INTRODUCTION
The liver serves as one of the principal detoxification organs in the human body, playing a critical role in the metabolism and elimination of toxins, drugs, and other deleterious substances. Nevertheless, the metabolic processing of drugs can also induce hepatic injury, a condition referred to as Drug-Induced Liver Injury (DILI)
5,6. Acetaminophen (APAP), a widely utilized component in cold medications, can provoke extensive hepatocyte necrosis when consumed in excessive amounts. The central mechanism underlying its hepatotoxicity is mitochondrial dysfunction
7. APAP is primarily metabolized by the cytochrome P450 enzyme CYP2E1, yielding the highly reactive and toxic metabolite N-acetyl-p-benzoquinone imine (NAPQI)
8. NAPQI covalently binds to mitochondrial inner membrane proteins, disrupting the electron transport chain (ETC), altering mitochondrial membrane permeability, and triggering the overproduction of reactive oxygen species (ROS)
9,10. These events culminate in mitochondrial impairment and subsequent hepatocyte necrosis. During the initial phase of APAP intoxication, the abundant availability of glutathione (GSH) facilitates the detoxification of NAPQI, thereby attenuating its direct deleterious effects on mitochondria
11. However, beyond 8-h post-APAP overdose, mitochondrial damage and ROS-driven oxidative stress initiate a cascade of extensive secondary injury to hepatocytes. This secondary damage cannot be mitigated by GSH administration beyond this critical 8-h window, leading to irreversible and widespread hepatocyte death and the onset of acute liver failure (ALF)
1. Therefore, elucidating the core pathological mechanisms underlying the secondary damage induced by APAP overdose is pivotal for developing strategies to reduce the incidence of APAP-induced ALF.
RESULTS
Cytoplasmic mtDNA activates ZBP1 to drive hepatocyte death
Mitochondrial damage-induced leakage of mitochondrial DNA (mtDNA) into the cytoplasm is a key trigger of immune responses and cell death, contributing to the pathogenesis of various diseases
12. Similarly, APAP-induced mitochondrial injury leads to the cytosolic release of mtDNA. In both APAP-treated murine livers (Supplementary Fig. S1a–c) and the hepatocyte cell line AML12 (Supplementary Fig. S1b, d), a substantial accumulation of cytosolic double-stranded DNA (dsDNA) signal was detected (Fig. 1a, b; Supplementary Fig. S1e, f), which was confirmed to be mtDNA-derived (Fig. 1a, b). This release occurred without a significant increase in overall mitochondrial mass (Supplementary Fig. S1g, h), indicating that the cytosolic mtDNA accumulation results from specific release rather than a generalized increase in mitochondrial mass.
Conventionally, aberrantly leaked cytosolic mtDNA is recognized by cyclic GMP-AMP synthase (cGAS), triggering innate immune signaling
13. However, in APAP-treated cells, cytosolic mtDNA failed to activate cGAS, as indicated by the absence of increased cGAMP production (Supplementary Fig. S1i, j) — the enzymatic product of cGAS activation, and the lack of phosphorylation of its downstream effector, TANK-binding kinase 1 (TBK1) (Supplementary Fig. S1k). Consistent with these findings, genetic ablation of
cGAS did not confer protection against APAP-induced liver injury in mice. Compared to wild-type (WT) controls, cGAS-deficient mice showed no improvement of liver function as indicated by serum markers such as alanine aminotransferase (ALT) and aspartate aminotransferase (AST) (Supplementary Fig. S1l), nor a reduction in TUNEL-positive hepatocytes or necrotic areas (Supplementary Fig. S1m–p). These findings may be attributed to the extremely low expression levels of key components of DNA-sensing pathways in hepatocytes
14,15, particularly the cGAS, as previously reported and also observed in our study.
Other cytosolic DNA sensors, including interferon gamma-inducible protein 16 (IFI16), DEAD-box helicase 41 (DDX41), and absent in melanoma 2 (AIM2), are known to activate inflammatory cytokine responses
16-18. However, 24-h post APAP treatment, no significant changes were observed in the levels of interferons or pro-inflammatory cytokines such as tumor necrosis factor (TNF), interleukin-1β (IL-1β), and interleukin-6 (IL-6) (Supplementary Fig. S1q). These findings suggest that APAP-induced cytosolic mtDNA does not trigger conventional innate immune signaling pathways.
Intriguingly, Z-DNA binding protein 1 (ZBP1), a non-canonical cytosolic dsDNA sensor that is constitutively expressed in the liver (Supplementary Fig. S1r), does not primarily induce type I interferon responses but is instead implicated in the direct initiation of cell death
19,20. Based on these observations, we hypothesized that cytosolic mtDNA released upon APAP treatment is sensed by ZBP1, which subsequently drives hepatocyte death.
In the APAP-induced liver injury model, genetic ablation of Zbp1 confers significant protection against hepatic dysfunction in mice as indicated by decreased levels of ALT and AST (Fig. 1c). Histopathological analysis using hematoxylin and eosin (H&E) staining demonstrated a substantial reduction in necrotic areas in the livers of Zbp1 knockout (KO) mice compared to those of WT controls (Fig. 1d, e). Quantification of TUNEL staining further revealed that the proportion of TUNEL-positive cells in Zbp1 KO livers decreased to 5%, indicating a marked reduction in cell death (Fig. 1f, g). Additionally, following APAP administration, the infiltration of immune cells, particularly neutrophils, was significantly attenuated in the livers of Zbp1 KO mice relative to WT mice (Supplementary Fig. S1s, t). In a survival study, treatment with a lethal dose of APAP resulted in a 100% mortality rate within 48 h in WT mice; however, Zbp1 KO reduced the mortality rate to 40% (Fig. 1h), highlighting the protective role of ZBP1 deficiency in mitigating APAP-induced lethality.
ZBP1 mediates hepatocyte apoptosis
Building on these findings, we further elucidated the mechanism through which ZBP1 drives hepatocyte cell death. Although ZBP1 is known to recruit RIPK3 via its RHIM domain to initiate necroptosis
21, this pathway is not operational in hepatocytes due to the absence of RIPK3 expression, and accordingly, necroptosis (pMLKL) was not observed in hepatocytes following APAP treatment (Supplementary Fig. S2a). Consistent with this, genetic ablation of
Ripk3 or
Mlkl failed to confer protection against APAP-induced liver injury (Supplementary Fig. S2b), ruling out necroptosis as a contributing mechanism. Beyond necroptosis, ZBP1 has been implicated in the regulation of apoptosis. In APAP-treated livers, robust caspase-3 cleavage was detected, whereas
Zbp1 KO livers exhibited no such activation of caspase-3 (Fig. 1i, j), suggesting a critical role for ZBP1 in APAP-induced apoptotic cell death. To address the potential confounding effects of non-hepatocyte contributions in liver tissue extracts, we validated these findings in a murine hepatocyte cell line, where APAP treatment similarly induced caspase-3 cleavage and apoptosis, and
Zbp1 knockdown effectively suppressed caspase-3 activation and apoptosis (Fig. 1k, l; Supplementary Fig. S2c). To extend these findings to a human model, we validated this mechanism in HepG2 cells, a human hepatocyte cell line that expresses significant levels of the key APAP-metabolizing enzyme CYP2E1 (Supplementary Fig. S2d). Since this cell line does not constitutively express ZBP1, we utilized IFN priming to induce its expression (Supplementary Fig. S2e). In IFN-primed HepG2 cells, APAP induced robust apoptotic cell death, evidenced by caspase-3 activation and propidium iodide (PI) staining (Fig. 1m, n; Supplementary Fig. S2e, f). Critically, sgRNA-mediated knockout of
Zbp1 in HepG2 nearly completely blocked APAP-induced cell death (Fig. 1o, p; Supplementary Fig. S2g).
