Acidic pH induces topoisomerase II-mediatedDNA damageHai Xiao*, Tsai-Kun Li*, Jin-Ming Yang*†, and Leroy F. Liu*†‡

*Department of Pharmacology, University of Medicine and Dentistry of New Jersey–Robert Wood Johnson Medical School, 675 Hoes Lane,Piscataway, NJ 08854-5635; and †Cancer Institute of New Jersey, 195 Little Albany & Somerset Street, New Brunswick, NJ 08901

Edited by James C. Wang, Harvard University, Cambridge, MA, and approved March 5, 2003 (received for review October 3, 2002)

Acidic pH plays an important role in various pathophysiologicalstates and has been demonstrated to be carcinogenic in animalmodels. Recent studies have also implicated acidic pH in thedevelopment of preneoplastic Barrett’s esophagus in human. How-ever, little is known about the molecular mechanism underlyingacidic pH-induced carcinogenesis. In the current study, we showthat acidic pH, like the topoisomerase II (TOP2) poison VP-16(demethylepipodophyllotoxin ethylidene-�-D-glucoside), inducestumors in 9,10-dimethyl-1,2-benzanthracene(DMBA)-initiatedmice. The following studies in tissue culture models have sug-gested that acidic pH acts like a TOP2 poison to induce TOP2-mediated DNA damage: (i) acidic pH induces TOP2-dependent DNAdamage signals as evidenced by up-regulation of p53 and Ser-139phosphorylation of H2AX [a substrate for ataxia telangiectasiamutated (ATM)�ATM and Rad3-related (ATR) kinases]; (ii) acidicpH-induced cytotoxicity in tumor cells is reduced in TOP2-deficientcells; (iii) acidic pH increases the mutation frequency of the hypo-xanthine phosphoribosyl transferase (HPRT) gene in a TOP2-dependent manner; and (iv) acidic pH induces reversible TOP2-mediated DNA strand breaks in vitro. We discuss the possibilitythat TOP2-mediated DNA damage may contribute to acidic pH-induced carcinogenesis.

Acidic pH plays an important role in cell death during variouspathophysiological states, including ischemia and cancer

(1–3). In addition to cell death, acidic pH has been suggested tobe involved in carcinogenesis. For example, Barrett’s esophagus(BE), which is a preneoplastic disorder of the esophagus (4), hasbeen tightly linked to gastroesophageal acid reflux (5). Studiesin rats have also demonstrated that duodenal-content refluxesophagitis induces the development of glandular metaplasiaand adenosquamous carcinoma (6). In addition, acid has beenshown to promote carcinogenesis in the hamster cheek pouch incombination with the tumor initiator 9,10-dimethyl-1,2-benzanthracene (DMBA) (7).

Despite the importance of acidic pH in carcinogenesis, little isknown about its underlying molecular mechanism. Several stud-ies have suggested that acidic pH may induce DNA damage incells. For example, in cultured Chinese hamster cells, acidic pHinduces chromosome aberrations including sister-chromatidexchanges and chromatid-type breaks and gaps (8, 9). AcidicpH, like DNA damaging agents, also up-regulates p53 andinduces p53-dependent apoptosis in human adenoma and car-cinoma cells (2, 10). These studies suggest that acidic pH mayinduce DNA damage, which could contribute to its carcinogenicactivity.

Human topoisomerase II (TOP2) isozymes are moleculartargets for many antitumor drugs (e.g., doxorubicin, etoposide(VP-16), mitoxantrone, and amsacrine; refs. 11–13). Theseantitumor drugs, referred to as TOP2 poisons, are highly effi-cient in inducing TOP2-mediated DNA damage in cells. Theyinterfere with the breakage�religation reactions of TOP2 bystabilizing the transient covalent reaction intermediates, theTOP2–DNA covalent complexes, often referred to as TOP2cleavable or cleavage complexes (14). TOP2-mediated DNAdamage is known to trigger tumor cell death (15), and induce

extensive gene deletion and rearrangements (16). Patientstreated with the TOP2 poison VP-16 exhibit a high frequency ofsecondary acute myeloid leukemia (AML), primarily due torearrangement of the MLL gene (17, 18).

Previous studies have demonstrated that both calf thymus andDrosophila TOP2s induce DNA cleavage at pH below 7.0, withan optimum about 5 (19, 20). In the current study, we show thatTOP2 is involved in acidic pH-induced DNA damage, mutagen-esis, and cytotoxicity. We discuss the possibility that TOP2-mediated DNA damage may contribute to acidic pH-inducedcarcinogenesis.

Materials and MethodsChemicals and Cell Lines. 4-Demethylepipodophyllotoxin thenyli-dene-�-D-glucoside (teniposide; VM-26) was obtained as a giftfrom the Bristol-Myers Squibb. 4,4-(2,3-Butanediyl)-bis(2,6-piperazinedione) (ICRF-193) was purchased from ICN. Allother chemicals were purchased from Sigma. These compoundswere dissolved in DMSO. The HL-60 cell line and its mitox-antrone-resistant (TOP2-deficient) variant, HL-60�MX2, wereobtained from ATCC (American Type Culture Collection).Human breast cancer cell lines MDA-MB-231, T47D, and theiretoposide-resistant cell lines (MDA-MB-231�3000 and T47D�VP1, respectively) were kindly provided by T. Fojo (MedicineBranch, Center for Cancer Research, National Cancer Institute,National Institutes of Health, Bethesda). Purification of recom-binant human TOP2 isozymes was performed following thepublished procedure (21). Cells were maintained in RPMImedium 1640 supplemented with 10% FCS. Different pH media(pH 4.0, 4.5, 5.0, 6.0, and 7.5) were buffered by citrate phosphate(15 mM). Antibodies against a synthetic peptide consisting of thelast nine amino acids (KATQASQEY) of H2AX with phospho-Serine-139 were obtained from D. Chen (Lawrence BerkeleyNational Laboratory, Berkeley, CA).

