Mitochondrial protein COXFA4L3 (C15ORF48) confers resistance to DNA-damaging anticancer agents by repressing mitochondrial DNA damage responses Authors: Yuki Takakura, Seika Kawamura, Yui Hashiguchi, Manami Yamanaka, Yasuyuki Kurihara, Taishin Akiyama, Hiroyuki Takano, Noritaka Yamaguchi Published: 2026 ABSTRACT DNA-damaging anticancer agents selectively kill highly proliferative cancer cells by inducing DNA damage, such as DNA double- or single-strand breaks, and by inhibiting DNA replication and transcription. These agents have been used for cancer chemotherapy for a long time; however, acquired resistance remains a serious problem that limits their effectiveness. Although mitochondria have their own DNA, which is distinct from nuclear DNA, the involvement of mitochondria and mitochondrial DNA (mtDNA) in resistance to DNA-damaging anticancer agents remains largely unknown. In this study, we found that the mitochondrial small protein cytochrome c oxidase-associated subunit FA4-like 3 (COXFA4L3), formerly known as C15ORF48, confers resistance to DNA-damaging anticancer agents. Mechanistically, DNA-damaging anticancer agents damage mtDNA and induce the cytosolic release of TFAM-unbound mtDNA via the mitochondrial permeability transition pore (mPTP), thereby promoting cell death via activation of the innate immune signalling cGAS–STING pathway. COXFA4L3 inhibits the cytosolic release of mtDNA by repressing mtDNA damage and mPTP opening. These results suggest that mitochondria affect sensitivity to DNA-damaging anticancer agents through COXFA4L3-mediated repression of mtDNA damage responses.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-54548-3. Introduction DNA-damaging anticancer agents including adriamycin (ADR), etoposide (ETP), cisplatin, bleomycin, and camptothecin have long been used in cancer chemotherapy1–7. These agents selectively kill highly proliferative cancer cells by inducing DNA damage, such as double- or single-strand breaks, and by inhibiting DNA replication and transcription via DNA intercalation, topoisomerase inhibition, or DNA crosslinking. Although these drugs improve the pathological conditions of patients with cancer, acquired resistance is a frequent and serious problem that limits the effectiveness of cancer chemotherapy6,8,9. Therefore, understanding the detailed molecular mechanisms of resistance to DNA-damaging anticancer agents is beneficial for overcoming this problem in patients with cancer. Because nuclear DNA (nDNA) is recognised as the primary target of DNA-damaging anticancer agents, DNA damage responses in the nucleus have been extensively studied8,10,11. nDNA breaks primarily activate the large nuclear kinase ataxia telangiectasia mutated (ATM). This kinase then phosphorylates the histone variant H2AX and activates nDNA damage responses or nDNA repair via the generation of γ-H2AX (a phosphorylated form of H2AX). Importantly, mitochondria have their own DNA (mtDNA), which is distinct from nDNA. mtDNA is a small circular DNA molecule containing genes that encode parts of the energy-producing machinery within the mitochondria12. Mitochondrial transcription factor A (TFAM) is a histone-like protein that packs mtDNA into a condensed structure13. During mtDNA replication and transcription, TFAM is released from the mtDNA, converting it into an uncondensed structure14. Because mtDNA, similar to nDNA, also depends on topoisomerases for replication and transcription, it is likely to be sensitive to DNA-damaging anticancer agents such as topoisomerase inhibitors and DNA intercalators15. However, the effects of DNA-damaging anticancer agents on mtDNA and the role of mitochondria in resistance to these agents in cancer cells are largely unknown. As mitochondria are located around the nucleus16, DNA-damaging anticancer agents may primarily target mtDNA before diffusing into the nucleus. Although the anthracycline type agents, such as ADR, show cardiotoxicity via reactive oxygen species and metabolic disruption17,18, other types agents, such as ETP, does not show so strong cardiotoxicity, suggesting that reactive oxygen species and metabolic disruption are not the primary common mechanisms of DNA-damaging anticancer agents- mediated cytotoxicity. Therefore, the relationship between DNA-damaging anticancer agents and mtDNA damage should be clarified. Cytochrome c oxidase-associated subunit FA4-like 3 (COXFA4L3), formerly C15ORF48, is a mitochondrial small protein that is an accessory subunit of the electron transport chain complex IV19–25. COXFA4L3 lowers cytochrome c oxidase activity and thereby reducing mitochondrial membrane potential22,24. Recent studies revealed that COXFA4L3 is highly expressed in various types of human cancers26,27. These observations strongly suggest that COXFA4L3 is closely associated with cancer progression. However, its role in cancer cells remains unclear. In this study, we examined the relationship between COXFA4L3 and sensitivity to DNA-damaging anticancer agents and found that COXFA4L3 confers resistance to these agents in cancer cells. Mechanistically, COXFA4L3 suppresses mtDNA damage responses, including the cytoplasmic release of mtDNA and activation of the innate immune signalling cyclic GMP–AMP synthase (cGAS)–stimulator of interferon genes (STING) pathway. These results suggest that mitochondria regulate sensitivity to DNA-damaging anticancer agents via COXFA4L3-mediated control of mtDNA damage responses and that COXFA4L3 is a useful marker for predicting the efficacy of cancer chemotherapy. Materials and methods Cell culture A549 and A431 cells were purchased from ATCC and cultured in Dulbecco’s modified Eagle’s medium supplemented with 5% foetal bovine serum at 37 °C. Stable cell pools were established as described previously24. siRNAs (5 nM) were transfected using the reverse transfection method with Lipofectamine RNAiMax (Thermo Fisher Scientific), according to the manufacturer’s instructions. Silencer select siRNAs specific to COXFA4L3 #1 (siL3 #1, s38981) and COXFA4L3 #2 (siL3 #2, s228367) were purchased from Thermo Fisher Scientific. siRNA