DNA-PK-mediated CRTC2 phosphorylation promotes NHEJ and suppresses antitumor immunity via relocation to repair complexes Authors: Fangdi Zou, Zhiqi Yao, Xiaohan Dong, Yang Liu, Zhanjie Xiu, Meixi Wang, Xiaofeng Liu, Xinlu Wang, Lin Sun, Jian Wang, Shuai Ma, Heng Zhang, Kaishun Hu, Yan Sun, Kexin Chen, Zhiyong Yuan, Lei Shi, Zhiqiang Wu, Zeyun Mi Published: 2026 ABSTRACT AbstractGenotoxic stress or exogenous DNA damage induces transcription arrest, enabling efficient DNA repair. Transcription activators directly participate in DNA damage repair (DDR), but the trans-regulatory mechanisms linking transcription and DDR remain elusive. Here we reveal that CRTC2 switches from a transcriptional coactivator to a DNA-damage responder. CRTC2 promotes non-homologous end joining (NHEJ) in vitro and in vivo. Mechanistically, PARP1 recruits CRTC2 to DNA breaks, where CRTC2 promotes DNA-PKcs enrichment and DNA-PK holoenzyme assembly, driving NHEJ. DNA-PK phosphorylates CRTC2 at Ser433, dissociating it from transcriptional complexes to suppress target gene transcription and promoting its incorporation into repair complexes, forming a positive feedback loop that enhances NHEJ. CRTC2 loss radiosensitizes liver cancer cells, potentiates irradiation-induced cGAS–STING activation, and promotes antitumor immunity and the abscopal effect. AAV8-mediated targeting of CRTC2 sensitizes tumors to radioimmunotherapy. Thus, CRTC2 couples transcriptional silencing to DNA repair, and its inhibition offers a promising strategy for radioimmunotherapy sensitization.Subject terms: Non-homologous-end joining, Radiotherapy, Cancer immunotherapy, Liver cancer Introduction Cells are continually subjected to genotoxic stress, both from endogenous sources such as reactive oxygen species and replication fork collapse, and from exogenous sources including ultraviolet radiation, ionizing radiation (IR), and chemotherapeutics1,2. This stress results in diverse forms of DNA damage. To cope with stress-induced DNA damage, cells activate the DNA damage response (DDR), a surveillance mechanism that coordinates DNA repair and transcriptional regulation to prevent transcription–repair conflicts3. Different types of DNA lesions elicit distinct transcriptional responses4,5. For instance, double-strand breaks (DSBs) trigger localized transcriptional arrest in the immediate vicinity of the lesion6, while most base adducts are readily bypassed by elongating RNA polymerase II (Pol II) and provoke only transient pausing7. In contrast, certain lesions, such as those induced by UV irradiation or platinum-based chemotherapeutics, not only impede Pol II progression along the transcribed strand but also trigger a global shutdown of transcription8,9. Importantly, pharmacological disruption of transcriptional silencing induced by DNA damage impairs the efficiency of DNA repair pathways10. These results indicate that targeted transcriptional silencing is essential for efficient DNA repair. The proper regulation of DNA damage-induced transcription stress is essential for maintaining genome integrity. DSBs represent one of the most lethal forms of DNA damage and are a leading cause of cell death. To counter this threat, eukaryotic cells primarily employ two major pathways to repair DSBs: non-homologous end joining (NHEJ) and homologous recombination (HR)11,12. Understanding the transcriptional response to such lesions remains a central challenge in DSB repair research10,13. Current evidence indicates that DSB-induced transcriptional silencing is multilayered, involving histone modifications and direct inhibition of RNA polymerase II (Pol II) at break sites, and this process is orchestrated primarily by ATM, DNA-PK and PARP110,13,14. Histone modifications help to establish a compacted chromatin configuration around break sites that impedes Pol II binding10,14. Negative regulators such as RECQL5 and the NELF complex contribute to repression by binding directly to Pol II and suppressing both initiation and elongation15,16. Additionally, PARP1-mediated poly (ADP-ribosylation) of the positive elongation factor P-TEFb disrupts its phase separation capacity, thereby inhibiting Pol II