INTRODUCTION
Our bodies are perpetually challenged by external menaces like viral/bacterial infections, as well as internal threats such as malignant cell transformation and oncogenesis. A prompt defense against these insults is vital for maintaining homeostasis
1-3. Host cells deploy an arsenal of innate immune signaling cascades to detect both exogenous and endogenous danger cues, triggering immune effectors for pathogen/tumor clearance
4. Pathogen-associated molecular patterns (PAMPs) released by microbes efficiently engage pattern recognition receptors on host cells, igniting innate immunity
1,5. Intriguingly, antitumor modalities like chemotherapy and radiotherapy also harness innate immune mechanisms to bolster immunosurveillance
6,7, but the underlying activation mechanisms remain incompletely understood.
Chemotherapy plays a pivotal role in the clinical management of malignant tumors. Accumulating evidence from previous studies has demonstrated that beyond its direct cytotoxic effect of eliminating tumor cells, chemotherapy can also elicit robust immunogenic responses that contribute to tumor regression
6,8,9. Specifically, the immunogenic responses induced by chemotherapy are multifaceted, encompassing two distinct yet interconnected mechanisms. On one hand, it directly potentiates the activity of the innate immune system by upregulating the expression of pattern recognition receptors, pro-inflammatory cytokines, and other immune-modulatory molecules in immune cells. On the other hand, it exerts a secondary effect triggered by the release of damage-associated molecular patterns (DAMPs) and tumor-associated antigens from dying tumor cells. These danger signals alert the host immune system to the presence of neoplastic cells, thereby initiating a cascade of adaptive immune responses characterized by the activation and proliferation of tumor-reactive T lymphocytes, which further reinforce the anti-tumor efficacy of chemotherapy
7,9,10. Type I interferons are primarily produced by cells that activate innate immune responses
6. Previously, anti-tumor innate immunotherapies mainly targeted the cGAS-STING pathway
11,12. However, decreased STING levels in cancer tissue consistently led to poor treatment outcomes when targeting STING. Nonetheless, chemotherapies that activate innate immunity can still yield positive results in tumors with low STING levels, indicating the existence of broader innate immunity activation pathways that do not rely on STING
13-19.
Our study uncovers a pivotal role for mitochondrial RNA (mtRNA)-derived danger signals in activating innate immunity during antitumor therapy. Chemotherapy elicits a robust immunostimulatory response by inducing the release of immunogenic mitochondrial double-stranded RNA (mt-dsRNA), a process that potently activates immune surveillance mechanisms. Mechanistically, this therapeutic modality promotes the translocation of mt-dsRNA into the cytoplasm through BAK/BAX-dependent pore formation, thereby amplifying antitumor immunity. Notably, the danger signals emanating from mtRNA engage the RIG-I/MDA5-MAVS signaling axis — a pathway ubiquitously expressed in tumor tissues. This ensures broad-spectrum activation of antitumor immune responses, effectively circumventing tumor-mediated immune evasion strategies.
Malignant cells employ sophisticated strategies to evade immune detection
20-22. Our data reveal that malignant cells upregulate PNPT1, an enzyme that degrades immunogenic mitochondrial nucleic acids, thereby dampening immune activation and evading immunosurveillance. Clinical analyses confirm that PNPT1 overexpression in tumor tissues correlates with immune escape and disease progression. Critically, PNPT1 knockdown potentiates anticancer immunity, synergizing with chemotherapy and PD-1 immunotherapy. Targeting the mtRNA-MAVS axis represents an unexploited therapeutic frontier. Addressing this unmet need, we have developed a therapeutic strategy targeting the PNPT1-mtRNA axis. Combinatorial therapy using an FDA-approved PNPT1 inhibitor and BH3-mimetic agents reactivates antitumor immunity, overcomes solid tumor resistance, and demonstrates favorable safety profiles in preclinical models. These findings highlight its translational potential as a safe and efficacious strategy to enhance cancer immunotherapy.
RESULTS
The MAVS pathway efficiently and broadly mediates chemotherapy-dependent immune responses.
Chemotherapy is the primary method for cancer treatment to kill malignant cells directly
23. Previous studies have shown that chemotherapies activate the innate immune pathway to enhance the anti-tumor immune response, with STING serving as the key regulator of this process by binding to DNA released from the nucleus or mitochondria. However, emerging evidence indicates that STING levels are decreased in various cancer types compared to normal tissues, and the targets of STING have failed to activate innate immunity in cancer cells
13,14,19.
The enrichment analysis shows that chemotherapy still induces innate immune responses in patient samples with lower STING expression in TCGA LIHC and LUAD datasets (Fig. 1a, b), while the GEO datasets demonstrated that cisplatin (CIPT) treatment elicits type I interferon and inflammatory responses in low-STING cell lines (Fig. 1c; Supplementary Fig. S1b). Using the genotoxic agent doxorubicin (DOX) and CIPT (both are commonly used chemotherapy drugs), we found that both drugs induced IFN-β mRNA expression in medium-STING-expressing human cervical cancer (HeLa) cells (Supplementary Fig. S1c), low-STING-expressing lung (A549) and liver (HepG2) cancer cells (Supplementary Fig. S1d), and murine melanoma (B16) cells (Supplementary Fig. S1e). DOX-mediated immune responses were only partially reduced in STING-knockout human keratinocytes (HaCaT), whereas the STING agonist cGAMP failed to induce responses (Supplementary Fig. S1f), confirming that chemotherapy triggers innate immunity independent of STING expression. These findings suggest that DNA-mediated cGAS-STING activation is not the universal pathway for chemotherapy-induced antitumor immunity, implying broader intracellular pathways exist.
