLa cell lines with AAG alkylated G residues (reviewed in ref. 7). Nevertheless, inhibitors protein levels reduced by at least 80% to 90% displayed a 5- to of MGMT have been tested in clinical trials as adjuncts to 10-fold increase in sensitivity to methyl methanesulfonate, chemotherapy (2, 8). 3-Alkyladenine-DNA glycosylase (AAG), also N-methyl-N-nitrosourea, and the chemotherapeutic drugs called 3-methyladenine DNA glycosylase, excises 3-alkyladenine temozolomide and 1,3-bis(2-chloroethyl)-1-nitrosourea. These from DNA and initiates the base excision repair pathway for cells showed no increase in sensitivity to UV light or ionizing alkylation damage (reviewed in refs. 9, 10). The 3-alkyladenine radiation. After treatment with methyl methanesulfonate, lesion protrudes into the minor groove of DNA where it blocks AAG knockdown HeLa cells were delayed in S phase but DNA replication (11). E. coli and S. cerevisiae mutants lacking AAG are very sensitive to killing but not mutagenesis by methylating accumulated in G2-M. Our data support the hypothesis that ablation of AAG activity in human tumor cells may provide a agents; thus, unrepaired DNA 3-methyladenine is cytotoxic (12, 13). / useful strategy to enhance the efficacy of current chemother- Different Aag murine cell lines display a range of methyl / apeutic regimens that include alkylating agents. (Cancer Res methanesulfonate (MMS) sensitivities. Aag embryonic stem 2005; 65(22): 10472-7) cells are hypersensitive to simple alkylating agents such as MMS and also to the chemotherapeutic agents BCNU and mitomycin C Introduction (14, 15). They undergo S-phase arrest on exposure to MMS, consistent with 3-methyladenine being a block to DNA replication Alkylating agents are used in the treatment of several human (16). Other Aag/ cell types, however, display lower sensitivities cancers. They are mostly methylating or chloroethylating agents or even resistance to methylating agents. Aag/ neurons, for such as temozolomide and dacarbazine or 1,3-bis(2-chloroethyl)-1- example, have a low sensitivity (17) whereas Aag/ mouse nitrosourea (BCNU). Cytotoxicity of these anticancer drugs results primary embryonic fibroblasts have either low or no sensitivity from the alkylation of DNA and is strongly attenuated by the ability (18–20). Moreover, Aag/ myeloid progenitor cells are resistant to of cells to repair DNA lesions (for reviews, see refs. 1–3). The major alkylation (21). These observations suggest that undifferentiated cytotoxic lesions generated by methylating agents are DNA 3- / 6 Aag embryonic stem cells are more sensitive to methylating methyladenine and DNA O -methylguanine, whereas BCNU induces agents than the other cell types. Several suggestions have been a more complex array of DNA damage including ethano bases and made to explain these varied responses. Because excision of a DNA inter- and intrastrand cross-links (4, 5). Besides being cytotoxic, modified base by a DNA glycosylase is the initial step of the base these agents are also mutagenic and the occurrence of secondary excision repair pathway, cleavage, excision, and filling-in of the resulting abasic site require several additional activities (9). Hence, varied responses of different cell types may be due to different relative levels of activities required to complete the base excision Note: J. Paik is presently at Department of Biomedical Sciences, College of Medicine Suite 2300-B, Florida State University, 1115 West Call Street, Tallahassee, repair pathway, once initiated by AAG. Alternatively, the cellular FL 32306-4300. T. Duncan is presently at MCD Biology, University of Colorado, capacity to remove replication blocking lesions by nucleotide Boulder, CO 80309. excision repair or recombination, or to tolerate them by DNA lesion Requests for reprints: Barbara Sedgwick, Clare Hall Laboratories, Cancer Research UK London Research Institute, South Mimms, Hertfordshire EN6 3LD, bypass, may influence the final response to alkylation (20, 21). United Kingdom. Phone: 44-20-7269-3982; Fax: 44-20-7269-3801; E-mail: barbara. To determine whether AAG may be a suitable target for [email protected]. I2005 American Association for Cancer Research. inhibition during chemotherapy, rather than using murine cells, doi:10.1158/0008-5472.CAN-05-1495 we have examined the role of AAG in the defense of two human