Immunoblotting Analysis. Cells were subjected to different treat-ment conditions, and then lysed with 100 �l of 2� SDS samplebuffer. Proteins were separated in 10% SDS�PAGE gel andelectrophoretically transferred onto a nitrocellulose membrane.All membranes were Ponceau-stained to confirm equal proteinloading. The membrane was blocked with 5% milk for 1 h.Immunoblotting was performed by using monoclonal mouseantibody against human p53 (Ab-6) (Oncogene) or antibodiesagainst phosphorylated H2AX. The secondary antibodies wereincubated for 1 h at room temperature. Bound secondary

This paper was submitted directly (Track II) to the PNAS office.

Abbreviations: TOP2, topoisomerase II; VM-26, 4-demethylepipodophyllotoxin thenyli-dene-�-D-glucoside (teniposide); VP-16, demethylepipodophyllotoxin ethylidene-�-D-glucoside (etoposide); CPT, camptothecin; DMSO, dimethyl sulfoxide; ICRF-193, 4,4-(2,3-butanediyl)-bis(2,6-piperazinedione); 6-TG, 2-amino-6-mercaptopurine (6-thioguanine);DMBA, 9,10-dimethyl-1,2-benzanthracene; TPA, phorbol 12-tetradecanoate 13-acetate;HPRT, hypoxanthine phosphoribosyl transferase; pHi, intracellular pH; pHe, extracellularpH; BE, Barrett’s esophagus; ATM, ataxia telangiectasia mutated; ATR, ATM andRad3-related.

‡To whom correspondence should be addressed. E-mail: lliu@umdnj.edu.

www.pnas.org�cgi�doi�10.1073�pnas.0935978100 PNAS � April 29, 2003 � vol. 100 � no. 9 � 5205–5210

CELL

BIO

LOG

Y

antibodies were detected by using the ECL Western procedure(Pierce).

TOP2-Mediated DNA Cleavage Assay. TOP2-mediated DNA cleav-age assay was performed by using purified recombinant humanTOP2 (10 ng each) and 3�-end 32P-labeled YEpG DNA (20 ngeach) in a reaction (20 �l each) containing 148 mM NaCl, 4.7mM KCl, 1.2 mM KH2PO4, 10 mM MgCl2, 2 mM CaCl2, and 20mM citrate phosphate (or Hepes) with different pHs. Incubationwas carried out at 37°C for 30 min. In reversal experiments, theincubated reactions were subjected to a second incubation atdifferent temperatures, in excess EDTA (10 mM at 37°C) or withpH adjusted to neutrality (pH 7.5 at 37°C) for another 30 min.The reactions were terminated by the addition of 5 �l of 5% SDSand 1 mg�ml proteinase K, and incubated for an additional 60min at 37°C. DNA samples were electrophoresed in 1% agarosegel containing 0.5� TPE buffer. Gels were then dried ontoWhatman 3MM chromatographic paper and autoradiographedat �80°C by using Kodak XAR-5 films.

Clonogenic Assay. For attached cells, 250 cells per well were platedin six-well plates and cultured overnight. Cells were treated withdifferent agents for 30 min, washed, and replenished with freshmedium. After 2 wk, cells were stained with 10 mg�ml methyleneblue (Sigma) in 50% methanol, and the colony number wascounted. For suspension culture (i.e., HL-60 and HL-60�MX2cells), cells were treated with different agents for 30 min.Subsequently, 250 cells were mixed with 0.3% agar (Sigma,A-9915) and plated on top of 0.5% agar in six-well plates. After2 wk, colonies were stained with 1% p-iodonitrotetrazoliumviolet (Sigma) and counted.

Mutation Frequency Measurement. MDA-MB-231 cells were pre-selected for 4 days in RPMI medium 1640 supplemented with theHAT medium to remove preexisting hypoxanthine phosphori-bosyl transferase (HPRT) mutants. Cells (1 � 106) were thenseeded in 100-mm dishes for 24 h before treatment with VM-26(0.5 �M), camptothecin (CPT; 0.5 �M), or a pH 6.0 medium for1 h in the presence or absence of ICRF-193 (25 �M). Treatedcells were trypsinized and plated in 150-mm dishes. Cells wereincubated for 7 days to allow the wild-type HPRT protein toturnover in mutant HPRT cells before 2-amino-6-mercaptopu-rine (6-thioguanine) (6-TG) selection. 6-TG selection was per-formed by adding 5 �g�ml 6-TG to cells, followed by plating in100-mm dishes. After 21 days, cells were stained with 10 mg�mlmethylene blue (Sigma) in 50% methanol, and the colonynumber was counted. For cloning efficiency, 250 cells wereplated in medium without 6-TG, and the colony number wascounted after 14 days of incubation.

