The Antitumor Enediyne C-1027 Alters Cell Cycle Progression and Induces Chromosomal Aberrations and Telomere Dysfunction

This study examined the extent of chromosome instability induced in cultured human colon carcinoma HCT116 cells by the antitumor radiomimetic enediyne antibiotic C-1027. Spectral karyotype analysis showed frequent intrachromosomal fusions and fragmentations 26 hours after addition of as little as 0.035 nmol/L C-1027. When the concentration was increased to 0.14 nmol/L C-1027, 92% of cells showed chromosomal aberrations compared with only 2.9% after treatment with an equivalent growth inhibitory dose of ionizing radiation (20 Gy). Thus, chromosome misrejoining was associated to a much greater extent with C-1027–induced than with ionizing radiation–induced cell growth inhibition. Despite these aberrations, a large fraction of C-1027–treated cells progressed into G₁. Comet analysis showed that these extensive chromosomal anomalies were not due to increased induction or reduced repair of C-1027–induced compared with ionizing radiation–induced strand breaks. Fluorescence in situ hybridization analysis showed that misrejoining of telomere repeats (i.e., chromosomes joined end to end at their telomeres or fused together after complete loss of telomere sequences) was observed within 26 hours of C-1027 addition. The extreme cytotoxicity of C-1027 may reflect both induction and erroneous repair of DNA double-strand break in the whole genome and/or in subgenomic targets such as telomere sequences.

Structural aberrations, including chromatid/chromosome breaks, fusion/translocations, and/or chromosome deletions (1), are caused by a wide variety of DNA reactive chemical agents and ionizing radiation (2, 3). Although these aberrations can drastically reduce cell survival over time, heritable mutations in the surviving cells can contribute to chromosome instability and cancer development (3).

The most important lesions responsible for the production of chromosome aberrations are DNA double-strand breaks (4), which can be induced either directly (by agents such as ionizing radiation, bleomycin, or neocarzinostatin; ref. 5) or indirectly (by alkylating agents that block replication fork progression; refs. 6, 7). Double-strand breaks signal phosphorylation of ATM and subsequent activation of cell cycle checkpoint proteins that delay cell cycle progression to promote DNA repair and to preserve chromosomal integrity (8).

The primary route for double-strand break repairs in all phases of the mammalian cell cycle is nonhomologous end joining (NHEJ; refs. 9–11), which joins any double-strand breaks end to end regardless of sequence. Both the frequency and clustering of double-strand breaks (12), as well as their specificity for unique chromosome regions, can modulate aberrant chromosome end joining (2). In addition, that double-strand breaks induced in unique chromosome regions, such as interstitial telomere repeats and centromeres, show a recombination frequency four to five times greater than that expected for chromosome regions of equivalent size suggests the existence of “hotspots” for recombination (13).

Whereas much of our current understanding of the mechanisms and consequences of aberrant NHEJ were obtained from studies with ionizing radiation (14, 15), double-strand breaks account for less than <10% of ionizing radiation–induced DNA damage (16),³

and lesions other than strand breaks may contribute to ionizing radiation–induced genotoxicity (18). By contrast, induction of sequence-specific DNA strand breaks is the primary cellular effect of the extremely potent enediyne antitumor antibiotics and is directly correlated with their cytotoxicity (19). Although their chemical structure, thiol dependence, or DNA sequence cleavage preference can vary, the enediyne chromophores all abstract hydrogen from deoxyribose on either one or two DNA strands leading either to single-strand breaks or double-strand breaks, respectively (20).

Recently, we showed that the double-strand break inducing enediyne C-1027 differed from ionizing radiation in its ability to promote an ATM-independent DNA damage response (21). In addition, C-1027–induced, unlike ionizing radiation–induced, hyperphosphorylation of replication protein A is not blocked by aphidicolin (22, 23), indicating that C-1027 effects on DNA replication and repair are independent of active replication fork progression. C-1027 also has a specific cleavage preference based on the DNA sequence (24), whereas ionizing radiation–induced double-strand breaks are not directed to any specific sequence but can be modulated by chromosomal conformation or orientation within the nucleus (12). These data suggest differences both in the incidence of and cellular response to DNA damage induced by ionizing radiation and C-1027.