Given the critical role of endogenous nucleic acids in activating immune responses
2, we analyzed all reported nucleic acid receptors and interferon-inducing pathways. Strikingly, the MAVS signaling axis and its upstream receptors RIG-I/MDA5 showed broader and higher expression in cancer cells compared to other receptors (Supplementary Fig. S1g, h). Unlike STING, which is significantly reduced in multiple types of cancer validated by the Human Protein Atlas (HPA)-based immunohistochemistry (IHC) of lung cancer samples (Fig. 1e, f; Supplementary Fig. S1a)
24, the RNA-responsive core protein MAVS (also known as VISA)
26 is either normally expressed or upregulated in tumors (Fig. 1d; Supplementary Fig. S1i), which is validated by strong MAVS expression in lung cancer via HPA IHC slides (Fig. 1i). HPA data further show MAVS expression exceeds that of STING in multiple cancer types (Fig. 1h), suggesting a broader role for MAVS in mediating antitumor immunity.
Functional studies in HeLa cells transfected with siRNA targeting MAVS or STING (Supplementary Fig. S1j, k) revealed that interfering with either partially inhibited chemotherapy-induced immune responses (Fig. 1i). In STING-knockout cells, chemotherapy still induced immune responses, which were abrogated by MAVS interference (Fig. 1j). In low-STING cancer cells, MAVS depletion alone suppressed chemotherapy-mediated immunity (Fig. 1k, l; Supplementary Fig. S1k). The above results suggest that MAVS can efficiently mediate chemotherapy-induced immune responses regardless of STING expression. Specifically, in cancers with concurrent STING and MAVS expression, both molecules contribute to the immune response, whereas in STING-negative cancers, MAVS serves as the primary mediator. Additionally, DOX was found to enhance IFN-β mRNA expression in the mouse tumor cell line B16, which was abrogated following MAVS interference (Fig. 1m; Supplementary Fig. S1k). Collectively, our findings demonstrate that the MAVS signaling pathway promotes immune responses during chemotherapy. Notably, the widespread expression of MAVS and its upstream RNA receptors in most cancer cells and tumor tissues allows these signals to initiate broad-spectrum antitumor immunity, thereby countering tumor immune evasion mechanisms
Mt-dsRNA release triggers MAVS-dependent immune response.
Our findings demonstrate that MAVS can initiate an immune response. Previous studies have shown that MAVS and its upstream receptors primarily mediate RNA-dependent immune responses, encompassing both exogenous pathogenic microorganism infections and the host's own RNA
27. As our system lacks exogenous infections, we focused on identifying which endogenous signals trigger the MAVS pathway. Host-derived RNA primarily originates from mitochondria and the nucleus. We first isolated mitochondrial and non-mitochondrial components from cells, purified their respective RNA, and transfected them into naïve cells to assess the immune-inducing capacity of RNA from different sources. The result shows that transfection of a low dose of mtRNA into recipient cells was sufficient to elicit robust immune responses. In striking contrast, even a tenfold higher dose of non-mtRNA failed to induce detectable immune activation. By integrating this observation into the context of our study, we further emphasized the critical role of mtRNA in mediating anti-tumor immunity, highlighting its potential as a key regulator in triggering tumor-reactive immune responses (Fig. 2a). Additionally, mtRNA from human tumor cells and primary mouse cells potently activated immune responses, indicating that the immunogenicity of mtRNA is an inherent property, unaffected by species or cellular background (Supplementary Fig. S2a).
Given that RNA receptors reside in the cytoplasm, we next evaluated whether antitumor therapy induces mtRNA release into the cytoplasm. We separated cytoplasmic components and other cellular precipitates (mitochondria and other insoluble materials) (Supplementary Fig. S2b), extracted cytoplasmic RNA (cRNAs), and quantified mtRNA content. Notably, mtRNAs such as cytochrome oxidase subunit 1 (Cox1) and cytochrome B (CytB) were significantly elevated in cytoplasmic extracts from chemotherapeutic drug-treated HeLa cells (Fig. 2b). Similar results were observed in HepG2 and B16 cells, where chemotherapy promoted Cox1 mtRNA release into the cytoplasm (Supplementary Fig. S2c, d). Collectively, these findings indicate that chemotherapy facilitates mtRNA release into the cytoplasm.
To investigate the role of cytoplasmic mtRNA in regulating immune responses, we isolated cRNAs from CIPT-treated (cRNA-CIPT) or untreated (cRNA-Ctrl) donor HeLa cells and transfected them into recipient HeLa cells. IFN-β levels remained unchanged in recipient cells transfected with cRNA-Ctrl but were significantly enhanced by cRNA-CIPT (Fig. 2c), suggesting that cytoplasmic mtRNA release contributes to immune response induction. Subsequently, since mtRNA transcription is mediated by mtRNA polymerase (POLRMT)
28, we treated cells with the POLRMT inhibitor IMT1
29. Chemotherapy-induced
Cox1 mtRNA release into the cytoplasm was significantly inhibited in IMT1-treated cells (Fig. 2d), paralleled by a reduction in chemotherapy-induced immune responses (Fig. 2e). Transfecting recipient HeLa cells with cRNAs from CIPT-treated cells (cRNA-CIPT) or CIPT plus IMT1-treated cells (cRNA-IMT1) further showed that IFN-β induction by cRNA was abrogated by IMT1 (Fig. 2f). The above results demonstrate that chemotherapeutic agents efficiently induce the release of mtRNA into the cytoplasm, thereby activating immune responses.