Carcinogenesis Assay Using the Mouse Skin Model. Female CD-1mice (3–5 wk old) were used in the carcinogenesis study (22, 23).The back of each mouse was shaved 2 days before differenttreatments. The tumor initiator DMBA (600 nmol in 100 �l ofDMSO) was applied only once during the first week. Onehundred microliters of phorbol 12-tetradecanoate 13-acetate(TPA; 10 nmol in 100 �l DMSO), VP-16 (5 �mol in 100 �lDMSO), or acidic pH medium (250 mM citrate phosphate, pH2.5, in RPMI medium 1640) was applied twice every week.Papillomas appearing on the skin were recorded every week. Inthe second experiment, a group of five mice were treated withVP-16 (5 �mol in 100 �l of DMSO) twice weekly for 10 wk. Atthe end of the 10th week, a single application of DMBA (600nmol in 100 �l of DMSO) was performed. Tumor formation wasmeasured at the end of the 16th week.

ResultsAcidic pH Induces Carcinogenesis in DMBA-Initiated Mice. We havetested whether acidic pH can induce tumors in the mouse skincarcinogenesis model. As shown in Fig. 1A, application of acidicpH buffer (100 �l of 250 mM citrate phosphate, pH 2.5) to themouse skin twice a week for 16 wk resulted in no tumorformation. However, if the mice were initiated with a singleapplication of DMBA followed by repeated application of acidicpH, tumor formation was readily observable (11 papillomas in agroup of five mice). As expected, a single application of DMBAby itself resulted in no tumor formation. Interestingly, theprototypic TOP2 poison VP-16 behaved similarly to acidic pH.Repeated application of VP-16 (5 �mol in 100 �l, twice weeklyfor 16 wk) resulted in no tumor formation. However, a singleapplication of DMBA followed by repeated application of VP-16resulted in significant tumor formation (14 papillomas in a groupof five mice; Fig. 1 A). In this case, we have switched the orderof application. We found that, whether DMBA was appliedbefore (Fig. 1 A) or after (Fig. 1B) VP-16 application, similarnumbers of tumors were observed (14 vs. 19 pappilomas). As apositive control, DMBA treatment followed by repeated appli-cation of the tumor promoter TPA was shown to result in 26papillomas in a group of five mice (Fig. 1 A). These studiessuggest that, like VP-16, acidic pH is potentially carcinogenic.

Acidic pH Induces p53 Up-Regulation and Ser-139 Phosphorylation ofH2AX. Previous studies have demonstrated that acidic pH inducesup-regulation of p53 in human glioblastoma cells (10). In thecurrent study, we show that acidic pH also induces up-regulationof p53 in breast cancer ZR75-1 cells. As shown in Fig. 2A,incubation of ZR75-1 cells in pH 6.0 medium for 1 h inducedup-regulation of p53. As positive controls, both VP-16 (a TOP2poison) and CPT (a TOP1 poison) were shown to induce morepronounced up-regulation of p53. In addition to p53, ataxiatelangiectasia mutated (ATM)�ATM and Rad3-related (ATR)

Fig. 1. Acidic pH and VP-16 induce tumors in the mouse skin carcinogenesismodel. (A) Acidic pH and VP-16 induce papillomas in DMBA-initiated mice.Groups of five mice were treated with different drugs (in DMSO) or pH 2.5medium as described in Materials and Methods. DMBA (600 nmol in 100 �l ofDMSO) was applied only once during the first week. TPA (10 nmol in 100 �l ofDMSO), VP-16 (5 �mol in 100 �l of DMSO), and pH 2.5 medium (250 mM citratephosphate-buffered RPMI medium 1640) were applied twice every week for16 wk. The total number of papillomas appearing on the mouse skin in eachgroup was measured. (B) DMBA induces papillomas in VP-16-initated mice. Agroup of five mice were treated with VP-16 (5 �mol in 100 �l of DMSO) twiceweekly for 10 wk. At the end of the 10th week, a single application of DMBA(600 nmol in 100 �l of DMSO) was performed. Tumor formation was measuredat the end of the 16th week.

5206 � www.pnas.org�cgi�doi�10.1073�pnas.0935978100 Xiao et al.

family kinases are also important DNA damage signaling mol-ecules (24). ATM�ATR kinases are known to phosphorylate alarge number of substrates, including H2AX, which is phosphor-ylated at serine 139 (25, 26). As shown in Fig. 2B, incubation ofmouse embryo fibroblast (MEF) cells in RPMI medium 1640(pH 5.0) for 1 h resulted in detectable phosphorylated H2AX. Amore pronounced increase of phosphorylated H2AX was ob-served in MEF cells treated with VM-26 (10 �M) and CPT (10�M). These results suggest that acidic pH induces DNA damagesignals.