for luciferase (sense: 5′-GCGCUGCUGGUGCCAACCCTT-3′ and antisense: 5′-GGGUUGGCACCAGCAGCGCTT-3′) was purchased from Hokkaido System Sciences and used as a control. ADR and cyclosporin A were purchased from Fujifilm Wako. ETP, cisplatin, bleomycin, and camptothecin were purchased from Tokyo Chemical Industry. C-176 was purchased from Selleck chemicals. Antibodies The following antibodies were used in this study: ATM (2873T, Cell Signaling Technology), phospho-ATM (5883T, Cell Signaling Technology), γ-H2AX (2577S, Cell Signaling Technology), TFAM (22586-1-AP, Proteintech), TOM20 (42406S, Cell Signaling Technology), STING (19851-1-AP, Proteintech), phospho-STING (19781T, Cell Signaling Technology), CCAAT/enhancer-binding protein homologous protein (CHOP; 15204-1-AP, Proteintech), and actin (M177-3, MBL). The COXFA4L3 monoclonal antibody (clone 5A10) was established as previously described23. Western blotting and subcellular fractionation The cells were washed with phosphate-buffered saline (PBS), lysed with Laemmli buffer, and used as whole-cell lysates. Subcellular fractionation was performed using Mammalian Mitochondria Isolation Kit for Tissue and Cultured Cells (K288-50, BioVision), according to the manufacturer’s instructions. Briefly, cells were gently lysed with hypotonic buffer, and nuclei were removed by gentle centrifugation (600 × g for 10 min at 4 °C). The collected supernatants (mixture of cytosol and mitochondrial fractions) were separated into cytosol (supernatants) and mitochondrial fractions (precipitates) by vigorous centrifugation (7 000 × g for 10 min at 4 °C). The mitochondrial fractions were resuspended in hypotonic buffer (the same amount as the cytosolic fractions). Samples were subjected to sodium dodecyl sulphate (SDS)–polyacrylamide gel electrophoresis and electrotransferred onto polyvinylidene difluoride membranes. The protein bands were treated with appropriate antibodies for detection and analysed using ChemiDoc XRS+ image analyser (Bio-Rad). The band intensity was measured using Quantity One software (Bio-Rad). Quantitative real-time PCR (qPCR) analysis For detection of mRNA expression levels, total RNA was isolated from cells using RNAiso Plus reagent (TaKaRa), and cDNA was synthesised from 0.5 µg of each RNA preparation using ReverTra Ace qPCR RT Kit (TOYOBO, FSQ-101), according to the manufacturer’s instructions. The following primers were used for PCR: GAPDH, 5′-GGA GCG AGA TCC CTC CAA AAT-3′ (sense) and 5′-GGC TGT TGT CAT ACT TCT CAT GG-3′ (antisense); DDIT3 (CHOP), 5′-GGAAACAGAGTGGTCATTCCC-3′ (sense) and 5′-CTGCTTGAGCCGTTCATTCTC-3′ (antisense); IL6, 5′-ACT CAC CTC TTC AGA ACG AAT TG-3′ (sense) and 5′-CCA TCT TTG GAA GGT TCA GGT TG-3′ (antisense); CXCL8 (IL8), 5′-ATG ACT TCC AAG CTG GCC GT-3′ (sense) and 5′-TTA CAT AAT TTC TGT GTT GGC-3′ (antisense); and IL1B, 5′-ATG ATG GCT TAT TAC AGT GGC AA-3′ (sense) and 5′-GTC GGA GAT TCG TAG CTG GA-3′ (antisense). After initial denaturation at 95 °C for 1 min, PCR was performed for 40 cycles (15 s at 95 °C and 45 s at 60 °C) using Thunderbird SYBR Green Polymerase Kit (TOYOBO, QPS-101) and Eco Real-Time PCR System (Illumina). Cell viability MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide) assay Cells were plated in 96-well microplates (20 000 cells/well) and incubated for 24 h. For knockdown experiments, cells were transfected with siRNAs using the reverse transfection method with Lipofectamine RNAiMax at the indicated times. After incubation, cells were treated with or without DNA-damaging anticancer agents for 24 h. After this additional incubation, the solution in each well was replaced with Dulbecco’s modified Eagle’s medium containing 0.5 mg/mL MTT (DOJINDO) and the cells were incubated for an additional 1 h. The resulting crystalised product was dissolved in 100 µL of 100% dimethyl sulfoxide and absorbance was automatically measured at 595 nm using Filter Max F5 microplate reader (Molecular Devices). The quantitative ratio of MTT absorbance in treated cells was normalised to that in untreated cells. The relative value of untreated cells was set at 1.0. These experiments were repeated three or four times using different samples. Results are shown as the mean ± SD of three samples. mtDNA copy number analysis For the analysis of mtDNA copy number in whole cells, total DNA (nDNA and mtDNA) was extracted using phenol–chloroform, precipitated with ethanol, and suspended in 50 µL of TE buffer (10 mM Tris–HCl and 1 mM EDTA; pH 8.0). mtDNA copy number in whole cells was detected by evaluating the relative copy number of mtDNA gene CYTB normalised to that of the nDNA gene FBXO15 by qPCR analyses using the following primers: CYTB, 5′-TGC AAC TAT AGC AAC AGC CTT CA-3′ (sense) and 5′-GAA CTA GGT CTG TCC CAA TGT ATG G-3′ (antisense); FBXO15, 5′-GCC AGG AGG TCT TCG CTG TA-3′ (sense) and 5′-AAT GCA CGG CTA GGG TCA AA-3′ (antisense). For the analysis of cytoplasmic release of mtDNA, DNA molecules in cytosolic fractions and mixtures of cytosolic and mitochondrial fractions prepared by subcellular fractionation were extracted by phenol–chloroform, precipitated with ethanol, and suspended in 50 µL of TE buffer. The mtDNA copy number in the cytosol was detected by evaluating the copy number of CYTB in cytosolic fractions normalised to that in mixtures of cytosolic and mitochondrial fractions via qPCR analyses. Chromatin immunoprecipitation (ChIP) assay ChIP assays were performed as previously described28. Briefly, the cells were suspended in PBS at 2 × 106 cells/mL and fixed with 1% formaldehyde for 10 min at 25 °C, after which fixation was halted by adding 150 mM glycine. Fixed cells were washed with PBS and lysed with 200 µL of lysis buffer (50 mM Tris–HCl, 10 mM EDTA, and 1% SDS; pH 7.4). Cell lysates were sonicated using a bath sonicator (Bioruptor) for three cycles (power setting: high; on, 30 s; off, 1 min; and 4 °C). The sonicated lysates were centrifuged for 10 min at 10 000 × g, and 1 800 µL of dilution buffer (50 mM Tris–HCl, 167 mM NaCl, 1.1% Triton X-100, and 0.11% sodium deoxycholate; pH 8.0) was added to the supernatants (chromatin solutions). The chromatin solution (2 mL) was pre-cleared with 80 µL of 50% protein G–Sepharose slurry pre-absorbed with 100 µg/mL sonicated salmon sperm DNA, aliquoted, and incubated with 4 µL of anti-TFAM antibody (22586-1-AP, Proteintech) or 4 µg of normal rabbit IgG (148–09551, Fujifilm Wako) overnight. Immunoprecipitates were recovered using 20 µL of 50% protein G–Sepharose/salmon sperm DNA for 2 h, washed sequentially with RIPA buffer (50 mM Tris–HCl, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, and 0.1% SDS; pH 8.0), RIPA buffer containing 500 mM NaCl, and TE buffer (10 mM Tris–HCl and 1 mM EDTA; pH 8.0), and then resuspended in 200 µL of elution buffer (10 mM Tris–HCl, 300 mM NaCl, 5 mM EDTA, and 0.5% SDS; pH 8.0). Beads and an input fraction saved before pre-clearing were incubated at 65 °C for at least 6 h. DNA was extracted with phenol–chloroform, precipitated with ethanol, and suspended in 50 µL of TE buffer. The extracted DNA was subjected to qPCR analysis as described above. The following PCR primers for mtDNA genes were used: ND1, 5′-CTACTACAACCCTTCGCTGAC-3′ (sense) and 5′-GGATTGAGTAAACGGCTAGGC-3′ (antisense); ND3, 5′-GCGGCTTCGACCCTATATCC-3′ (sense) and 5′-AGGGCTCATGGTAGGGGTAA-3′ (antisense); ND5, 5′-TCGAATAATTCTTCTCACCC-3′ (sense) and 5′-TAGTAATGAGAAATCCTGCG-3′ (antisense); and ND6, 5′-GTAGGATTGGTGCTGTGG-3′ (sense) and 5′-GGATCCTCCCGAATCAAC-3′ (antisense). Percent input values were calculated by comparing Ct values of the input and immunoprecipitated fractions and are shown as ratios relative to those of the control samples. mtDNA damage analysis mtDNA damage was detected using long-range PCR. Briefly, DNA molecules were extracted from whole cells using phenol–chloroform, precipitated with ethanol, and suspended in 50 µL of TE buffer. mtDNA fragments were amplified by PCR using KOD-Multi & Epi-DNA polymerase (TOYOBO), according to the manufacturer’s instructions. Primer sequences were as follows: ND1–ND3 long-range fragments, 5′-GCG GCT TCG ACC CTA TAT CC-3′ (ND3_sense) and 5′-GGA TTG AGT AAA CGG CTA GGC-3′ (ND1_antisense); and CYTB short-range fragments, 5′-TGC AAC TAT AGC AAC AGC CTT CA-3′ (sense) and 5′-GAA CTA GGT CTG TCC CAA TGT ATG G-3′ (antisense). The amplification of the long fragments was carried out with the following thermocycling conditions: initial denaturation at 94 °C for 2 min, 20 touchdown cycles (98 °C for 10 s, 68 °C for 10 s, and 68 °C for 6 min) (with the annealing temperature lowered by 0.5 °C in each cycle), and 10 amplification cycles (98 °C for 10 s, 55 °C for 10 s, and 68 °C for 6 min). The amplification of short fragments was performed similarly, except that the elongation time was 30 s. The amplified fragments were analysed on a 1% agarose gel containing ethidium bromide. The bands were detected using ChemiDoc XRS+ image analyser. The band intensity was measured using Quantity One software. Immunocytochemistry Images were obtained using LSM 780 confocal laser scanning microscope with a 63×/1.40 NA oil-immersion objective (Zeiss). A single planar (xy) slice (1.0-µm thickness) is shown for all experiments. The cells were fixed in 4% paraformaldehyde for 20 min at room temperature and permeabilised in PBS containing 0.2% Triton X-100 and 3% bovine serum albumin at room temperature. Cells were subsequently incubated with anti-TFAM (22586-1-AP, Proteintech) or anti-dsDNA (sc-58749, Santa Cruz Biotechnology) antibody for 1 h, washed with PBS containing 0.1% saponin, and stained with Alexa Fluor 488- or 594-secondary antibody for 1 h. Mitochondria were stained with 100 nM MitoTracker CMXRos (Thermo Fisher Scientific). The stained cells were mounted using ProLong Antifade reagent (Thermo Fisher Scientific). Mitochondrial permeability transition pore (mPTP) opening detection The opening of the mPTP was determined using calcein–CoCl2 staining method and confocal microscopy, according to a previously reported protocol29. Briefly, treated cells were incubated with 2 µM calcein acetoxymethyl ester (calcein-AM) and 100 nM MitoTracker CMXRos for 30 min in the dark. The cells were then washed twice with PBS and exposed to 2 mM CoCl2 for 10 min. Calcein fluorescence remains compartmentalised within the mitochondria until the opening of the mPTP permits the distribution of cobalt into the mitochondria, which results in the quenching of calcein fluorescence in the mitochondrial matrix. After three washes with PBS, fluorescence imaging of the cells was performed using LSM 780 confocal laser scanning microscope. Digital images were analysed using ZEN software (Zeiss), and the degree of mPTP opening was reflected by the red (MitoTracker)/green (calcein) fluorescence ratio. Trypan blue exclusion Viable and dead cells were counted using trypan blue and the percentage of cell death was calculated as follows: cell death (%) = (dead cell number/total cell number) × 100. Quantification and statistical analysis Graphs are presented as mean ± SD, as indicated in the figure legends. Two-tailed unpaired Student’s t-test were used for comparison between two groups. Comparisons between more than two groups were performed using Tukey’s multiple comparison test. Statistical significance was set at p < 0.05. P values less than 0.001 are shown as p < 0.001. Results COXFA4L3 expression confers resistance to DNA-damaging anticancer agents in cancer cells To analyse the effects of COXFA4L3 expression on cancer progression, we established human lung cancer A549 cells stably expressing COXFA4L3 (A549-L3) and treated them with various DNA-damaging anticancer agents, such as ADR, ETP, cisplatin, bleomycin, and camptothecin. A549-L3 cells showed higher viability than that of control cells after treatment with these anticancer agents (Fig. 1a, b). A549-L3 cells showed reduced cell death in the presence of ADR or ETP (Fig. 1c). We knocked down endogenous COXFA4L3 in human epidermal cancer A431 cells, which have higher COXFA4L3 expression than that of A549 cells (Fig. 1d, e), and found that COXFA4L3 knockdown significantly reduced the viability of ADR- and ETP-treated A431 cells (Fig. 