To confirm MAVS dependency, we transfected mtRNA from B16 cells into wild-type (WT) and
MAVS-knockout (
MAVS–/–) MEFs. MtRNA induced IFN-β in WT MEFs but not in MAVS
–/– MEFs (Supplementary Fig. S2e), establishing MAVS as crucial for mtRNA-triggered immune responses during antitumor therapy. These results collectively demonstrate that chemotherapy promotes mtRNA release into the cytoplasm, driving MAVS-dependent immune activation. Previous studies indicate that intracellular RNA receptors like TLR3, MDA5, and RIG-I recognize specific RNA structures were typically generated during viral infections
27,30. We hypothesized that mtRNA forms similar immunostimulatory structures. Using a monoclonal antibody (J2) to detect dsRNA
31, we observed dsRNA in normal uninfected cells, though less abundant than in vesicular stomatitis virus (VSV)-infected positive controls (Fig. 2g). Subcellular fractionation and J2 immunoprecipitation confirmed that mt-dsRNA localizes to mitochondria in untreated HeLa cells, not the cytoplasm (Fig. 2h), indicating basal mt-dsRNA presence in mitochondria. To test whether mtRNA immunogenicity relies on its double-stranded structure, we transfected mtRNA with or without dsRNA-specific RNase III treatment. Untreated mtRNA induced
IFN-β mRNA in HeLa and MEF cells, while RNase III-treated mtRNA did not (Fig. 2i), confirming that dsRNA is essential for IFN-β induction. Under normal conditions, immunogenic nucleic acids are sequestered in mitochondria, separated from cytoplasmic receptors. Confocal microscopy showed that mt-dsRNAs localize to mitochondria under basal conditions but are released into the cytoplasm following CIPT or H
2O
2 treatment (Fig. 2j, k). RNA immunoprecipitation (RIP) validated that CIPT induces the release of
Cox1 and
CytB mt-dsRNA into the cytoplasm of HeLa cells (Fig. 2l). Transfecting recipient cells with J2-purified dsRNA from chemotherapy-treated HeLa cytoplasm, but not J2-depleted RNA, significantly induced interferon expression, confirming that chemotherapy releases mt-dsRNA into the cytoplasm to trigger immune responses.
Finally, we explored the mechanism of chemotherapy-induced mt-dsRNA release. Damaged cells often activate BAK/BAX pores, which facilitate mtDNA release, while mitochondrial permeability transition pores (mPTPs) and ANT2 pores may also play roles
32,33. However, mPTP inhibition by Cyclosporin A (CSA) did not block
Nd1/CytB mtRNA release in DOX-treated HeLa cells or
Cox1/CytB/Nd5 mtRNA release in CIPT-treated cells, nor did it affect post-chemotherapy immune responses (Supplementary Fig. S3a–c).
ANT2 interference similarly did not impact chemotherapy-induced mtRNA release (ease (Fig. 2n; Supplementary Fig. S3d), ruling out mPTPs and ANT2 pores in mt-dsRNA release. Conversely, the BAX inhibitor peptide V5 (BIP-V5) blocked CIPT-induced
Cox1/CytB mtRNA release (Supplementary Fig. S3e) and IFN-β induction (Supplementary Fig. S3f), while BAX knockdown inhibited mtRNA release and subsequent immune responses (Fig. 2o, p; Supplementary Fig. S3g). These data establish that BAK/BAX pores are essential for mtRNA release and immune activation during chemotherapy. In conclusion, our study reveals that chemotherapy stimulates mitochondrial dsRNA release through BAK/BAX pores, triggering a MAVS-dependent immune response that combats tumor cells.
Tumor-induced PNPT1 suppresses the mtRNA-dependent immune response
We previously reported the critical role of mtRNA release in antitumor immunity and that chemotherapy activates the MAVS pathway via mtRNA release to enhance antitumor immunity. However, many patients show poor chemotherapy response and develop chemoresistance clinically. This suggests that tumors may evade the mtRNA-MAVS pathway through specific mechanisms to gain chemoresistance, despite its role in chemotherapy-induced antitumor immunity. Accordingly, this study aims to identify and elucidate the mechanisms underlying tumor escape from the mtRNA-MAVS pathway during chemoresistance development.
Due to intrinsic chemosensitivity differences across tumor cell lines, direct comparisons are confounded by genetic background and inherent biological variations. To eliminate such confounders and ensure reliable screening, we used chemosensitive and chemoresistant sublines derived from the same parental cell line. This design enables focused analysis of chemoresistance- and immune escape-related differences, minimizing inter-cell line background interference.