Acidic pH-Induced DNA Damage Signaling and Cytotoxicity Is Reducedin TOP2-Deficient Cells. To determine whether acidic pH-inducedDNA damage signaling involves TOP2, acidic pH-induced phos-phorylation of H2AX in a pair of TOP2 wild-type (HL-60) andTOP2-deficient cells (HL-60�MX2) was investigated (see Fig. 3Bfor TOP2 isozyme levels in HL-60 and HL-60�MX2; refs. 27 and28). As shown in Fig. 3A, incubation of HL-60 cells in RPMImedium 1640 (pH 5.0) for 1 h resulted in increased phosphor-

ylation of Ser-139 of H2AX. RPMI medium 1640 (pH 4.0 or pH6.0) was not as effective as RPMI medium 1640 (pH 5.0).Importantly, acidic pH (pH 5.0)-induced phosphorylation ofH2AX was significantly diminished in TOP2-deficient HL-60�MX2 cells, suggesting that acidic pH-induced DNA damagesignaling is TOP2-mediated. As a positive control, VM-26-induced phosphorylation of H2AX was shown to be similarlyreduced in HL-60�MX2 cells as compared with HL-60 cells. Bycontrast, CPT (a TOP1 poison)-induced phosphorylation ofH2AX was little changed in HL-60�MX2 cells as compared withHL-60 cells, suggesting that the reduced DNA damage signalingwas specific for TOP2-mediated DNA damage.

To determine whether acidic pH-induced cytotoxicity is alsoTOP2-mediated, acidic pH-induced cytotoxicity was measuredin several pairs of wild-type and TOP2-deficient cells, HL60 vs.HL-60�MX2, MDA-MB-231 vs. MDA-MB-231�3000, and T47Dvs. T47D�VP1. If TOP2-DNA covalent complexes are respon-sible for acidic pH-induced cytotoxicity, these TOP2-deficientcells are expected to be cross-resistant to acidic pH. As shown inFig. 4A, TOP2-deficient MDA-MB-231�3000 cells were moreresistant to a brief 30-min treatment of an acidic pH medium (pH4.5) as measured by a clonogenic assay. VM-26, a TOP2 poison,and CPT, a TOP1 poison, served as positive and negativecontrols, respectively. VM-26-induced cytotoxicity was similarlyreduced in TOP2-deficient MDA-MB-231�3000 cells whereasCPT-induced cytotoxicity was little affected (Fig. 4A). Similarly,TOP2-deficient T47D�VP1 and HL-60�MX2 cells were shownto be more resistant to a brief 30-min treatment of an acidicmedium (pH 4.5) than their respective parental cells (T47D andHL-60 cells) as determined by clonogenic assays (Fig. 4 B and C).These results suggest that acidic pH-induced cytotoxicity, likeacidic pH-induced DNA damage signaling, involves TOP2.

Acidic pH-Induced Mutagenesis Is Antagonized by a TOP2 CatalyticInhibitor ICRF-193. Resistance to 6-TG, which is primarily due tomutations in the HPRT gene, has been used extensively formeasuring the mutation frequency in mammalian cells (29). Asshown in Fig. 5, a brief exposure (1 h) of MDA-MB-231 breastcancer cells to VM-26 (0.5 �M), CPT (0.5 �M), or acidic pH (pH6.0) medium resulted in an increase in 6-TG resistance. To testwhether acidic pH (and VM-26)-stimulated mutagenesis isTOP2-mediated, the TOP2 catalytic inhibitor ICRF-193 (25�M) was used to antagonize the action of TOP2. As shown inFig. 5, ICRF-193 was able to antagonize the mutagenic activityof both VM-26 and acidic pH, but not CPT. In the case of VM-26and acidic pH, the number of 6-TG resistant colonies wasreduced by �50% in the presence of ICRF-193 (Fig. 5). Theseresults suggest that acidic pH-induced mutagenesis also involvesTOP2.

Acidic pH Induces Reversible TOP2-Mediated DNA Strand Breaks inVitro. To test whether acidic pH induces TOP2-mediated DNAdamage, we have performed DNA cleavage assay using purifiedhuman TOP2 isozymes and 32P-end-labeled linearized plasmidDNA. As shown in Fig. 6A, extensive cleavage of DNA into smallDNA fragments occurred between pH 5.0 and 7.0 in thepresence of human TOP2�. Above pH 7.5 or below pH 4.0, littleDNA cleavage was detectable. The strongest cleavage wasobserved at pH 5.0, consistent with the results obtained withboth calf thymus and Drosophila TOP2s (19, 20). Similar resultswere obtained with human TOP2� (data not shown). Unlikeclinically useful TOP2 poisons such as VP-16 and doxorubicin,which are stimulated 30- to 100-fold by 1 mM ATP (30), acidicpH-induced DNA cleavage was insensitive to ATP (data notshown). The DNA cleavage pattern, which is reflective of DNAcleavage specificity, was also slightly different between acidicpH- and VM-26-induced DNA cleavages (the differences were

Fig. 2. Acidic pH induces up-regulation of p53 and phosphorylation ofH2AX. (A) Acidic pH induces up-regulation of p53. Breast cancer ZR75-1 cellswere treated with VP-16 (25 �M), CPT (25 �M), or RPMI medium 1640 (pH 6.0)for 1 h. Treated cells were lysed directly with SDS sample buffer and preparedfor immunoblotting by using anti-p53. (B) Acidic pH induces phosphorylationof H2AX. Mouse embryo fibroblast (MEF) cells were treated with VM-26 (10�M), CPT (10 �M), and different pH media for 1 h. Cells were then lysed andprocessed for detection of the phosphorylated H2AX by immunoblottingusing antibodies against the phosphorylated H2AX epitope.

Fig. 3. Acidic pH-induced H2AX phosphorylation is defective in a TOP2mutant cell line. (A) HL-60 and HL-60�MX2 cells were treated with VM-26 (10�M), CPT (10 �M), and different pH media for 1 h. Cells were then lysed andprocessed for detection of the phosphorylated H2AX. (B) HL-60 and HL-60�MX2 cells were lysed and processed for detection of TOP2� and TOP2�.

