Inhibiting Mitochondrial DNA Ligase IIIα Activates Caspase 1–Dependent Apoptosis in Cancer Cells

Among the ATP-dependent DNA ligases encoded by the mammalian LIG genes, LIG1, LIG3, and LIG4 (1), DNA ligase I (LigI) is primarily responsible for joining Okazaki fragments during nuclear DNA replication. However, DNA ligase IIIα (LigIIIα) is essential for DNA replication in LigI-deficient cells (2–5). LigI and LigIIIα also appear to have overlapping functions in the repair of base damage and single-strand breaks (3–8). Although DNA ligase IV is predominantly responsible for the repair of nuclear DNA double-strand breaks (DSB) by nonhomologous end joining (NHEJ), LigI and Lig IIIα participate in alternative (alt) NHEJ pathways (9, 10).

Unlike the nucleus, only one DNA ligase is present in mitochondria (3, 4, 11). Mitochondrial (mito) and nuclear (nuc) versions of LigIIIα are generated by alternative translation initiation (11). Although mito LigIIIα is required to maintain mitochondrial DNA and is essential for cell viability under normal culture conditions, this lethality can be rescued by either addition of pyruvate and uridine to the culture media or expression of mitochondrially targeted, heterologous DNA ligases, including the NAD-dependent E. coli LigA (3, 4, 12).

A subset of DNA ligase inhibitors preferentially sensitized cancer cells to DNA-damaging agents (13). Subsequently, it was shown that BCR-ABL1–positive cell lines and samples from patients with chronic myeloid leukemia, in particular leukemia cells that had acquired resistance to imatinib, were hypersensitive to the LigI/III inhibitor L67 in combination with a PARP inhibitor (14). A similar hypersensitivity was observed in breast cancer cell lines with either intrinsic or acquired resistance to antiestrogens (15). Because LigIIIα knockdown had the same effect as L67 in combination with a PARP inhibitor, it appears that L67 exerts its cancer cell–specific effect by inhibition of LigIIIα (14, 15). The hypersensitivity to the combination of L67 and a PARP inhibitor correlated with elevated expression of both LigIIIα and PARP1, and increased dependence on PARP1- and LigIIIα-dependent alt NHEJ (9, 14–16).

Although the repair inhibitor combination does inhibit alt NHEJ (14, 15), the observed synergy is unlikely to be due to the inhibition of two enzymes in the same pathway (9). Because LigIIIα has nuclear and mitochondrial functions, we examined the mechanism of L67-induced cytotoxicity. These studies revealed that L67 preferentially targets mito LigIIIα, resulting in mitochondrial dysfunction. Surprisingly, cancer cell mitochondria were more susceptible to L67 than mitochondria in nonmalignant cells. The disruption of mitochondrial function in cancer cells resulted in elevated levels of mitochondrially generated reactive oxygen species (ROS) and activation of a caspase 1–dependent apoptotic pathway that is involved in inflammatory responses induced by pathogenic microorganisms (17). In nonmalignant cells, there was no increase in mitochondrially generated ROS, but oxidative phosphorylation (OXPHOS) was completely uncoupled and the cells became senescent.

Human cervical (HeLa, 2012), colorectal (HCT116, 2006 and 2016), and breast (MDA-MB-231, 2008) cancers cell lines were purchased from the ATCC and grown in the recommended media. A HeLa cell line that stably expresses mitochondrially targeted E. coli LigA (mitoLigA; ref. 4) after transfection with the plasmid pCAG-mitoLigAYFP-Neo that encodes E. coli LigA fused at its N terminus to the LigIIIα mitochondrial targeting sequence and at its C terminus to EYFP. The telomerase-immortalized human fibroblast cell line HCA-Ltrt from Dr. Murnane (2010) was grown in DMEM/F12 medium with 10% FBS. Normal breast epithelium MCF10A cells from Dr. Rassool (2012) were grown using recommended medium and mixture of additives (Lonza/Clonetics Corporation) with 5% horse serum and 100 ng/mL cholera toxin. Cell lines lacking mitochondria DNA (Rho minus cells) were established as described (18, 19). The identity of commercially available cell lines was confirmed by STR profiling with the PowerPlex 1.2 System (Promega), most recently in 2016.