To identify factors contributing to resistance to antitumor therapy, we analyzed GEO datasets including multiple chemotherapy-resistant cancer cell lines. GSEA revealed that pathways related to mitochondrial gene expression were significantly enriched in chemotherapy-resistant ovarian cancer cells (Fig. 3a). Further coupling analysis revealed that the mitochondrial gene polyribonucleotide nucleotidyltransferase 1 (PNPT1), which is involved in the mitochondrial gene expression pathway, was the only significantly upregulated gene across 3 chemotherapeutic-resistant cell lines (Fig. 3b, c). Meanwhile, the analysis indicated that in the three resistant cell lines, there was no significant change in MAVS, and the expression trend of BAX was not uniform; only PNPT1 was significantly upregulated in all resistant cells, suggesting that PNPT1 is a common gene for tumor chemotherapy resistance (Supplementary Fig. S4a). While previous research highlighted PNPT1's role in mitochondrial nucleic acid metabolism
34,35, its specific influence on tumor development and therapy resistance remains unclear. Analysis of online data revealed significantly higher levels of
PNPT1 mRNA in liver cancer tissue compared to nontumor tissues. Both unpaired and paired analyses revealed that PNPT1 expression was markedly elevated in liver tumor tissues compared with adjacent normal tissues in the GEO dataset (Fig. 3d). Furthermore, IHC of matched patient samples from the HPA-LIHC datasets validated the higher expression of PNPT1 in tumor tissue (Fig. 3e). Additionally, upregulated Pnpt1 expression was observed in an N-RAS/Akt-induced mouse model of primary liver cancer (Fig. 3f). These findings suggest that PNPT1 is significantly upregulated in tumor tissues, indicating its role in tumor progression. Our findings indicate that mtRNA plays a crucial role in antitumor immunity (Fig. 2), while tumors upregulate PNPT1, which implies that PNPT1, as an mtRNA nuclease, may regulate the anti-tumor immune response. An analysis of IHC scores for PNPT1 and IFNA1 from hepatocellular carcinoma (HCC) patients in the HPA database revealed a negative correlation between PNPT1 and IFNA1 (Fig. 3g), suggesting that PNPT1 negatively regulates interferon expression. To investigate the role of PNPT1 in regulating anticancer immunity, we generated stable cell lines expressing shRNAs targeting
PNPT1 (sh
PNPT1-1 and sh
PNPT1-2). Both cell lines presented reduced PNPT1 protein levels, which remained consistently low over several generations (Fig. S4b). In B16 cells with stable sh
PNPT1 expression, we observed downregulation of
PNPT1 mRNA and upregulation of
CXCL10,
IFN-β, and
IL-6 mRNAs (Fig. 3h). Given that
PNPT1 knockdown promotes immune responses, we hypothesized that reducing PNPT1 levels in tumors could enhance chemotherapy effectiveness. Notably, in sh
PNPT1 B16 cells, DOX significantly increased the mRNA expression of
Cxcl10,
IFN-β,
IL-1β,
IL-6,
IL-8, and
TNF-α, whereas control cells presented only slight responses (Fig. 3i). Additionally, CIPT treatment induced significantly higher levels of
IFN-β,
IL-1β, and
IL-6 mRNA in sh
PNPT1 B16 cells than in control cells (Fig. 3j). In human HepG2 cells stably expressing sh
PNPT1, DOX significantly induced
IFN-β expression, whereas control cells did not (Fig. 3k, l). These results collectively indicate that
PNPT1 knockdown promotes immune responses in both mouse tumor cells and human cancer cells following chemotherapy. Our previous findings demonstrated that PNPT1 suppresses immune responses and promotes therapy resistance. We next evaluated the role of PNPT1 in regulating mitochondrial nucleic acid release and immune responses. Notably,
PNPT1 knockdown led to cytoplasmic dsRNA accumulation in B16 cells (Supplementary Fig. S4c). Additionally,
PNPT1 knockdown increased mtRNA release into the cytoplasm of B16 cells (Supplementary Fig. S4d), and this release was further amplified by DOX treatment (Supplementary Fig. S4e). The same experiment also yielded the same results in HeLa cells (Supplementary Fig. S4f). Lastly,
PNPT1 knockdown-promoted immune responses in B16 were abrogated by IMT1 treatment (Supplementary Fig. S4g), indicating that the regulation of immune responses by PNPT1 depends on mtRNA. Overall, these findings suggest that tumors upregulate PNPT1 to suppress mt-dsRNA release, thereby attenuating immune surveillance and inhibiting chemotherapy-mediated immune responses.
PNPT1 suppresses antitumor immunity in solid tumors
The solid tumor microenvironment drives immune escape and treatment resistance, underscoring the critical heterogeneity between tumor and normal cells
22. Building on our discovery that tumor-induced PNPT1 suppresses mtRNA-dependent immune responses in human and murine cancer cells (Fig. 3; Supplementary Fig. S4), we have explored its role in antitumor immunity within solid tumors. Using a B16 subcutaneous tumor model — characterized as an immunologically "cold" tumor with limited immune infiltration — we implanted stable sh
PNPT1-expressing B16 cells into immunodeficient NSG mice (severely immunodeficient mice) and immunocompetent C57BL/6 mice (Fig. 4a). In NSG mice, sh
PNPT1 and shNC tumors showed comparable volumes, weights, and sizes (Fig. 4b). In contrast, C57BL/6 mice bearing sh
PNPT1 tumors exhibited significantly reduced tumor volumes and weights relative to controls (Fig. 4c), indicating heightened sensitivity to host anticancer immunity. Notably, sh
PNPT1-treated tumors displayed fewer viable tumor cells and enhanced cytotoxicity (Fig. 4d), accompanied by increased infiltration of CD4
+ and CD8
+ T cells (Fig. 4e), suggesting that PNPT1 restricts immune invasion and tumor cell killing. Analysis via the UALCAN portal revealed elevated PNPT1 expression in metastatic SKCM and liver hepatocellular carcinoma (LIHC) tissues, with intermediate levels in primary tumors and low expression in normal tissues (Fig. 4f, g)
36. C57BL/6 mice injected with sh
PNPT1-B16 cells showed markedly reduced lung metastasis, fewer metastatic clones, lower lung weights, and decreased tumor nodules (Fig. 4h–l), demonstrating that
PNPT1 knockdown suppresses tumor metastasis
in vivo.
To assess the role of MAVS in this antitumor response, we used shRNA targeting MAVS (shMAVS) in B16 cells, which achieved near-complete depletion of MAVS mRNA (Fig. 4m). In stable shPNPT1-expressing cells, IFN-β and IL-6 mRNA levels were significantly upregulated, an effect abrogated by shMAVS co-expression. In mice, shPNPT1-B16 tumors showed reduced growth, but this inhibition was reversed when shMAVS was co-introduced (Fig. 4n, o), indicating that MAVS knockdown counteracts the antitumor effects of PNPT1 depletion.