ZBP1-mediated apoptosis is primarily executed via RIPK1–FADD–caspase-8 complex
22. Accordingly, in APAP-treated liver samples, activation of caspase-8 but not caspase-9 was observed (Supplementary Fig. S3a, b). The obvious activation of both caspase-8 and downstream caspase-3 within 8 h post APAP suggests a direct role for caspase-8-mediated apoptosis in the initial cell death cascade (Supplementary Fig. S3c, d). Genetic ablation of
Zbp1 abrogated APAP-induced caspase-8 and caspase-3 activation in both liver and hepatocyte cell line AML12 (Fig. 1i, j; Supplementary Figs. S3e, f), confirming the dependence of ZBP1 for this apoptotic pathway. Furthermore,
Ripk3–/–Casp8–/– double-KO mice were highly resistant to APAP-induced liver injury, hepatocyte apoptosis and mouse lethality (Fig. 2a–h). As hepatocytes lack RIPK3 and that
Ripk3 knockout alone is not protective (Supplementary Fig. S2b), these results point to caspase-8 as the key mediator of ZBP1-dependent apoptosis in hepatocytes. This conclusion was further substantiated by the marked protection observed in mice with AAV-mediated, hepatocyte-specific
Casp8 knockout (Fig. 2i–o).
However, genetic deletion of
Ripk1 or
Fadd did not mitigate APAP-induced apoptotic signaling in hepatocyte cell line AML12 (Supplementary Fig. S4a, b), suggesting that ZBP1-triggered apoptosis occurs independently of the canonical RIPK1–FADD complex. Supporting this notion,
Ripk1,
Fadd,
Ripk3 triple-knockout (TKO) mice were not protected from APAP-induced liver injury (Supplementary Fig. S4c). This finding aligns with a prior report that hepatocyte-restricted
Ripk1 knockout did not protect mice from APAP-induced liver toxicity
23. We thus considered an alternative mechanism.
Given that the mitochondrial adaptor MAVS can activate caspase-8 to induce apoptosis
24,25 and has been shown to functionally interact with ZBP1
26, we hypothesized a role for MAVS in this context. We first confirmed that MAVS can induce the activation of both caspase-8 and its downstream effector caspase-3, thereby validating its pro-apoptotic function (Supplementary Fig. S4d–f). Furthermore, consistent with prior reports
24, we found that MAVS-induced caspase-3 activation is dependent on caspase-8 (Supplementary Fig. S4e). We also confirmed the physical interaction between ZBP1 and MAVS by co-immunoprecipitation (co-IP; Supplementary Fig. S4g). Additional co-IP experiments revealed that ZBP1 with mutated RHIM domains (RHIM1 and RHIM2) exhibited attenuated binding to MAVS compared with WT-ZBP1 (Supplementary Fig. S4h). Functionally, knockout of
Mavs in AML12 cells significantly decreased APAP-induced activation of caspase-8 and caspase-3, similar to
Zbp1 knockout, supporting that MAVS is required for APAP-induced apoptosis (Supplementary Fig. S4i–k). More importantly,
Mavs KO mice were significantly protected from APAP-induced liver injury, hepatocyte cell death and mouse lethality (Fig. 2p–u), comparable to
Zbp1 KO mice. Critically, this protection was associated with a complete abolition of both caspase-8 and caspase-3 activation, without affecting ZBP1 expression (Fig. 2v, w; Supplementary Fig. S4l). To determine whether MAVS and ZBP1 operate in the same pathway, we performed a genetic epistasis analysis. In AML12 cells, knockout of
Mavs in a
Zbp1-deficient background did not produce an additive reduction in APAP-induced caspase-3 cleavage compared to
Zbp1 knockout alone (Supplementary Fig. S4m, n). This indicates that ZBP1 and MAVS function within the same signaling pathway in the APAP hepatotoxicity. Furthermore, using proximity ligation assay (PLA), we detected a significant interaction between ZBP1 and MAVS in APAP-treated AML12 cells, reinforcing the conclusion that ZBP1 engages MAVS to activate caspase-8 and downstream apoptosis (Supplementary Figs. S4o, p). We noted that the ZBP1-MAVS pathway appears to be specifically engaged during APAP-induced injury, as knockdown of MAVS had no effect on established ZBP1-dependent cell death induced by splicing inhibition
27,28 (Supplementary Fig. S4q, r). Collectively, these data support a model wherein ZBP1 activates caspase-8-dependent apoptosis via MAVS, independently of RIPK1-FADD axis.
We also explored the possibility of pyroptosis in this context. However, no evidence of Gasdermin D-mediated pyroptosis was detected in APAP-treated livers (Supplementary Fig. S4s), further corroborating the lack of AIM2 activation despite the presence of cytosolic mtDNA. Additionally, Gasdermin E (GSDME), whose cleavage is caspase-3-dependent, remained intact following APAP treatment (Supplementary Fig. S4s), reinforcing the notion that APAP-induced caspase-3 activation specifically drives apoptosis rather than Gasdermin E-mediated pyroptosis. Collectively, these findings indicate that the critical pathological mechanism underlying APAP-induced liver injury involves the release of mtDNA into the cytoplasm, which subsequently activates the ZBP1-mediated apoptotic pathway, leading to hepatocyte death.
ZBP1 binds to Z-DNA derived from released mitochondrial DNA
ZBP1 functions as a critical sensor of left-handed Z-form nucleic acids (Z-DNA/Z-RNA)
29-33, raising the possibility that APAP-induced hepatocyte injury triggers the formation of Z-nucleic acids (ZNA). To investigate this hypothesis, we employed the Z22 antibody, which selectively recognizes ZNA, including both Z-DNA and Z-RNA, to detect their presence in APAP-treated hepatocytes. As shown in Fig. 3a–c, significant accumulation of ZNA was observed in both liver tissues,
ex vivo hepatocyte and cultured hepatocyte AML12 cells following APAP exposure. Notably, the ZNA signal detected in the livers of APAP-treated mice, primary hepatocyte and APAP-stimulated AML12 cells was abolished by DNase I treatment but remained unaffected by RNase A (Fig. 3a–f), confirming that the observed ZNA signal was primarily derived from DNA. The APAP-induced Z-DNA accumulation was comparable between WT and
Zbp1 KO mice (Supplementary Fig. S5a, b), indicating that Z-DNA formation occurs independently of ZBP1.
A substantial fraction of Z-DNA colocalized with the mitochondrial marker TOM20, while a portion was detected outside mitochondria (Fig. 3b, c), suggesting that the Z-DNA may originate from leaked mtDNA. The majority of ZNA signals in the cytosol did not co-localize with mitochondrial-derived vesicles (MDVs) (Supplementary Fig. S5c, d), suggesting that the mtDNA is largely released freely into the cytosol rather than remaining vesicle-encapsulated. To directly determine the source of Z-DNA that engages ZBP1 upon APAP stimulation, we performed ZBP1 immunoprecipitation (IP) followed by qPCR analysis to distinguish nuclear vs mtDNA (Fig. 3g). The results demonstrated that, following APAP exposure, ZBP1 was specifically bound to mtDNA, as indicated by the enrichment of mitochondrial markers Nd1 and Nd2, while markers of nuclear DNA, including LINE1 and RNA18S, were absent (Fig. 3g). Consistent with this, depletion of mtDNA using ethidium bromide (EtBr) (Supplementary Fig. S5e) abolished APAP-induced Z-DNA formation (Fig. 3h, i). Furthermore, EtBr-mediated mtDNA depletion significantly attenuated APAP-induced apoptotic signaling (Fig. 3j, k), reinforcing the critical role of mtDNA-derived Z-DNA in promoting cell death pathways. Furthermore, nearly all APAP-induced dead cells (TUNEL-positive cells) were also positive for Z-DNA (Supplementary Fig. S5f, g). This strong correlation reinforces that Z-DNA formation is a driving force of the cell death pathway in this model.