Cancer Res 2005; 65: (22). November 15, 2005 10472 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2005 American Association for Cancer Research. siRNA Knockdown of Human AAG carcinoma cell lines against alkylation toxicity. The exploitation Flow cytometry. Cells were exposed to 1 mmol/L MMS for 1 hour and of small interfering RNAs (siRNA) to selectively inhibit gene washed and fresh culture medium was added. After incubation for 6 to expression has been used previously to knockdown DNA repair 48 hours, cells were harvested, washed with PBS, and fixed in 70% ethanol. A activities and to examine their biological importance (22). In this After washing twice with PBS, they were treated with 100 g/mL RNase and DNA was stained with 40 Ag/mL propidium iodide. At each time point, article, we have used vector-based RNA interference to stably 20,000 cells were analyzed using a FACSCalibur analytic cytometer (BD knockdown AAG protein levels in cervical and ovarian carcinoma Biosciences Clontech) and the CellQuest program. cell lines and determined whether these cells are sensitized to alkylating agents, including those used in chemotherapy. Results 3-Alkyladenine-DNA glycosylase protein and antibodies. Materials and Methods Three plasmids (IMAGE clones 359782, 1839537, and 1723180) 3-Alkyladenine-DNA glycosylase antibodies and Western blot encoding three human AAG cDNAs were obtained from the IMAGE analysis. AAG cDNAs were obtained from the IMAGE consortium and consortium and sequenced. All three cDNA sequences were identical sequenced. The AAG coding sequence of IMAGE clone 359782 was to a previously reported colon adenocarcinoma AAG cDNA subcloned into the pET29a expression vector (Merck Biosciences Novagen, (accession no. L10752; ref. 25) except for two nucleotides (GC Nottingham, United Kingdom). Expression constructs were transformed instead of CG) at positions 570 and 571 of the coding sequence. into BL21-CodonPlus(DE3) (Stratagene, La Jolla, CA) and overexpressed as These two AAG isoforms result in two amino acid differences (Gln- 10 histidine-tagged AAG proteins (tagged at the amino terminus), which Leu instead of His-Val at amino acids 190 and 191) and have been were purified by affinity chromatography on Ni2+-nitrilotriacetic acid agarose columns (Qiagen, Crawley, United Kingdom). Polyclonal antibodies previously observed (accession no. P29372). The predicted amino- were raised against the purified protein by immunization of rabbits. terminal sequence of AAG encoded by the three IMAGE consortium Proteins (10-16 Ag) in whole-cell extracts were resolved by SDS-PAGE and cDNAs also agreed with the result of Vickers et al. (25). A liver cDNA transferred to a nitrocellulose membrane [Whatman (Schleicher & Schuell), encoding AAG with a different amino terminus may be a splice Brentford, United Kingdom] using standard protocols. AAG was detected variant (25, 26). Plasmids were constructed that encoded either full- using the rabbit antisera at 1:100 dilution. Tubulin was also monitored using length AAG or this protein deleted for its amino-terminal 74 amino antitubulin antibodies to check for equal loading of cell extracts. The acids (AAGD74). Both proteins with 10 his tags were overexpressed immunoblot was developed using horseradish peroxidase–conjugated anti- in E. coli and purified by affinity chromatography. 10hisAAGD74 is rabbit immunoglobulins (Amersham Biosciences, Chalfont St. Giles, United more thermostable (27) and also more active in in vitro assays (data Kingdom) and a chemiluminescent substrate (Pierce, Rockford, IL). not shown) than the full-length protein. Polyclonal antibodies were Generation of stable cell lines with low levels of 3-alkyladenine DNA D glycosylase protein. Palindromic oligonucleotides (64-mer) that will code raised in rabbits against recombinant 10hisAAG 74 protein. These for hairpin RNAs targeted to nucleotides 424 to 442 (CCGAGGCATGTT- antibodies recognized full-length recombinant AAG and native AAG CATGAAG9) or 670 to 688 (CAAGAGCTTTGACCAGAGG) of the AAG open in crude HeLa cell extracts on Western blots (Fig. 1). They detected reading frame (882 nucleotides) were cloned into pSUPER (23). The two closely migrating protein bands in HeLa cell extracts (Fig. 1). resulting plasmids were sequenced to confirm correct insertion of the The same two protein bands were also recognized by a different oligonucleotides and are referred to as pSUPER-AAG-siRNA424 and pSUPER-AAG-siRNA670. HeLa Ohio and A2780-SCA5 (clone A2780 in ref. 24) cells were transfected with the pSUPER-AAG-siRNA constructs together with the pPUR vector (BD Biosciences Clontech, Oxford, United Kingdom) that carries a selectable puromycin resistance marker. The cells were serially diluted 48 hours posttransfection and cultured for 3 weeks in DMEM containing 2 Ag/mL puromycin. Twelve independent clones from each transfection were expanded. These clones were screened for reduced levels of AAG protein by Western blotting and cell lines with low AAG expression were established. Stable puromycin-resistant derivatives of cells transfected with the empty pSUPER vector together with pPUR were also cloned and had unchanged AAG protein levels. Cytotoxicity assays. MMS and BCNU (carmustine) were obtained from Sigma (Poole, United Kingdom). N-Methyl-N-nitrosourea (MNU) and temozolomide were gifts from Peter Swann (University College London, London, United Kingdom) and Julie Silvester (Cancer Research UK, London, United Kingdom), respectively. Solutions of 0.1 mol/L MNU in 10 mmol/L potassium acetate (pH 4.8) and 0.5 mol/L BCNU in cold absolute ethanol were freshly prepared for each experiment. Aliquots of a stock solution of 0.15 mol/L temozolomide in DMSO were stored at 80jC. Cells were seeded in duplicate or triplicate in 10-cm dishes at 1,000 to 3,000 per dish. At 16 to 20 hours after seeding, the cells were treated with an alkylating agent or exposed to UV at 254 nm or to g-radiation. Cells were treated with MMS, Figure 1. Knockdown of AAG protein in HeLa cells. A, regions of AAG temozolomide, or BCNU for 1 hour, washed, and cultured in fresh medium. mRNA targeted by siRNAs expressed from pSUPER-AAG-siRNA424 and MNU was left in the culture overnight, during which time the MNU pSUPER-AAG-siRNA670. B, cell extracts of HeLa cells stably transfected with pSUPER, pSUPER-AAG-siRNA424, or pSUPER-AAG-siRNA670 were analyzed decomposes. Cells were exposed to UV in PBS, washed, and cultured in fresh by Western blotting. The nitrocellulose filter was cut above the 41 kDa marker. medium. To monitor sensitivity to g-radiation, cells were washed, The upper half was probed with antitubulin antibody and the lower half with resuspended in 10 mL PBS, and exposed to g-rays from a IBL437C Cs-137 anti-AAG polyclonal antisera. By comparison with molecular weight markers, source. The g-irradiated cells were plated in triplicate (1,000-3,000 per dish) these proteins migrated according to their known molecular sizes, as indicated. In the uncut blot, the 34 kDa doublet was the major band detected by our AAG and cultured in fresh medium. After treatments, cells were cultured for 10 to antiserum and was the only protein not recognized by preimmune serum from 14 days and colonies were stained with Giemsa. the same rabbit. www.aacrjournals.org 10473 Cancer Res 2005; 65: (22). November 15, 2005