Building on our prior findings that mtRNA triggers antitumor immune responses and PNPT1 negatively regulates this pathway in cell and subcutaneous tumor models, we investigated in vivo immune surveillance in primary liver cancer. Using a hydrodynamic injection model of NRasV12/AKT-driven HCC (Supplementary Fig. S5a), we delivered adeno-associated viruses carrying shRNA against murine Pnpt1 (shmPnpt1) or control shRNA (shNC). shPnpt1 treatment significantly reduced Pnpt1 mRNA and protein levels in the livers (Supplementary Fig. S5b, c). Six weeks post-injection, WT mice developed liver tumors, whereas shPnpt1 mice showed reduced tumor burden, lower liver weights, and decreased hepatic damage (Supplementary Fig. S5d–f), confirming that PNPT1 inhibition suppresses primary liver cancer development. Collectively, these data demonstrate that PNPT1 knockdown enhances anticancer immunity, restricts tumor metastasis, and suppresses tumor growth, highlighting PNPT1 as a key mediator of immune evasion and resistance to antitumor immunity.
Loss of PNPT1 overcomes therapeutic resistance
Despite significant advancements in cancer therapeutics, many malignancies develop drug resistance and evade immune surveillance, presenting substantial challenges to treatment efficacy. Previous studies have shown that tumor-induced PNPT1 expression suppresses the immune response, prompting us to further evaluate its role in therapeutic resistance.
GSEA analysis has revealed a negative correlation between PNPT1 expression and IFN-γ response pathways, which are pivotal for antitumor immune responses (Fig. 5a). Additionally, PNPT1 levels were negatively associated with CD8
+ T-cell scores in patients with SKCM (Fig. 5b). Furthermore, in SKCM, prostate adenocarcinoma (PRAD), and liver LIHC, PNPT1 expression inversely correlated with the expression of granzyme B (GZMB)
37, a key lymphocyte effector protein (Fig. 5c–e). These findings suggest that PNPT1 suppresses lymphocyte infiltration and antitumor function within tumors.
Analysis of TCGA data has demonstrated significantly elevated PNPT1 mRNA levels in various common tumors compared to non-tumor tissues. Both unpaired and paired analyses confirmed markedly higher PNPT1 expression in tumor tissues vs adjacent normal tissues (Supplementary Fig. S6a). Consistently, CPTAC data showed elevated PNPT1 protein levels in tumors relative to non-tumor tissues (Supplementary Fig. S6b). Lower PNPT1 expression was associated with favorable prognoses in LIHC patients (Supplementary Fig. S6c) and in adrenal cortex carcinoma (ACC), esophageal carcinoma (ESCA), chromophobe renal cell carcinoma (KICH), and low-grade glioma (LGG) (Supplementary Fig. S6d). Collectively, these results indicate that PNPT1 expression correlates positively with cancer incidence and progression.
Given the close relationship between tumor immune infiltration and antitumor therapy efficacy, we investigated how PNPT1 modulation affects cancer treatment. Following DOX administration, tumor volume remained stable in shNC-treated groups, whereas shPNPT1-treated tumors showed significant regression (Fig. 5f). In C57BL/6 mice, shPNPT1 combined with DOX for 14 days led to marked reductions in tumor weight and volume (Fig. 5g). Flow cytometry revealed increased tumor-infiltrating lymphocytes (TILs), including CD4+ and CD8+ T cells, in PNPT1-knockdown tumors after DOX treatment (Fig. 5h). These findings indicate that PNPT1 knockdown enhances DOX-mediated tumor growth inhibition, highlighting its role as a chemotherapeutic resistance factor and its negative correlation with immune cell infiltration.
Finally, we evaluated the impact of PNPT1 on PD-1 checkpoint blockade therapy. While PD-1 blockade has shown limited efficacy in solid tumors
38,39, sh
PNPT1-treated tumors exhibited reduced volume compared to shNC controls. Notably, the combination of sh
PNPT1 and PD-1 antibody treatment synergistically inhibited tumor growth, whereas single-agent treatments showed modest effects (Fig. 5i–k). Tumor weights in the combined treatment group were significantly lower than in all other groups, demonstrating that
PNPT1 knockdown enhances the efficacy of PD-1 antibody therapy.
Collectively, these results establish that PNPT1 suppresses antitumor immune responses and promotes tumor progression. Moreover, PNPT1 knockdown overcomes resistance to both chemotherapy and immune checkpoint blockade, positioning PNPT1 as a potential therapeutic target to enhance cancer treatment efficacy.
Pharmacological inhibition of PNPT1 in synergy with the BH3 mimetic drugs results in robust antitumor immunity and overcomes therapeutic resistance
Our previous findings suggest that mtRNA efficiently activates antitumor immunity, whereas the tumor microenvironment upregulates PNPT1 to inhibit this response. As noted previously, the clinical translation of STING agonists has been substantially hindered. To date, no therapeutics targeting the mtRNA-MAVS signaling axis have been reported. In response to this unmet need, we have pioneered a novel therapeutic strategy targeting the PNPT1-mtRNA axis. We first investigated lanatoside C (Lanc), an FDA-approved drug clinically used for treating chronic heart failure that was recently reported to inhibit PNPT1
40. However, its role in antitumor therapy has not been proven. While Lanc induced a weak immune response in macrophages, it had little effect on tumor cells, indicating that simply inhibiting PNPT1 may not be sufficient (Supplementary Fig. S7a–c). One possible reason for the ineffectiveness of PNPT1 inhibitors alone is the overexpression of antiapoptotic BCL-2 family members (such as BCL-2, BCL-XL, and MCL-1) in tumor cells and tissues
41,42, which prevent the formation of BAK/BAX pores necessary for mtRNA release (Fig. 2). Therefore, we designed a combination therapy using BH3-mimetic drugs
43-45, which inhibit the survival of BCL-2 family members and promote apoptosis by opening BAK/BAX pores (Fig. 6a). This class of drugs has been clinically used to treat blood cancers, but it is less effective against solid tumors. We selected Navitoclax (ABT-263) and Venetoclax (ABT-199) for our study. The former targets several antiapoptotic proteins, such as BCL-2, BCL-XL, and BCL-W, while the latter, already approved by the FDA, is used to treat chronic lymphocytic leukemia (CLL) and acute myelocytic leukemia (AML)
43,46. While neither Lanc nor BH3-mimetic drugs alone activated an immune response, their combination significantly induced one (Fig. 6b). Notably, ABT-263 demonstrated enhanced efficacy, potentially due to its broader targeting spectrum. The combined treatment further amplified the immune response triggered by DOX (Fig. 6c, d) and facilitated mtRNA release into the cytoplasm (Fig. 6e), with similar results observed in macrophages (Fig. 6f–h). This combination also effectively induced immune responses in various human cancer cells (Fig. 6i, j). Crucially, the combination did not compromise the apoptosis-inducing ability of BH3-mimetic drugs (Supplementary Fig. S7d) and maintained their tumor-killing efficacy (Supplementary Fig. S7e, f). These results indicate that combining BH3-mimetic drugs with the Lanc significantly activates the antitumor immune response while preserving the tumoricidal effects of BH3 mimetics.