The N-terminal Zα domain of ZBP1 is essential for its interaction with dsDNA, and point mutations in the Zα1 and Zα2 subdomains effectively disrupted this binding. Consistent with the mitochondrial origin of the Z-DNA, this mutation completely abrogated the enrichment of mtDNA in the ZBP1 immunoprecipitation (Fig. 3g; Supplementary Fig. S6a), confirming that mtDNA specifically binds to the Zα domains of ZBP1. To definitively confirm the functional requirement of the Zα domains, we reconstituted Zbp1-KO AML12 cells with either WT ZBP1 or a Zα mutant. Expression of WT ZBP1 fully restored cellular sensitivity to APAP-induced apoptosis, whereas the Zα mutant did not (Supplementary Fig. S6b). We further validated this finding in a human context by expressing ZBP1 in human HepG2 hepatocytes, where we observed a similar requirement for the Zα domains to confer APAP sensitivity (Fig. 3l, m; Supplementary Fig. S6c). These results conclusively demonstrate that the ZNA sensing function of ZBP1 is essential for initiating the cell death pathway.
In the APAP-induced liver injury model, mice expressing ZBP1 Zα mutants exhibited hepatoprotection comparable to that observed in Zbp1 KO mice (Supplementary Fig. S6d). This protection was characterized by reduced apoptosis (Supplementary Fig. S6e–h), diminished necrotic regions (Supplementary Fig. S6i, j), and decreased immune cell infiltration (Supplementary Fig. S6k, l). These findings indicate that ZBP1-mediated recognition of Z-DNA plays a pivotal role in APAP-induced hepatocyte death and liver injury, underscoring the pathogenic significance of mtDNA-derived Z-DNA in APAP toxicity.
Oxidative stress induces mtDNA fragmentation and cytoplasmic leakage in APAP-induced liver injury
In the context of APAP-induced oxidative stress, excessive ROS mediate mtDNA oxidation, leading to its fragmentation (Fig. 4a; Supplementary Figs. S1b, 7a). 8-Oxoguanine (8-hydroxyguanine, 8-oxoG) is one of the most common DNA lesions resulting from ROS. Using an 8-oxoG-specific antibody, we observed a significant elevation in cytoplasmic oxidized DNA levels in the APAP-induced liver injury model (Fig. 4b; Supplementary Fig. S1b, c). Additionally, APAP-treated hepatocytes exhibited an approximately 16-fold increase in cytoplasmic oxidized DNA (Fig. 4c; Supplementary Fig. S1b, d). The excess accumulation of 8-oxoG in mtDNA following APAP treatment was further validated by mass spectrometry (Supplementary Figs. S1b, 7b, c). Notably, treatment with the mitochondrial-targeted ROS scavenger MitoQ markedly reduced cytoplasmic oxidized DNA levels (Fig. 4d–f; Supplementary Fig. S7d). This reduction was accompanied by a substantial decrease in both intra-mitochondrial and cytoplasmic Z-DNA (Fig. 4g, h), suggesting that mtDNA oxidation promotes Z-DNA formation. Consistently, ROS depletion by MitoQ also significantly attenuated APAP-induced apoptotic signaling (Fig. 4i, j).
Next, we sought to elucidate the mechanism underlying oxidized mtDNA release. Previous studies have proposed that oxidized mtDNA undergoes cleavage by Flap Endonuclease 1 (FEN1) to generate short oxidized mtDNA fragments before their release into the cytoplasm
34. However, in APAP-treated hepatocytes, downregulation of FEN1 did not affect either the cytoplasmic leakage of oxidized mtDNA (Supplementary Fig. S7e–g) or the extent of APAP-induced hepatocyte apoptosis (Supplementary Fig. S7h, i). These findings suggest that mtDNA fragmentation in APAP-treated cells occurs independently of FEN1-mediated processing.
Additionally, it has been suggested that short oxidized mtDNA fragments are transported into the cytoplasm via the mitochondrial permeability transition pore (mPTP)-VDAC1 transport axis
35. In our study, pharmacological inhibition of mPTP using cyclosporine A (CsA) resulted in a partial reduction in cytoplasmic oxidized mtDNA levels (Fig. 4k, l; Supplementary Fig. S7j) and a corresponding attenuation of APAP-induced apoptosis (Fig. 4m, n). While previous reports indicate that mtDNA promotes VDAC1 oligomerization, thereby forming ion channels that facilitate mtDNA leakage from the mitochondrial intermembrane space to the cytoplasm
35, treatment with the VDAC1 oligomerization inhibitor VBIT-4 failed to reduce cytoplasmic oxidized mtDNA levels (Supplementary Fig. S7k, l) or mitigate APAP-induced apoptosis (Supplementary Fig. S7m, n).
These findings suggest that APAP-induced mtDNA fragmentation and cytoplasmic leakage occur through a distinct mechanism. We hypothesize that severe oxidative stress triggered by APAP leads to extensive mtDNA oxidation, promoting fragmentation in a nuclease-independent manner. Furthermore, the resulting oxidized mtDNA fragments may translocate into the cytoplasm via a pathway that is not fully dependent on the mPTP-VDAC1 transport system.
Oxidative modifications of DNA bases facilitate the B-to-Z DNA conformational transition
The administration of MitoQ for ROS clearance resulted in a reduction in Z-DNA levels, establishing a direct link between oxidative stress and the B-to-Z conformational transition of DNA
36. In AML12 cells, APAP treatment induced a significant increase in PLA signal compared to controls, demonstrating robust proximity between Z-DNA and 8-oxoG (Fig. 5a, b). This result supports the notion that oxidative modifications drive the B-to-Z DNA transition.
To determine whether oxidative modifications alone are sufficient to induce the B-to-Z transition, we synthesized short dsDNA sequences (12 bp) containing six consecutive GC repeats, incorporating defined oxidation levels (0, 1, or 3 8-oxoG substitutions per strand) (Fig. 5c). The Z-DNA conformation was assessed using the A260/295 absorbance ratio as a readout, which is markedly reduced in Z-DNA relative to the canonical B-form
37. The unmodified dsDNA initially displayed an A260/295 ratio of approximately 6, consistent with the B-form conformation (Fig. 5d). Upon substitution of guanine (G) with 8-oxoG, an inverse relationship emerged between the oxidation level and the A260/295 ratio, signifying a progressive transition from the B-form to the Z-conformation (Fig. 5d). Notably, while the canonical B-to-Z transition typically requires high-salt condition (≥ 2 M NaCl), this oxidation-driven structural shift occurred at low-salt concentration (20 mM NaCl), underscoring the unique ability of oxidative modifications to induce Z-DNA formation under physiologically relevant condition (Fig. 5d). The B-to-Z transition induced by oxidative modification was further validated through electrophoretic mobility shift assays (EMSA), which revealed specific binding of 8-oxoG-modified DNA to ZBP1 (Fig. 5e). Notably, this interaction was strictly dependent on the Zα domain of ZBP1, as evidenced by experiments using a mutated Zα domain that had lost its ability to bind Z-DNA (Fig. 5f). This domain-specific binding provides definitive evidence that the oxidized DNA adopts a left-handed Z-conformation, rather than merely representing a chemically modified B-form structure. Of note, the efficiency of Z-DNA formation was markedly enhanced in DNA strands containing three 8-oxoG substitutions compared to those with a single substitution (Fig. 5e), further underscoring the critical role of cumulative oxidative modifications in driving the B-to-Z transition.
