Single Chemical Modifications of the C-1027 Enediyne Core, A

Home , Calicheamicin, Enediyne, Esperamicin, Neocarzinostatin

Research Article

Single Chemical Modifications of the C-1027 Enediyne Core, a Radiomimetic Antitumor Drug, Affect Both Drug Potency and the Role of Ataxia-Telangiectasia Mutated in Cellular Responses to DNA Double-Strand Breaks Daniel R. Kennedy,1 Loretta S. Gawron,1 Jianhua Ju,2 Wen Liu,2 Ben Shen,2,3,4 and Terry A. Beerman1

1Department of Pharmacology and Therapeutics, Roswell Park Cancer Institute, Buffalo, New York and 2Division of Pharmaceutical Sciences, 3University of Wisconsin National Cooperative Drug Discovery Group, and 4Department of Chemistry, University of Wisconsin-Madison, Madison, Wisconsin

Abstract mediates activation of subsequent cell cycle progression inhibitors, resulting in a G2-M checkpoint (5). The ATM kinase cascades The radiomimetic enediyne C-1027 induces almost exclusively 15 DNA double-strand breaks (DSB) and is extremely cytotoxic. also induce phosphorylation of p53 at Ser , which is involved Unique among radiomimetics, ataxia-telangiectasia mutated in both the G1-S and G2-M checkpoints (3). Phosphorylation of 139 g (ATM) is dispensable for cellular responses to C-1027-induced the histone variant H2AX at Ser (commonly known as H2AX), which recruits DNA DSB repair proteins to the sites of the breaks, DNA damage. This study explores the biological activity of three recently bioengineered C-1027 analogues: 700-desmethyl- is also activated via ATM (6). Cells without ATM function C-1027 (desmethyl), 20¶-deschloro-C-1027 (deschloro), and 22¶- cannot efficiently activate checkpoint responses to IR-induced deshydroxy-C-1027 (deshydroxy). Each compound maintains DNA strand breaks and are hypersensitive to IR treatment (7–9). the characteristic ability of radiomimetics to cleave DNA in Moreover, tumor cells are generally more susceptible than cell-free systems, varying in activity from 2-fold (deschloro) to normal cells to IR treatment in part because they frequently 55-fold (desmethyl) less than C-1027. The induction of cellular are defective in their DNA repair and cell cycle checkpoint func- DNA breaks based on pulsed field gel electrophoresis, comet tions (10). analysis, and ;H2AX activation was in the same rank order as In addition to IR, DNA DSBs can be induced by radiomimetic cell-free DNA break induction, although the amount of breaks enediynes, which generate free radicals that abstract hydrogen atoms from the backbone of DNA while undergoing a Bergman induced by desmethyl is greatly reduced compared with the other analogues. Despite the disparity in inducing DNA DSBs, cycloaromatization reaction (11). There are two major groups of enediynes: the protein chromophore family, which includes neo- all of the analogues produced G2-M cell cycle arrest and activated DNA DSBdamage response proteins, such as p53- carzinostatin and C-1027, where the chromophore is surrounded Ser15 and Chk2-Thr68, at concentrations in concordance with by a hydrophobic protein, and the chromophore enediynes, such as calicheamicin, which lack a protein conjugate (12). Unlike their ability to inhibit cell growth. Interestingly, of the three f analogues, only the desmethyl-induced DNA damage response IR, which induces 100 single-strand breaks (SSB) for every was similar to C-1027, as it did not cause hypersensitive cell DSB, enediynes induce lower ratios of SSBs to DSBs, about 5:1 growth inhibition in the absence of ATM nor require the for neocarzinostatin and 2:1 of DSBs for C-1027 (13–15). In kinase to phosphorylate p53 or Chk2. These findings show that addition, unlike IR, which induces random DNA breaks, simple modifications of the chromophore of C-1027 can result enediynes induce preferential DNA damage at distinct sequences in varied induction of, and cellular response to, DNA DSBs. that differ for each drug (11). Despite the diversity of enediynes, a common characteristic is the rapid induction of DNA breaks [Cancer Res 2007;67(2):773–81] under cell-free and cellular conditions (16). The induction of cellular DNA breaks correlates well with cytotoxicity, and cells Introduction usually respond to enediyne-induced damage as they would to Induction of DNA double-strand breaks (DSB) by ionizing IR-induced damage (17–19). For example, treatment of cells with radiation (IR) is often used as a treatment option in cancer enediynes typically results in ATM-dependent activation of DNA therapy. In response to DNA DSBs, cells activate DNA damage damage response proteins accompanied by G2-M cell cycle response pathways to impede progression of the cell cycle, arrest (20–23). allowing cells additional time to either repair the damaged DNA Recent studies have shown that, unlike IR and other radio- or activate apoptotic pathways (1–3). Following IR treatment, the mimetics, cells treated with the enediyne C-1027 can robustly ataxia-telangiectasia mutated (ATM) kinase is activated, thereby activate damage responses in the absence of ATM (24, 25). The initiating kinase cascades (4). For example, ATM activates the mechanism responsible for the divergence between the damage 68 transducer kinase Chk2, which, when phosphorylated at Thr , responses of C-1027-treated cells and those treated with IR or another radiomimetic is currently unknown. Whether the unique- ness of C-1027 extends to analogues has not been tested because the synthesis of protein chromophore enediynes by traditional Requests for reprints: Terry A. Beerman, Department of Pharmacology and chemistry is limited by complexity of the chromophore as well as Therapeutics, Roswell Park Cancer Institute, Elm and Carlton Streets, Buffalo, NY 14263. an inability to recreate the apoprotein. Phone: 716-845-3443; Fax: 716-845-1575; E-mail: [email protected]. I2007 American Association for Cancer Research. Recently, a breakthrough in C-1027 biosynthesis and engineer- doi:10.1158/0008-5472.CAN-06-2893 ing has allowed the generation of novel C-1027 analogues by www.aacrjournals.org 773 Cancer Res 2007; 67: (2). January 15, 2007 Cancer Research