Increased sensitivity of HeLa cells with low 3-alkyladenine DNA glycosylase protein levels to methyl methanesulfonate but not to UV or ;-radiation. To determine whether AAG protein protects HeLa cells against the toxicity of a methylating agent, we examined the sensitivity of the two AAG knockdown HeLa cell lines to MMS. Cells were exposed to MMS and survival was estimated by clonogenic assays. Both AAG knockdown cell lines were 4- to 5-fold more sensitive to killing by MMS than the control (HeLa cells stably transfected with pPUR and the empty pSUPER vector; Fig. 2). The two knockdown cell lines express siRNAs targeted to different regions of the AAG mRNA. Hence, increased MMS sensitivity of both lines indicates that this phenotype is due to low AAG protein levels and not a result of nonspecific changes of a different activity. As a further control, sensitivities of the cells to 254-nm UV light and g-radiation were examined. The two AAG knockdown lines and the control displayed similar sensitivities to UV light or g-rays (Fig. 2). These observations support the conclusion that MMS sensitivity is due to the specific depletion of AAG and that other DNA repair pathways such as base excision repair of oxidative damage, nucleotide excision repair, and recombination remain intact. Sensitivity of 3-alkyladenine DNA glycosylase knockdown HeLa cells to the SN1 methylating agents N-methyl-N-nitrosourea and temozolomide. The major toxic methylated DNA base generated by MMS, an SN2 methylating agent, is 3-methyladenine (10% of total alkylation) whereas O6-methylguanine is a minor product (0.3% of total alkylation). In contrast, MNU, an SN1 agent, Figure 2. Sensitivity of HeLa cell lines with reduced cellular levels of AAG protein to MMS, UV light, and g-radiation. The stable HeLa transfectants with low AAG protein levels were exposed to various doses of MMS, UV light, or g-radiation and cell survival was monitored by colony formation. Open and closed symbols, repeat experiments using different dose ranges. A, MMS; B, UV light; C, g-radiation. The UV and g-ray survival data are the average of two or three independent experiments. HeLa/pSUPER (o, .); HeLa/pSUPER-AAG-siRNA424 (5); HeLa/pSUPER-AAG-siRNA670 (4, E). polyclonal AAG antiserum (a gift from T. O’Connor, Biology Division, Beckman Research Institute, Duarte, CA; data not shown), indicating that they are both forms of AAG. Furthermore, a doublet AAG protein band was observed also when overexpressed AAG was monitored in human cell extracts using a monoclonal antibody (28). The two protein bands might correspond to the products of two AAG mRNA splice variants that have been found in several cell lines and tissues (25, 29), might have resulted from an uncharacterized posttranslational modification of AAG, or, as proposed by Rinne et al. (28), might be due to proteolytic removal of a small number of amino acids from a terminus. Knockdown of 3-alkyladenine DNA glycosylase in HeLa cells. Two derivatives of pSUPER were constructed that express siRNAs targeted to two different regions of AAG mRNA (nucleotides 424-442 or 670-688 of the 882-nucleotide coding sequence). The constructs, pSUPER-AAG-siRNA424 or pSUPER-AAG-siRNA670, were cotrans- fected with pPUR into HeLa cells and stable puromycin-resistant clones were selected. The level of AAG protein in these clones was monitored by Western blotting. After transfection with either construct, 2 of 12 clones had significantly reduced AAG protein levels. The stable cell lines referred to as HeLa/pSUPER-AAG- siRNA424 and HeLa/pSUPER-AAG-siRNA670 had AAG protein levels reduced by 80% and 90%, respectively (Fig. 1). Both bands recognized Figure 3. Sensitivity of AAG knockdown HeLa cell lines to MNU, temozolomide by the polyclonal antibodies were reduced, suggesting that they (TMZ), and BCNU. HeLa cells and derivatives with low AAG protein levels were both forms of the AAG protein. The AAG protein level was were exposed to various doses of alkylating agents and cell survival was monitored by colony formation. For all agents, the data shown are the average of two unchanged in stable puromycin-resistant clones of HeLa cells independent experiments. A, MNU; B, temozolomide; C, BCNU. HeLa/pSUPER transfected with the empty pSUPER vector. (o); HeLa/pSUPER-AAG-siRNA424 (5); HeLa/pSUPER-AAG-siRNA670 (4).