To measure colony formation, cells were cultured in triplicate in 6-well plates overnight and then were incubated with different concentrations of L67 for 14 days prior to staining with crystal violet. To measure cell growth, cells incubated with L67 for 48 to 72 hours. Genomic DNA was then stained with the CyQUANT NF reagent and quantitated according to the manufacturer's protocol.

Human Lig IIIα/His-tagged XRCC1 complex was expressed in and purified from insect cells (20). E. coli DNA ligase (LigA) was purchased from New England Biolabs. A labeled duplex oligonucleotide with a single ligatable nick was prepared as described previously (21). Lig IIIα/XRCC1 (0.05 pmol) and LigA (0.1 units) were incubated with L67 in the presence of either ATP (for Lig IIIα/XRCC1) or NAD (for LigA) at room temperature for 10 minutes prior to the addition of the DNA substrate (0.1 pmol) and incubation in buffer containing 50 mmol/L Tris-Cl pH 7.5, 12.5 mmol/L NaCl, 6 mmol/L MgCl₂, 3% glycerol, 1 mmol/L DTT, 0.25 mg/mL BSA, and 1% DMSO at 37°C for 20 minutes in a final volume of 20 μL. After the addition of SDS to a final concetario of 0.5%, DNA was recovered with Streptavidin MagneSphere beads (Promega), suspended in formamide/dye solution and electrophoresed through a 12.5% denaturing polyacrylamide gel. Ligated and unligated DNAs were detected by autoradiography and quantitated by ImageJ (NIH).

Cells were lysed in 20 mmol/L Tris-HCl, pH 7.5, 100 mmol/L NaCl, 1 mmol/L EDTA, 5% glycerol, and 1% Triton X-100 containing a protease inhibitor cocktail (Sigma). Proteins were detected by immunoblotting using the following antibodies: actin, phospho-KAP1(serine 824), (Abcam); Chk2, KAP1 (GeneTex), phospho-Chk2, p53, phospho-p53(serine 15), PARP1, cleaved PARP1 (Cell Signaling Technology); H2AX (R&D Systems); and γ-H2AX(serine 139) (Millipore). ATM (Cell Signaling Technology) and phospho-ATM (serine 1981) (Abcam) were detected as described (22).

Overnight cultures were incubated in fresh medium containing DMSO with different concentrations of L67 for 24 hours in the absence or presence of 5 mmol/L NAC. γ-H2AX and mitochondrial superoxide (mSOX) were detected by flow cytometry using an Alexa Fluor 647–conjugated antibody specific for γ-H2AX phosphorylated on serine139 and MitoSOX Red (Molecular Probes), respectively.

In experiments performed on a Seahorse XF24 Flux analyzer, cells were maintained at 5% CO₂ in DMEM media at 37°C. Cells (40,000) were plated on XF24 plates using Cell-Tak Cell and Tissue Adhesive (Corning), and growth medium was replaced with bicarbonate-free modified DMEM, the “assay medium.” In experiments performed on a Seahorse XF96e Flux analyzer, cells (20,000) were placed into each well using Cell-Tak. After incubation for another 60 minutes in a 37°C incubator without CO₂, oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were measured essentially as previously described (18).

Cells were seeded on slides and grown overnight in phenol-free DMEM medium prior to incubation in fresh medium containing either DMSO or 30 μmol/L L67 for 24 hours. Mitochondria were then stained by incubation with 250 nmol/L Mitotracker Red CMXRos for 20 minutes at 37°C. After washing, cells were maintained in phenol-free medium prior to imaging with a Zeiss LSM510 confocal system on an Axio Observer inverted microscope with a 63 × 1.2 NA water immersion objective. Mitotracker Red was excited with a 543 nm HeNe laser and fluorescence emission collected with 560 nm LP filter.