Figure 1. A, structures of the C-1027 chromophore and the locations (enclosed rectangles) of the substitutions on the chromophore that produce 700-desmethyl-C-1027 on the benzoxazolinate moiety and 20¶-deschloro-C-1027 and 22¶-deshydroxy-C-1027 on the h-amino acid moiety, respectively. B, C-1027 analogues cleave cell-free plasmid DNA based on forms conversion analysis. Plasmid DNA was treated with either a standard enediyne [neocarzinostatin (NCS) or C-1027] or one of the analogues for 30 min at 37jC. The DNA was then electrophoresed on a 1% agarose gel and visualized by staining with ethidium bromide to assess conversion of the forms of the plasmid DNA from superhelical form I to circular nicked form II or linear form III DNA. C, each enediyne induces conversion of form I to II or III in a concentration-dependent manner. The percentages of the three forms relative to the total DNA are plotted as a function of drug concentration for neocarzinostatin, C-1027, and the three analogues. The DNA was visualized by staining with ethidium bromide and quantitated as described in Materials and Methods. rational manipulations of the C-1027 biosynthetic machinery in zoxazolinate moiety (Fig. 1A).5 Little is known about the biological Streptomyces globisporus (26). The C-1027 chromophore is derived activity of the analogues, except that they are active against biosynthetically from four basic biochemical units: a benzoxazo- Micrococcus luteus (26).5 linate, a deoxyamino hexose, a h-amino acid, and an enediyne In this study, we have systematically evaluated the three novel core (26). Systematical introduction of permutations into the analogues in comparison with the parent C-1027 for their ability genes that encode the biosynthesis of these units resulted in the to induce DNA strand breaks into cell-free and cellular DNA, isolation of recombinant strains that produce various C-1027 verifying that these compounds maintain this quintessential pro- analogues. For example, inactivation of the sgcC or the sgcC3 perties of enediynes (11). Analogue-induced cell growth inhibition gene, which encode a hydroxylase and halogenase, respectively, and interference with cell cycle progression were also determined yielded the mutant strains that produce the deshydroxy and along with the ability to activate the DNA damage response proteins deschloro analogues, generating structural variations at the gH2AX, Chk2-Thr68, and p53-Ser15. Finally, we establish whether h-amino acid moiety of the C-1027 chromophore (Fig. 1A; DNA damage responses are ATM dependent or independent. refs. 26, 27). Meanwhile, inactivation of sgcD4, which encodes a methyltransferase, afforded a mutant strain that produces the desmethyl analogue, exemplifying structural alteration at the ben- 5 Unpublished data.

Cancer Res 2007; 67: (2). January 15, 2007 774 www.aacrjournals.org Responses to C-1027 Analogue-Induced DNA Damage

Materials and Methods Chk2-Thr68 (Cell Signaling Technology, Beverly, MA), as well as anti-h-actin (Sigma, St. Louis, MO). Protein bands were visualized by enhanced Chemicals. Fermentation, production, isolation, and purification of chemiluminescence and quantitated using a Personal Densitometer SI C-1027, desmethyl, deschloro, and deshydroxy were carried out as described (Amersham Biosciences, Piscataway, NJ) and ImageQuant software. previously (26, 27). The yields for C-1027 from the S. globisporus wild-type Immunofluorescent staining. After drug treatment, cells were per- strain, desmethyl analogue from the S. globisporus SB1009 strain (i.e., meabilized and washed with 0.5% Triton X-100 in PBS followed by fixation D 5 sgcD4 mutant), deschloro analogue from the S. globisporus SB1008 strain with 4% formaldehyde in PBS. The extraction-resistant gH2AX was stained D (i.e., sgcC3 mutant; ref. 27), and deshydroxy analogue from the S. with anti-gH2AX primary antibody and fluorescein-conjugated secondary D globisporus SB1006 strain (i.e., sgcC mutant; ref. 26) were estimated to be antibodies. Hoechst 33342(2 Amol/L) was used to stain DNA. gH2AX foci 60, 20, 60, and 60 mg/L, respectively. Neocarzinostatin was a gift from were examined using an Olympus BX40 microscope with a Regita Exi digital Bristol-Myers Squibb Co. (Syracuse, NY). All compounds were stored at camera and Image-Pro Plus software. A j 200 mol/L in double-distilled water at 20 C. Flow cytometry. Exponentially growing cells were drug treated contin- Cells. Isogenic ATM-null (AT169A) and ATM-restored (YZ510B) human uously for 24 h, rinsed with PBS, and harvested. The cells were fixed in j fibroblast cell lines (gift of Drs. P. Lu and Y. Shiloh) were grown at 37 C and 70% ethanol, stained with propidium iodide solution (0.1% propidium iodide, 5% CO2 in DMEM supplemented with 10% fetal bovine serum plus 0.1% sodium citrate, 0.4% NP40, and 0.2mg/mL RNase A), and analyzed by A 50 g/mL hygromycin to maintain selective pressure on the transfected flow cytometry on a FACScan instrument (Becton Dickinson, San Jose, CA) constructs (28). All experiments were done with the YZ510B cell line unless using WinList software (Verity Software House, Inc., Topsham, ME). otherwise specified. Forms conversion. pBR322 plasmid DNA (100 ng; New England Biolabs, Ipswich, MA) was incubated with C-1027, desmethyl, deschloro, Results deshydroxy, or neocarzinostatin in 10 mmol/L Tris (pH 7.5) for 30 min at 37jC. DTT (0.5 mmol/L) was included only in the neocarzinostatin C-1027 analogues cleave cell-free DNA. An intrinsic property reaction buffer as an activating agent, as we found it was not necessary of enediynes is their ability to cleave DNA in a cell-free system. The for C-1027 (29) or for these analogues (data not shown). Samples were C-1027 analogues (Fig. 1A) were compared with the parent electrophoresed on a 1% agarose gel and then the gels were stained with compound and another protein chromophore enediyne, neo- ethidium bromide, photographed using the Gel Doc XR (Bio-Rad Corp., carzinostatin, for their ability to cleave an isolated supercoiled Hercules, CA), and analyzed by ImageQuant software (Molecular plasmid DNA. In this assay, supercoiled DNA (form I) can be Dynamics, Piscataway, NJ). converted to a nicked circular form (form II) by a SSB, whereas a 5 Growth inhibition assay. Cells were seeded at 2.5 10 per well in six- DSB results in a full-length linear duplex (form III) as shown by well dishes and drug treated 24 h later. Following a 3-day incubation, dishes neocarzinostatin and C-1027 (Fig. 1B). The rapid increase in the were washed and cells were counted by Coulter counter (Beckman Coulter, appearance of form III DNA after treatment with C-1027, but not Inc., Fullerton, CA). Cell growth inhibition was calculated by comparing the number of treated to nontreated control cells. neocarzinostatin, indicates that a high percentage of the total C- Pulsed field gel electrophoresis. Cells were treated with drugs for 2h, 1027-induced DNA breaks were DSBs (Fig. 1B and C), consistent washed, and resuspended in 0.5% low melting temperature agarose, poured with the observation that C-1027 induces primarily DSBs whereas into a plug former, and kept at 4jC until solidified. Plugs containing neocarzinostatin induces a mixture of SSBs and DSBs (29). Similar approximately 3 105 cells were incubated with 1 mg/mL proteinase K in to the parent compound, the three analogues also rapidly convert lysis buffer [10 mmol/L Tris, 2% sodium sarcosinate, 0.5 mol/L EDTA (pH the plasmid into form III DNA, indicating that they also induce a 8.0)] for 24 h at 50jC. The plugs were loaded into wells of a 0.8% agarose gel. high percentage of DSBs (Fig. 1B and C).6 With regard to activity, Pulsed field gel electrophoresis was conducted in 0.5 Tris borate/EDTA for each of the analogues was able to induce cell-free DNA breaks, with 18 h at 200 V using a ChefDR apparatus (Bio-Rad), and damage was a potency of approximately 55X, 2X, and 20X for desmethyl, quantitated using ImageQuant software. deschloro, and deshydroxy, respectively, where X is the concentra- Comet analysis. Following a 30-min drug treatment, cells were tion of C-1027 required to reduce the amount of form I DNA by resuspended in PBS at a concentration of 1 107/mL and then analyzed by Comet as described elsewhere with minor modifications (30). Briefly, 50% (Fig. 1C). cells were diluted into low melting temperature agarose and placed on an At very high drug concentrations that result in extensive DNA agarose-covered slide for 2h at 4 jC. The slide was then immersed in cleavage, deschloro and deshydroxy induce mobility retardation in alkaline lysis buffer overnight at 4jC and then placed in electrophoresis all three forms of the plasmid DNA (Fig. 1B). The levels at which buffer for 20 min before electrophoresis at 27 V for 25 min at 4jC. The slides the mobility changes first appear are 600 and 2,000 nmol/L for were then placed in 0.4 mol/L Tris (pH 7.5) thrice for 5 min followed by deschloro and deshydroxy, respectively (Fig. 1B), and the mobility immersion in 100% methanol and then ethanol. The slides were dried and retardation increased in a concentration-dependent manner for stained with 15 Ag/mL ethidium bromide, and DNA damage was assessed those analogues (data not shown). Furthermore, the desmethyl as described (30). compound also induced mobility retardation in a concentration- j Immunoblotting. Cells were incubated with drugs at 37 Cfor1h, dependent manner, starting f14,000 nmol/L (data not shown). harvested, and washed in PBS. Total cellular extracts were prepared by This binding may be related to a reversible DNA interaction with incubating cells on ice in lysis buffer [50 mmol/L HEPES (pH 7.6), 150 mmol/L NaCl, 5 mmol/L EDTA, 5 mmol/L EGTA, 0.5% NP40, 0.5% the postactivated chromophore that has previously been described sodium deoxycholate, 0.5% Triton X-100, 50 mmol/L sodium fluoride, for the parent at very high concentrations (32), but a complete 1 mmol/L sodium o-vanadate, 1 mmol/L h-glycerophosphate, 1 mmol/L phenylmethylsulfonyl fluoride, protease inhibitors] for 15 min. The cell lysates were cleared by centrifugation, and protein content was determined using the Bradford method (Bio-Rad). Equal amounts of protein were 6 We also examined the ability of the analogues to induce abasic sites, which are a electrophoresed on SDS-polyacrylamide gels and transferred to a poly- common lesion induced by enediynes, including C-1027 (14). To measure abasic sites, vinylidene difluoride membrane. The membranes were probed with primary the drug-treated DNA was incubated with 100 mmol/L putrescine for 1 h at 37jC, antibodies followed by secondary antibodies conjugated with horseradish which results in conversion of abasic sites to DNA strand breaks (31). Similar to the parent compound, putrescine treatment resulted in a 15% to 20% increase in the peroxidase. The primary antibodies used were anti-gH2AX (Upstate, amount of DSBs induced by each of the analogues, suggesting that they all induce a Charlottesville, VA), anti-phosphorylated p53-Ser15 and anti-phosphorylated similar ratio of abasic sites to direct DNA DSBs (data not shown). www.aacrjournals.org 775 Cancer Res 2007; 67: (2). January 15, 2007 Cancer Research