To further determine whether oxidative modification of DNA alone is sufficient to induce Z-DNA formation in cells, we transfected oxidized dsDNA into
Zbp1 KO mouse embryonic fibroblasts (MEFs). These cells lack both known Zα-binding proteins, ZBP1 and ADAR1-p150, with the latter exhibiting minimal expression in basal conditions
38. In alignment with our biochemical data, the oxidized dsDNA adopted the Z-conformation intracellularly in the absence of ZBP1 and ADAR1-p150 (Fig. 5g–i). Thus, Z-DNA formation by oxidized dsDNA occurred independently of the stabilization by ZNA binding protein. Consistent with the
in vitro results (Fig. 5e), Z-DNA formation in cells was significantly more efficient in DNA strands containing three 8-oxoG substitutions compared to those with a single substitution (Fig. 5g), further underscoring the ability of oxidative modifications to promote the B-to-Z transition in a cellular context. Notably, transfection of cells with 8-oxoG-modified Z-DNA was sufficient to induce ZBP1-dependent cell death (Fig. 5j, k). Together, these findings robustly demonstrate that oxidative modifications alone are sufficient to drive the functional B-to-Z transition, both
in vitro and in cells.
To investigate the reversibility of oxidative modifications in B-to-Z DNA transitions, we employed TH10785, a recently identified agonist of OGG1, to activate this DNA glycosylase that specifically removes oxidized guanine residues
4. In
Zbp1 KO MEF cells transfected with oxidized dsDNA, OGG1 activation led to efficient removal of oxidative modifications, as evidenced by the loss of 8-oxoG signals (Fig. 5l, m). Concomitantly, DNAs were reverted from the Z-form to the B-form, as indicated by the disappearance of Z-DNA signals (Fig. 5l, n). These findings provide compelling evidence that oxidative modifications are sufficient to drive the B-to-Z DNA transition, and that this structural shift can be reversed through enzymatic deoxidation by OGG1.
Deoxidation facilitates Z-to-B DNA transition, protecting against APAP-induced liver failure and mortality
To assess the pathological relevance of oxidation-driven B-to-Z DNA transitions in APAP-induced liver injury, we examined hepatocyte cell lines following APAP treatment. APAP exposure promoted the formation of oxidized Z-DNA, while OGG1 agonist administration — whose mitochondrial accession we confirmed — efficiently reversed these oxidative modifications, restoring DNA to its B-form conformation (Fig. 6a–c; Supplementary Fig. S8a, b). This intervention significantly reduced cytoplasmic oxidized mtDNA levels and diminished Z-DNA content in both mitochondrial and cytoplasmic fractions (Fig. 6a). Notably, these molecular changes resulted in substantially attenuated cellular apoptosis (Supplementary Fig. S8c, d).
Consistently, in an APAP-induced liver injury model, treatment with the OGG1 agonist TH10785 substantially decreased cytoplasmic oxidized mtDNA levels (Fig. 6d; Supplementary Fig. S8e), which was accompanied by a corresponding reduction in Z-DNA (Fig. 6e, f). This molecular effect was associated with reduced apoptosis (as indicated by TUNEL staining and cleaved caspase-3) (Fig. 6g–j), diminished necrotic regions (Fig. 6k, l), decreased immune cell infiltration (Supplementary Fig. S8f–i), and overall preservation of liver function (Fig. 6m). Notably, the OGG1 agonist did not affect APAP metabolism in cells or the liver, as indicated by CYP2E1 (the key APAP-metabolizing enzyme) levels and GSH levels (Supplementary Fig. S9a–d). Additionally, it did not influence ROS production (Supplementary Fig. S9e, f) or ZBP1 expression levels (Supplementary Fig. S9g), confirming that its protective effects were specifically mediated through the direct removal of oxidative modifications from guanine in DNA. While TH10785 efficiently protected WT mice from APAP-induced hepatocyte apoptosis and liver injury, it completely failed to confer protection in Ogg1 KO mice (Supplementary Fig. S9h–m) and Ogg1-knockdown cells (Supplementary Fig. S9n–p). This genetic evidence from both in vivo and in vitro models conclusively validates the on-target specificity of TH10785.
To evaluate the therapeutic potential of the OGG1 agonist in APAP-induced liver injury, we first examined its efficacy at a lethal APAP dose. Treatment with the OGG1 agonist conferred 90% protection against mortality (Supplementary Fig. S10a, b). To simulate a clinically relevant scenario, we employed a delayed-treatment model in which intervention was initiated 10 h post APAP administration (Fig. 6n). In this setting, the OGG1 agonist maintained a 90% survival rate, significantly surpassing the efficacy of N-acetylcysteine (NAC), the current clinical standard, which provided less than 50% protection (Fig. 6o). Strikingly, combining the OGG1 agonist with NAC resulted in 100% survival (Fig. 6o), demonstrating a synergistic therapeutic effect and underscoring the potential of this dual approach for clinical translation.
We note that TH10785 retained a milder, yet significant, protective effect in Zbp1 KO and ZBP1-Zα mutant mice (Supplementary Fig. S10c), indicating that TH10785 also protects against ZBP1-independent cytotoxicity in APAP-induced liver injury.
DISCUSSION
In this study, we identify oxidative modification-driven B-to-Z transitions within mtDNA as the central pathological driver of APAP-induced ALF. Upon APAP exposure, oxidized mtDNA fragments leak into the cytosol, where they are recognized by the Zα domain of ZBP1, activating a downstream apoptotic signaling cascade (Supplementary Fig. S10d). During the preparation of this manuscript, another study demonstrated that amyloid-β (Aβ)-induced oxidative stress also promotes B-to-Z transition of mtDNA in microglia and contributes to neuroinflammation in Alzheimer's disease (AD)
39. This finding, together with the current study, reinforces oxidative modification as a driving force for the B-to-Z conformational transition of DNA and implicates oxidized Z-DNA in the pathogenesis of various diseases. Our genetic data support a model in which ZBP1 engages MAVS to trigger caspase-8-dependent apoptosis, a pathway that operates independently of the canonical RIPK1 and FADD adaptors. This ZBP1-MAVS-caspase-8 axis is the principal cause of APAP-induced liver damage in our experimental model.
In addition to our finding that ZBP1 engages MAVS in the APAP context, we note that in other settings, ZBP1 signaling exhibits species- and context-dependent preferences — for instance, in murine cells, ZBP1 can induce necroptosis by directly complexing with RIPK3, whereas in human cells it may require RIPK1 to form a stable ZBP1–RIPK3 complex
20. A key question arising from our work is how APAP stress directs ZBP1 toward this unconventional MAVS-dependent pathway. Our data showed that MAVS is not required for canonical ZBP1-dependent cell death induced by splicing inhibition (Supplementary Fig. S4q, r). Furthermore, using transfection of the Z-DNA oligonucleotide 8-oxodG
3dC, we observed that this Z-DNA ligand alone induces canonical ZBP1-RIPK3/RIPK1-dependent cell death in MEFs, independent of MAVS (data not sown). These observations indicate that the ZBP1-MAVS-caspase-8 axis is not dictated by the ligand (Z-DNA) alone, but rather by the unique redox microenvironment of hepatocytes during APAP overdose. We propose that APAP-induced mitochondrial ROS, or its electrophilic metabolite NAPQI — which forms protein adducts with cellular sulfhydryl groups — may alter the subcellular context. This could involve changes in MAVS oligomerization status, re-localization to mitochondria-associated membranes, or post-translational modifications of ZBP1, MAVS, or RIPK1, thereby biasing ZBP1 toward MAVS recruitment instead of RIPK1/FADD. Thus, the downstream signaling modality appears to be cell type- and context-dependent, not ligand-determined. The precise mechanisms governing ZBP1's choice of signaling partner — RIPK1, RIPK3, or MAVS — remain an important subject for future investigation.