This study determined the type and extent of C-1027–induced chromosome aberrations (i.e., breaks, deletions, fragmentation, and fusion/translocation/recombination). Spectral karyotyping was used to quantitatively and qualitatively analyze chromosome changes in mitotic spreads of cells treated with C-1027 or ionizing radiation during interphase. The type and amount of C-1027–induced chromosome aberrations were compared with those observed after treatment with ionizing radiation at a dose that caused equivalent cell growth inhibition. The association of the C-1027–induced aberrations with induction and repair of DNA strand breaks, and with altered cell cycle progression, was also examined. Because the preferred DNA sequence for C-1027 double-strand breaks (GTTA) is contained within the telomere repeat sequence (TTAGGG; ref. 25), fluorescence in situ hybridization (FISH) was used to determine whether telomeres were targets for C-1027–induced DNA damage and misrejoining. We show that the extremely cytotoxic C-1027 exhibits highly potent clastogenic activity.

Chemicals. C-1027, a generous gift from Taiho Pharmaceutical, Co., Ltd. (Saitama, Japan), was adjusted to 1 mg/mL in H₂O and stored at −20°C. All other chemicals were of reagent grade. A Siemens therapeutic X-ray machine operated at 250 kV and 15 mA at a dose rate of 123 rad/min was used for irradiation.

Cell culture. Human colon carcinoma HCT116 cells (American Type Culture Collection, Manassas, VA) were maintained in monolayer culture in McCoys medium with 10% fetal bovine serum in a humidified atmosphere of 5% CO₂.

Fluorescence-activated cell sorting analysis. Cell cycle distribution was determined by fluorescence-activated cell sorting analysis of propidium iodide–stained ethanol-fixed cells using a FACScan flow cytometer (Becton Dickinson, Franklin Lakes, NJ). Cell cycle compartments were determined using WinList and ModFit software (Verity Software House, Inc., Topsham, ME).

Spectral karyotyping analysis. Metaphase chromosome spreads were prepared using standard hypotonic treatment and air drying methods. After sequential digestion with RNase and pepsin according to the procedure recommended by Applied Spectral Imaging, Inc. (Carlsbad, CA), the chromosome preparations were denatured in 70% formamide and hybridized with the human spectral karyotyping paint probes (i.e., a mixture of individual chromosome DNAs prepared by flow sorting and then tagged with different combinations of digoxigenin, Texas red, Cy5, biotin/streptavidin, and Cy5.5). The detailed methods, using a variable ratio and combination of a number of individual dyes, were described elsewhere (26, 27). The spectral karyotyping images were captured using a Nikon microscope equipped with a spectral cube and interferometer module. Karyotypes were prepared using SKY View software (version 1.62). Chromosomes and chromosomal rearrangements or alterations, including simple balanced translocation or unbalanced (or nonreciprocal) translocation resulting in dicentric (telomere) fusion, were sorted according to the classification described by Smogorzewska et al. (28).

Comet assay. Cells were treated with a DNA-damaging agent and harvested by trypsinization. After one wash in PBS, 2 × 10⁴ cells were analyzed by Comet assay as described elsewhere with minor modifications (29). Briefly, cells were suspended in low gelling temperature agarose, placed on a slide, and covered with a coverslip. Once the agarose had hardened, cells were lysed in alkaline buffer and subjected to alkaline unwinding of DNA followed by electrophoresis. A tail (Comet) on cells with damaged DNA was visualized by ethidium bromide staining and immunofluorescence microscopy.

DNA damage was quantitated after electrophoresis of the Comet slides as described elsewhere (30). A minimum of 50 cells per sample were examined and the extent of damage to each cell was assigned a score of 0 to 4 as follows: 0, no damage–no DNA migration; 1, dense nucleus with slight migration of nuclear material; 2, the Comet tail has progressed to its full length, but the Comet width does not exceed that of the nucleus; 3, the width of the Comet tail is greater than the width of the nucleus, and the nucleus is less dense; 4, the nucleus and the tail are completely separated.

Telomere detection. Telomeres were detected by FISH with FITC-labeled peptide-nucleic acid telomere probes. (DAKOCytomation, Carpenteria, CA). DNA was counterstained with 4′,6-diamidino-2-phenylindole (DAPI). FISH and DAPI signals were visualized by immunofluorescent microscopy.

DNA strand breaks induced by ionizing radiation and radiomimetics, like the enediyne neocarzinostatin, cause chromosome aberrations and cell death (5). This study examined whether, and to what extent, C-1027–induced DNA damage caused chromosome aberrations.