Xiao et al. PNAS � April 29, 2003 � vol. 100 � no. 9 � 5207

CELL

BIO

LOG

Y

marked by arrow heads on the side of the cleavage lanes inFig. 6A).

Like DNA cleavages induced by TOP2 poisons, acidic pH-induced DNA cleavage was reversible. As shown in Fig. 6B,neutralization of the reaction (compare lane 8 with lane 14),addition of excess EDTA (10 mM EDTA final concentration;compare lane 8 with lane 13), or shifting the temperature from37°C to 75°C (compare lane 8 with lane 12) in a secondincubation (30 min) before termination of the reaction with 1%SDS resulted in essentially no cleavage. As a control, VM-26-induced DNA cleavage was shown to be reversed in a secondincubation with either excess EDTA (compare lane 2 with lane7) or at 75°C (compare lane 2 with lane 6). The reversibility ofthe DNA cleavage is the hallmark of topoisomerase cleavagecomplexes (31). These results thus suggest that acidic pH stabi-lizes reversible TOP2 cleavage complexes in vitro. Previous

studies have demonstrated that acidic pH-induced DNA breaksare primarily single-strand breaks (20). As shown in Fig. 6C,acidic pH induced both single-strand breaks as evidenced by thenicked DNA (marked N) and double-strand breaks as evidencedby the linear DNA (marked L). At pH 5.5, TOP2-mediated DNAcleavage generated more nicked DNA than linear DNA (Fig.6C), suggesting that single-strand breaks are the predominantproduct of acidic pH-induced DNA damage.

DiscussionAcidic pH is known to cause genomic instability and is causallylinked to carcinogenesis (8, 9, 32). In the current study, we haveshown that acidic pH can induce tumors in DMBA-initiatedmice.

To study the molecular basis for acidic pH-induced carcino-genesis, we have investigated the possibility that acidic pH mayinduce DNA damage in tissue culture models. Indeed, acidic pH,like DNA damaging agents, up-regulates p53 in breast cancerZR75-1 cells. This result is consistent with a previous study inwhich acidic pH (pH 6.5) has been shown to up-regulate p53 inhuman glioblastoma cells (10). The possibility that acidic pHinduces DNA damage is further supported by our studies onATM�ATR family kinases. Acidic pH was shown to activateATM�ATR family kinases as evidenced by increased phosphor-ylation on Serine 139 of histone H2AX (26). It is particularlyinteresting to note that the optimal pH for activation of ATM�ATR kinases was pH 5.0. These studies were performed byadjusting the extracellular pH (pHe) in the tissue culture me-dium. Intracellular pH (pHi) is known to be regulated bymultiple mechanisms when cells are exposed to a nonphysiologi-cal pHe environment (33–35). The influence of pHe on pHi hasbeen well studied. For example, when HL-60 cells are exposedto acidic pH medium in the range of 6.1 to 6.8 (pHe), the pHiin HL-60 cells has been determined to range from 6.5 to 7.2(about 0.4 unit higher than pHe) within 1 h (36). Although thepHi was not determined in our studies, it seems likely that theobserved DNA damaging signals were due to intracellularacidification.

Acidic pH-induced DNA damage signaling seems to involveTOP2 because acidic pH-induced activation of ATM�ATRkinases (as evidenced by phosphorylation of H2AX at Serine139) was significantly reduced in TOP2-deficient HL-60�MX2cells. Studies on acidic pH-induced cytotoxicity have also pro-vided indirect support for a role of TOP2 in acidic pH-inducedDNA damage. In three independent TOP2-deficient cell lines(HL-60�MX2, MDA-MB-231�3000, and T47D�VP1), acidic pHwas shown to induce significantly greater cytotoxicity in their

Fig. 4. Acidic pH cytotoxicity is reduced in TOP2-deficient cells. Cytotoxicitywas measured by using the clonogenic assay as described in Materials andMethods. (A) Acidic pH cytotoxicity is reduced in TOP2-deficient MDA-MB-231�3000 cells. MDA-MB-231 and TOP2-deficient MDA-MB-231�3000 (at-tached cells) were treated with 50 �M VP-16, 1 �M CPT, pH 4.5 medium, or pH7.5 medium for 30 min. Cells were then replenished with drug-free neutralmedium and incubated for �2 wk. Clonogenic survival was measured bycounting colonies on plates. (B) Acidic pH cytotoxicity is reduced in TOP2-deficient T47D�VP1 cells. T47D and TOP2-deficient T47D�VP1 cells (attachedcells) were treated with 50 �M VP-16 or pH 4.5 medium for 30 min. Cells werethen incubated in drug-free neutral medium for 2 wk. Clonogenic survival wasalso measured as described in A. (C) Acidic pH cytotoxicity is reduced inTOP2-deficient HL-60�MX2 cells. HL-60 and TOP2-deficient HL-60�MX2 cells(suspension cells) were treated with 2 �M VM-26 or pH 4.5 medium for 30 min.Cells were then resuspended in soft agar (in drug-free neutral medium), andclonogenic survival in soft agar was measured after �2 wk as described inMaterials and Methods.