Genomic DNA was isolated using the QIAamp DNA Kit (QIAGEN) and then quantitated with the Quant-iT PicoGreen dsDNA Assay Kit (Molecular Probes). Short (220 bp) and long (8.9 kb) fragments of mitochondrial DNA and a 4 kb fragment of the nuc gene encoding DNA polymerase δ were amplified using 3 ng of total genomic DNA using the primers (Supplementary Table S2), amplification conditions (Supplementary Table S3), and a KAPA LongRange HotStart PCR kit (KAPA BIOSYSTEMS). Mitochondrial DNA copy number was calculated as described (19).

Apoptotic cells were detected by flow cytometry using the PE Annexin V apoptosis detection Kit (BD Pharmingen) according to the manufacturer's instructions. Where indicated, a caspase 1 inhibitor (50 μmol/L Ac-YVAD-cmk; Sigma-Aldrich), a caspase 3 inhibitor (50 μmol/L Z-DEVD-fmk; ApexBio), or a pan caspase inhibitor (50 μmol/L Z-VAD-FMK; R&D Systems) was added to the media with L67 prior to incubation for 24 hours.

Expression of β-galactosidase was detected using the Senescence β-Galactosidase Staining Kit (Cell Signaling Technology) according to the manufacturer's instructions.

Caspase 1 activity was measured by flow cytometry using a FLICA 660 in vitro caspase-1 detection Kit (Immunochemistry Technologies) according to the manufacturer's instructions.

Data are expressed as mean ± SEM. For comparison of groups, we used the Student two-tailed t test. A level of P < 0.05 was regarded as statistically significant.

To determine the mechanisms underlying the cytotoxicity of the LigI/III inhibitor L67 (13), derivatives of cell lines established from cervical (HeLa; Supplementary Fig. S1A), breast (MDA-MB-231), and colon (HCT116) cancers that lack mito DNA (Rho minus) were selected by growth in the presence of ethidium bromide, pyruvate, and uridine (18). HeLa cells with abnormal mitochondria that lack mitochondrial DNA (Supplementary Fig. S1A and S1B) were more resistant to L67 in colony forming (Fig. 1A) and cell proliferation assays (Fig. 1B) with the absence of mitochondrial DNA increasing the IC₅₀ from 8.2 μmol/L to 29.7 μmol/L (Supplementary Table S1). The differential effect of L67 on HeLa and HeLa Rho minus cells was also observed when HeLa cells were grown in the media containing pyruvate and uridine (Supplementary Fig. S1C). MDA-MB-231 and HCT 116 cells lacking mitochondrial DNA were also more resistant to L67 compared with their respective parental cells (Fig. 1C and D; and Supplementary Table S1), suggesting that, at lower concentrations, L67 causes cell death by targeting mitochondrial function.

To confirm that the effect of L67 is dependent upon inhibition of mito LigIIIα, we stably expressed mitoLigA in HeLa cells (Supplementary Fig. S2A). Although this enzyme can substitute for mitoLigIIIα (4), it is at least 6-fold more resistant to L67 than LigIIIα (Supplementary Fig. S2B). Notably, expression of mitoLigA significantly increased the resistance of HeLa cells to L67 (Fig. 1A and B) albeit not to same level as the HeLa Rho minus cells (Supplementary Table S1). This may be due in part to a dominant-negative effect of inhibiting endogenous LigIIIα.

The differential effect of L67 on the growth of parental and Rho minus cancer cells was most pronounced in a concentration range (5 to 15 μmol/L; Fig. 1A–D) that encompasses the IC₅₀ values for LigI (6 μmol/L) and LigIII (8 μmol/L; ref. 13). Although LigI and LigIIIα participate in nuc DNA replication (2, 5), incubation of synchronized cells with L67 at concentrations up to 15 μmol/L for 24 hours had only minor effects on cycle distribution (Supplementary Fig. S3A). Furthermore, incubation with L67 concentrations up to 10 μmol/L for 24 hours only reduced bromodeoxyuridine (BrdUrd) incorporation by 10% to 15% (Supplementary Fig. S3B). At higher concentrations of L67, there was an accumulation of HeLa cells, the derivative expressing mitoLigA, and, to a lesser extent, Rho minus cells in S phase (Supplementary Fig. S3A) that correlated with reduced BrdUrd incorporation (Supplementary Fig. S3B).