Because metaphase cells would be used to determine the magnitude of C-1027–induced effects on higher-order genomic structure (i.e., chromosomes), HCT116 cells were treated with different C-1027 concentrations and incubations times to determine conditions optimal for the accumulation of cells in G₂-M. HCT116 cells were treated for 16 hours⁴

with 0.014 to 1.43 nmol/L C-1027, and examined by flow cytometry as described in Materials and Methods (see Fig. 1A). Compared to control (untreated) cells, a greater accumulation of cells in G₂-M was observed in cells treated with as little as 0.014 nmol/L C-1027. With 0.14 nmol/L C-1027, G₂-M accumulation was maximal (≥80%). With 1.43 nmol/L, a large proportion of the cells had accumulated in the S phase as well as in the G₂-M phase. Figure 1B shows cell cycle progression at increasing times after addition of 0.14 nmol/L C-1027. Ten hours after C-1027 addition, a G₂-M block was indicated by the large fraction of C-1027–treated cells that had accumulated in G₂-M. By 24 hours after addition of 0.14 nmol/L C-1027, viability, as assayed by trypan blue exclusion, remained high (≥85%; data not shown), and a G₁ peak was also observed, indicating that some cells were able to overcome the G₂ block and progress through mitosis into G₁. Thus, metaphase spreads were prepared from cells treated for ≤26 hours with 0.14 nmol/L C-1027.

To determine whether C-1027 induced chromosome aberrations, spectral karyotyping analysis of metaphase chromosome spreads as described in Materials and Methods was done. Figure 2A shows spectral karyotyping analysis of a metaphase chromosome spread of an untreated HCT116 cell.⁵

Differentiation of each chromosome by a single unique color made it possible to identify endogenous chromosome translocations. Figure 2B and C shows chromosome spreads from cells treated with 0.035 and 0.14 nmol/L C-1027, respectively. Aberrant chromosome rejoining, evidenced by chromosome structures consisting of more than one color, was seen after treatment with as little as 0.035 nmol/L C-1027 (see Fig. 2B). Portions of different chromosomes (e.g., 2 and 4; 3 and 7; 8 and 18) seemed to have fused, whereas smaller chromosome fragments were also apparent. When cells were treated with 0.14 nmol/L C-1027 (Fig. 2C), many more aberrant chromosome fusion events, including misrejoining of more than two chromosome fragments, as well as chromosome fragmentations, were observed. That the exchange involved chromosomes, rather than chromatid fragments, suggested that generation of C-1027–induced aberrations was likely not limited to cells traversing the S phase. The spectral karyotyping data was quantitated as shown in Table 1. The percent of cells with chromosomal rearrangements was much higher after treatment with 0.14 nmol/L C-1027 (92%) than with an equivalent growth-inhibitory dose (i.e., 20 Gy)⁶ of ionizing radiation (2.9%). Whereas ionizing radiation primarily caused fragmentation, C-1027–treated cells exhibited both fragmentation and recombination. Thus, chromosomes were misrejoined to a much greater extent after damage induced by C-1027 than by ionizing radiation.

Whether the larger numbers of chromosome aberrations observed after C-1027 compared with ionizing radiation treatment reflected greater numbers of C-1027–induced double-strand breaks distributed throughout the cell population, or of double-strand breaks that were induced only in a limited number of cells, was examined using alkaline Comet (single-cell gel electrophoresis) analysis (32) as described in Materials and Methods. Figure 3A shows Comet analysis of cells incubated for 30 minutes with C-1027. At the lowest concentration (0.14 nmol/L), damage to nuclear DNA, as evidenced by a Comet tail, was detected in <10% of the cells. By contrast, more damage was observed after treatment with 20 Gy ionizing radiation. When the C-1027 concentration was increased to 1.43 nmol/L, some damage was detected in all of the cells and the extent of this damage continued to increase up to 2.9 nmol/L. Figure 3B shows the DNA damage score from three separate experiments determined as described in Materials and Methods. When C-1027 was increased from 0.14 to 2.9 nmol/L, the damage score increased dramatically. This was in agreement with earlier pulsed-field gel electrophoresis data from our laboratory that showed minimal detectable damage to genomic DNA with 0.1 nmol/L C-1027 (33). As was evident from the images in Fig. 3A, 0.14 nmol/L C-1027 induced less damage than did 20 Gy ionizing radiation (i.e., a damage score of 0.03 compared with 0.4). Thus, the difference in chromosomal aberrations shown in Table 1 was not due to increased DNA damage in C-1027–treated compared with ionizing radiation–treated cells.