Cancer Res 2005; 65: (22). November 15, 2005 10474 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2005 American Association for Cancer Research. siRNA Knockdown of Human AAG generates both 3-methyladenine (9% of total alkylation) and O6- methylguanine (6.3% of total alkylation) as major products (4). Sensitivity of the AAG knockdown cells to MNU was examined. Strikingly, sensitivity of these cell lines to MNU was 10-fold greater than the control (Fig. 3A). Temozolomide is used in chemotherapy and generates the same alkylating intermediate (the methyldiazo- nium ion) as MNU and, consequently, the same array of methylation products (2). It was therefore of interest to verify whether AAG can protect HeLa cells against this drug. Sensitivity of both AAG knockdown cell lines to temozolomide was also increased 10-fold (Fig. 3B). These observations strongly indicate that AAG affords a high degree of resistance against both MNU and temozolomide in HeLa cells. MGMT activity was undetectable in crude cell extracts of control HeLa cells and of the two AAG knockdown derivatives (data not shown). Differences in MGMT activity are therefore unlikely to Figure 5. Knockdown of AAG in A2780-SCA cells and sensitivity of the explain the enhanced sensitivity of the AAG knockdown cells to AAG-depleted cells to MMS. A, cell extracts of A2780-SCA cells stably MNU or temozolomide by comparison with the control cells. Loss transfected with pSUPER or pSUPER-AAG-siRNA constructs were analyzed by Western blotting. B, A2780-SCA cells and derivatives with low AAG protein of MGMT gene expression is frequently associated with epigenetic levels were exposed to various doses of MMS and cell survival was monitored by silencing (30), has been previously observed in HeLa sublines (31), colony formation. A2780-SCA (o); A2780-SCA/AAG knockdown 1 (4); 5 and may account for the lack of MGMT activity in the HeLa Ohio A2780-SCA/AAG knockdown 2 ( ). cells employed here. Sensitivity of 3-alkyladenine DNA glycosylase knockdown knockdown cells (Fig. 2). Progression of the cells through the cell HeLa cells to 1,3-bis(2-chloroethyl)-1-nitrosourea. Chloronitro- cycle was monitored by flow cytometry at 6, 12, 24, 36, and 48 hours soureas are used as anticancer drugs and induce multiple lesions, after MMS treatment. Six hours after treatment, control cells were including larger and more complex adducts than those induced by delayed in S phase but progressed to return to the same methylating agents. BCNU generates chloroethyl and hydroxyethyl asynchronous distribution as the untreated cells at 24 hours. The base derivatives, ethano bases, and DNA intra- and interstrand AAG knockdown cells were more markedly delayed in S phase and cross-links (4, 5). Aag/ murine stem cells are sensitive to cell then accumulated in G2-M (Fig. 4). These cells remained in G2-M at killing by MMS, MNU, and BCNU (14, 15). To determine whether 24, 36, and 48 hours after MMS treatment (Fig. 4 and data not AAG protects HeLa cells against BCNU toxicity, sensitivity of the shown). AAG knockdown cells was examined. Both AAG knockdown cell Knockdown of 3-alkyladenine DNA glycosylase in A2780 lines showed a 10-fold increased sensitivity to the cytotoxicity of cells. To determine whether AAG depletion sensitizes a different BCNU compared with the control cell line (Fig. 3C). tumor cell line to alkylation, AAG knockdown derivatives of an Cell cycle progression after methyl methanesulfonate ovarian carcinoma cell line, A2780, were isolated. A2780 cells have treatment. Asynchronous populations of the HeLa/pSUPER- MGMT activity and are proficient in mismatch repair and p53 AAG670 knockdown cell line and control HeLa/pSUPER cells were responses (24). AAG protein levels of two isolates were reduced by treated with 1 mmol/L MMS for 1 hour and then allowed to grow in 80% to 90% (Fig. 5A). MMS sensitivity of the AAG knockdown the absence of alkylating agent. At this MMS dose, 80% of A2780 cells was examined and found to be increased by at least the control cells survive compared with only 0.2% of the AAG 2-fold at doses >1 mmol/L (Fig. 5B). Discussion By expressing siRNAs targeted to AAG mRNA, we have established stable HeLa cell lines with very low levels of the AAG DNA repair enzyme. These cells have a 5- to 10-fold increased sensitivity to the simple methylating agents MMS, MNU, and temozolomide and also to the chloroethylating agent BCNU (carmustine). They were not sensitive to UV light or g-radiation. AAG-depleted A2780 ovarian carcinoma cells were also sensitized to MMS. The phenotype of HeLa cells is comparable with that of the murine Aag/ knockout embryonic stem cells, which are also sensitive to MMS, MNU, temozolomide, and BCNU (14, 15), but contrasts with those of murine Aag/ embryonic fibroblasts, neurons, and bone marrow progenitor cells, which are less sensitive or even resistant to alkylation (17–20). Hence, in murine stem cells and human HeLa cells, as well as in E. coli and S. cerevisiae (32, 33),