We further tested the efficacy of the drug combination (ABT-263 + Lanc) in vivo using immunocompetent C57BL/6 mice implanted with B16 cells (Fig. 6k). The results indicated that while single drug treatments had limited effects, the combination significantly inhibited tumor growth (Fig. 6l–o). Remarkably, even at reduced dosage, substantial antitumor effects were still produced (Fig. 6l). Furthermore, the combined treatment inhibited tumor growth, reduced tumor size, and significantly prolonged survival (Fig. 6m–o). Additionally, this combination enhanced immune responses within the tumor tissue (Fig. 6p), and tumors presented fewer tumor cells, greater cytotoxicity and greater CD8+ T-cell infiltration in drug-treated tumors than in control tumors (Fig. 6q). Throughout the treatment course, the combination only slightly affected the body weight of the mice initially; however, the body weight returned to control levels thereafter (Supplementary Fig. S7g). Pathological assessments revealed no significant organ damage from the treatments (Supplementary Fig. S7h), confirming the safety of the drug combination in the mouse model. The alternative combination of ABT-199 and Lanc also resulted in tumor growth inhibition and significantly prolonged survival (Fig. 6r–t). In a clinical context where multimodal therapies are common, we observed that tumor volume remained largely unchanged with DOX or anti-PD-1 monotherapy. However, combining ABT-263 + Lanc significantly reduced tumor volume when combined with these treatments and significantly prolonged survival (Fig. 6u–w). These results suggest that the ABT-263 + Lanc combination can effectively overcome cancer resistance to anti-cancer therapy. Finally, we evaluated the potential of the combination drugs in treating primary liver cancer (Fig. 6x). The control mice developed liver tumors, whereas those treated with the combination presented a reduced tumor presence, significantly lower liver weights and significantly prolonged survival (Fig. 6y), and a reduced liver damage area (Fig. 6z), indicating that the combination effectively suppressed primary liver cancer development.
Our results highlight several advantages of the combination drug approach. First, Lanc inhibits PNPT1 and promotes the production of immunogenic mtRNA, whereas BH3-mimetic drugs effectively open BAK/BAX pores, facilitating mtRNA release into the cytoplasm to activate immune responses. Second, Lanc alone has poor antitumor effects, and while BH3-mimetic drugs are effective in hematologic tumors, they are less effective in solid tumors because of their immunosuppressive apoptotic effects, which can induce tumor antigen tolerance
47-49. By combining these drugs, we not only preserve the tumor-killing ability of BH3 mimetics but also effectively enhance the immune response, transforming "cold" tumors into "hot" tumors. This strategy overcomes the limitations of individual therapies, significantly improving overall antitumor efficacy, and both drugs have previously been approved for clinical use with good safety profiles. In summary, our data confirm that pharmacological inhibition of PNPT1 synergizes with BH3-mimetic drugs to induce robust antitumor immunity and overcome therapeutic resistance in solid tumors, highlighting that this combinatorial approach holds considerable promise for clinical translation.
DISCUSSION
The innate immune system serves as the body's indispensable first line of defense against microbial pathogens, while also maintaining tissue homeostasis. It rapidly activates in response to PAMPs and DAMPs, orchestrating immune clearance mechanisms. Beyond infections, the body confronts diverse stressors — including DNA damage, oxidative stress, and oncogene activation
50 — that can lead to the accumulation of damaged cells, elevating cancer risk. Thus, effective immune activation under such stressful conditions is critical for eliminating aberrant cells and preventing malignant transformation. Innate immunity also plays a pivotal role in tumor therapy: radiotherapy and chemotherapy enhance cancer patient survival not only by directly eradicating tumor cells but also by potentiating immune responses and sensitizing tumors to immunotherapies
12. However, the mechanisms underlying immune activation during antitumor therapies require deeper exploration to unlock their full therapeutic potential.
Our research illuminates the central role of mitochondria-derived nucleic acids in triggering immune responses during antitumor interventions (Fig. 7). Chemotherapy induces the release of mtRNA into the cytoplasm, a process that activates innate immunity through the formation of dsRNA structures. In normal cells, dsRNA is compartmentalized within mitochondria, preventing autoimmunity while priming the cell for rapid immune responses to future stimuli. During chemotherapy, mitochondrial dsRNA is released via BAX/BAK channels, activating the MAVS signaling pathway and inducing the expression of interferons and inflammatory factors. This immune cascade is essential for recruiting immune cells and facilitating tumor clearance.
Notably, the cGAS-STING signaling pathway — traditionally targeted for its role in detecting tumor DNA
11,12,51 — has faced therapeutic challenges. Evidence shows STING expression is downregulated in numerous tumor types, and certain chemotherapies fail to activate this pathway, highlighting the limitations of targeting cGAS-STING alone
19. In contrast, the widespread expression of MAVS-related components across tumor tissues suggests that mtRNA-derived danger signals can effectively trigger antitumor immunity in a broad range of cancers.