Repair of C-1027–induced DNA strand damage has not been reported. However, chromosome misrejoining, such as that observed after C-1027 treatment (see Fig. 2), is indicative of NHEJ repair of DNA double-strand breaks (34). The extent of repair of C-1027–induced HCT116 cell DNA damage was determined by Comet analysis (Fig. 4). Cells were treated for 30 minutes with 1.43 nmol/L C-1027, followed by incubation for 0.5 to 26 hours in drug-free medium. DNA damage was increased 30 minutes, compared with zero time, after C-1027 removal. However, 2 and 26 hours after removal, DNA damage was reduced to near control levels, indicating that the majority of C-1027–induced DNA breaks were rejoined. A similar ability to repair DNA lesions was observed in cells that were incubated continuously in the presence of C-1027 (i.e., without drug removal; data not shown). Thus, rejoining of C-1027–induced DNA damage was rapid and persisted for at least 26 hours.

We next examined whether C-1027–induced chromosome aberrations reflect selective damage to specific chromosome domains. Because telomere shortening can lead to chromosome aberrations (35, 36) and the preferred double-strand break cleavage site for C-1027, GTTA/CAAT, is contained within the telomere repeat sequence (i.e., 5′-GGT TAG-3′), FISH was used to determine whether C-1027 could damage telomere DNA. Figure 5 shows damage to telomere structures (arrows) in a metaphase cell 30 hours after addition of 0.14 nmol/L C-1027. Aberrations observed included chromosomes joined end to end with partial or complete loss of telomeres, as well as large amounts of chromosome fragmentation and fusions and telomere loss. The image shown is typical of other metaphase cells treated with C-1027 and suggests that C-1027 can target telomere sequences.

C-1027–induced lesions resulted in extensive amounts of aberrant chromosome recombination and damage to telomere sequences. As well as being the first to show rapid repair of C-1027–induced genomic DNA damage, this report showed that at least some C-1027–induced DNA lesions were misrepaired.

The data show that C-1027 is a much more potent inducer of chromosome fragmentation and recombination than ionizing radiation. Chromosome aberrations are also induced by other radiomimetics, notably neocarzinostatin and bleomycin, but at much higher concentrations (i.e., 2 and 1 nmol/L, respectively; ref. 3) than by C-1027 (0.035 nmol/L; see Fig. 2A). More than 80% of lesions induced by neocarzinostatin and bleomycin are single-strand breaks (5), whereas C-1027 causes primarily double-strand breaks (37). Thus, increased amounts of C-1027–induced chromosome aberrations compared with neocarzinostatin- or bleomycin-induced chromosome aberrations may result from lack of fidelity of repair of greater numbers of double-strand breaks.

The primary mechanism for DNA double-strand breaks repair in mammalian cells is NHEJ (10, 38), which can cause extensive chromosome misrejoining in the presence of multiple double-strand breaks. However, despite causing far less damage to intracellular DNA, C-1027 caused much more extensive chromosome recombination than ionizing radiation.

The likelihood of NHEJ-induced chromosome misrejoining may be enhanced if clusters of double-strand breaks are directed to specific regions of the genome (11, 34). Whereas ionizing radiation induces double-strand breaks at random sequences (39), C-1027, as well as neocarzinostatin and bleomycin, can target damage to specific intracellular DNA sequences (33, 40, 41). If the limited number of strand breaks induced by C-1027 compared with ionizing radiation were directed to more localized regions of the genome as well as to the genome as a whole, clustering of double-strand breaks would occur and misrejoining likely would be enhanced.

Double-strand telomere DNA, consisting of multiple copies of the sequence 5′-GGT TAG-3′ (42), presents multiple sites for closely spaced C-1027–induced double-strand breaks at its preferred cleavage sequence, 5′-GTTA-3′. Because the average telomere length in HCT116 cells is 4,000 bp,⁷

the number of possible C-1027 cleavage sites can be >600 per telomere. Whether all these sites are equally accessible to C-1027 cleavage (e.g., whether C-1027 cleavage at one site can promote or inhibit cleavage at adjacent sites, such as those in adjacent telomere repeats) is unknown. In addition, the spatial distribution of telomeres within the nucleus differs with different cell cycle compartments (43) and may modulate C-1027–induced damage. For example, telomeres are widely distributed throughout the nucleus in G₀-G₁ and S, but become aligned along a central plane in late G₂, forming a telomere disc. Such reorientation of telomere DNA during G₂ might alter the affinity of C-1027 for cleavage at some or all of its preferred sites on telomere repeats.