1f). COXFA4L3 knockdown also increased cell death of A431 cells in the presence of ADR and ETP (Fig. 1g). These results suggest that COXFA4L3 expression confers resistance to DNA-damaging anticancer agents. Fig. 1COXFA4L3 expression confers resistance to DNA-damaging anticancer agents in cancer cells. (a) A549 cells stably expressing COXFA4L3 (A549-L3) or an empty vector (control) were lysed and subjected to western blotting with the indicated antibodies. (b) A549-L3 or control cells (20 000 cells/well) were treated with or without the indicated anticancer agents for 24 h and subjected to an MTT assay. The quantitative ratio of MTT absorbance in treated cells normalised to that in untreated cells is shown as the mean ± SD. Statistical significance was calculated using a two-tailed unpaired Student’s t-test (n = 4 biological replicates). (c) A549-L3 and control cells were treated with ADR or ETP for 24 h. The percentage of dead cells was determined using the trypan blue exclusion method. Statistical significance was calculated using a two-tailed unpaired Student’s t-test (n = 3 biological replicates). (d) A549 and A431 cells were lysed and subjected to western blotting using the indicated antibodies. (e) A431 cells were transfected with the indicated siRNAs for 48 h, lysed, and subjected to western blotting with the indicated antibodies. (f) A431 cells (20 000 cells/well) were transfected with the indicated siRNAs for 48 h. After incubation, the cells were treated with or without ADR or ETP for 24 h and subjected to MTT assay. The quantitative ratio of MTT absorbance in treated cells normalised to that in untreated cells is shown as the mean ± SD. Statistical significance was calculated using Tukey’s multiple comparison test (n = 3 biological replicates). (g) A431 cells were transfected with the indicated siRNAs for 48 h and then treated with ADR or ETP for 24 h. The percentage of dead cells was determined using the trypan blue exclusion method. Statistical significance was calculated using Tukey’s multiple comparison test (n = 3 biological replicates). COXFA4L3 expression confers resistance to DNA-damaging anticancer agents in cancer cells. (a) A549 cells stably expressing COXFA4L3 (A549-L3) or an empty vector (control) were lysed and subjected to western blotting with the indicated antibodies. (b) A549-L3 or control cells (20 000 cells/well) were treated with or without the indicated anticancer agents for 24 h and subjected to an MTT assay. The quantitative ratio of MTT absorbance in treated cells normalised to that in untreated cells is shown as the mean ± SD. Statistical significance was calculated using a two-tailed unpaired Student’s t-test (n = 4 biological replicates). (c) A549-L3 and control cells were treated with ADR or ETP for 24 h. The percentage of dead cells was determined using the trypan blue exclusion method. Statistical significance was calculated using a two-tailed unpaired Student’s t-test (n = 3 biological replicates). (d) A549 and A431 cells were lysed and subjected to western blotting using the indicated antibodies. (e) A431 cells were transfected with the indicated siRNAs for 48 h, lysed, and subjected to western blotting with the indicated antibodies. (f) A431 cells (20 000 cells/well) were transfected with the indicated siRNAs for 48 h. After incubation, the cells were treated with or without ADR or ETP for 24 h and subjected to MTT assay. The quantitative ratio of MTT absorbance in treated cells normalised to that in untreated cells is shown as the mean ± SD. Statistical significance was calculated using Tukey’s multiple comparison test (n = 3 biological replicates). (g) A431 cells were transfected with the indicated siRNAs for 48 h and then treated with ADR or ETP for 24 h. The percentage of dead cells was determined using the trypan blue exclusion method. Statistical significance was calculated using Tukey’s multiple comparison test (n = 3 biological replicates). COXFA4L3 protects mtDNA from DNA damages caused by DNA-damaging anticancer agents As the primary target of DNA-damaging anticancer agents is nDNA, we assessed the activation of nDNA damage responses by detecting the activation (phosphorylation) of the nuclear kinase ATM and the subsequent phosphorylation of histone H2AX (γ-H2AX). ADR- and ETP-induced activation of nDNA damage responses occurred similarly in A549-L3 and control cells (Fig. 2a), suggesting that COXFA4L3 is not involved in nDNA damage responses. Next, we examined whether COXFA4L3 affects DNA damage in mtDNA. Although the amplification of short-range mtDNA fragments (CYTB fr: 122 bp) was not affected, amplification of long-range fragments (ND1-ND3 fr: 10 kbp) was impaired in the presence of ADR or ETP (Fig. 2b). Interestingly, COXFA4L3 expression ameliorated the amplification of long-range fragments (Fig. 2c, d), suggesting that DNA-damaging anticancer agents induce mtDNA damage and that COXFA4L3 protects mtDNA from this damage. Fig. 2COXFA4L3 protects mtDNA from DNA damages caused by DNA-damaging anticancer agents. (a) A549-L3 and control cells were treated with ADR or ETP for the indicated times. After incubation, cells were lysed and subjected to western blotting with the indicated antibodies. (b) Schematics of mtDNA and the loci of ND1, ND3, and CYTB are shown. Red and green lines represent long-range fragment of ND1–ND3 (ND1–ND3 fr.) and short-range fragments of CYTB (CYTB fr.), respectively. (c) A549-L3 and control cells were treated with or without ADR or ETP for 3 h. After incubation, DNA was extracted from the whole cells and subjected to PCR to detect mtDNA damage. (d) Quantitative ratios of long-range to short-range fragments were calculated based on the data shown in (c) and are shown as ratios relative to those of untreated cells. Statistical significance was calculated using Tukey’s multiple comparison test (n = 3 biological replicates). COXFA4L3 protects mtDNA from DNA damages caused by DNA-damaging anticancer agents. (a) A549-L3 and control cells were treated with ADR or ETP for the indicated times. After incubation, cells were lysed and subjected