Tumor microenvironments exist in a state of heightened stress due to intrinsic metabolic dysregulation and environmental pressures, enabling tumors to suppress immune responses and evade surveillance
20-22. While chemotherapeutic agents induce cellular stress to exert antitumor effects, solid tumors often exhibit profound drug resistance, reflecting their functional heterogeneity compared to normal cells. Our findings reveal that the mtRNA metabolic enzyme PNPT1 is upregulated in tumor tissues, where it specifically inhibits the formation and release of immunogenic mitochondrial dsRNA, thereby suppressing innate immunity and immune-mediated clearance. Knocking down
PNPT1 in both normal and tumor cells enhances dsRNA release, accelerating immune elimination of damaged cells. Clinical data confirm elevated
PNPT1 mRNA and protein levels in various malignancies, implicating PNPT1 in dampening antitumor immune responses
. In vivo studies further demonstrate that reducing PNPT1 inhibits tumor growth and synergizes with chemotherapy and immune checkpoint therapies, underscoring its role in promoting tumor progression and therapeutic resistance.
In summary, our research uncovers a novel mechanism whereby mitochondrial nucleic acids act as stress-associated molecular patterns during tumorigenesis and antitumor therapy, triggering immune surveillance and clearance. The formation and release of mitochondrial dsRNA emerge as key drivers of innate immune activation in response to stress and infection. Malignant cells counteract this by upregulating PNPT1, which suppresses dsRNA-mediated signaling to facilitate immune evasion and treatment resistance. Critically, pharmacological inhibition of PNPT1 in combination with BH3-mimetic drugs potently activates mtRNA-mediated antitumor immunity, overcoming resistance in solid tumors without systemic toxicity. These findings establish a promising combinatorial strategy to enhance cancer immunotherapy with potential for safe and effective clinical translation.
MATERIALS AND METHODS
Key resources table
Animal studies
C57BL/6 WT mice were obtained from the Hubei Research Center of Laboratory Animals (Wuhan, China). MAVS–/– mice on a C57BL/6 background were generously provided by Dr. Hongbin Shu from Wuhan University (Wuhan, China). NSG mice were sourced from Gempharmattech Co., Ltd (Nanjing, China). All animal experiments were approved by the IACUC of the College of Life Sciences, Wuhan University. The studies adhered to the Animal Welfare Act and the National Institutes of Health guidelines for the care and use of experimental animals in biomedical research.
For the CIPT experiment, six- to eight-weeks-old mice were intraperitoneally injected with 15–25 mg/kg of CIPT and monitored daily for durability. After 48–72 h, serum samples were collected from the mice to assess indicators of inflammatory response and organ damage. Additionally, mouse tissues were harvested for IHC and quantitative real-time PCR (qRT-PCR) analyses.
For tumor challenge, 5 × 105–2 × 106 B16 cells were suspended in Hank’s balanced salt solution and subcutaneously injected into the C57 mice right underarm on day 0. For NSG mice, 2 × 105 B16 cells were suspended in Hank’s balanced salt solution and subcutaneously injected into the right underarm on day 0. After the initial formation of the tumor, the length and width of the tumor were tested every two days until the end of the experiment. For the tumor chemotherapy experiment, the mice were injected with 4 mg/kg DOX intraperitoneally on the 6th and 9th day, respectively. For the immune checkpoint blocking experiment, the mice were intraperitoneally injected with 100 µg PD-1 antibody on days 6, 9 and 12. The tumor volume was estimated using the formula (length × width × width)/2. After euthanasia, tumor tissues were harvested for immunohistochemistry and flow cytometry analyses.
Mouse liver cancer was induced by hydrodynamic injection of Sleeping Beauty transposon-overexpressing myr-AKT/NRASV12. Each mouse was injected with plasmid containing pT3-myr-AKT-HA (10 µg), pT/Caggs-NRASV12 (10 µg), and pCMV (CAT)T7-SB100 (2 µg). After 6–12 days, the livers were harvested for IHC, quantitative immune cell infiltration and N-Ras clearance. After 5–8 weeks, the livers were harvested for IHC and quantitative assessment of cancer progression.
For IHC experiments, different organs of mice were fixed in 4% paraformaldehyde, embedded in paraffin, sectioned, and stained with a color stain or antibody, then observed under a microscope.
For the flow cytometry experiment, the mice’s tumor tissues were cut, digested with collagenase at 37 °C for 30 min, and the cell suspension was separated. The number of cells was counted, stained with antibodies, and then analyzed on a flow cytometer (Beckman).
Cell cultures and treatment
HeLa cells, MEFs, Vero cells and HEK293 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, Grand Island, NY, USA) supplemented with 10% Fetal Bovine Serum, 100 U/mL penicillin, and 100 μg/mL streptomycin sulfate. All cells were cultured in an incubator at 37 °C in 5% CO2.
For drug treatment, cells at approximately 60–70% confluency were treated with 0–5 μM Dox, 0–20 μM CIPT and 0–100 μM H2O2 for 6–24 h, which were then collected for subsequent analysis.
Lentivirus
Plko.1-shCrtl, Plko.1-shPNPT1 and Plko.1-shMAVS were constructed, and the stably expressed viral products were generated. In short, the constructed plasmid and packaging plasmid were transfected into HEK293T cells. After 48 h, the cell supernatant was collected and then filtered with a 0.45-um filter to obtain lentivirus, which was used to infect the target cells for 48 h. Puromycin was used to screen positive cells. The sequence of shRNA was obtained from Sigma.