That C-1027 treatment caused chromosome misrejoining at telomere junctions, as well as a reduction or complete loss of telomere DNA from chromosome ends (see Fig. 5), suggested that C-1027 may preferentially cleave telomere DNA. Such telomere damage could cause chromosome aberrations and/or rearrangements in several ways. First, multiple C-1027–induced double-strand breaks might be rejoined with less fidelity than double-strand breaks dispersed throughout the genome as has been reported for high numbers of closely aligned ionizing radiation–induced double-strand breaks (11, 34). Also, telomeres possess a 3′ overhang (28, 44) that is essential for stabilizing chromosome ends (45), as well as for efficient telomere priming (46) and extension by telomerase (47, 48). C-1027 cleavage may reduce telomerase function in telomerase-competent cells (e.g., tumor cells) by eliminating the 3′ overhang necessary for efficient telomere elongation. Thus, the nearly blunt-ended double-strand breaks induced by C-1027 at its preferred cleavage site within the telomere repeat may be refractory to repair and may cause shortening or even a complete loss of telomere DNA.⁸

DNA ligase IV–dependent NHEJ can fuse chromosomes with shortened telomeres to double-strand breaks located elsewhere in the genome (28), leading to chromosome instability (49). In the present study, such incorrect misrejoining was observed after treatment with as low as 0.035 nmol/L C-1027 (see Fig. 2B). Lastly, other workers have reported that, like other repetitive DNA sequences (50), telomere repeat-like sequences may be preferred sites for enhanced intrachromosomal rearrangements and fragility (13). Such rearrangements might be increased by C-1027–induced double-strand breaks directed to these sequences.

Telomere dysfunction can lead to senescence (51), and senescence has been reported after treatment of human hepatoma and breast carcinoma cell lines with low levels of lidamycin (C-1027, 0.01-1 nmol/L; ref. 52). In the present study, HCT116 cells also exhibited a senescent phenotype 10 days after treatment with 0.14 nmol/L C-1027 (data not shown), lending credence to the possibility that C-1027 may target telomeres. By contrast, we (data not shown) and others (53) found that apoptosis also can be induced, but only at very high concentrations (i.e., 2 nmol/L) of C-1027. However, these nanomolar concentrations also caused such extensive chromosomal fragmentation that identification of damage to specific chromosomal regions, such as telomeres, was not possible.

Regardless of whether damage is more focused on discrete sequences or evenly distributed throughout the genome, cell cycle progression is generally slowed to allow time for repair of DNA lesions (54). In the present study, reduced progression through S phase (i.e., replication inhibition) was observed within 2 hours after 0.14 nmol/L C-1027 addition (data not shown). Earlier reports from our laboratory showed that double-strand breaks induced by C-1027, as well as by neocarzinostatin, rapidly inhibited replication initiation via hyperphosphorylation of replication protein A (55). Other workers showed that phosphorylation of replication protein A by the NHEJ DNA-PKcs proteins inhibited homologous recombination (56). That C-1027 can induce an inhibitor of homologous recombination further suggests that NHEJ is the repair pathway responsible for the formation of C-1027–induced chromosome aberrations.

After 10-hour treatment with 0.14 nmol/L C-1027, ≥80% of cells were in G₂-M, indicating activation of a G₂ checkpoint. Other workers have shown that a G₂ checkpoint is necessary to allow nonhomologous chromosome fusions resulting from misrepair to segregate into individual chromosomes, and that cells that bypass this checkpoint exhibit entangled chromosome regions (57). In the present study, C-1027–treated cells progressed through mitosis without resolution of chromosome misrejoining, whereas ionizing radiation–treated cells that progressed into mitosis showed far fewer aberrations. Despite these differences in repair fidelity, that 20 Gy ionizing radiation and 0.14 nmol/L C-1027 cause equivalent reductions in cell colony formation (data not shown) suggested that chromosome misrejoining may play a greater role in C-1027–induced than in ionizing radiation–induced cell growth inhibition.

The extreme cytotoxicity of C-1027 compared with other enediynes as well as to other classes of DNA-damaging agents may reflect its ability both to induce DNA double-strand breaks and to promote erroneous DNA repair in the whole genome and/or in subgenomic targets, such as telomere-like repeat sequences. Additional studies are under way to elucidate the mechanism and consequences of C-1027 induction of chromosome aberrations.

Grant support: National Cancer Institute grants CA 106312 and CA 16056 (T.A. Beerman).

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 with 18 U.S.C. Section 1734 solely to indicate this fact.