Our results further demonstrate that the OGG1 agonist TH10785 provides robust protection in WT mice, with efficacy superior to the standard clinical intervention NAC. Notably, it also retains a significant, though milder, protective effect in ZBP1-deficient models, indicating it mitigates both ZBP1-dependent and -independent cytotoxicity. As an agonist of OGG1, TH10785 promotes the repair of all oxidized DNA, including both mitochondrial and nuclear DNA. Our model proposes that APAP-induced mtDNA oxidation drives Z-DNA formation, its cytosolic release, and ZBP1-dependent cell death. However, the widespread oxidative stress induced by APAP also can cause nuclear DNA damage
40, which can trigger ZBP1-independent cytotoxicity through multiple well-established pathways, including PARP1 hyperactivation
41, ATM/ATR-mediated DNA damage responses
42, and transcriptional stress
43. By globally reducing oxidative DNA damage, TH10785 simultaneously blocks both mtDNA-driven ZBP1-dependent death and nuclear DNA-driven ZBP1-independent death pathways. This dual action likely explains its superior efficacy compared to strategies targeting ZBP1 alone. Moreover, the liver possesses remarkable regenerative capacity. The initial response to excessive protein adduct formation after APAP overdose includes robust induction of autophagic pathways to clear damaged proteins and organelles
6. An additional adaptive mechanism is mitochondrial biogenesis, which occurs selectively in surviving hepatocytes around areas of injury and is critical for liver recovery
6. By preventing massive hepatocyte loss early after injury, TH10785 preserves tissue integrity and creates a permissive environment for these endogenous repair processes — including mitophagy and subsequent mitochondrial biogenesis — to operate effectively.
In summary, our findings establish the pivotal role of oxidation-driven B-to-Z DNA conformation transition in APAP hepatotoxicity. Our findings also highlight that promoting deoxidation via OGG1 activation facilitates the Z-to-B DNA transition, effectively mitigating hepatotoxicity and reducing mortality. This strategy represents a promising therapeutic avenue for improving clinical outcomes in APAP-induced ALF.
MATERIALS AND METHODS
Animal experiments
All animal experiments were approved by the Animal Care and Use Committee of Zhejiang University.
Zbp1–/–21,
Ripk3–/–44,
Mlkl–/–45 mice were same as used
30.
Zbp1Za1,2Mut/Za1,2Mut mice were a gift from Dr. Jonathan Maelfait.
cGAS–/– mice and
Mavs–/– mice were initially from The Jackson Laboratory.
Ogg1–/– mice were purchased from GemPharmatech (Strain No. T014260, GemPharmatech Co., Ltd.).
Ripk3,
Casp8 DKO were generated by crossing
Ripk3–/– mice with
Casp8–/– mice
46.
Casp8f/f mice
47 were used as described.
Ripk1,
Fadd,
Ripk3 TKO mice were generated by crossing
Ripk3–/– mice with
Ripk1+/– mice
48 and
Fadd+/– mice
49. C57BL/6J mice (6- to 8-week-old) were purchased from Shanghai SLAC Laboratory Animal Co., Ltd. All mice were on a C57BL/6J genetic background and were bred, housed, and analyzed in the same specific facility at the Laboratory Animal Center of Zhejiang University. They were fed with a standard laboratory diet (Research Diets, D12450K, USA) and had free access to water in a constant temperature of 23 °C in a 12-h light/dark cycle. To induce liver injury with APAP, age-matched male mice were used; littermate WT controls were used in all experiments except those shown in Supplementary Fig. 4c. They were fasted for 16 h and injected intraperitoneally (i.p.) with either saline or APAP (Yuanye Bio-Tech Co., Shanghai, China) at 300, or 550 mg/kg body weight. The animals were anesthetized via isoflurane induction at various timepoints after APAP dosing for analyses. For all terminal (24 h) experiments, all injected mice were included in the analysis; no animals were excluded. To treat the animals with OGG1 agonist TH10785
4, the compound was dissolved in DMSO (Sigma-Aldrich) at a concentration of 100 μg/uL as a stock solution and diluted with 40% PEG300 and 5% Tween-80 to 0.5 mg/mL before use. TH10785 was injected i.p. at 5 μg/g. Following the planned treatment, the mice were anesthetized. Blood samples were obtained from the retroorbital venous plexus for analysis, and liver tissues were collected. Whole blood was centrifuged at 3,000×
g and 4 °C for 10 min to separate the serum. Serum levels of ALT and AST were measured using commercial kits (Sigma-Aldrich, Shanghai, China). Liver tissues were fixed in 4% paraformaldehyde, dehydrated, and embedded in paraffin. Sections were cut at a thickness of 7 µm and collected for further examination. For histological analysis, tissue sections were stained with H&E. All histological assessments of tissue damage were performed by investigators blinded to the genotypes. Necrotic areas in H&E-stained sections were outlined and quantified using Adobe Photoshop under a 10× objective, and the percentage of necrotic area was calculated accordingly. For survival studies, mice were allowed to reach a natural endpoint, and the time post APAP injection was recorded.
Cell culture and stimulation
MEFs, AML12 and HepG2 cells were obtained in the lab. Cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 1% (v/v) penicillin/streptomycin in an incubator supplied with a humidified atmosphere of 5% CO
2 at 37 °C. For primary hepatocytes, the mouse liver was perfused with 0.05% collagenase type IV (Sigma-Aldrich) to obtain primary hepatocytes, which were plated in round coverslips in DMEM with 10% FBS and 1% PenStrep for 4 h for attachment. Subsequently, the cells were used in immunofluorescence staining. AML12 cells and HepG2 cells were incubated with APAP (10 mM or 20 mM) for 24 h to establish an
in vitro DILI model. APAP was dissolved into serum free DMEM directly. Cells were seeded in six-well plates before being subjected to treatments. Supernatants and cell lysates were collected for ELISA and immunoblot (IB) analyses. Protein sample of proliferating human hepatocytes (ProliHHs) was a gift from Dr. Lijian Hui
50.
Generation of KO and knockdown cell lines
The KO AML12 or HepG2 cell pools were generated using CRISPR/Cas9 methods. The target sequence for mouse Ripk1 was: 5′-AGAAGAAGGGAACTATTCGC-3′. The target sequence for mouse Fadd was: 5′-TAGATCGTGTCGGCGCAGCG-3′. The target sequence for mouse Zbp1 was: 5’-CAGGTGTTGAGCGATGACGG-3’. The target sequence for human ZBP1 was: 5’- ATGTGAACCGAGACTTGTAC-3’. The target sequence for mouse Mavs was: 5′-AGAGTCCCCAGAGTGTGTCC-3’. The target sequence for mouse Casp8 was:5′-GCAGGTCCCACCGACTGATG-3’. Zbp1 knockdown by shRNA in AML12 cells was also used. Sequences of specific shRNAs used in this study were obtained from the MISSION shRNA Library (Sigma). gRNA and shRNA were transduced into indicated cell line by lentiviral delivery. Cells were then subjected to blasticidin or puromycin selection.
siRNA transfection in AML12
For siRNA transfection in 24-well plates, a mixture containing 0.75 μL of siRNA duplexes (20 μM), 1.5 μL of Lipofectamin RNAiMAX reagent (Thermo Fisher Scientific, 13778150), and 75 μL of Opti-MEM (Gibco, 31985070) was incubated for 5 min and then added into AML12 culture for 24 h before treatment. The following oligonucleotide sequences were used for knockdown: siOgg1,5’-AUAAGUGACAUCAUCAAGCUGTT-3’; siFen1, 5’-CGUGCUAAUGCGACACUUATT-3’; siNC, 5’-UUCUCCGAACGUGUCACGUTT-3’; siMavs, 5’-AGAUUCUAAAGGACGAAUACU-3’.
Lentivirus preparation and infection
HEK293T cells were transfected via calcium phosphate precipitation with lentiviral vectors encoding the cDNA or gRNA of interest, together with lentiviral packaging plasmids (pMD2.G and psPAX2). After 12 h, the culture medium was replaced. Viral supernatants were harvested 36 h post-transfection and used for subsequent infections. Target cells were incubated with lentivirus-containing supernatant supplemented with 10 µg/mL Polybrene and centrifuged at 2,500 rpm for 30 min to enhance transduction efficiency. The medium was refreshed 12 h after infection.