C-1027 analogue treatment of cells induces primarily DNA DSBs. C-1027 has a SSB/DSB ratio of approximately 2:1, so neutral pulsed field gel electrophoresis is a suitable measure of total DNA breaks (15). However, because the ratio of SSB/DSB produced by the analogues in cells is unknown, we also determined total DNA breaks (SSB and/or DSBs) based on alkaline comet analysis (Fig. 3B). In the comet assay, the nucleus can be seen as the head of the comet, and on alkaline electrophoresis, DNA that was damaged by either a SSB or DSB migrates out of the nucleus, forming a tail. Smaller, more rapidly migrating DNA fragments move further away from the nucleus than do less damaged fragments so that the amount of DNA damage is proportional to the length of the comet tail. After treatment with C-1027, deschloro, or deshydroxy, the total DNA breaks were consistent with the pulsed field analysis (DSBs alone), as the comet tails formed after an initial limited amount of damage and then the comet tails lengthen in a Figure 2. C-1027 analogues induce cell growth inhibition. After 3 d of treatment with C-1027, an analogue or neocarzinostatin at 37jC, the percentage growth concentration-dependent manner (Fig. 3). Similarly, treatment with inhibition of YZ510B cells was calculated based on cell number relative to the desmethyl analogue also was consistent with the pulsed field the control at each drugconcentration. Inset, IC50 value for each compound study, as it resulted in an initial limited amount of damage at examined and their relative potencies compared with C-1027. 50 nmol/L, which did not seem to increase even at 800 nmol/L (Fig. 3B). Thus, the disparity of cellular DNA DSBs induced by work-up of thermodynamic DNA-binding properties would be necessary to determine this conclusively. C-1027 analogues inhibit cellular growth. To determine whether the analogues are active in a cellular environment, we tested their ability to induce cell growth inhibition. Cells were treated with C-1027 analogues, and growth inhibition was measured after 3 days of continuous drug treatment. The IC50s for growth arrest were 24 pmol/L for C-1027, 1,412 pmol/L for desmethyl, 72pmol/L for deschloro, and 174 pmol/L for deshydroxy, which is approximately 60-, 3-, and 7-fold higher than C-1027, respectively (Fig. 2). The enediyne neocarzinostatin itself, a potent inhibitor of cell growth inhibition, was included in this study and showed that even the weakest C-1027 analogue retains considerable potency in cells. C-1027 analogues induce cellular DNA DSBs. Neutral pulsed field gel electrophoresis, which provides a direct measure of DNA DSBs by fractionating DNA according to molecular weight, was used to test the ability of C-1027 and the analogues to induce cellular DNA breaks. In this assay, undamaged cellular DNA is typically too large to enter the gel and is retained in the wells. However, induction of a small amount of DNA DSBs results in a band of high molecular weight (HMW) material at the top of the gel (Fig. 3A, arrow), although further induction of DSBs reduces this band to a range of smaller and therefore more rapidly migrating pieces of DNA that resolve as a smear below the HMW band. Treatment of cells with the deschloro or deshydroxy analogue for 1 h showed a DNA cleavage pattern comparable with 1 nmol/L C-1027 (Fig. 3A), although as in the cell-free DNA study higher drug concentrations, 8 and 36 times, respectively, were needed. Interestingly, although desmethyl initially resulted in limited Figure 3. A, C-1027 analogues induce cellular DNA DSBs. The ability of amounts of DNA leaving the well, the amount of DNA did not C-1027 analogues to cleave cellular DNA was measured by pulsed field gel increase significantly with drug concentration nor was DNA electrophoresis. DNA DSBs were measured after treatment of YZ510B cells for migrating below the HMW band (Fig. 3A). Thus, even at the 1hat37jC with either C-1027 or an analogue over a range of concentrations that was expected to induce similar damage profiles. Following pulsed field highest concentration (800 nmol/L), desmethyl-induced DSBs were gel electrophoresis to size fractionate the DNA, the gels were stained with far less than what was observed with 1 nmol/L C-1027 even if ethidium bromide to visualize the DNA that entered the lane versus total DNA (DNA in the lane + DNA in the well). Arrow, HMW band of DNA enteringthe normalized for cytotoxicity. To test if the low levels of DNA DSBs gel after drug treatment. B, C-1027 analogues induce a low ratio of SSBs/DSBs were a result of a slower rate of DNA cleavage by the desmethyl in cells. FollowingC-1027 or analoguetreatment under the concentrations analogue, increased treatment times of up to 4 h were examined. used above, total DNA strand breaks were assessed by alkaline comet analysis. After drugtreatment, YZ510B cells were harvested and placed on a microscope However, the proportion of damaged DNA remained similar to that slide, lysed, and electrophoresed, and after addition of ethidium bromide, of the parent (data not shown). DNA comets were detected by fluorescence microscopy.