Fig. 5. Acidic pH increases 6-TG resistance. MDA-MB-231 cells were treatedwith 0.5 �M VM-26, 0.5 �M CPT, pH 6.0 medium, or pH 7.4 medium (control)for 1 h in the presence or absence of ICRF-193 (25 �M), followed by recoveryin RPMI medium 1640 (pH 7.4) for 7 days. Selection with 6-TG was performedas described in Materials and Methods.

5208 � www.pnas.org�cgi�doi�10.1073�pnas.0935978100 Xiao et al.

corresponding wild-type cells. All these studies have pointed toa role for TOP2 in mediating acidic pH-induced DNA damage.

Our studies have also suggested that TOP2-mediated DNAdamage contributes to acidic pH-induced genomic instability.We have demonstrated in the current study that both acidic pHand VM-26 stimulated mutagenesis as measured by 6-TG resis-tance. Most strikingly, acidic pH- and VM-26-stimulated mu-tagenesis was antagonized by ICRF-193, a TOP2 catalytic in-hibitor known to antagonize TOP2 cleavage complex formation(37). By contrast, CPT-stimulated mutagenesis was not affectedby ICRF-193. These results are consistent with the notion thatacidic pH-stimulated mutagenesis is in part due to the formationof TOP2 cleavage complexes.

All these results can be explained by the formation of TOP2cleavage complexes in acidotic cells. We have made numerousattempts using different techniques [e.g., the in vivo complex ofenzyme (ICE) and band depletion methods; refs. 38 and 39] todirectly demonstrate the presence of TOP2 cleavage complexesin acidic pH-treated cells. However, we have been unable todemonstrate reproducibly and convincingly the presence ofTOP2 cleavage complexes in cells possibly due to the low levelof TOP2 cleavage complexes induced by acidic pH. As analternative, we have studied TOP2 cleavage complex formationin vitro by using purified TOP2. Our studies using purifiedhTOP2� have demonstrated that acidic pH in the range of 5.0 to7.0 can induce TOP2-mediated DNA cleavage. The possibilitythat DNA depurination may contribute to acidic pH-inducedDNA cleavage (40, 41) was ruled out because acidic pH-treatedDNA (under our assay conditions) did not induce DNA cleavagein a subsequent incubation with TOP2 at neutral pH (data notshown). The result from our reversal experiment in whichpH-induced DNA cleavage was completely abolished on neu-tralization also supports this conclusion. The reversibility ofDNA cleavage suggests that the acidic pH-induced DNA cleav-age most certainly reflects the formation of TOP2 cleavagecomplexes. The simplest explanation for acidic pH-inducedDNA cleavage is that the religation reaction of TOP2 is inhibitedby acidic pH. The pH profile of TOP2-mediated DNA cleavagecould suggest the involvement of a histidine residue. We havemutated the nonessential histidine, His-795 (42), which is locatednear the active site tyrosine (Tyr-805) of TOP2�, into alanine,and found the pH profile of TOP2-mediated DNA cleavage tobe unchanged. Clearly, further studies are necessary to establishthe molecular basis for acidic pH-induced formation of TOP2cleavage complexes.

Based on our in vitro studies, we can estimate the efficiency ofacidic pH in trapping TOP2 cleavage complexes as comparedwith VM-26. The amount of TOP2 cleavage complexes trappedin a pH 5.0 reaction buffer is about the same as that trapped by1 �M VM-26 in the absence of ATP (Fig. 6A). In the presenceof ATP (which is a closer approximation to the in vivo situation),the amount of TOP2 cleavage complexes trapped by VM-26 isstimulated 30- to 100-fold (30) whereas that trapped by acidic pHis minimally affected (this study). Consequently, acidic pH maytrap the same amount of TOP2 cleavage complexes as 10–30 nMVM-26. Neither the band depletion assay nor the ICE methodhas the sensitivity to detect TOP2 cleavage complexes inducedby 10–30 nM VM-26.

Fig. 6. Acidic pH induces TOP2-mediated DNA breaks in vitro. (A) Acidic pHinduces TOP2-mediated DNA cleavage in vitro. 3�-end 32P-labeled linearizedYEpG DNA and hTOP2� were incubated in reactions with different pHs (lanes4–13) for 30 min at 37°C as described in Materials and Methods. Reactionswere terminated with SDS�proteinase K. VM-26 (0, 0.5, and 5 �M) was used ascontrol in a standard cleavage reaction at pH 7.5 (lanes 1–3). (B) AcidicpH-induced DNA cleavage is reversible. DNA cleavage in the presence ofhTOP2� was performed as described in A at pH 6.0 (lanes 8–14), pH 7.5(control), or pH 7.5 with 5 �M VM-26 (lanes 2–7). Before termination, thereactions were subjected to a second incubation under different reversalconditions (indicated on top of each lane) for 30 min. The reversal conditions

for the second incubation include different temperatures (lanes 3–6 and9–12), neutralization of the reactions to pH 7.5 (lane 14), and excess EDTA (10mM; lanes 7 and 13). (C) Acidic pH induces both single- and double-strandedDNA breaks. Supercoiled pBluescript SK(�) DNA was used to react withhTOP2� in various pH conditions as described in A. The reactions were termi-nated by SDS�proteinase K and subjected to electrophoresis in 1.0% agarosegel containing 0.5 �g�ml ethidium bromide. N, nicked DNA; L, linear DNA; SC,supercoiled DNA.