Cell death assay
For cell death assay by PI incorporation: briefly, 2 × 104 cells were seeded in 96-well plates. After 12 h, the cells were treated with reagents for the indicated durations. The cell death was also analyzed by the incorporation of PI. Briefly, the PI and Hoechst were added into medium when treatment. After treatment, three independent images per condition were captured by microscope, analyzed by Image J and the percent of PI-positive cells were averaged.
For cell apoptosis assay, an FITC Annexin V Apoptosis Detection Kit (Yeasen Biotech) was used. Briefly, cells were seeded in 12-well plates and cultured overnight. After indicated treatments, cells were harvested by mild trypsin digestion. The FITC-Annexin V and propidium iodide were used for double staining in accordance with the manufacturer’s instructions, followed by analysis with Attune NxT flow cytometry (Thermo Fisher Scientific).
Caspase activity assay
Caspase-3 activity was measured using the Caspase-Glo® 3/7 Assay Kit (Promega, G8091), following the manufacturer's protocol. Briefly, 2.0 × 104 cells were seeded per well in white-walled 96-well plates (Nunc) and allowed to adhere for 12 h. Following experimental treatments, an equal volume of room temperature-equilibrated reagent was added directly to the culture medium. Plates were gently agitated for 5 min and then incubated at room temperature for 30 min. Luminescence was recorded using a microplate reader (Feyond-A300, ALLSHENG, China).
Subcellular fractionation of cells and tissues
In brief, cells were washed with ice-cold PBS while fresh tissues were minced into small pieces on ice. Both sample types were then homogenized using a glass Dounce homogenizer in Mitochondrial Preparation Buffer (0.225 M mannitol, 0.075 M sucrose, 20 mM HEPES, pH 7.4). The homogenates were centrifuged at 1,000× g for 10 min at 4 °C to pellet the nuclear fraction. The resulting post-nuclear supernatants were carefully transferred to new tubes and subjected to centrifugation at 12,000× g for 15 min at 4 °C, yielding cytosolic supernatants and pellets containing the mitochondrial fraction. The cytosolic fractions were reserved for subsequent analysis by ELISA or nucleic acid extraction. The mitochondrial pellets were either processed directly for DNA extraction using standard protocols or further purified for proteomic analysis by mass spectrometry.
Immunofluorescence staining
AML12 or MEF cells cultured in round coverslips were fixed in 4% PFA for 10 min and washed with PBS. After being permeabilized and blocked with 0.4% Triton X-100 and 3% BSA, cells were incubated with primary antibodies including 8oxodG (1:1,000, Abcam, ab48508), ZNA (1:1,000, Absolute Antibody, Ab00783-23.0), dsDNA (1:1,000, Abcam, ab27156), Tom20 (1:1,000, Proteintech, 11802-1-AP), TOM20 (1:1,000, Abclonal, A27799), PDHX (1:1,000, Proteintech, 10951-1-AP), ZNA (1:1,000, Absolute Antibody, Ab00783-3.0), ZNA (1:1,000, Novus Biologicals, NB100-749) antibody overnight at 4 °C. Then the cells were washed with PBS, incubated with secondary antibody for 1 h at RT. Images were captured by Olympus FV3000 or High Intelligent and Sensitive SIM (HIS-SIM).
Tissue immunofluorescence
The liver tissues were fixed in 4% paraformaldehyde, dehydrated in 15% and 30% sucrose, and embedded in optimal cutting temperature compound at −20 °C. The tissues were then sectioned (12 μm) using a cryostat. Afterwards, the sections were washed with PBS three times, blocked with 3% BSA and 0.4% Triton X-100 at RT for 1 h, and incubated with primary antibodies including CD45 (1:200, BD, 550539), Ly6G (1:200, Biolegend, 127601), cleaved Caspase-3 (1:1,000, CST, 9664S), Cleaved Caspase8 (1:1,000, CST, 8592) overnight at 4 °C. For ZNA staining in liver tissues, the tissues were pre-dried at 55 °C for 30 min and then subjected to proteinase K treatment (20 µg/mL) for 5 min at RT. When acquired, RNase A (1 mg/mL in PBS, Solarbio, 9001-99-4) or DNase I (25 U/mL in PBS, Invitrogen, AM2222) was used after proteinase K treatment at 37 °C for 1 h. The sections were incubated with primary antibody ZNA (1:1,000, Absolute Antibody, Ab00783-23.0) overnight at 4 °C, washed with PBS three times, and incubated with secondary antibody for 1 h at RT in the dark. Images were captured by Olympus FV3000 (fluorescence).
TUNEL staining
The principle of TDT-mediated dUTP nick-end labelling (TUNEL) to detect cell apoptosis is that the exposed 3′-OH of broken DNA can be catalysed by terminal deoxynucleotidyl transferase (TdT) with FITC-labelled dUTP, which can be detected using fluorescence microscopy. The specific steps were performed according to the kit’s instructions (Vazyme, A113-03). Images were obtained by Olympus FV3000 (fluorescence). Digital images were recorded and analyzed using Image J software (NIH).
PLA
A PLA was employed to visualize protein–protein interactions, specifically assessing the associations between ZBP1 and MAVS, as well as the proximity between ZNA and 8-oxodG. Briefly, AML12 cells grown on coverslips in 24‑well plates were treated as indicated. After fixation with 4% paraformaldehyde and permeabilization with 0.05% Triton X‑100 in PBS, cells were incubated overnight at 4 °C with species-matched primary antibody pairs against the targets of interest. The following antibodies were used (all at 1:100 for PLA): anti-ZBP1 (AdipoGen, AG-20B-0010-C100), anti-MAVS (HUABIO, HA721310), anti-ZNA (Absolute Antibody, Ab00783-23.0), anti-8oxodG (Abcam, ab48508).
Following primary antibody incubation, cells were probed with species-specific PLA secondary antibodies conjugated to oligonucleotides (rabbit MINUS, Sigma DUO92004, 1:5; mouse PLUS, Sigma DUO92002, 1:5). Subsequent ligation and amplification steps were performed according to the manufacturer’s protocol (Sigma DUO92008). Samples were mounted using DAPI-Fluoromount-G (Beyotime, P0131) and imaged with Olympus FV3000 (fluorescence). Image processing and analysis were conducted using ImageJ software.
Intensity line measurement
Representative images for intensity measurement were captured by Olympus FV3000, and pixel intensity was assessed with Fiji (ImageJ) to measure indicated staining (ZNA/8oxodG/FAM) intensity across the dotted lines at baseline and after APAP-treated AML12 cells or FAM-8oxo-dG3dC duplexes-transfected cells.
Automatic image analysis
For the quantification of mitochondrial superoxide, the intensity of the mitochondrial superoxide indicator MitoSOX was analyzed within cells using Hoechst as the nuclear counterstain and cells were defined by propagation of the nuclear objects to the cellular periphery on the basis of MitoSOX staining. For the quantification of cytoplasmic dsDNA and ZNA, the intensity of the cytoplasmic dsDNA and ZNA fluorescence intensity was analyzed within cells using DAPI as the nuclear counterstain and cells were defined by propagation of the nuclear objects to the cellular periphery on the basis of dsDNA staining. Statistical analysis of CD45 and Ly6G fluorescence was performed using ImageJ to calculate the percentage of area positive for each marker within the field of view. To quantify the colocalization of FAM/ZNA or FAM/8oxodG, the Manders' colocalization coefficient (MCC) was calculated using ImageJ software. First, dual-channel fluorescence images under a 60× objective were acquired under identical acquisition settings to ensure consistency. The images were then opened in ImageJ, and background subtraction was performed to minimize noise interference. The "Coloc JACop" plugin in ImageJ was utilized for the analysis.