to western blotting with the indicated antibodies. (b) Schematics of mtDNA and the loci of ND1, ND3, and CYTB are shown. Red and green lines represent long-range fragment of ND1–ND3 (ND1–ND3 fr.) and short-range fragments of CYTB (CYTB fr.), respectively. (c) A549-L3 and control cells were treated with or without ADR or ETP for 3 h. After incubation, DNA was extracted from the whole cells and subjected to PCR to detect mtDNA damage. (d) Quantitative ratios of long-range to short-range fragments were calculated based on the data shown in (c) and are shown as ratios relative to those of untreated cells. Statistical significance was calculated using Tukey’s multiple comparison test (n = 3 biological replicates). COXFA4L3 inhibits mPTP-mediated cytosolic release of mtDNA in the presence of DNA-damaging anticancer agents We first confirmed that COXFA4L3 expression did not affect mtDNA copy number in whole cells (Fig. 3a). Next, we analysed the cytosolic release of mtDNA, which is induced by various mitochondrial stressors30,31. qPCR analyses showed that ADR or ETP treatment significantly increased the cytosolic release of mtDNA and that this increase was repressed by COXFA4L3 expression (Fig. 3b). Anti-double stranded DNA (dsDNA) antibodies showed that ADR or ETP treatment significantly increased dsDNA signals in the cytosol and that these signals were reduced in A549-L3 cells (Fig. 3c, d), supporting the hypothesis that COXFA4L3 suppresses ADR- or ETP-induced cytosolic release of mtDNA. We examined the involvement of the mitochondrial inner membrane pore mPTP in the cytoplasmic release of mtDNA. The mPTP inhibitor cyclosporin A (CycA) significantly repressed the cytosolic release of mtDNA (Fig. 3e). We further examined mPTP opening by detecting calcein fluorescence in the mitochondria, which was quenched by cobalt ions flowing into the mitochondria via the mPTP. ADR and ETP treatments significantly reduced calcein fluorescence in the mitochondria of control cells, whereas this reduction was repressed in A549-L3 cells (Fig. 3f, g). These results suggest that ADR or ETP treatment opens the mPTP and that COXFA4L3 suppresses mPTP opening. Fig. 3COXFA4L3 inhibits mPTP-mediated cytosolic release of mtDNA in the presence of DNA-damaging anticancer agents. (a) Total DNA was extracted from A549-L3 and control cells and subjected to qPCR analyses to determine mtDNA copy number in whole cells. Data are shown as ratios relative to those of the control cells (n = 12 biological replicates). (b) A549-L3 and control cells were treated with or without ADR or ETP for 3 h. After incubation, cytosolic fractions were prepared by subcellular fractionation. DNA molecules were extracted from these fractions and subjected to qPCR analysis to determine the mtDNA copy number in the cytosol. Data are shown as ratios relative to those of the untreated control cells. Statistical significance was calculated using Tukey’s multiple comparison test (n = 3 biological replicates). (c) A549-L3 and control cells were treated with or without ADR (0.3 µg/mL, 3 h). After incubation, the cells were fixed and stained with an anti-dsDNA antibody and MitoTracker. Representative images from two independent experiments are shown (left). The number of dsDNA dots in the cytosol of each cell was calculated and shown as a box plot. Statistical significance was calculated using Tukey’s multiple comparison test (n = 25 cells from two independent experiments) (right). (d) As in (c), except that ETP (20 µM) was used. (e) A549-L3 and control cells were pre-treated with or without CycA for 12 h and then treated with or without ADR (left) or ETP (right) for 3 h. After incubation, the mtDNA copy number in the cytosol were examined, as shown in (b). Statistical significance was calculated using Tukey’s multiple comparison test (n = 3 biological replicates). (f) A549-L3 and control cells were treated with or without ADR (0.3 µg/mL, 3 h). After incubation, the cells were subjected to mPTP opening detection using calcein-AM and MitoTracker CMXRos. Representative images from two independent experiments are shown (left). The degrees of mPTP opening are reflected by the red (MitoTracker)/green (calcein) fluorescence ratios and are shown as values relative to those of the untreated control cells in box plots. Statistical significance was calculated using Tukey’s multiple comparison test (n = 15 cells from two independent experiments) (right). (g) As in (e), except that ETP (20 µM) was used. COXFA4L3 inhibits mPTP-mediated cytosolic release of mtDNA in the presence of DNA-damaging anticancer agents. (a) Total DNA was extracted from A549-L3 and control cells and subjected to qPCR analyses to determine mtDNA copy number in whole cells. Data are shown as ratios relative to those of the control cells (n = 12 biological replicates). (b) A549-L3 and control cells were treated with or without ADR or ETP for 3 h. After incubation, cytosolic fractions were prepared by subcellular fractionation. DNA molecules were extracted from these fractions and subjected to qPCR analysis to determine the mtDNA copy number in the cytosol. Data are shown as ratios relative to those of the untreated control cells. Statistical significance was calculated using Tukey’s multiple comparison test (n = 3 biological replicates). (c) A549-L3 and control cells were treated with or without ADR (0.3 µg/mL, 3 h). After incubation, the cells were fixed and stained with an anti-dsDNA antibody and MitoTracker. Representative images from two independent experiments are shown (left). The number of dsDNA dots in the cytosol of each cell was calculated and shown as a box plot. Statistical significance was calculated using Tukey’s multiple comparison test (n = 25 cells from two independent experiments) (right). (d) As in (c), except that ETP (20 µM) was used. (e) A549-L3 and control cells were pre-treated with or without CycA for 12 h and then treated with or without ADR (left) or ETP (right) for 3 h. After incubation, the mtDNA copy number in the cytosol were examined, as shown in (b). Statistical significance was