Cancer Res 2007; 67: (2). January 15, 2007 776 www.aacrjournals.org Responses to C-1027 Analogue-Induced DNA Damage

the initial focalization of gH2AX that appeared at the lower concentration (Fig. 4B). Overall, the induction of gH2AX is consistent with the measurements of DNA DSBs made by pulsed field analysis (Figs. 3A and 4). C-1027 analogue treatment induces cell cycle checkpoints. Typically, cell cycle checkpoints are activated to impede cell cycle progression following induction of DNA damage. At low concentrations, C-1027, similar to IR and other radiomimetic treatments, induces a strong G2-M arrest 24 h after treatment (Fig. 5A; refs. 3, 21). To test if C-1027 analogues also induce G2-M arrest, cells were treated with equicytotoxic concentrations of C-1027, desmethyl, deschloro, or deshydroxy. Based on fluorescence-activated cell sorting analysis, the desmethyl and deschloro analogues induce a strong G2-M block. Although deshydroxy also blocks cells at G2- M, a somewhat greater proportion of cells resided in G1 compared with the other compounds (Fig. 5A).

Figure 4. A, C-1027 analogues induce gH2AX phosphorylation. YZ510B cells were treated with approximately equicytotoxic concentrations of either C-1027 or an analogue at 37jC for 0.5 h, and the cellular extracts were analyzed by Western blotting. Immunoblots were then probed with an antibody specific for phosphorylated gH2AX. B, C-1027 analogues induce gH2AX focalization. Cells were grown on microscope slide coverslips before treatment at the conditions described above. Cells were then washed in PBS, fixed with formaldehyde, and probed with a monoclonal antibody against gH2AX. A fluorescein-labeled secondary antibody and Hoechst 33342 were used to detect the foci and nucleus, respectively. desmethyl is not because it is inducing a higher ratio of SSBs to DSBs in a cellular environment. C-1027 analogue treatment induces cellular DNA damage responses. Similar to IR or other radiomimetics, treatment of cells with C-1027 has previously shown that H2AX is phosphorylated to gH2AX and recruited to sites of DSBs (33). To test whether treatment of cells with the C-1027 analogues also induces gH2AX, cells were treated with approximately equicytotoxic concentrations of C-1027 or an analogue for 30 min and analyzed by Western blotting. At low levels of C-1027 treatment (0.1 nmol/L), gH2AX is detected and increases in a concentration- dependent manner (Fig. 4A). At low, approximately equicytotoxic levels of deschloro or deshydroxy, gH2AX is activated to a similar extent as the parent compound and the amount of activation increased in a concentration-dependent manner (Fig. 4A). Treatment with a low equicytotoxic level of desmethyl showed an initial activation of gH2AX, but the level did not increase with concentration (Fig. 4A). Because gH2AX localization is necessary for recruitment of DNA repair proteins, such as 53BP1, NBS1, and BRCA1, to DNA DSBs, focalization of gH2AX after analogue treatment was also examined (34). To test for gH2AX focalization, cells were treated for 30 min with the lowest and highest concentration of each drug used in Fig. 4A and analyzed by microscopic immunofluorescence. After treatment with low, approximately equicytotoxic concentrations of C-1027, gH2AX foci were easily detectable, although the high level of C-1027 showed a dramatic increase in foci intensity (Fig. 4B). Figure 5. A, C-1027 analogues induce a G2-M cell cycle arrest and activate Similarly, using the low, approximately equicytotoxic concentration DNA damage response proteins. YZ510B cells were treated with equicytotoxic of the three C-1027 analogues, each induced a modest level of concentrations of either C-1027 or an analogue for 24 h at 37jC, fixed in ethanol, g stained with propidium iodide, and analyzed by flow cytometry. Cell cycle H2AX foci similar to C-1027. When the concentrations of the compartments were determined usingWinList software. B, C-1027 analogues analogues were increased to correspond to higher levels of C-1027, activate the damage response proteins p53 and Chk2 in proportion to their the deschloro and deshydroxy analogues also showed a dramatic cytotoxicity. YZ510B cells were treated with either C-1027 or an analogue for1hat37jC, and the cellular extracts were analyzed by Western blotting. increase in the intensity of gH2AX foci, whereas the intensity of Immunoblots were then probed with an antibody specific for phosphorylated the desmethyl-induced gH2AX foci did not increase notably from p53-Ser15 or Chk2-Thr68. www.aacrjournals.org 777 Cancer Res 2007; 67: (2). January 15, 2007 Cancer Research