Figure 4. Cell cycle distribution of AAG knockdown HeLa cells after MMS 3-methyladenine DNA glycosylase is a major determinant of MMS treatment. Asynchronous cell cultures of HeLa/pSUPER (Control) and resistance whereas in the other murine cell types and also in S. HeLa/pSUPER-AAG-siRNA670 (AAG knockdown) were exposed to 1 mmol/L pombe (34), other processes such as nucleotide excision repair, MMS for 1 hour. The cell cycle distribution was monitored at various times after treatment by flow cytometric analysis of DNA content; 20,000 cells were recombinational repair, or DNA polymerases, which can bypass analyzed in each histogram. DNA damage, may play a significant role (20, 21). www.aacrjournals.org 10475 Cancer Res 2005; 65: (22). November 15, 2005

Figure 6. BCNU alkylation of adenine and guanine in DNA. A, 1-chloroethyladenine and 1-hydroxyethyladenine are primary products of alkylation of adenine residues in DNA by BCNU. The 1-chloroethyl adduct can react with the N 6-amino group to generate tricyclic 1, N 6-ethanoadenine. This DNA lesion is a known substrate of AAG (41). B, other potential AAG substrates include 7-chloroethylguanine, the most abundant product of guanine alkylation, 7-hydroxyethylguanine, and the secondary reaction product 7-diguanyl DNA intrastrand cross-links (4). The labile lesions 3-chloroethyladenine and 3-hydroxyethyladenine are presumably formed and may also be AAG substrates.