calculated using Tukey’s multiple comparison test (n = 3 biological replicates). (f) A549-L3 and control cells were treated with or without ADR (0.3 µg/mL, 3 h). After incubation, the cells were subjected to mPTP opening detection using calcein-AM and MitoTracker CMXRos. Representative images from two independent experiments are shown (left). The degrees of mPTP opening are reflected by the red (MitoTracker)/green (calcein) fluorescence ratios and are shown as values relative to those of the untreated control cells in box plots. Statistical significance was calculated using Tukey’s multiple comparison test (n = 15 cells from two independent experiments) (right). (g) As in (e), except that ETP (20 µM) was used. DNA-damaging anticancer agents preferentially target TFAM-unbound forms of mtDNA As TFAM-bound mtDNA does not activate inflammatory responses due to its degradation by autophagy in the cytosol32, we examined whether DNA-damaging anticancer agents selectively induce the cytosolic release of TFAM-bound or -unbound forms of mtDNA. TFAM protein levels were unaffected by ADR or ETP treatment in control and A549-L3 cells (Fig. 4a). Furthermore, immunocytochemical and subcellular fractionation analyses showed that ADR or ETP treatment did not increase TFAM protein levels in the cytosol of control and A549-L3 cells (Fig. 4b, c), suggesting that TFAM proteins are not released into the cytosol by DNA-damaging anticancer agents. To evaluate the possibility that anticancer agents promote dissociation between TFAM and mtDNA, we performed ChIP assays between TFAM and mtDNA. ChIP assays showed that ADR or ETP treatment did not induce separation between TFAM and mtDNA (Fig. 4d). These results suggest that DNA-damaging anticancer agents preferentially target TFAM-unbound forms of mtDNA and promote their release into the cytosol. Fig. 4DNA-damaging anticancer agents preferentially target TFAM-unbound forms of mtDNA. (a) A549-L3 and control cells were treated with or without ADR (top) or ETP (bottom) for 3 h, lysed, and subjected to western blotting with the indicated antibodies. (b) A549-L3 and control cells were treated with or without ADR (left) or ETP (right) for 3 h. After incubation, the cells were fixed and stained with an anti-TFAM antibody and MitoTracker. Representative images from two independent experiments are shown. (c) A549-L3 and control cells were treated with or without ADR (left) or ETP (right) for 3 h. After incubation, subcellular fractionation was performed and the cytosolic and mitochondrial fractions were subjected to western blotting with the indicated antibodies. (d) A549 cells were treated with or without ADR (top) or ETP (bottom) for 3 h. After incubation, ChIP assays were performed to examine the association between TFAM and mtDNA. The percentage input values are shown as ratios relative to those of the control immunoprecipitates from untreated cells. Statistical significance was calculated using Tukey’s multiple comparison test (n = 3 biological replicates). DNA-damaging anticancer agents preferentially target TFAM-unbound forms of mtDNA. (a) A549-L3 and control cells were treated with or without ADR (top) or ETP (bottom) for 3 h, lysed, and subjected to western blotting with the indicated antibodies. (b) A549-L3 and control cells were treated with or without ADR (left) or ETP (right) for 3 h. After incubation, the cells were fixed and stained with an anti-TFAM antibody and MitoTracker. Representative images from two independent experiments are shown. (c) A549-L3 and control cells were treated with or without ADR (left) or ETP (right) for 3 h. After incubation, subcellular fractionation was performed and the cytosolic and mitochondrial fractions were subjected to western blotting with the indicated antibodies. (d) A549 cells were treated with or without ADR (top) or ETP (bottom) for 3 h. After incubation, ChIP assays were performed to examine the association between TFAM and mtDNA. The percentage input values are shown as ratios relative to those of the control immunoprecipitates from untreated cells. Statistical significance was calculated using Tukey’s multiple comparison test (n = 3 biological replicates). Cytosolic mtDNA induces cell death via cGAS–STING pathway As cytosolic DNA activates the innate immune signalling cGAS–STING pathway33, we examined the activation of this pathway after treatment with ADR or ETP. STING phosphorylation, a hallmark of STING activation34, was significantly increased by ADR and ETP treatment in control cells. In A549-L3 cells, this phosphorylation was not elevated even in the presence of ADR or ETP (Fig. 5a). Interestingly, the STING inhibitor C-176 significantly recovered the ADR- and ETP-dependent reductions in viability and cell death in control cells (Fig. 5b, c). In A549-L3 cells, C-176 did not affect the cell viability or death. ADR and ETP promoted the expression of the pro-apoptotic transcription factor CHOP in control cells35, and CHOP induction was significantly repressed in A549-L3 cells (Fig. 5d, e). Interestingly, ADR- and ETP-induced CHOP expression was significantly suppressed by CycA and C-176 (Fig. 5f). These results suggest that DNA-damaging anticancer agents activate the cGAS–STING pathway via the cytosolic release of mtDNA and then induce cell death, probably through the cGAS–STING pathway-mediated induction of CHOP. Fig. 5Cytosolic mtDNA induces cell death via cGAS–STING pathway. (a) A549-L3 and control cells were treated with ADR (left) or ETP (right) for the indicated times. After incubation, cells were lysed and subjected to western blotting with the indicated antibodies. (b) A549-L3 and control cells (20 000 cells/well) were pre-treated with or without C-176 for 3 h. After incubation, the cells were treated with or without ADR (left) or ETP (right) for 24 h and subjected to the MTT assay. The quantitative ratios of MTT absorbance in treated cells normalised to those in untreated cells are shown as the mean ± SD. Statistical significance was calculated using Tukey’s multiple comparison test (n = 4 biological replicates). (c) A549-L3 and