Xiao et al. PNAS � April 29, 2003 � vol. 100 � no. 9 � 5209

CELL

BIO

LOG

Y

Our carcinogenesis studies have shown that both VP-16 andacidic pH induce tumors in DMBA-initiated mice. Consideringtheir differential potency in inducing TOP2 cleavage complexes,their comparable potency in inducing tumors is surprising. Thisresult could suggest that acidic pH may induce tumors by adifferent mechanism (e.g., increased reactive oxygen speciesproduction and�or altered calcium homeostasis) than VP-16. Onthe other hand, the dose-response of VP-16 in inducing tumorshas not been determined. Being both an anticancer drug and acarcinogen, the dose-response of VP-16 in tumor formation maybe complex. The concentration of VP-16 used in our currentstudy could have exceeded the optimal concentration for stim-ulation of carcinogenesis. Consequently, the comparison be-tween the potency of VP-16 and acidic pH might not be relevant.The acidic pH used in our carcinogenesis studies is quite low (pH2.5). However, the exposure of the susceptible cells in the mouseskin to this low pH solution is likely to be quite transient becausethe solution dries out within a few minutes. The exact pHi in thesusceptible cells on the mouse skin cannot be properly estimated.Consequently, the potential role of TOP2 in carcinogenesisneeds to be rigorously investigated. The conditional TOP2knockout mice could be useful for establishing the role of TOP2in acidic pH-induced carcinogenesis.

We have shown that neither VP-16 (also acidic pH) norDMBA alone can induce tumors in mice. However, VP-16 (alsoacidic pH) induces large numbers of tumors in DMBA-initiatedmice. It seems that VP-16 and acidic pH behave as tumorpromoters in the mouse skin carcinogenesis model. In Fig. 1B,we have also shown that the treatment sequence of VP-16 andDMBA is not important in stimulating the formation of tumors,a result distinctly different from that of tumor promoters.

However, tumor promotion has been subdivided into stage-I andstage-II (43). In contrast to the stage-II tumor promotion, whichis associated with increased cell proliferation, the stage-I tumorpromotion is irreversible and may be due to permanent alter-ation of the genetic materials (44). In addition, tumor formationis not affected by the reversal of the sequence of treatment withthe initiator and the stage-I tumor promoter (45). It seemspossible that VP-16 and acidic pH may act as stage-I tumorpromoters.

DMBA, a tumor initiator, is known to be a strong pointmutagen and is associated with oncogenic ras mutations (46). Onthe other hand, VP-16 is known to induce deletions and DNArearrangements (47). Based on our current results, acidic pH,being a TOP2 poison, is likely to be a rearrangement mutagen.It will be interesting to test whether DNA sequence rearrange-ments could be the underlying mechanism for stage-I tumorpromotion. Clearly, further studies are necessary to establish themolecular mechanism of TOP2-mediated carcinogenesis.

Our results obtained in mice could be relevant to the etiologyof BE. BE is tightly associated with gastroesophageal acid reflux(5). Based on our results, acid reflux could introduce mutationsto cells in the esophagus due to acidic pH-induced DNA damage.However, acid reflux also causes inflammatory responses thatare known to contribute to carcinogenesis. It will be interestingto determine whether acidic pH-induced DNA damage plays animportant role in the development of BE.

We are grateful to Dr. David Chen for providing us with anti-H2AXantibodies, and Dr. Tito Fojo for some of the TOP2-deficient cell lines.This work was supported by National Institutes of Health GrantsGM27731 and CA39662.

1. Kubasiak, L. A., Hernandez, O. M., Bishopric, N. H. & Webster, K. A. (2002)Proc. Natl. Acad. Sci. USA 99, 12825–12830.

2. Williams, A. C., Collard, T. J. & Paraskeva, C. (1999) Oncogene 18, 3199–3204.3. Yamamoto, D., Kiyozuka, Y., Uemura, Y., Yamamoto, C., Takemoto, H.,

Hirata, H., Tanaka, K., Hioki, K. & Tsubura, A. (2000) J. Cancer Res. Clin.Oncol. 126, 191–197.

4. Chevfec, G., Schnell, T. & Sontag, S. (1992) Am. J. Clin. Pathol. 98, 5–7.5. Shaheen, N. & Ransohoff, D. F. (2002) J. Am. Med. Assoc 287, 1972–1981.6. Pera, M., Brito, M. J., Poulsom, R., Riera, E., Grande, L., Hanby, A. & Wright,

N. A. (2000) Carcinogenesis 21, 1587–1591.7. Adams, J., Heintz, P., Gross, N., Andersen, P., Everts, E., Wax, M. & Cohen,

J. (2000) Arch. Otolaryngol. Head Neck Surg. 126, 405–409.8. Morita, T., Nagaki, T., Fukuda, I. & Okumura, K. (1992) Mutat. Res. 268,

297–305.9. Morita, T. (1995) Mutat. Res. 334, 301–308.

10. Ohtsubo, T., Wang, X., Takahashi, A., Ohnishi, K., Saito, H., Song, C. W. &Ohnishi, T. (1997) Cancer Res. 57, 3910–3913.