Interestingly, incubation of HeLa, but not HeLa Rho minus cells, with L67 at 10 and 15 μmol/L for 24 hours resulted in a dose-dependent increase in the formation of nuclear γH2AX foci (Supplementary Fig. S4A) and steady-state levels of γH2AX (Fig. 2A), consistent with replication stress and DSB induction in the nucleus (23). Expression of mitoLigA reduced γH2AX foci formation (Supplementary Fig. S4A) and the steady-state levels (Fig. 2A) induced by L67 compared with the parental cells, but not to the extent observed in Rho minus cells (Fig. 2A and Supplementary Fig. S4A). Because mitochondria are the major cellular source of reactive oxygen species (ROS) and disruption of mitochondrial function can lead to elevated levels of ROS (24, 25), we examined the effect of L67 on mSOX levels. Incubation with L67 for 24 hours resulted in a concentration-dependent increase in mSOX levels in HeLa cells, whereas expression of mitoLigA reduced the levels of mSOX induced by L67 and, as expected, there was no induction of mSOX in HeLa Rho minus cells (Fig. 2B).

Because γH2AX can be generated in response to other types of cellular stress in addition to DSBs (26), we examined the phosphorylation status of the ataxia telangiectasia mutated (ATM) kinase, which is recruited to and activated by DSBs, and ATM substrates (27). Expression of mitoLigA greatly attenuated L67-induced autophosphorylation of ATM and phosphorylation of γH2AX and several other ATM substrates but not p53, which was constitutively phosphorylated (Fig. 2C, compare left and right plots). While expression of mitoLigA attenuated the effects of L67 on cell survival and mSOX levels, mitochondrial function was abnormal (Fig. 4), suggesting that p53 may be activated in response to metabolic stress in these cells (28). Although ATM can be activated by DSB-independent mechanisms, including direct activation by ROS (27, 29), incubation with L67 resulted in phosphorylation of KAP1, whose phosphorylation by ATM is DSB-dependent (Fig. 2C; ref. 30). As expected, the free radical scavenger N-acetyl cysteine (NAC) reduced mSOX (Fig. 2E), ATM activation (Supplementary Fig. S4B), and γH2AX formation (Fig. 2D) induced by L67 in HeLa cells. Together, these results indicate that the nuclear DNA damage induced by L67 in HeLa cells is due to increased generation of ROS by mitochondria as result of inhibition of mito LigIIIα.

Similar to the cancer cell lines (Fig. 1), proliferation of a Rho minus derivative of a telomerase-immortalized fibroblast (HCA-Ltrt) was more resistant to L67 than the parental cell line (Fig. 3A). Surprisingly, while L67 had similar effects on cell-cycle distribution (Supplementary Figs. S3A and S5A) and the proliferation of cancer cell lines (Fig. 1A–D) and nonmalignant cell lines established from normal tissues (Fig. 3A), elevated levels of mSOX were not detected in nonmalignant cells, including HCA-Ltrt Rho minus cells (Fig. 3B and Supplementary Fig. S5B). Accordingly, L67 did not induce detectable phosphorylation of H2AX and KAP1 in HCA-Ltrt cells, and ATM autophosphorylation was only observed with 30 μmol/L L67 (Fig. 3C). Interestingly, phosphorylation of p53 and Chk2 occurred in the absence of detectable ATM activation (Fig. 3C), suggesting a response to a stress signal other than DNA damage (28).