a concentration-dependent increase of p53-Ser15 phosphorylation (Fig. 5B). These results show that all three C-1027 analogues induce cell cycle arrest at G2-M and robustly activate cell cycle checkpoint proteins (Fig. 5). DNA DSBdamage responses to C-1027 analogue treatment vary in their ATM dependence. Cells typically activate DNA DSB damage response proteins through an ATM-dependent kinase cascade after treatment with IR and radiomimetics, such as calicheamicin and neocarzinostatin (22, 23, 35). However, we have previously shown that, in response to C-1027-induced damage, ATM-null cells can phosphorylate both Chk2-Thr68 and p53-Ser15 as robustly as ATM wild-type cells (24, 25), a result that is recapitulated in Fig. 6A. Similarly, ATM-null cells treated with desmethyl phosphorylate Chk2-Thr68 and p53-Ser15 as readily as ATM wild-type cells (Fig. 6A). In contrast, treatment of ATM-null cells with deschloro resulted in a notable decrease in the activation of both Chk2and p53 (Fig. 6 A) and a very large decrease in the phosphorylation of these proteins after deshydroxy treatment (Fig. 6A). The relative ratio of Chk2and p53 phosphorylation between ATM wild-type and ATM-null cells was determined by dividing the phosphorylation level of each protein in the wild-type by the level of the ATM null in response to each drug. The ratio of C-1027 of f1 is indicative for ATM independence with regard to phosphorylation of both Chk2and p53 (Fig. 6 B). Similarly, desmethyl treatment also activates these DNA DSB damage response proteins independently of ATM as it also resulted in a ratio of f1 for both proteins (Fig. 6B). In contrast, the relative ratio of neocarzinostatin (f2) was determined to use as a reference point for a typical ATM-dependent damage response (Fig. 6B). Figure 6. C-1027 and desmethyl induce ATM-independent DNA damage Deschloro treatment also resulted in a ratio of f2, establishing responses, whereas deschloro and deshydroxy do not. A, AT169A (ATM null) that, like neocarzinostatin, this analogue induces ATM-dependent and YZ510B (ATM wild-type) cells were treated with equicytotoxic concentrations of either C-1027 or an analogue for 1 h at 37jC, and the cellular damage responses (Fig. 6B). Furthermore, in response to treat- extracts were analyzed by Western blotting. Immunoblots were then probed ment with deshydroxy, a ratio of 4 was determined for both 68 15 with an antibody specific for phosphorylated Chk2-Thr or p53-Ser . B, the proteins, suggesting that this agent induces more extreme ATM- protein band in each lane was normalized to a loadingcontrol ( h-actin) and then quantitated usingImageQuant software. The fold difference between these dependent damage responses in comparison with other radio- values was used to determine the ratio of Chk2-Thr68 and p53-Ser15 activation mimetics (Fig. 6B). between ATM wild-type and ATM-null cells. The values of at least three independent experiments were analyzed for each drug. C, cell growth inhibition Because cells lacking ATM are unable to completely induce of the ATM wild-type and ATM-null cells was similar for C-1027 and desmethyl DNA damage responses to DNA DSBs induced by IR and but not for deschloro, deshydroxy, or neocarzinostatin. After 3 d of drug radiomimetics (except C-1027), they are more susceptible to treatment, the IC was calculated based on concentration of drugrequired to 50 growth inhibition or cell death (4). We therefore tested whether reduce cell growth by 50%. The fold differences in the IC50s were used to determine the ratio of growth inhibition between ATM wild-type and ATM-null hypersensitivity to growth inhibition would vary according to the cells. ATM dependence of checkpoint activation. The IC50s for growth inhibition of ATM wild-type and ATM-null cells were measured after treatment with C-1027 analogues. The IC50 for the ATM Activation of the G2-M checkpoint in response to DNA DSBs wild-type was divided by the IC50 for ATM-null cells to determine involves the activation of damage response proteins, including a hypersensitivity ratio. The ratio of C-1027 of f1 is indicative phosphorylation of Chk2at Thr 68 (2, 3). At low equicytotoxic for ATM independence with regard to cell growth inhibition concentrations of drug treatment, modest levels of Chk2-Thr68 (Fig. 6C). Similar to C-1027, the desmethyl analogue also has a phosphorylation were observed for all the compounds (Fig. 5B). hypersensitivity ratio of f1(Fig.6C). Ratios >1, such as that The amount of Chk2-Thr68 phosphorylation increased in a induced by neocarzinostatin (f2.4), are typical of DNA damage concentration-dependent manner even after treatment with the responses that are dependent on ATM. Treatment with either desmethyl analogue, which did not induce gH2AX in a concentra- deschloro or deshydroxy resulted in cellular hypersensitivity with tion-dependent manner (Figs. 4 and 5B). regard to cell growth inhibition with ratios of 2.2 and 3.6, Phosphorylation of p53 at Ser15, another downstream DNA respectively (Fig. 6C). Thus, the growth inhibition data are damage response protein, is involved in both G1-S and G2-M arrest consistent with the ATM-dependent activation of DNA damage (3). Phosphorylation of p53-Ser15 was consistent with that of Chk2- response proteins Chk2and p53. Therefore, both deschloro and Thr68 after treatment with C-1027 or desmethyl, deschloro, and deshydroxy act like IR and other radiomimetics in that they induce deshydroxy, as low equicytotoxic concentrations result in modest a classic ATM-dependent damage response, whereas desmethyl, like levels of p53 activation (Fig. 5B). In addition, similar to Chk2-Thr68, C-1027, induces an ATM-independent damage response to DNA increasing drug concentrations of each of the analogues resulted in DSBs (Fig. 6).

Cancer Res 2007; 67: (2). January 15, 2007 778 www.aacrjournals.org Responses to C-1027 Analogue-Induced DNA Damage