11. Wang, J. C. (2002) Nat. Rev. Mol. Cell Biol. 3, 430–440.12. Li, T. K. & Liu, L. F. (2001) Annu. Rev. Pharmacol. Toxicol. 41, 53–77.13. Liu, L. F. (1989) Annu. Rev. Biochem. 58, 351–375.14. D’Arpa, P. & Liu, L. F. (1989) Biochim. Biophys. Acta 989, 163–177.15. Ritke, M. K., Rusnak, J. M., Lazo, J. S., Allan, W. P., Dive, C., Heer, S. &

Yalowich, J. C. (1994) Mol. Pharmacol. 46, 605–611.16. Chen, C. L., Fuscoe, J. C., Liu, Q., Pui, C. H., Mahmoud, H. H. & Relling, M. V.

(1996) J. Natl. Cancer Inst. 88, 1840–1847.17. Felix, C. A. (2001) Med. Pediatr. Oncol. 36, 525–535.18. Lovett, B. D., Strumberg, D., Blair, I. A., Pang, S., Burden, D. A., Megonigal,

M. D., Rappaport, E. F., Rebbeck, T. R., Osheroff, N., Pommier, Y. G. et al.(2001) Biochemistry 40, 1159–1170.

19. Liu, L. F., Rowe, T. C., Yang, L., Tewey, K. M. & Chen, G. L. (1983) J. Biol.Chem. 258, 15365–15370.

20. Zechiedrich, E. L., Christiansen, K., Andersen, A. H., Westergaard, O. &Osheroff, N. (1989) Biochemistry 28, 6229–6236.

21. Wasserman, R. A., Austin, C. A., Fisher, L. M. & Wang, J. C. (1993) CancerRes. 53, 3591–3596.

22. Brooks, G., Evans, A. T., Aitken, A. & Evans, F. J. (1989) Carcinogenesis 10,283–288.

23. Chouroulinkov, I., Lasne, C., Lowy, R., Wahrendorf, J., Becher, H., Day, N. E.& Yamasaki, H. (1989) Cancer Res. 49, 1964–1969.

24. Shiloh, Y. (2001) Curr. Opin. Genet. Dev. 11, 71–77.25. Ward, I. M. & Chen, J. (2001) J. Biol. Chem. 276, 47759–47762.26. Burma, S., Chen, B. P., Murphy, M., Kurimasa, A. & Chen, D. J. (2001) J. Biol.

Chem. 276, 42462–42467.27. Harker, W. G., Slade, D. L., Drake, F. H. & Parr, R. L. (1991) Biochemistry 30,

9953–9961.28. Harker, W. G., Slade, D. L., Parr, R. L. & Holguin, M. H. (1995) Cancer Res.

55, 4962–4971.29. Kulling, S. E. & Metzler, M. (1997) Food Chem. Toxicol. 35, 605–613.30. Wang, H., Mao, Y., Zhou, N., Hu, T., Hsieh, T. S. & Liu, L. F. (2001) J. Biol.

Chem. 276, 15990–15995.31. Zhang, H., D’Arpa, P. & Liu, L. F. (1990) Cancer Cells 2, 23–27.32. LeBoeuf, R. A. & Kerckaert, G. A. (1986) Carcinogenesis 7, 1431–1440.33. Clausen, T. (1992) Can. J. Physiol. Pharmacol. 70, Suppl., S219–S222.34. Deitmer, J. W. & Rose, C. R. (1996) Prog. Neurobiol. 48, 73–103.35. Hackam, D. J., Grinstein, S. & Rotstein, O. D. (1996) Shock 5, 17–21.36. Park, H. J. (1995) Yonsei Med. J. 36, 473–479.37. Ishida, R., Miki, T., Narita, T., Yui, R., Sato, M., Utsumi, K. R., Tanabe, K.

& Andoh, T. (1991) Cancer Res. 51, 4909–4916.38. Subramanian, D., Kraut, E., Staubus, A., Young, D. C. & Muller, M. T. (1995)

Cancer Res. 55, 2097–2103.39. Desai, S. D., Liu, L. F., Vazquez-Abad, D. & D’Arpa, P. (1997) J. Biol. Chem.

272, 24159–24164.40. Sabourin, M. & Osheroff, N. (2000) Nucleic Acids Res. 28, 1947–1954.41. Cline, S. D., Jones, W. R., Stone, M. P. & Osheroff, N. (1999) Biochemistry 38,

15500–15507.42. Okada, Y., Ito, Y., Kikuchi, A., Nimura, Y., Yoshida, S. & Suzuki, M. (2000)

J. Biol. Chem. 275, 24630–24638.43. Slaga, T. J., Klein-Szanto, A. J., Fischer, S. M., Weeks, C. E., Nelson, K. &

Major, S. (1980) Proc. Natl. Acad. Sci. USA 77, 2251–2254.44. Lahiri-Chatterjee, M., Katiyar, S. K., Mohan, R. R. & Agarwal, R. (1999)

Cancer Res. 59, 622–632.45. Kinzel, V., Furstenberger, G., Loehrke, H. & Marks, F. (1986) Carcinogenesis

7, 779–782.46. Dandekar, S., Sukumar, S., Zarbl, H., Young, L. J. & Cardiff, R. D. (1986) Mol.

Cell. Biol. 6, 4104–4108.47. Marchetti, F., Bishop, J. B., Lowe, X., Generoso, W. M., Hozier, J. & Wyrobek,

A. J. (2001) Proc. Natl. Acad. Sci. USA 98, 3952–3957.

5210 � www.pnas.org�cgi�doi�10.1073�pnas.0935978100 Xiao et al.