As mentioned previously, expression of mitoLigA in HeLa cells resulted in altered mitochondrial function. The OCR was about 50% lower compared with the parental HeLa cells (Fig. 4A), whereas the proportion of oligomycin-insensitive oxygen consumption, indicative of uncoupled OXPHOS, was higher (∼40%) compared with the parental HeLa cells (∼20%). Incubation with L67 resulted in a similar reduction in OCR (40%–50%) in both cell lines with about 80% of the remaining OCR due to uncoupled OXPHOS (Fig. 4A). The changes in OCR in the HeLa cells expressing mitoLigA resulted in a compensatory increase in ECAR, a measure of glycolysis. As expected, there was an increase in ECAR following incubation with L67 (Fig. 4B).

Because reduced mito DNA has been linked with elevated mSOX and increased sensitivity to DNA-damaging agents (31), we examined the effect of L67 on mito DNA copy number. Incubation of HeLa cells with 5 and 10 μmol/L L67 resulted in about a 25% reduction in mito DNA, whereas there was no reduction in mito DNA in derivatives expressing mitoLigA (Fig. 4C). Furthermore, incubation with 30 μmol/L L67 disrupted the filamentous network of mitochondria observed in untreated HeLa cells, resulting in a similar appearance to HeLa Rho minus cells with large irregular clumps of mitochondria and diffuse cell staining (Fig. 3D; Supplementary Fig. S1B). Thus, while expression of mitoLigA resulted in reduced OCR (Fig. 4A) and increased ECAR (Fig. 4B), it prevented the reduction of mito DNA and abnormal mitochondrial morphology induced by L67 (Fig. 4D). Notably, incubation of telomerase-immortalized fibroblasts with 10 μmol/L L67 did not cause a reduction in OCR, but OXPHOS was totally uncoupled (high OCR in presence of both L67 and oligomycin; Fig. 4E) and there was a large compensatory increase in glycolysis (Fig. 4F). In addition, there were no reductions in mitochondrial DNA copy number (Fig. 4C) or alterations in mitochondrial morphology (Fig. 4G) in nonmalignant cells incubated with L67.

Following incubation with 10 μmol/L L67 for 24 hours, there were greater than 12-fold more apoptotic (7-AAD positive, PE Annexin V positive) HeLa and HeLa mitoLigA cells than apoptotic HeLa Rho minus cells (Fig. 5A), indicating that the majority of cell death induced by L67 is dependent upon mito DNA and presumably mitochondrial function. Although, at 10 μmol/L L67, the extent of apoptosis was similar in the HeLa and HeLa mitoLigA cells (Fig. 5A), cell death was caspase-dependent in HeLa cells (Fig. 6) and caspase-independent in the HeLa mitoLigA cells (Fig. 6A and Supplementary Fig. S6). There was greater apoptosis in the HeLa cells compared with HeLa cells expressing mitoLigA at 100 μmol/L L67 with apoptotic cells constituting about 50% of the HeLa cell population compared with 23% of the HeLa mitoLigA population (Fig. 5A), indicating that expression of mitoLigA reduces the cytotoxicity of L67. In contrast, very low levels of apoptosis were detected in HCA-Ltrt cells incubated with L67 (Fig. 5A). Because L67 reduced the proliferation of nonmalignant cells (Fig. 3A), we asked whether L67 induced senescence. As shown in Fig. 5B, incubation of HCA-Ltrt and MCF10A cells with L67 resulted in β-galactosidase expression, a marker of senescence. Thus, nonmalignant and cancer cells respond differently to inhibition of mito LigIIIα with nonmalignant cells becoming senescent, whereas cancer cells undergo apoptosis.