Immunoblotting
Cells were lysed in a buffer containing 50 mm Tris-HCl (pH 7.5), 0.5 mm EDTA, 150 mM NaCl, 1% NP40 and 1% SDS, then protease inhibitor and a phosphatase inhibitor (Roche) were added to the lysate, which was briefly ultrasonicated or overturned at 4°C for 1 h and centrifuged to collect the supernatant. Then, protein buffer was added to the supernatant for electrophoresis, denatured at 100 °C for 5 min, and SDS-PAGE was used for electrophoresis. Following this, the membrane was transferred, and the indicated antibodies were used to detect the target protein.
qRT-PCR and ELISA
Total RNAs were isolated from the tissues or cells using Trizol reagent (Invitrogen). Then, cDNA was generated from 1 µg of isolated RNA by reverse transcription with a reverse transcription mix (Vazyme Biotech Co., Ltd., China). Specific primers and ChamQ SYBR qPCR Master Mix (Vazyme Biotech Co., Ltd., China) were used for the RT-PCR reactions. For ELISA experiments, an ELISA kit was used to determine the levels of cytokines in cell supernatant or mouse serum. All subsequent experimental steps were performed according to the manufacturer's instructions.
Immunofluorescence and confocal microscopy analysis
Cells were cultured in confocal dishes for 24–48 h, after which the supernatant was removed and the cells were fixed with 4% paraformaldehyde for 15 min. Following three washes with PBS, 0.1% Triton X-100 was added to permeabilize the cell membrane for 5 min, then the cells were blocked with 5% BSA for 45 min. A specific primary antibody was added, and the cells were kept at 4 °C overnight. After washing, the cells were incubated with a secondary antibody for 1 h and then washed three times. The nuclei were labeled with DAPI (1 µg/mL), after which the cells were observed under a confocal microscope.
Mitochondrial extraction and mtRNA treatment
Mitochondria were isolated from cultured cells using a Mitochondrial Isolation Kit (Thermo Fisher Scientific, 89874). Trizol reagent was added to the purified mitochondria to extract RNA from mitochondria. After obtaining mtRNA, a total amount of 2–5 µg mtRNA was transfected into new untreated cells. After 12 h, the treated cells’ RNAs were extracted for subsequent analysis. RNase III (M0245S, NEB) was used for RNase treatment according to the reagent instructions, followed by RNase-treated mtRNA transfection and analysis.
Detection of mtRNA in cytosolic extracts and cRNA treatment
The digitonin extraction method was used to extract cRNA. Briefly, the cells were treated with digitonin (20 μg/mL). The purity of the cytosolic components was detected by western blot. Then, the cRNA was extracted, and mtRNA-specific primers were used for qPCR. For the cRNA transfection, a total amount of 2–5 µg cRNA was transfected into new untreated cells. After 12 h, RNAs were extracted from the treated cells for subsequent analysis.
For the RNase treatment experiment, RNase III (M0245S, NEB) and RNase H (D7089, Beyotime) were used to treat cRNA and R-loop according to the manufacturer’s instructions, and then RNase-treated cRNA and R-loop were transfected and analyzed.
Immunoprecipitation of cytoplasm mtdsRNA and mt R-loop
Cells were treated with digitonin (20 μg/mL). Cell lysate was prepared and transferred to 1.5 mL of a fresh RNA-free enzyme tube and centrifuged at 4 °C at 12,000 rpm for 10 min. The supernatant was transferred to a fresh tube, dsRNA J2 and R-loop S9.6 antibody were added, and kept on a rotor at 4 °C for 1–2 h. The sample tubes were added with Protein G and then kept on the rotor at 4 °C for 2-h centrifugation. The supernatant was removed, washed twice, and J2 antibody-bound dsRNA was extracted with Trizol reagent. S9.6 antibody-bound R-loop was isolated by phenol chloroform extraction, then mt-dsRNA and mitochondrial R-loop were detected by qPCR using the mtRNA-specific primers.
ROS detection
A ROS Kit (Beyotime, s0033s) was used to detect ROS in the cultured cells (> 10,000 cells) using flow cytometry following the manufacturer’s instructions.
TCGA data and online data analysis
Gene expression profiles of patient samples were downloaded from the TCGA dataset. Clinical data were downloaded from the University of California Santa Cruz Xena dataset. Patient proteome data were obtained from the CPTAC dataset. RNA-Seq data of drug-resistant cell lines (GSE270030, GSE140077, GSE222187) and drug-treated cell lines (GSE223698, GSE81878, GSE235908, GSE108214) were available on the GEO database. We analyzed the expression of multiple nucleic acid sensors in different tissues or cell types using public databases such as HPA. Gene expression levels in different tissue cells were obtained from the HPA. We used the RNA dataset from the HPA, including RNA single-cell type data and RNA HPA tissue gene data. IHC data from HPA were available on the website. The RNA-seq data were normalized using the DESeq2 package or converted to TPM values. The immune cell infiltration levels of each SKCM sample were evaluated using Cibersortx. For gene-set enrichment analysis, the gene expression matrix of SKCM was divided into two parts based on the PNPT1 expression level. The pre-ranked data were uploaded to GSEA 4.1.0, and the enrichment of MSigDB C2 gene sets was analyzed with 1,000 random permutations to obtain P values, q values, and NES. For analyzing the association of PNPT1 expression with survival and tumor metastasis, the online UALCAN portal and GEPIA were used to assess the impact of PNPT1 expression on patient survival rates and metastasis.
Statistical analysis
The related results were expressed as the mean ± SEM or SD. Statistical methods are described in detail in the figure legend. The Prism v9 software was used for statistical analysis. P < 0.05 was statistically significant.
DATA AND MATERIALS AVAILABILITY
All data are available in the main text or the supplementary file.
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/).