progression after damage17. In addition, RNA: DNA hybrids (R-loops) accumulate at DSBs within transcribed genomic regions as a consequence of transcriptional shutdown18–20. Despite these advances, the potential role of transcriptional activators in coordinating DSB repair with transcriptional silencing remains elusive. Emerging evidence demonstrates that a substantial proportion of transcription activators are recruited to DSBs and facilitate repair. In prostate cancer, the elongation factor ELL2 mediates the recruitment and retention of KU70/KU80 at DSBs and promotes NHEJ21. The E4F transcription factor 1 (E4F1) undergoes PARP1-dependent recruitment to breaks, where it enhances ATR/CHK1 signaling, resection, and HR22. Additionally, the circadian rhythm regulators BMAL1/CLOCK complex localize to DSBs, stimulate DNA end-resection to generate single-stranded DNA (ssDNA), and facilitate HR23. These observations support a model in which certain transcriptional activators dynamically transition from canonical transcriptional complexes to dedicated DDR machinery, leading to transient suppression of their gene expression programs. How these factors precisely coordinate repair with transcriptional regulation remains an open question. CREB-regulated transcription coactivator 2 (CRTC2) serves as a transcriptional coactivator for CREB and integrates signaling inputs from cAMP and Ca²⁺ pathways across diverse tissues24. CRTC2 is well-established as a key metabolic regulator in the liver, governing the expression of enzymes such as PEPCK, G6Pase, and SREBP225,26. Our prior work showed that CRTC2 activates P-TEFb through 7SK snRNP disassembly to promote Pol II transcriptional elongation27. CRTC2 further recruits CBP/p300 to deposit activating histone marks (H3K27ac) at inflammatory genes, enhancing cytokine expression28. Despite extensive evidence of CRTC2’s transcriptional roles, its potential involvement in DSB repair remains unexplored. Intriguingly, our recent study demonstrated that CRTC2 upregulation in hepatocellular carcinoma (HCC) via copy number amplification promotes tumorigenesis and lenvatinib resistance29. Parallel studies revealed that CRTC2 promoter hypomethylation elevates its expression in HCC, with high levels correlating with poor prognosis30. Collectively, these findings highlight the clinical value of targeting CRTC2 in liver cancer treatment strategies. In this study, we identify a previously unrecognized, transcription-independent role for CRTC2 in DSB repair. We show that upon DNA damage, PARP1 activation rapidly recruits CRTC2 to break sites. CRTC2 promotes DNA-PKcs enrichment and facilitates DNA-PK holoenzyme assembly and activation, driving NHEJ repair. Notably, the activated DNA-PK complex phosphorylates CRTC2 at Ser433, reducing its affinity for transcriptional complexes and consequently inhibiting transcriptional activity. Concurrently, this phosphorylation enhances CRTC2 co-localization with the DNA repair complex at damage sites, thereby amplifying NHEJ efficiency. Furthermore, elevated CRTC2 expression in liver cancer confers radioresistance and suppresses irradiation-induced activation of the cGAS–STING pathway and abscopal immune responses. Our results position CRTC2 as a dual-functional regulator that couples transcriptional silencing to DNA repair, and we propose its targeting as a strategy for overcoming radioimmunotherapy resistance. Results CRTC2 deficiency impairs genomic integrity and DNA damage repair To investigate the role of CRTC2 in DSB repair, we first assessed its contribution to genomic integrity maintenance by quantifying spontaneous chromosomal and chromatid breaks in wild-type (Crtc2+/+) versus Crtc2-knockout (Crtc2−/−) mouse embryonic fibroblasts (MEFs). CRTC2-deficient MEFs exhibited a significant increase in such breaks (Supplementary Fig. 1a), suggesting impaired DNA integrity. This finding was corroborated by elevated DNA damage signals, as indicated by γH2AX level, across multiple tissues from Crtc2−/− mice compared to controls after IR (Supplementary Fig. 1b, c). Furthermore, compared to Wt controls, Crtc2-/- mice showed accelerated mortality and a reduced long-term