Measurement of B-DNA and Z-DNA conformation using a ratio of A260/295
Conformation of Z-DNA and B-DNA was assessed using the absorbance ratio of 260 to 295 nm as previously described
51,52. We used short dsDNA sequences (12 bp) with six consecutive GC repeats with defined oxidation levels (0, 1, or 3 8-oxodG substitutions per strand) (100 ng/μL) for conformation analysis. Poly(dG:dC) was diluted in annealing buffer (20 mM NaCl, 20mM Tris-HCL, 0.1mM EDTA) to induce Z-DNA conformation. A260/295 was measured with a NanoDrop (Thermo Fisher Scientific) using annealing buffer as a blank. Each incubation was performed in triplicates. The values were then plotted as the ratio of A260/295.
dsDNA transfection
For dsDNA transfection in 96-well plates, 0.1 μg of DNA duplexes were mixed with Lipofectamine 2000 (Thermo Fisher Scientific, 11668030) and brought up to 20 μL in OptiMEM and allowed to sit at room temperature for 20 min. The mixture was then added to the MEF cells and allowed to incubate at 37 °C for 8 h. The medium was changed 8 h post transfection.
Generation of mtDNA-depleted AML12 cells
To establish mtDNA-depleted AML12 cells, the AML12 cell line was cultured for six days in medium supplemented with 400 ng/mL EtBr (Sangon, A500328) before proceeding with the indicated treatments. To evaluate the efficiency of mtDNA depletion, total DNA was extracted and subjected to real-time quantitative PCR. The expression levels of mitochondrial genes (D-loop-1, D-loop-2, Nd1, and Nd2) were measured, and the values from each replicate were normalized to the nuclear-encoded gene Tert.
EMSA
The binding assays were performed by incubating 2.5 µM of FAM-dGdC, FAM-8oxo-dG1dC, and FAM-8oxo-dG3dC oligonucleotides in 10 mM HEPES buffer (pH 7.5, containing 10 mM MgCl2) with either reconstituted Human-ZBP1-Zα1 or its mutant form (Human-ZBP1-Zα1 with N46A and Y50A mutants) at specified DNA:protein molar ratios. The incubation was carried out for 2.5 h at 25 °C. Following incubation, the DNA–protein complexes were separated by native polyacrylamide gel electrophoresis using 15% TBE gels. DNA visualization was achieved through post-staining withEtBr, and the resulting bands were captured using the Magic SHST‘s Gel doc system.
ELISA
For the measurement of cGAMP, AML12 cells were digested and collected into centrifuge tubes, and cells were counted using the Countstar automated after two washes with cold PBS. Cells were lysed by repeated freeze-thaw cycles in PBS at a ratio of 100 μL per million cells. Mice were euthanized, and the liver tissues were carefully dissected. A total of 1 mL of PBS was used for every 100 mg of fresh mouse liver tissue, which were then thoroughly disrupted using Tissuelyser-II (Shanghai Jingxin). Tissue homogenates were placed into a 3D shaker at 4 °C for 5 min before centrifuging. The Pierce BCA Protein Assay was used to normalize the protein concentration of the supernatants. The 2′3′-cGAMP concentrations were measured using ELISA (COIBO, CB15107-Mu) according to the manufacturer’s instructions. For the measurement of Ox-mtDNA, purified mtDNA was extracted from the cytosolic or mitochondrial fractions as indicated. The 8-OH-dG content was then quantified using 8-hydroxy 2-deoxyguanosine ELISA Kit (COIBO BIO, CB10013) according to manufacturer’s instructions.
Immunoblotting
For mouse liver tissues, samples were collected from mice following cardiac perfusion with PBS, tissues were homogenized using metal beads at 60 Hz for 60 s in ice-cold RIPA lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 1% protease inhibitor cocktail, 1 mM PMSF, and 1% phosphatase inhibitor) with a TissueLyser II (QIAGEN), and centrifuged at 12,000× g at 4 °C for 30 min. Then the protein concentrations were adjusted to 1 mg/mL based on BCA. Proteins were blotted following a standard protocol. Antibodies against the following proteins were used for immunoblotting: ZBP1 (AdipoGen, AG-20B-0010-C100, 1:1,000), ZBP1 (Rabbit polyclonal antibody, generated in-house against the Zα domain of human ZBP1, 1:3,000), p-TBK1 S172 (CST, 5483S, 1:1,000), TBK1 (CST, 3013, 1:1,000), Cleaved Caspase-3 (CST, 9664S, 1:1,000), Caspase-3 (CST, 9662S, 1:1,000), p-MLKL (phospho S345)(Abcam, ab196436, 1:1,000), MLKL (Proteintech, 66675-1-lg, 1:1,000), GSDME (Abcam, ab215191,1:1,000), GSDMD (Abcam, ab209845, 1:1,000), RIPK1 (CST, 3493, 1:1,000), MAVS (HUABIO, HA721310, 1:1,000), Cleaved Caspase8 (CST, 8592, 1:1,000), Caspase8 (CST, 4790, 1:1,000), Cleaved Caspase9 (CST, 9509, 1:1,000), Caspase9 (CST, 9508S, 1:1,000), CYP2E1 (HUABIO, R1511-7, 1:1,000), Tom20 (Proteintech, 11802-1-AP, 1: 2,000), Lamin B (Proteintech, 66095-1-Ig, 1:5,000), HA (HUABIO, HA721750, 1:3,000), FLAG (Sigma, f3165, 1:3,000), Actin (Proteintech, 66009-1-Ig, 1:5,000), GAPDH (HUABIO, ET1601-4, 1:5,000), Vinculin (HUABIO, ET1705-94, 1:5,000) and Tubulin (Millipore, 05-829, 1:10,000). The signals were detected by Immobilon ECL Ultra Western HRP Substrate (Millipore).
Co-IP
Cells were harvested and lysed in a non-denaturing lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 5% Glycerol, 1% Triton X-100) supplemented with complete protease and phosphatase inhibitor cocktails. At 4 °C with rotation for 30 min, then the cell lysates were clarified by centrifugation at 12,000 rpm for 15 min at 4°C. For each immunoprecipitation reaction, a total of 500 µg to 1 mg of clarified lysate was used. Anti-FLAG M2 Magnetic Beads were equilibrated by washing three times with lysis buffer. The pre-cleared lysate was then incubated with 10 µL of the bead slurry overnight at 4 °C with constant rotation. The following day, the bead-bound complexes were isolated by placing the tube on a magnetic rack. The supernatant was carefully aspirated, and the beads were washed stringently four times with 1 mL of ice-cold lysis buffer. Finally, the specifically bound proteins were eluted from the beads by boiling in 2× SDS sample buffer for 10 min. The eluted proteins were then resolved by SDS-PAGE and analyzed by immunoblotting with antibodies against FLAG and HA.
Cytosolic DNA immunoprecipitation
Liver tissue samples were lysed on ice for 10 minutes using 1 mL of digitonin lysis buffer (150 mM NaCl, 20 mM HEPES, pH 7.4, 25 μg/mL digitonin). The lysates were then centrifuged twice at 13,000× g for 5 min at 4 °C to separate the soluble supernatant from the pellet, which contains the heavy membrane fraction. For ZBP1 immunoprecipitation, antibodies against ZBP1 and control IgG were used. Following overnight incubation at 4 °C with rotation, the samples were washed extensively. Co-precipitated DNA was extracted using phenol‑chloroform and quantified by real-time quantitative PCR. Results are expressed as relative expression levels compared to untreated controls.