No activation of caspases 7 and 9 that are involved in extrinsic and intrinsic apoptotic pathways was detected in HeLa cells incubated with L67 (Supplementary Fig. S6). Instead, caspase 1, which is activated in the inflammatory response elicited by pathogens (17) and by the release of ROS and mito DNA into the cytoplasm (32–34), was activated in HeLa cells but not in either the Rho minus or mitoLigA-expressing derivatives (Fig. 6A). Notably, incubation with L67 also activated caspase 1 in breast (MDA-MB-231) and colorectal (HCT116) cancer cell lines but not in nonmalignant cell lines (Fig. 6B). To determine whether cell death is dependent upon caspase 1 activation, we examined the effect of the caspase inhibitors on L67-induced apoptosis in HeLa cells. As shown in Fig. 6C, apoptosis was reduced by a caspase 1 inhibitor and a pan caspase inhibitor but not a caspase 3 inhibitor, confirming that cell death was mediated by caspase 1. Together, these results demonstrate that inhibition of mito LigIIIα selectively induces cell death in cancer cells by activating a caspase 1–dependent apoptotic pathway.

Previous studies have shown that therapy-resistant forms of breast cancer and leukemia with elevated levels of LigIIIα and PARP-1 are hypersensitive to inhibition of LigIIIα in combination with a PARP inhibitor (14, 15). Here, we have shown that the LigIIIα inhibitor, L67, used in those studies preferentially targets mito LigIIIα and, at the concentrations used to preferentially kill therapy-resistant leukemia and breast cancer cells in combination with a PARP inhibitor (14, 15), had no effect on nuclear DNA replication. Importantly, while L67 affected mitochondrial function in both cancer and nonmalignant cells, the effects and outcomes were very different. In cancer cell lines, incubation with L67 resulted in elevated levels of mitochondrially generated ROS, abnormal mitochondrial morphology, reductions in OCR and mitochondrial DNA, and caspase 1–dependent cell death. Expression of L67-resistant mitoLigA in HeLa cells partially suppressed the increase in mSOX, and completely prevented changes in mitochondrial DNA copy number and morphology, and activation of caspase 1, demonstrating that the effects of L67 were dependent upon inhibition of mito LigIIIα. In contrast to cancer cells, L67 did not alter mitochondrial morphology, mitochondrial DNA copy number, or mSOX levels and did not activate caspase 1 in nonmalignant cells. Instead of apoptosis, L67 induced significant uncoupling of mitochondria with a compensatory increase in glycolysis and senescence in nonmalignant cells. Thus, cancer cells are less able to mitigate the deleterious effects caused by disruption of mitochondrial DNA metabolism than nonmalignant cells.

Because inhibition of mito LigIIIα did not cause an increase in steady-state levels of mSOX in nonmalignant cells even though there was no reduction in the OCR, we suggest that initial increases in the levels of mito DNA damage and/or mSOX activate antioxidant responses in nonmalignant cells that prevent further increases in mSOX. Notably, we found that OXPHOS was totally uncoupled in the nonmalignant cell line incubated with L67. Because the uncoupling of oxygen consumption from electron transport reduces the generation of ROS by the electron transport chain (35), this alteration in nonmalignant cells presumably attenuates increases in mSOX. In addition, nonmalignant cells may be able to prevent the increase in mitochondrially generated ROS by removing dysfunctional, damaged mitochondria via autophagy (32, 34, 36). Although mito dysfunction caused by inhibition of mito LigIIIα did not result in elevated levels of ROS, p53 was activated, presumably in response to the alterations in cellular energy metabolism, and may contribute to cellular senescence (28). The increase in glycolysis may either be an effort to compensate for the uncoupled OXPHOS or an indicator of senescence (37).

There are conflicting reports as to the effect of LigIIIα knockdown on mitochondrial DNA (12, 38). In accord with the normal appearance of mitochondria in L67-treated nonmalignant cells, these cells were able to maintain their mitochondrial DNA copy number. Recent studies have shown that there are at least two modes of mitochondrial DNA replication, one of which involves a conventional coupled leading and lagging strand, whereas the other one involves RNA incorporation throughout the lagging strand (39). Because the latter mechanism does not require the joining of multiple Okazaki fragments (39), it is possible that nonmalignant cells are able to maintain mitochondrial DNA copy number by switching to this relatively DNA ligase–independent replication mechanism.