Discussion showed a corresponding increase in the amount of gH2AX phos- This study examined the DNA cleavage activity of, and cellular phorylation and focalization; however, desmethyl did not (Figs. 3A DNA damage responses to, three biosynthetic analogues of the and 4). DNA DSBs induced by IR and radiomimetics increase enediyne C-1027, a potent inducer of DNA DSBs (14). Although the linearly with drug concentration, so the finding that desmethyl- biological activity of truncated versions of the enediynes calichea- induced DNA DSBs do not increase with drug concentration is atypical (29). micin and esperamicin has previously been characterized (36, 37), Analysis of cellular DNA breaks by alkaline comet, which detects this is the first study to examine single substitutions on an total DNA strand breaks (DSBs + SSBs), was found to be consistent enediyne chromophore. All of the analogues maintain the ability to with the neutral pulsed field gel analysis and gH2AX data (Fig. 3B). cleave cell-free DNA as measured by forms conversion analysis of Therefore, the drugs all induce primarily DSBs under both cell-free drug-treated superhelical plasmid DNA (Fig. 1B). C-1027 was the and cellular conditions (Figs. 1C and 3). This is consistent with most potent followed by deschloro, deshydroxy, and, finally, another enediyne, calicheamicin, which also produces primarily desmethyl, with cleavage activity varying by up to 55-fold. DSBs under cellular conditions, mimicking what is observed in cell- Furthermore, the DNA cleavage activities compare favorably with free systems (17). Thus, the unusual mechanism of action of the the truncated calicheamicin and esperamicin analogues, which desmethyl compound is likely not explained by it producing a were 10- to 1,000-fold weaker than their respective parent greater amount of SSBs in cells. compounds (36, 37). The cytotoxicity of the analogues was C-1027>deschloro>deshy- Substitutions on the C-1027 chromophore may alter the droxy>desmethyl, which is in the same rank order as their ability to location of the biradical within the DNA minor groove, which induce DNA DSBs (Fig. 2). This is unsurprising, as, regardless of previous studies have shown can change not only the potency of differences in the DNA damage induced by treatment with IR or a a compound but also the ratio of SSBs/DSBs induced (32, 38). The radiomimetic, such as C-1027, calicheamicin, neocarzinostatin, and proportion of SSBs to DSBs induced by each enediyne are related macromomycin/auromomycin, it is the induction of DSBs that best to the efficiency of the two radicals in abstracting hydrogen from correlates with the induction of cytotoxicity (17–19, 29). However, DNA. For example, the biradicals of enediynes that induce a high the relationship between desmethyl-induced DSBs and cytotoxicity proportion of DSBs, such as C-1027 and calicheamicin, are is more complex, as, even when normalized for cytotoxicity, positioned in a conformation favorable for each radical to desmethyl failed to induce a similar level of DNA DSBs in com- abstract hydrogen from the DNA backbone (39). In contrast, parison with the other analogues (Fig. 3A). Moreover, treatment of neocarzinostatin induces primarily SSBs because one radical is cells with increasing amounts of desmethyl resulted in an increase positioned such that abstraction of hydrogen from other sources, in cellular cytotoxicity (Fig. 2), yet the amount of DSBs induced such as water, is more favorable than from the DNA sugar did not increase accordingly (Fig. 3A). Nevertheless, the desmethyl (40, 41). Similar to the parent compound, all three analogues compound is still extremely potent at inducing growth inhibition, maintain a low ratio of cell-free DNA SSBs/DSBs because form III as its IC50 was an order of magnitude lower than neocarzinostatin DNA is readily observable in drug-treated lanes, whereas an (Fig. 2). Furthermore, desmethyl, like the other analogues, induces abundant amount of form I DNA still remains, implying that the 7 aG2-M cell cycle block at low equicytotoxic concentrations chromophore substitutions induce little, if any, changes in the (Fig. 5A) despite its limited ability to induce cellular DNA DSBs. positioning of the biradical (Fig. 1C). Therefore, it is possible that the cellular cytotoxicity and cell cycle Neutral pulsed field gel analysis of DNA isolated from drug- block induced by desmethyl treatment is not limited to DSB treated cells provides direct evidence that all of the analogues were induction but could involve other mechanisms. For example, also capable of inducing DNA DSBs in cells. Based on the some enediynes can also cleave RNA, albeit at concentrations well appearance of the HMW band of DNA (Fig. 3A, arrow), which above those needed to cleave DNA (42). Thus, perhaps certain appears when sufficient DNA DSBs are induced to allow the DNA modifications on the benzoxazolinate moiety (desmethyl ana- to enter the gel, the rank order of activity was consistent with the logue) of the C-1027 chromophore increase its affinity for non- cell-free experiments. Similar to other radiomimetics, increasing DNA targets. levels of deschloro and deshydroxy treatment continued to reduce Consistent with the activation of a G2-M cell cycle block, it was DNA size, resulting in a range of lower molecular weight material not surprising that each analogue induces Chk2-Thr68 and p53- that appears as a smear below the HMW band within each lane Ser15. However, these proteins are also thought to be increasing by (Fig. 3A; ref. 29). Because the cellular DNA strand scission activity phosphorylated as the level of DNA DSBs increases (23, 24). of these two analogues reflects their strand scission activity under Interestingly, although DNA DSBs do not increase in a concentra- cell-free conditions, the uptake and stability of these compounds tion-dependent manner after desmethyl treatment, Chk2and p53 are likely similar in cells. activation was concentration dependent (Figs. 3A and 5B). However, unlike C-1027 and the other analogues, increasing Certainly, phosphorylation of p53 can occur in response to a concentrations of desmethyl did not result in further degradation variety of cellular stresses besides strand breaks, such as that of the HMW DNA band (Fig. 3A). In fact, even at longer treatment caused by UV irradiation or hypoxia (43). However, phosphoryla- times, or levels >100-fold higher (2,000 nmol/L) than those tion of Chk2, like gH2AX, is considered to be more specific to the expected to result in sufficient DSBs to reduce the HMW band, cellular response to DNA DSBs (23, 44). Yet, unlike gH2AX, Chk2 we found no evidence of further DNA damage (data not shown). Consistent with the pulsed field measurements, all of the ana- 7 logues induced detectable gH2AX phosphorylation and focaliza- Distinctly from C-1027 and from the other analogues, a notable G1 population tion at concentrations even lower than those at which the HMW remained after deshydroxy treatment (Fig. 5A). Perhaps deshydroxy induces both G1-S band was first detected (Figs. 3A and 4). Similarly, increasing and G2-M arrest at approximately equal concentration of treatment as opposed to the other analogues, which require 3- to 5-fold higher concentrations to activate the G1-S concentrations of C-1027, as well as deschloro and deshydroxy, checkpoint (data not shown; ref. 22). www.aacrjournals.org 779 Cancer Res 2007; 67: (2). January 15, 2007 Cancer Research was activated in a concentration-dependent manner after in response to C-1027 treatment is consistent with the idea that desmethyl treatment (Figs. 4A and 5B). This is consistent with this radiomimetic induces cellular DNA ISCs in addition to DNA recent studies that suggest that Chk2can be activated in response DSBs (25). to cellular cytotoxicity, not just by DNA DSBs (8, 45, 46). Moreover, Furthermore, if desmethyl treatment results in cellular DNA for desmethyl, the observation that Chk2and p53 are activated and ISCs, it would also help explain the ability of the analogue to cell growth inhibition is induced in a concentration-dependent induce DNA damage response protein activation and cell growth manner without a correlating increasing of DNA DSBs is further inhibition in a concentration-dependent manner (Figs. 2and 5 B) evidence that the DNA damage induced by desmethyl is likely not without inducing a correlative increase in the amount of DNA limited to induction of DSBs. DSBs (Figs. 3A and 4). Whether DNA ISCs are induced in cells The role of the phosphatidylinositol 3-kinase–like protein by C-1027 and possibly by the other analogues is under study, as kinase (PIKK) family member ATM in activating other DNA well as the relative importance of cellular DNA ISCs versus DSBs damage response proteins and avoiding cellular hypersensitivity with regard to their reliance on ATM and ATR to activate DNA to IR- or drug-induced DNA DSBs is well established (8, 24, 47). damage responses. Selective modifications of the C-1027 chro- The notable exception is that, in response to C-1027-induced DNA mophore will help define the structure-activity relationship of damage, cells activate Chk2-Thr68 and p53-Ser15 similarly the ATM-dependent and ATM-independent DNA damage res- regardless of ATM status, and cells lacking the kinase are not ponse pathways. hypersensitive to C-1027 treatment with regard to cell growth Overall, the current work clearly showed that structural inhibition or cell death (24, 25). Surprisingly, deschloro and permutations on the C-1027 chromophore have a profound effect deshydroxy, which generate comparable amounts of DNA DSBs on the mode of action of the resultant C-1027 chromoprotein with C-1027 when normalized for cytotoxicity, induce ATM- complex. It further suggests that the location, not just the presence, dependent damage responses, behaving similar to IR and other of a single modification to the C-1027 chromophore is important in radiomimetics, such as neocarzinostatin (Fig. 6). In contrast, determining not only the ability of the drug to induce DNA DSBs desmethyl induces ATM-independent DNA damage responses, yet but also the ATM dependency of the cellular DNA damage it induced very low levels of DNA DSBs (Fig. 6). Thus, it is not yet response. Experiments are in progress to produce additional known how both C-1027, which is the most potent DNA DSB C-1027 analogues to further decipher the mechanism(s) of action of inducer of the C-1027 family, and desmethyl, which is the the C-1027 family of antitumor antibiotics and their ability to weakest, activate ATM-independent damage responses, whereas invoke unique cellular DNA damage responses. the other analogues do not. Possibly, additional types of DNA damage could trigger activation of DNA damage response Acknowledgments proteins in the absence of ATM. For example, under anaerobic conditions in cell-free systems, C-1027 has been shown to induce Received 8/4/2006; revised 9/25/2006; accepted 11/13/2006. Grant support: National Cancer Institute grants CA106312and CA16056 DNA interstrand cross-links (ISC), whereas other enediynes (T.A. Beerman), NIH grants CA078747 and CA113297 (B. Shen), and NIH Training (neocarzinostatin and calicheamicin) do not (41). The ability of grant CA09072-30 (D.R. Kennedy). C-1027 to induce cellular DNA ISCs has not yet been studied. 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 However, cells use another PIKK family member, the ATM- and with 18 U.S.C. Section 1734 solely to indicate this fact. Rad3-related (ATR) kinase, to induce damage responses to other We thank Dr. Y. Li (Institute of Medicinal Biotechnology, Chinese Academy of Medical Sciences, Beijing, China) for the wild-type S. globisporus strain; Drs. Adam agents that induce ISCs (48). Thus, the observation that cells can Karpf, Athena Lin, and Mary McHugh for critical reading of the manuscript; and Renee use ATR to activate Chk2and p53 while avoiding hypersensitivity Triplet for her technical assistance.