MMS, MNU, and temozolomide methylate DNA at several Knockdown of AAG protein sensitized HeLa cells to the different sites (4) but they all induce substantial amounts of chemotherapeutic methylating agents temozolomide and BCNU. 3-methyladenine, which is the major substrate of AAG (35). This is the first indication that targeted inhibition of human Hence, defective repair of DNA 3-methyladenine in the AAG AAG activity might sensitize tumor cells to cytotoxic drugs. knockdown cell lines is consistent with their hypersensitivity Other anticancer drugs that generate the same DNA lesions as to these three agents. Formation of the analogous lesion 3- temozolomide and BCNU include the SN1 methylating agents hydroxyethyladenine by BCNU does not seem to have been dacarbazine, procarbazine, and streptozotocin and chloronitro- investigated. BCNU induces a complex array of DNA damage soureas such as 1-(2-chloroethyl)-3-cyclohexyl-L-nitrosourea including several chloroethylated and hydroxyethylated bases, (lomustine). Hence, inhibition of AAG might increase the efficacy four different tricyclic ethano bases, and two types of DNA cross- of all these agents. Inhibitors of MGMT used as adjuncts to links (4, 5). Both the E. coli AlkA and S. cerevisiae MAG 3- temozolomide chemotherapy have reached clinical trials (2, 8). methyladenine DNA glycosylases excise N 7-chloroethylguanine Although inhibition of MGMT may increase cell killing by (the most abundant lesion) and N 7-hydroxyethylguanine from methylating drugs, it will also be expected to increase the risk of DNA, and E. coli AlkA has also been reported to remove mutagenesis and the occurrence of secondary tumors (7, 30). In diguanosinylethane (from N 7-diguanyl DNA cross-links), N 2,3- contrast, inhibition of AAG could increase the cytotoxicity of ethanoguanine, 1,N 6-ethanoadenine, and 3,N 4-ethanocytosine (33, alkylation with a lesser risk of carcinogenicity. Simple assays for 36–39). Activity of human AAG on these DNA lesions has thus far DNA glycosylases are available; thus, they are practical targets been observed only for 1,N 6-ethanoadenine (40, 41). This lesion for screening of potential inhibitors. We have initiated a search probably arises by reaction of the chloroethyl adduct of 1- for small chemical inhibitors of AAG using a high-throughput chloroethyladenine with the N 6-amino group to form the tricyclic screen;1 more potent inhibitors are being sought by further ethano base (Fig. 6). 1,N 6-Ethanoadenine interferes with base screening and modeling of these compounds into the active site pairing and blocks in vitro DNA synthesis by both DNA of the AAG three-dimensional structure (45). The level of polymerases a and h (42). Hence, deficient repair of this modified sensitization of the AAG-depleted HeLa cells to the alkylating base is consistent with the sensitivity of AAG knockdown HeLa agents tested was high. Doses that allowed f80% survival of the cells to BCNU. In this case, resistance of Aag/ murine control cell line killed the AAG-depleted cells to 0.1% to 0.2% embryonic fibroblasts to BCNU could result from replication past survival. AAG-depleted A2780 ovarian carcinoma cells were also these lesions by polymerase D (42). It may now be of interest to sensitized but not as dramatically as the HeLa cells. AAG examine the ability of AAG to repair other BCNU-induced lesions, expression has been observed to be high in cervical neoplasia such as N 3-chloroethyladenine, N 3-hydroxyethyladenine, N 7- (46). Future work must therefore determine whether extreme chloroethylguanine, N 7-hydroxyethylguanine, and N7-diguanyl sensitization is a particular characteristic of HeLa cells, human DNA intrastrand cross-links, which are formed by interaction of papillomavirus–transformed tumor cells, or rapidly proliferating abundant N 7-chloroethylguanine lesions (Fig. 6). Inhibition of malignant carcinoma cells. Further cell lines derived from apurinic endonuclease or treatment with methoxyamine that different tumor types as well as human tumor explants in binds to abasic sites also sensitizes cells to BCNU, providing further evidence that lesions induced by this drug are repaired by the base excision repair pathway (43, 44). 1 T. Hammonds, T. Duncan, T. Lindahl, B. Sedgwick, unpublished data.

Cancer Res 2005; 65: (22). November 15, 2005 10476 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2005 American Association for Cancer Research. siRNA Knockdown of Human AAG immunosuppressed mice should be examined. Secondary tumors Acknowledgments induced by S 1 chemotherapeutic methylating agents, such as N Received 4/29/2005; revised 7/27/2005; accepted 9/7/2005. temozolomide, and some primary tumors such as hereditary Grant support: Cancer Research Technology, Cancer Research UK, London (J. Paik nonpolyposis colorectal cancer are often deficient in mismatch and T. Duncan). repair and resistant to alkylating agents (47, 48). A further The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance relevant extension of this work is, therefore, to determine with 18 U.S.C. Section 1734 solely to indicate this fact. whether AAG depletion will resensitize mismatch repair– We thank Peter Robins, Qian An, Peter MacPherson, and Gary Warnes for deficient tumor cells to alkylation. These questions may be experimental assistance and Peter Karran, Nick Wells, and Philip Masterson for discussions. answered more succinctly when potent low-molecular weight The authors dedicate this work to the memory of Nicole Winkelmann, a graduate chemical inhibitors of AAG are available. student at the Clare Hall Laboratories who died recently from cancer.