control cells were pre-treated with or without C-176 for 3 h and then treated with ADR (left) or ETP (right) for 24 h. The percentage of dead cells was determined using the trypan blue exclusion method. Statistical significance was calculated using Tukey’s multiple comparison test (n = 3 biological replicates). (d) A549-L3 and control cells were treated with or without ADR (left) or ETP (right) for 3 h. After incubation, cells were lysed and subjected to western blotting with the indicated antibodies. (e) A549-L3 and control cells were treated with or without ADR (left) or ETP (right) for 3 h. After incubation, total RNA was extracted, and the expression levels of CHOP mRNA were measured using reverse transcription and subsequent qPCR analysis. GAPDH mRNA expression was used to normalise the data. The expression level of CHOP mRNA in the untreated control cells was set to 1. The result indicates the mean ± SD. Statistical significance was calculated using Tukey’s multiple comparisons test (n = 3 biological replicates). (f) A549-L3 and control cells were pre-treated with CycA for 12 h or C-176 for 3 h and then treated with or without ADR or ETP for 3 h. After incubation, cells were lysed and subjected to western blotting with the indicated antibodies. Cytosolic mtDNA induces cell death via cGAS–STING pathway. (a) A549-L3 and control cells were treated with ADR (left) or ETP (right) for the indicated times. After incubation, cells were lysed and subjected to western blotting with the indicated antibodies. (b) A549-L3 and control cells (20 000 cells/well) were pre-treated with or without C-176 for 3 h. After incubation, the cells were treated with or without ADR (left) or ETP (right) for 24 h and subjected to the MTT assay. The quantitative ratios of MTT absorbance in treated cells normalised to those in untreated cells are shown as the mean ± SD. Statistical significance was calculated using Tukey’s multiple comparison test (n = 4 biological replicates). (c) A549-L3 and control cells were pre-treated with or without C-176 for 3 h and then treated with ADR (left) or ETP (right) for 24 h. The percentage of dead cells was determined using the trypan blue exclusion method. Statistical significance was calculated using Tukey’s multiple comparison test (n = 3 biological replicates). (d) A549-L3 and control cells were treated with or without ADR (left) or ETP (right) for 3 h. After incubation, cells were lysed and subjected to western blotting with the indicated antibodies. (e) A549-L3 and control cells were treated with or without ADR (left) or ETP (right) for 3 h. After incubation, total RNA was extracted, and the expression levels of CHOP mRNA were measured using reverse transcription and subsequent qPCR analysis. GAPDH mRNA expression was used to normalise the data. The expression level of CHOP mRNA in the untreated control cells was set to 1. The result indicates the mean ± SD. Statistical significance was calculated using Tukey’s multiple comparisons test (n = 3 biological replicates). (f) A549-L3 and control cells were pre-treated with CycA for 12 h or C-176 for 3 h and then treated with or without ADR or ETP for 3 h. After incubation, cells were lysed and subjected to western blotting with the indicated antibodies. COXFA4L3 represses cGAS–STING pathway-dependent inflammatory responses in the presence of DNA-damaging anticancer agents Because the cGAS–STING pathway activates inflammatory responses, we examined the expression of proinflammatory cytokines. ADR and ETP treatment significantly induced the expression of IL6, CXCL8, and IL1B in control cells, and this induction was reduced in A549-L3 cells (Fig. 6a, b). ADR- or ETP-induced expression of IL6 and CXCL8 was significantly repressed by CycA or C-176 (Fig. 6b, c), suggesting that cytokine expression depends on the cGAS–STING pathway induced by mtDNA in the cytosol and that COXFA4L3 suppresses the activation of this inflammatory response. Fig. 6COXFA4L3 represses cGAS–STING pathway-dependent inflammatory responses induced by DNA-damaging anticancer agents. (a) A549-L3 and control cells were treated with or without ADR for 24 h. After incubation, total RNA was extracted, and the expression levels of the indicated genes were measured using reverse transcription and subsequent qPCR analysis. GAPDH mRNA expression was used to normalise the data. The expression level of each mRNA in the untreated control cells was set to 1. The result indicates the mean ± SD. Statistical significance was calculated using Tukey’s multiple comparisons test (n = 3 biological replicates). (b) As in (a), except that ETP was used. (c) A549 cells were pre-treated with or without CycA for 12 h and then treated with or without ADR or ETP for 3 h. After incubation, total RNA was extracted and the expression levels of the indicated genes were measured using reverse transcription and subsequent qPCR analysis, as described in (a). (d) As in (c), except that C-176 was used for 3 h. COXFA4L3 represses cGAS–STING pathway-dependent inflammatory responses induced by DNA-damaging anticancer agents. (a) A549-L3 and control cells were treated with or without ADR for 24 h. After incubation, total RNA was extracted, and the expression levels of the indicated genes were measured using reverse transcription and subsequent qPCR analysis. GAPDH mRNA expression was used to normalise the data. The expression level of each mRNA in the untreated control cells was set to 1. The result indicates the mean ± SD. Statistical significance was calculated using Tukey’s multiple comparisons test (n = 3 biological replicates). (b) As in (a), except that ETP was used. (c) A549 cells were pre-treated with or without CycA for 12 h and then treated with or without ADR or ETP for 3 h. After incubation, total RNA was extracted and the expression levels of the indicated genes were measured using reverse transcription and subsequent qPCR analysis, as described in (a). (d) As in (c), except that C-176 was used for 3 h. Discussion In this study, we found that the mitochondrial small protein, COXFA4L3, confers resistance to