References in response to different forms of DNA damage. J Biol extracellular simian virus 40 DNA with the ene- Chem 2005;280:1186–92. diyne anticancer drug C-1027. Biophys Chem 1997; 1. Abraham RT. PI 3-kinase related kinases: ‘big’ players 9. Ziv Y, Jaspers NG, Etkin S, et al. Cellular and molecular 63:201–9. in stress-induced signaling pathways. DNA Repair characteristics of an immortalized ataxia-telangiectasia 16. Nicolaou KC, Smith AL, Yue EW. Chemistry and (Amst) 2004;3:883–7. (group AB) cell line. Cancer Res 1989;49:2495–501. biology of natural and designed enediynes. Proc Natl 2. Shiloh Y. ATM and ATR: networking cellular responses 10. Choudhury A, Cuddihy A, Bristow RG. Radiation and Acad Sci U S A 1993;90:5881–8. to DNA damage. Curr Opin Genet Dev 2001;11:71–7. new molecular agents part I: targeting ATM-ATR 17. Elmroth K, Nygren J, Martensson S, Ismail IH, 3. Iliakis G, Wang Y, Guan J, Wang H. DNA damage checkpoints, DNA repair, and the proteasome. Semin Hammarsten O. Cleavage of cellular DNA by calichea- checkpoint control in cells exposed to ionizing radia- Radiat Oncol 2006;16:51–8. micin g1. DNA Repair (Amst) 2003;2:363–74. tion. Oncogene 2003;22:5834–47. 11. Povirk LF. DNA damage and mutagenesis by 18. Beerman TA, Goldberg IH. The relationship between 4. Shiloh Y. ATM and related protein kinases: safeguard- radiomimetic DNA-cleaving agents: bleomycin, neo- DNA strand-scission and DNA synthesis inhibition in ing genome integrity. Nat Rev Cancer 2003;3:155–68. carzinostatin, and other enediynes. Mutat Res 1996; HeLa cells treated with neocarzinostatin. Biochim 5. Li L, Zou L. Sensing, signaling, and responding to DNA 355:71–89. Biophys Acta 1977;475:281–93. damage: organization of the checkpoint pathways in 12. Smith AL, Nicolaou KC. The enediyne antibiotics. 19. Beerman TA. Cellular DNA damage by the antitumor mammalian cells. J Cell Biochem 2005;94:298–306. J Med Chem 1996;39:2103–17. protein macromomycin and its relationship to cell 6. Fernandez-Capetillo OH-TC, Celeste A, Ward I, et al. 13. Sutherland BM, Bennett PV, Sidorkina O, Laval J. growth inhibition. Biochem Biophys Res Commun 1978; DNA damage-induced G2-M checkpoint activation by Clustered damages and total lesions induced in DNA by 83:908–14. histone H2AX and 53BP1. Nat Cell Biol 2002;4:993–7. ionizing radiation: oxidized bases and strand breaks. 20. Turenne GA, Paul P, Laflair L, Price BD. Activation of 7. Barlow C, Brown KD, Deng CX, Tagle DA, Wynshaw- Biochemistry 2000;39:8026–31. p53 transcriptional activity requires ATM’s kinase Boris A. Atm selectively regulates distinct p53-depen- 14. Xu YJ, Zhen YS, Goldberg IH. C1027 chromophore, domain and multiple N-terminal serine residues of dent cell-cycle checkpoint and apoptotic pathways. Nat a potent new enediyne antitumor antibiotic, induces p53. Oncogene 2001;20:5100–10. Genet 1997;17:453–6. sequence-specific double-strand DNA cleavage. Bio- 21. McHugh MM, Gawron LS, Matsui S, Beerman TA. 8. Helt CE, Cliby WA, Keng PC, Bambara RA, O’Reilly MA. chemistry 1994;33:5947–54. The antitumor enediyne C-1027 alters cell cycle Ataxia telangiectasia mutated (ATM) and ATM and 15. Kirk CA, Goodisman J, Beerman TA, Gawron LS, progression and induces chromosomal aberrations and Rad3-related protein exhibit selective target specificities Dabrowiak JC. Kinetics of cleavage of intra- and telomere dysfunction. Cancer Res 2005;65:5344–51.