References nine DNA glycosylase. Proc Natl Acad U S A 1997;94: 34. Memisoglu A, Samson L. Contribution of base 13087–92. excision repair, nucleotide excision repair, and DNA 1. Tentori L, Graziani G. Pharmacological strategies to 19. Elder RH, Jansen JG, Weeks RJ, et al. Alkylpurine- recombination to alkylation resistance of the fission increase the antitumor activity of methylating agents. DNA-N-glycosylase knockout mice show increased yeast Schizosaccharomyces pombe. J Bacteriol 2000;182: Curr Med Chem 2002;9:1285–301. susceptibility to induction of mutations by methyl 2104–12. 2. Middleton MR, Margison GP. Improvement of che- methanesulfonate. Mol Cell Biol 1998;18:5828–37. 35. Roy R, Kennel SJ, Mitra S. Distinct substrate motherapy efficacy by inactivation of a DNA-repair 20. Sobol RW, Kartalou M, Almeida KH, et al. Base preference of human and mouse N-methylpurine-DNA pathway. Lancet Oncol 2003;4:37–44. excision repair intermediates induce p53-independent glycosylases. Carcinogenesis 1996;17:2177–82. 3. Drablos F, Feyzi E, Aas PA, et al. Alkylation damage in cytotoxic and genotoxic responses. J Biol Chem 2003; 36. Carter CA, Habraken Y, Ludlum DB. Release of 7- DNA and RNA–repair mechanisms and medical signif- 278:39951–9. alkylguanines from haloethylnitrosourea-treated DNA icance. DNA Repair (Amst) 2004;3:1389–407. 21. Roth RB, Samson L. 3-Methyladenine DNA glyco- by E. coli 3-methyladenine-DNA glycosylase II. Biochem 4. Singer B, Grunberger D. In: Molecular biology of sylase-deficient Aag null mice display unexpected Biophys Res Commun 1988;155:1261–5. mutagens and carcinogens: reactions of directly acting bone marrow alkylation resistance. Cancer Res 2002; 37. Habraken Y, Carter CA, Kirk MC, Ludlum DB. Release agents with nucleic acids. New York: Plenum Press; 1983. 62:656–60. of 7-alkylguanines from N-(2-chloroethyl)-NV-cyclo- p. 45–96. 22. Miller H, Grollman AP. DNA repair investigations hexyl-N-nitrosourea-modified DNA by 3-methyladenine 5. Ludlum DB. DNA alkylation by the haloethylnitro- using siRNA. DNA Repair (Amst) 2003;2:759–63. DNA glycosylase II. Cancer Res 1991;51:499–503. soureas: nature of modifications produced and their 23. Brummelkamp TR, Bernards R, Agami R. A system 38. Ludlum DB, Li Q, Matijasevic Z. Role of base excision enzymatic repair or removal. Mutat Res 1990;233:117–26. for stable expression of short interfering RNAs in repair in protecting cells from the toxicity of chlor- 6. Bignami M, O’Driscoll M, Aquilina G, Karran P. mammalian cells. Science 2002;296:550–3. oethylnitrosoureas. IARC Sci Publ 1999;271–7. Unmasking a killer: DNA O6-methylguanine and the 24. Massey A, Offman J, Macpherson P, Karran P. DNA 39. Guliaev AB, Singer B, Hang B. Chloroethylnitro- cytotoxicity of methylating agents. Mutat Res 2000;462: mismatch repair and acquired cisplatin resistance in sourea-derived ethano cytosine and adenine adducts are 71–82. E. coli and human ovarian carcinoma cells. DNA Repair substrates for E. coli glycosylases excising analogous 7. Gerson SL. MGMT: its role in cancer aetiology and (Amst) 2003;2:73–89. etheno adducts. DNA Repair (Amst) 2004;3:1311–21. cancer therapeutics. Nat Rev Cancer 2004;4:296–307. 25. Vickers MA, Vyas P, Harris PC, Simmons DL, Higgs 40. Li Q, Wright SE, Matijasevic Z, et al. The role of 8. Quinn JA, Pluda J, Dolan ME, et al. Phase II trial of DR. Structure of the human 3-methyladenine DNA human alkyladenine glycosylase in cellular resistance to carmustine plus O6-benzylguanine for patients with glycosylase gene and localisation close to the 16p the chloroethylnitrosoureas. Carcinogenesis 2003;24: nitrosourea-resistant recurrent or progressive malignant telomere. Proc Natl Acad Sci U S A 1993;90:3437–41. 589–93. glioma. J Clin Oncol 2002;20:2277–83. 26. Samson L, Derfler B, Boosalis M, Call K. Cloning and 41. Guliaev AB, Hang B, Singer B. Structural insights by 9. Barnes DE, Lindahl T. Repair and genetic consequen- characterisation of a 3-methyladenine-DNA glycosylase molecular dynamics simulations into differential repair ces of endogenous DNA base damage in mammalian cDNA from human cells whose gene maps to chromo- efficiency for ethano-A versus etheno-A adducts by the cells. Annu Rev Genet 2004;38:445–76. some 16. Proc Natl Acad Sci U S A 1991;88:9127–31. human alkylpurine-DNA N-glycosylase. Nucleic Acids 10. Sedgwick B. Repairing DNA-methylation damage. 27. O’Connor TR. Purification and characterization of Res 2002;30:3778–87. Nat Rev Mol Cell Biol 2004;5:148–57. human 3-methyladenine-DNA glycosylase. Nucleic Acids 42. Hang B, Chenna A, Guliaev AB, Singer B. Miscoding 11. Fronza G, Gold B. The biological effects of N3- Res 1993;21:5561–9. properties of 1,N 6-ethanoadenine, a DNA adduct methyladenine. J Cell Biochem 2004;91:250–7. 