survival plateau (Supplementary Fig. 1d), suggesting that Crtc2 loss increases IR sensitivity in mice. Given the critical role of hepatic CRTC2 dysregulation in liver cancer29–31, we focused on models of this malignancy. Using CRISPR/Cas9, we generated stable CRTC2-knockout liver cancer cell lines (HCCLM3 and SK-HEP1, Supplementary Fig. 1e). Following irradiation, sgCRTC2 cells displayed persistent γH2AX retention compared to sgControl cells (Supplementary Fig. 1e-g), indicating unresolved DNA damage. Comet assays further quantified this repair defect, revealing significantly increased DSBs in irradiated sgCRTC2 cells, evidenced by elevated tail moment and tail DNA percentage (Supplementary Fig. 1h). Conversely, CRTC2-overexpressing cells showed markedly reduced γH2AX post-irradiation (Supplementary Fig. 2a-d). Corroborating this, comet assays demonstrated significantly decreased DSBs in these cells (Supplementary Fig. 2e). These findings showed that CRTC2 deficiency impairs DNA damage repair efficiency. To validate the clinical relevance of CRTC2 in human liver cancer, we analyzed a published proteomic/phosphoproteomic dataset OEP000321, which includes 159 patients (56 with aristolochic acid (AA) exposure and 103 non-AA-associated cases)32. Since HBV-related HCC with an AA signature represents a distinct subgroup characterized by unique clinicopathological and molecular features32 and marked DNA repair deficiency33, we focused our subsequent analysis on the 103 non-AA-associated patients. Pearson correlation analysis of this cohort revealed a significant negative correlation between γH2AX and CRTC2 expression (Supplementary Fig. 2f). Western blotting of 20 human liver cancer samples confirmed this inverse relationship: high CRTC2 expression correlated with low γH2AX, and vice versa (Supplementary Fig. 2g, h). This supports a reduced DSB burden in CRTC2-high tumors. Consistent with these findings, univariate and multivariate regression analyses of DDR-related genes in TCGA-liver cancer data revealed significantly higher DDR scores in CRTC2-high liver cancers compared to CRTC2-low liver cancers (Supplementary Fig. 2i). In summary, our data reveal that CRTC2 is critical for preserving genomic integrity and efficient DNA repair. High CRTC2 expression correlates with fewer DSBs and potentially stimulates the DDR pathway in liver cancer, while its deficiency exacerbates DNA damage. CRTC2 specifically promotes NHEJ-dependent DNA repair in vitro and in vivo To define the specific role of CRTC2 in the DSB repair pathway, we employed DR-GFP (HR reporter) and EJ5-GFP (NHEJ reporter) systems in U2OS cells. In these reporter assays, DSBs induced by I-SceI endonuclease are repaired through HR or NHEJ, leading to restored GFP expression that is quantified by flow cytometry (Supplementary Fig. 3a; Fig. 1a). CRTC2 overexpression significantly increased the proportion of GFP + cells in the NHEJ assay while reducing that in the HR assay (Fig. 1b; Supplementary Fig. 3b). Conversely, CRTC2 knockdown impaired NHEJ-mediated repair, reducing GFP+ cells in NHEJ while increasing them in HR assays (Fig. 1c; Supplementary Fig. 3c). Moreover, Homologous Recombination Deficiency (HRD) analysis using TCGA data linked high CRTC2 expression to significantly elevated HRD scores, indicating HR repair deficiency (Supplementary Fig. 3d). Given the established role of 53BP1 in promoting NHEJ by limiting DNA end resection34, we examined its dynamics as a proxy for NHEJ efficiency. CRTC2 deficiency significantly delayed the recruitment dynamics of 53BP1 foci (Fig. 1d, e), indicating impaired NHEJ repair. CRTC2 promotes NHEJ repair in vitro and in vivo. a Schematic diagram of I-SceI-induced NHEJ repair in EJ5-GFP U2OS cells. b NHEJ repair efficiency (left) in response to CRTC2 overexpression, and western blot analysis of I-SceI and CRTC2 expression (right) in EJ5-GFP U2OS cells (n = 3 biologically independent experiments). c NHEJ repair efficiency (left) in response to CRTC2 knockdown, and western blot analysis of I-SceI and CRTC2 expression (right) in EJ5-GFP U2OS cells (n = 3 biologically independent experiments). Immunofluorescence staining (d) and