Cancer Res 2007; 67: (2). January 15, 2007 780 www.aacrjournals.org Responses to C-1027 Analogue-Induced DNA Damage

22. Uziel T, Lerenthal Y, Moyal L, Andegeko Y, Mittelman genetic toxicology testing. Environ Mol Mutagen bistranded DNA lesions by the antitumor antibiotics L, Shiloh Y. Requirement of the MRN complex for ATM 2000;35:206–21. bleomycin, neocarzinostatin, and calicheamicin. Chem activation by DNA damage. EMBO J 2003;22:5612–21. 31. Povirk LF, Houlgrave CW. Effect of apurinic/apyr- Res Toxicol 1992;5:311–32. 23. Ismail IH, Nystrom S, Nygren J, Hammarsten O. imidinic endonucleases and polyamines on DNA treated 40. Meschwitz SM, Schultz RG, Ashley GW, Goldberg IH. Activation of ataxia telangiectasia mutated by DNA with bleomycin and neocarzinostatin: specific formation Selective abstraction of 2H from C-1¶ of the C residue in strand break-inducing agents correlates closely with the and cleavage of closely opposed lesions in complemen- AGC.ICT by the radical center at C-2of activated number of DNA double strand breaks. J Biol Chem 2005; tary strands. Biochemistry 1988;27:3850–7. neocarzinostatin chromophore: structure of the drug/ 280:4649–55. 32. Xu YJ, Xi Z, Zhen YS, Goldberg IH. Mechanism of DNA complex responsible for bistranded lesion forma- 24. Dziegielewski J, Beerman TA. Cellular responses to formation of novel covalent drug.DNA interstrand cross- tion. Biochemistry 1992;31:9117–21. the DNA strand-scission enediyne C-1027 can be links and monoadducts by enediyne antitumor anti- 41. Xu Y, Zhen Y, Goldberg IH. Enediyne C1027 independent of ATM, ATR, and DNA-PK kinases. J Biol biotics. Biochemistry 1997;36:14975–84. induces the formation of novel covalent DNA Chem 2002;277:20549–54. 33. Liu JS, Kuo SR, Beerman TA, Melendy T. Induction of interstrand cross-links and monoadducts. J Am Chem 25. Kennedy DR, Beerman TA. The radiomimetic ene- DNA damage responses by adozelesin is S phase-specific Soc 1997;119:1133–4. diyne C-1027 induces unusual DNA damage responses and dependent on active replication forks. Mol Cancer 42. Battigello JM, Cui M, Roshong S, Carter BJ. Enediyne- to double-strand breaks. Biochemistry 2006;45:3747–54. Ther 2003;2:41–7. mediated cleavage of RNA. Bioorg Med Chem 1995;3: 26. LiuW,ChristensonSD,StandageS,ShenB. 34. Fernandez-Capetillo O, Lee A, Nussenzweig M, 839–49. Biosynthesis of the enediyne antitumor antibiotic C- Nussenzweig A. H2AX: the histone guardian of the 43. Gomez-Lazaro M, Fernandez-Gomez FJ, Jordan J. 1027. Science 2002;297:1170–3. genome. DNA Repair (Amst) 2004;3:959–67. p53: twenty five years understanding the mechanism of 27. Van Lanen SG, Dorrestein PC, Christenson SD, et al. 35. BaninS,MoyalL,ShiehS,etal.Enhanced genome protection. J Physiol Biochem 2004;60:287–307. Biosynthesis of the h-amino acid moiety of the enediyne phosphorylation of p53 by ATM in response to DNA 44. Sedelnikova OA, Rogakou EP, Panyutin IG, Bonner antitumor antibiotic C-1027 featuring h-amino acyl-S- damage. Science 1998;281:1674–7. WM. Quantitative detection of (125)IdU-induced DNA carrier protein intermediates. J Am Chem Soc 2005;127: 36. Walker S, Landovitz R, Ding WD, Ellestad GA, Kahne double-strand breaks with g-H2AX antibody. Radiat Res 11594–5. D. Cleavage behavior of calicheamicin g1 and calichea- 2002;158:486–92. 28. Ziv Y, Bar-Shira A, Pecker I, et al. Recombinant ATM micin T. Proc Natl Acad Sci U S A 1992;89:4608–12. 45. Li J, Stern DF. Regulation of CHK2by DNA- protein complements the cellular A-T phenotype. 37. Sugiura Y, Uesawa Y, Takahashi Y, Kuwahara J, Golik dependent protein kinase. J Biol Chem 2005;280: Oncogene 1997;15:159–67. J, Doyle TW. Nucleotide-specific cleavage and minor- 12041–50. 29. McHugh MM, Yin X, Kuo SR, Liu JS, Melendy T, groove interaction of DNA with esperamicin antitumor 46. Wang XQ, Redpath JL, Fan ST, Stanbridge EJ. ATR Beerman TA. The cellular response to DNA damage antibiotics. Proc Natl Acad Sci U S A 1989;86:7672–6. dependent activation of Chk2. J Cell Physiol 2006;208: induced by the enediynes C-1027 and neocarzinostatin 38. Gao X, Stassinopoulos A, Rice JS, Goldberg IH. 613–9. includes hyperphosphorylation and increased nuclear Structural basis for the sequence-specific DNA strand 47. Bakkenist CJ, Kastan MB. DNA damage activates retention of replication protein a (RPA) and trans cleavage by the enediyne neocarzinostatin chromo- ATM through intermolecular autophosphorylation and inhibition of DNA replication. Biochemistry 2001;40: phore. Structure of the post-activated chromophore- dimer dissociation. Nature 2003;421:499–506. 4792–9. DNA complex. Biochemistry 1995;34:40–9. 48. Niedernhofer LJ, Lalai AS, Hoeijmakers JH. Fanconi 30. Tice RR, Agurell E, Anderson D, et al. Single cell 39. Dedon PC, Goldberg IH. Free-radical mechanisms anemia (cross)linked to DNA repair. Cell 2005;123: gel/comet assay: guidelines for in vitro and in vivo involved in the formation of sequence-dependent 1191–8.

www.aacrjournals.org 781 Cancer Res 2007; 67: (2). January 15, 2007