28. Rinne M, Caldwell D, Kelley MR. Transient adenoviral derived from reaction with the antitumor agent 1,3- 12. Karran P, Lindahl T, Ofsteng I, Evensen GB, Seeberg E. N -methylpurine DNA glycosylase overexpression bis(2-chloroethyl)-1-nitrosourea. Mutat Res 2003;531: Escherichia coli mutants deficient in 3-methyladenine- imparts chemotherapeutic sensitivity to human breast 191–203. DNA glycosylase. J Mol Biol 1980;140:101–27. cancer cells. Mol Cancer Ther 2004;3:955–67. 43. Silber JR, Bobola MS, Blank A, et al. The apurinic/ 13. Xiao W, Chow BL. Synergism between yeast nucleo- 29. Pendlebury A, Frayling IM, Santibanez Koref MF, apyrimidinic endonuclease activity of Ape1/Ref-1 con- tide and base excision repair pathways in the protection Margison GP, Rafferty JA. Evidence for simultaneous tributes to human glioma cell resistance to alkylating against DNA methylation damage. Curr Genet 1998; expression of alternatively spliced alkylpurine N- agents and is elevated by oxidative stress. Clin Cancer 33:92–9. glycosylase transcripts in human tissues and cells. Res 2002;8:3008–18. 14. Engelward BP, Dreslin A, Christensen J, et al. Repair- Carcinogenesis 1994;15:2957–60. 44. Liu L, Yan L, Donze JR, Gerson SL. Blockage of abasic deficient-3-methyladenine DNA glycosylase homozy- 30. Esteller M, Herman JG. Generating mutations but site repair enhances antitumor efficacy of 1,3-bis-(2- gous mutant mouse cells have increased sensitivity to providing chemosensitivity: the role of O 6-methylgua- chloroethyl)-1-nitrosourea in colon tumor xenografts. alkylation-induced chromosome damage and cell killing. nine DNA methyltransferase in human cancer. Onco- Mol Cancer Ther 2003;2:1061–6. EMBO J 1996;15:945–52. gene 2004;23:1–8. 45. Lau AY, Scharer OD, Samson L, Verdine GL, 15. Allan JM, Engelward BP, Dreslin AJ, et al. Mammalian 31. Danam RP, Howell SR, Brent TP, Harris LC. Ellenberger T. Crystal structure of a human alkylbase- 3-methyladenine DNA glycosylase protects against the Epigenetic regulation of O6-methylguanine-DNA meth- DNA repair enzyme complexed to DNA: mechanisms for toxicity and clastogenicity of certain chemotherapeutic yltransferase gene expression by histone acetylation and nucleotide flipping and base excision. Cell 1998;95:249–58. DNA cross-linking agents. Cancer Res 1998;58:3965–73. methyl-CpG binding proteins. Mol Cancer Ther 2005; 46. Sohn TJ, Kim NK, An HJ, et al. Gene amplification 16. Engelward BP, Allan JM, Dreslin AJ, et al. A chemical 4:61–9. and expression of the DNA repair enzyme, N-methyl- and genetic approach together define the biological 32. Chen J, Derfler B, Samson L. Saccharomyces cerevi- purine-DNA glycosylase (MPG) in HPV-infected cervical consequences of 3-methyladenine lesions in the mam- siae 3-methyladenine DNA glycosylase has homology to neoplasias. Anticancer Res 2001;21:2405–11. malian genome. J Biol Chem 1998;273:5412–8. the AlkA glycosylase of E. coli and is induced in 47. Bignami M, Casorelli I, Karran P. Mismatch repair 17. Kisby GE, Lesselroth H, Olivas A, et al. Role of response to DNA alkylation damage. EMBO J 1990;9: and response to DNA-damaging antitumour therapies. nucleotide- and base-excision repair in genotoxin- 4569–75. Eur J Cancer 2003;39:2142–9. induced neuronal cell death. DNA Repair (Amst) 2004; 33. Matijasevic Z, Boosalis M, Mackay W, Samson L, 48. Taverna P, Liu L, Hanson AJ, Monks A, Gerson SL. 3:617–27. Ludlum DB. Protection against chloroethylnitrosourea Characterisation of MLH1 and MSH2 DNA mismatch 18. Engelward BP, Weeda G, Wyatt MD, et al. Base cytotoxicity by eukaryotic 3-methyladenine DNA glyco- repair proteins in cell lines of the NCI anticancer drug excision repair deficient mice lacking the Aag alkylade- sylase. Proc Natl Acad Sci U S A 1993;90:11855–9. screen. Cancer Chemother Pharmacol 2000;46:507–16.

www.aacrjournals.org 10477 Cancer Res 2005; 65: (22). November 15, 2005

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2005 American Association for Cancer Research. Sensitization of Human Carcinoma Cells to Alkylating Agents by Small Interfering RNA Suppression of 3-Alkyladenine-DNA Glycosylase

Johanna Paik, Tod Duncan, Tomas Lindahl, et al.

Cancer Res 2005;65:10472-10477.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/65/22/10472

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2005/11/18/65.22.10472.DC1

Cited articles This article cites 40 articles, 17 of which you can access for free at: http://cancerres.aacrjournals.org/content/65/22/10472.full#ref-list-1

Citing articles This article has been cited by 6 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/65/22/10472.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/65/22/10472. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.