DNA-damaging anticancer agents. Mechanistically, these agents damage mtDNA and induce the cytosolic release of TFAM-unbound mtDNA via mPTP, thereby promoting cGAS–STING pathway-dependent cell death. COXFA4L3 induces resistance to anticancer agents by protecting mtDNA from damage and inhibiting mPTP opening. These results suggest that mitochondria affect sensitivity to DNA-damaging anticancer agents through COXFA4L3-mediated suppression of mtDNA damage responses (Fig. 7). Fig. 7Schematic model of COXFA4L3-mediated repression of mtDNA damage responses in the presence of DNA-damaging anticancer agents. DNA-damaging anticancer agents preferentially target TFAM-unbound forms of mtDNA and induce their cytoplasmic release through the mPTP, thereby activating cGAS–STING pathway-dependent cell death and inflammatory responses. COXFA4L3 represses mtDNA damage responses, including mtDNA damage and cytoplasmic release, and protects cancer cells from cell death. Schematic model of COXFA4L3-mediated repression of mtDNA damage responses in the presence of DNA-damaging anticancer agents. DNA-damaging anticancer agents preferentially target TFAM-unbound forms of mtDNA and induce their cytoplasmic release through the mPTP, thereby activating cGAS–STING pathway-dependent cell death and inflammatory responses. COXFA4L3 represses mtDNA damage responses, including mtDNA damage and cytoplasmic release, and protects cancer cells from cell death. COXFA4L3 is an accessory protein of the mitochondrial electron transport chain complex IV that lowers mitochondrial membrane potential by reducing cytochrome c oxidase activity. Although the molecular mechanisms by which COXFA4L3 regulates mtDNA damage and mPTP opening remain unclear, we hypothesised that reduced mitochondrial activity is involved in these phenomena. Our present data show that DNA-damaging anticancer agents preferentially target TFAM-unbound forms of mtDNA. During mtDNA replication or transcription, TFAM dissociates from mtDNA and converts its structure into an uncondensed form14. Therefore, the COXFA4L3-mediated reduction in mitochondrial activity might lower the replication or transcription of mtDNA and increase TFAM-bound condensed forms of mtDNA, which are thought to be insensitive to anticancer agents. Additionally, because the mPTP is associated with electron transport chain complexes and related proteins36, the COXFA4L3-mediated reduction in mitochondrial activity is likely to be closely associated with mPTP opening. Our data show that DNA-damaging anticancer agents damage mtDNA and induce mtDNA damage responses, including the cytoplasmic release of mtDNA via mPTP and activation of the cGAS–STING pathway. Our data also show that the TFAM-unbound forms of mtDNA were released into the cytosol. Because TFAM-bound condensed mtDNA is degraded by autophagy in the cytosol32, TFAM-unbound mtDNA is likely released into the cytosol to activates the cGAS–STING pathway in the presence of DNA-damaging anticancer agents. We speculate that mitochondria have a mechanism that promotes the cytosolic release of TFAM-unbound forms of damaged mtDNA. A component of the mPTP may selectively recognise damaged mtDNA and promote cytoplasmic release. Analyses of mPTP proteins may be helpful for understanding the mechanisms of cytosolic release of damaged mtDNA. COXFA4L3 is highly expressed in various types of cancer tissues by unidentified mechanisms26,27. Previous studies have shown that COXFA4L3 is a target of the transcription factor NF-κB and that NF-κB is constitutively active in various cancer cells24,37,38. Therefore, high expression of COXFA4L3 likely depends on constitutively active NF-κB in cancer cells. NF-κB is known to cause resistance to anticancer agents by inducing anti-apoptotic proteins39. Therefore, COXFA4L3-mediated repression of mtDNA damage responses may represent an additional mechanism underlying NF-κB-induced resistance to anticancer agents in cancer cells. Our data show that the cytoplasmic mtDNA-activated cGAS–STING pathway is associated with cell death in the presence of DNA-damaging anticancer agents. Although the detailed molecular relationship between the cGAS–STING pathway and cell death is unclear, the pro-apoptotic transcription factor CHOP has been suggested to be involved in cell death mediated by this pathway. Recent studies have revealed that STING can activate Z-DNA binding protein 1-mediated necroptosis40. Therefore, mtDNA-activated cGAS–STING pathway may induce necroptosis concurrently with CHOP-dependent apoptosis. It has been thought that DNA-damaging anticancer agents kill cancer cells by damaging nDNA. Our data have shown that mtDNA damage is also involved in these anticancer agents-mediated cell death. DNA damage-dependent cell death is likely to be a system to prevent abnormal cell survival. Because nDNA has repair mechanisms that repress nDNA damage41, nDNA repairment has a potential to contribute to abnormal cell survival. Therefore, we hypothesize that mtDNA damage is a compensatory mechanism that warrant abnormal cell death upon DNA damage stress. Our data suggest that TFAM-unbound mtDNA is selectively released into the cytosol via mPTP. Although a molecular mechanism of this phenomenon is still enigmatic, we hypothesize that an unidentified protein binds to TFAM-unbound mtDNA in response to damaging stress and thereby promotes association between mtDNA and mPTP. Proteomic analysis of proteins associating with TFAM-unbound mtDNA may be useful to detect a protein that promotes cytosolic release of mtDNA. In conclusion, we found that COXFA4L3 induces resistance to DNA-damaging anticancer agents in cancer cells by repressing mtDNA damage and cytoplasmic release. Our data suggest that COXFA4L3 is a useful therapeutic marker for predicting the effectiveness of anticancer agent-based chemotherapy in patients with cancer. Furthermore, medications that repress the role of COXFA4L3 are expected to overcome resistance to these anticancer agents in patients with cancer. Electronic Supplementary Material Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1