Quantitative real-time PCR
Total RNA was isolated from mice liver using TRIzol reagent (Life Technologies). RNA concentration was measured using the Nanodrop spectrophotometer (Thermo Fisher Scientific). cDNA was prepared using 1 μg of RNA with HiScript III room temperature SuperMix kit (Vazyme), and reverse transcribed into cDNA. qPCR was performed with ChamQ Universal SYBR qPCR Master Mix (Vazyme) by the CFX Connect Real-Time PCR Detection System (Bio-Rad). Data were analyzed according to the ΔΔCT method. Actin (encoding β-actin) was used as the reference gene for accurate normalization of qPCR data. To measure the abundance of cytosolic mtDNA, 10 ng/μL of template DNA was used for qPCR analysis, and expression values of each replicate (D-loop-1, D-loop-2, Nd1, and Nd2) were normalized against nuclear-encoded Tert. The sequences of gene-specific primers used for PCR are shown below.
Zbp1-F:5’-TTGAGCACAGGAGACAATCTG-3’,
Zbp1-R:5’-TTCAGGCGGTAAAGGACTTG-3’;
Ifnb1-F: 5’-CAGCTCCAAGAAAGGACGAAC-3’,
Ifnb1-R: 5’-GGCAGTGTAACTCTTCTGCAT-3’;
Tnf-F: 5’-CCCTCACACTCAGATCATCTTCT-3’,
Tnf-R: 5’- GCTACGACGTGGGCTACAG-3’;
Il6-F: 5’-TAGTCCTTCCTACCCCAATTTCC-3’,
Il6-R: 5’-TTGGTCCTTAGCCACTCCTTC-3’;
Il1b-F: 5’-GCAACTGTTCCTGAACTCAACT-3’,
Il1b-R: 5’-ATCTTTTGGGGTCCGTCAACT-3’;
Usp18-F: 5’-TTGGGCTCCTGAGGAAACC-3’,
Usp18-R: 5’-CGATGTTGTGTAAACCAACCAGA-3’;
Isg15-F: 5’-CCCCCATCATCTTTTATAACCAAC-3’,
Isg15-R:5’-CACAGTGATCAAGCATTTGCG-3’;
Oasl1-F: 5’-TCCTTCGGTTGGTCAAACAC-3’,
Oasl1-R: 5’-CAGGCATAGACAGTGAGCAG-3’;
Irf7-F: 5’-TGTTTGGAGACTGGCTATTGG-3’,
Irf7-R:5’-ATCCCTACGACCGAAATGCT-3’;
Tert-F: 5’-CTAGCTCATGTGTCAAGACCCTCTT-3’,
Tert-R:5’-GCCAGCACGTTTCTCTCGTT-3’;
D-loop-1-F: 5’-AATCTACCATCCTCCGTGAAACC-3’,
D-loop-1-R: 5’-TCAGTTTAGCTACCCCCAAGTTTAA-3’;
D-loop-2-F: 5’-TCCTCCGTGAAACCAACAA-3’,
D-loop-2-R: 5’-AGCGAGAAGAGGGGCATT-3’;
Nd1-F: 5’- CTAGCAGAAACAAACCGGGC-3’,
Nd1-R: 5’-CCGGCTGCGTATTCTACGTT-3’;
Nd2-F: 5’- CCATCAACTCAATCTCACTTCTATG-3’,
Nd2-R: 5’-GAATCCTGTTAGTGGTGGAAGG-3’;
Fen1-F: 5’-TTCACGGCCTTGCCAAACTAA-3’,
Fen1-R: 5’-ACAGCAATCAGGAACTGGTAGA-3’;
LINE1-F: TAGGAAATTAGTTTGAATAGGTGAGAGGGT,
LINE1-R: TCCAGAAGCTGTCAGGTTCTCTGGC;
RNA18S-F: GTAACCCGTTGAACCCCATT,
RNA18S-R: CCATCCAATCGGTAGTAGCG;
Actin-F: 5’-GGCTGTATTCCCCTCCATCG-3’,
Actin-R: 5’-CCAGTTGGTAACAATGCCATGT-3’;
Ogg1-F: 5’-TGGTTTTAGCTTCTGGACAGTC-3’,
Ogg1-R: 5’-TGGTTTTAGCTTCTGGACAGTC-3’.
Measurement of mitoROS
MitoSOX™ Red dye (Invitrogen, USA) is a cell-permeable probe that selectively accumulates in mitochondria, where it is oxidized by superoxide to yield red fluorescence (λex = 396 nm; λem = 610 nm). AML12 cells were seeded in 35-mm dishes and treated with 10 mM APAP for 24 h. After treatment, the medium was removed and the cells were washed three times with PBS. Subsequently, 1 mL of 5 μM MitoSOX working solution was added, and cells were incubated at 37 °C for 30 min in the dark. Following incubation, cells were washed three times with PBS and counterstained with 1 μg/mL Hoechst for 10 min. Fluorescence images were acquired using an Olympus FV3000 confocal microscope.
Measurement of GSH
For GSH quantification in liver tissue, samples were homogenized in extraction buffer at a 1:10 (w/v) ratio. The homogenate was centrifuged at 10,000× g and 4 °C for 10 min to collect the supernatant. Total protein concentration was determined using a BCA assay kit. Subsequently, GSH levels were measured according to the manufacturer’s protocol with a commercial GSH detection kit (S0053, Beyotime Biotechnology).
For cellular GSH analysis, cells were plated in 6‑well plates and allowed to adhere overnight. Following the indicated treatments, cells were harvested and counted. An equal number of cells (2 × 106 per sample) were used for GSH measurement with the same commercial kit (S0053, Beyotime Biotechnology), following the provided instructions.
Mass spectrometry analysis
LC-MS/MS analysis was conducted using SCIEX QTRAP 6500+ system triple quadrupole mass spectrometer connected with Exion LC system. Chromatographic separation was achieved on an ACQUITY UPLC HSS T3 column (100 mm × 2.1 mm, 1.8 μm) at 40 °C. The mobile phase consisted of 0.1% formic acid in water (A) and 0.1% formic acid acetonitrile (B) at a flow rate of 0.4 mL/min. The gradient of mobile phase B was 0% in 1 min, 0% to 1% in 1 min, 1% to 6% in 1 min, held at 6% for 0.5 min, then 6% to 50% in 1 min, 50% to 75% in 1 min, held at 75% for 1.5 min, then 75% to 0% in 0.5 min, and held at 0% for 3.5 min. The sample volume injected was 5 μL. Mass spectrometer was operated in electrospray ionization-positive mode using the following settings: nitrogen, with a purity of 99.9%, served as the desolvation gas at a manipulation temperature of 500 °C. The curtain gas was set at 35 L/min, the ion source gas 1 pressure was maintained at 55 psi, and ion source gas 2 pressure was maintained at 50 psi. The capillary voltage was set to 5500V. Mass transitions monitored were m/z 284.2 -> 168.1 and m/z 284.2 -> 140.1 for 8-oxodeoxyguanosine, m/z 268.1 -> 152.3 and m/z 268.1 -> 135.1 for deoxyguanosine, m/z 268.4 -> 186.1 and m/z 268.4 -> 169.2 for TH10785.
Quantification and statistical analysis
In vitro data are presented as mean ± SEM from a representative experiment with the indicated number of independent replicates. For in vivo studies, results are shown as mean ± SEM, with the sample size (n) specified for each group. Detailed statistical information — including exact n values, post-hoc comparisons, and significance levels — is provided in the corresponding figures and legends. All statistical analyses were conducted in GraphPad Prism 8.0, applying two-tailed unpaired Student’s t-tests, one-way ANOVA, or two-way ANOVA as appropriate and noted in the figure legends. Survival curves were compared using the Mantel–Cox log-rank test. P value below 0.05 was considered statistically significant.
DATA AVAILABILITY
All data are available in the main text or the supplementary materials.
The Author(s) 2026. Published by Higher Education Press. This is an Open Access article distributed under the terms of the CC BY license (https://creativecommons.org/licenses/by/4.0/).