It is well established that cancers driven by oncogenes, such as Ras, c-Myc, BCR-ABL1, and FLT3-ITD, have elevated levels of ROS (40–42). While this increase in ROS activates signaling pathways that promote cell proliferation and metastasis and inhibit cell death pathways (43), redox homeostasis is presumably deregulated as part of the adaptation to constitutively higher ROS levels. Because treatment with an exogenous oxidizing agent induces elevated levels of mSOX and degradation of mitochondrial DNA (44–46) and reduced levels of LigIIIα increase the ROS-induced degradation of mitochondrial DNA (12), our results suggest that the elevated mSOX induced by L67 in cancer cells results from an inability to activate mechanisms, such as OXPHOS uncoupling and mitophagy, that reduce mitochondrially generated ROS (35, 36). Consequently, this initiates a vicious cycle in which elevated mSOX levels cause more DNA damage, resulting in reduced mitochondrial DNA copy number and more mito dysfunction that in turn generates more mSOX (31, 45, 47). Interestingly, expression of mitoLigA in HeLa cells did not totally prevent the increase in mSOX induced by L67, but fully maintained mitochondrial DNA copy number. Thus, our results support the idea that a combination of increased mSOX and inhibition of mito LigIIIα results in the degradation of mito DNA in HeLa cells.

L67-induced cell death in cancer cells occurred via a caspase 1–dependent apoptotic pathway. Although caspase 1 can be activated by either inflammasomes or pyroptosomes (17), it seems likely that activation of caspase 1 by L67 is mediated by the inflammasome because ROS and mitochondrial DNA have been shown to activate the NLRP3 inflammasome (32–34). Thus, we suggest that the L67-induced mitochondrial dysfunction triggers the mitochondrial permeability transition (48) in cancer cells with the subsequent release of superoxide and mito DNA into the cytoplasm, activating the inflammasome and caspase 1–dependent apoptosis (32–34).

Here, we have shown that cancer cells are unable to effectively respond to the deleterious effects caused by the inhibition of LigIIIα, a key enzyme in mitochondrial DNA metabolism. Notably, the elevated levels of mitochondrially generated ROS in cancer cells incubated with L67 result in increased nuclear DNA damage, suggesting that L67-induced mitochondria-dependent nuclear DNA damage underlies the synergistic activity of L67 with PARP inhibitors that reduce nuclear DNA repair (14, 15). Furthermore, our studies showing that regulation of mitochondrial function is abnormal in cancer cells not only provide further evidence that strategies targeting cancer cell mitochondria may have utility in cancer therapy (49, 50), but may lead to the identification of novel therapeutic approaches that exploit these differences to selectively eliminate cancer cells by activating caspase 1–dependent apoptosis.

A.E Tomkinson is Associate Editor at DNA Repair Journal, has ownership interest (including patents) in University of Maryland US patent 8,445.537 and US patent 9,132,120, and is a consultant/advisory board member for University of Maryland Cancer Center. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A. Sallmyr, V. Roginskaya, B. Van Houten, A.E. Tomkinson

Development of methodology: A. Sallmyr, Y. Matsumoto, B. Van Houten

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Sallmyr, V. Roginskaya, B. Van Houten

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Sallmyr, V. Roginskaya, B. Van Houten, A.E. Tomkinson

Writing, review, and/or revision of the manuscript: A. Sallmyr, B. Van Houten, A.E. Tomkinson

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): V. Roginskaya

Study supervision: A. Sallmyr, B. Van Houten, A.E. Tomkinson

We thank Dr. Maria Jasin for the plasmid construct encoding mitoLigA and Dr. George Greco for the synthesis of L67.

This work was supported by US NIH Grants (R01 ES012512 and P01 CA92584 to A.E. Tomkinson). Flow cytometry and microscopy were carried out in University of New Mexico Cancer Center Shared Resources supported by NCI Cancer Center Support Grant P30 CA11800. This work was also supported in part by PA CURE from the Department of Health, PA (B. Van Houten).

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.