quantitative analysis (e) of 53BP1 foci formation at 0, 0.5, 2, 4, 8, and 12 h after IR (6 Gy) in control and CRTC2-knockout HCCLM3 cells (n = 40 cells per group). Scale bars, 10 µm. f Schematic diagram of VDJ rearrangement and primer locations in pro-B cells. g Pro-B cell sorting process with CD19, B220, IgM, and CD43 from spleen of Crtc2+/+ and Crtc2−/− mice (left) and mRNA level of distal and proximal VDJ rearrangement detection by qPCR (right) (n = 3 mice per group). h Schematic diagram of mouse immunoglobulin class switch recombination (CSR) in mature B cells. i Serum IgM, IgG1, IgG2a, IgG2b, IgG3, and IgA levels in Crtc2+/+ and Crtc2−/− mice (n = 10 mice per group). j Splenic B cells cultured under LPS or LPS/IL4 stimulation following CSR to IgG3 (left) and IgE (right) (n = 3 mice per group). All data are shown as the mean ± SEM. Comparisons between two groups were analyzed by Student’s t test. Comparisons between three or more groups with two or more conditions were assessed by two-way ANOVA. All statistical tests were conducted as two-sided. Source Data are provided with this paper. V(D)J recombination in jawed vertebrates generates antibody and T cell receptor diversity by assembling V, D, and J gene segments35. This process requires creating and repairing DSBs, primarily mediated by the NHEJ pathway35 (Fig. 1f). To validate CRTC2’s role in physiological NHEJ, we quantified V(D)J recombination at the IgH locus in splenic pro-B cells. Crtc2−/− mice exhibited significantly fewer distal and proximal VH-DH-JH recombination events than did wild-type controls (Fig. 1g), confirming CRTC2’s in vivo role in NHEJ-dependent recombination. We next evaluated CRTC2’s role in class switch recombination (CSR) in mature B cells, another NHEJ-dependent process (Fig. 1h). Crtc2−/− mice showed normal serum IgM but significantly reduced IgG1, IgG2a, IgG2b, IgG3, and IgA (Fig. 1i). We isolated naive B lymphocytes from the spleens of Crtc2+/+ or Crtc2−/− mice, which were stimulated with lipopolysaccharide (LPS) alone or LPS plus interleukin-4 (IL-4) to test the CSR to IgG3 or IgE. Our results showed that CRTC2 deficiency did not affect AID protein level or live cell numbers (Supplementary Fig. 3e, f), but class switching to both IgG3 and IgE was decreased in Crtc2−/− B cells following cytokine stimulation (Fig. 1j). These findings indicate that CRTC2 is essential for regulating immunoglobulin CSR. We also examined B220 + CD19 + B cell or CD4 + CD8a + T cell population in spleen or thymus between Crtc2+/+ and Crtc2−/− mice. There were no differences in lymphocyte composition and differentiation in the spleen or thymus (Supplementary Fig. 3g, h), indicating that impaired CSR in Crtc2−/− mice is not due to immune system defects. These results indicated that CRTC2 regulates CSR independent of lymphocyte development. Excessive NHEJ activity diverts DNA repair toward error-prone pathways, propagating misjoined DNA breaks that generate loss of heterozygosity (LOH), large-scale state transitions (LST), and telomeric allelic imbalances (TAI)36–38. These lesions collectively elevate the whole-genome instability index (wGII) through progressive genomic destabilization. Consistent with this mechanism, TCGA liver hepatocellular carcinoma (LIHC) data analysis revealed that elevated CRTC2 expression significantly correlates with increased frequencies of LOH, LST, and TAI, and is concurrently associating with higher wGII (Supplementary Fig. 3i, j), suggesting that CRTC2 promotes error-prone NHEJ repair in human liver cancer. Taken together, these data demonstrate that CRTC2 specifically promotes NHEJ-mediated DNA repair in vitro and in vivo. CRTC2 recruitment to DSB sites requires the activity of PARP1 To determine whether CRTC2 directly participates in the NHEJ pathway, we monitored its recruitment to DNA lesions using laser micro-irradiation coupled with live-cell imaging. CRTC2 accumulated at break sites within 4 min post-damage and dissociated after 16 min (Fig. 2a), indicating its dynamic recruitment. This was confirmed using a LacI-LacO targeting system (Fig. 2b, c), a method designed to monitor DSB dynamics at specific engineered genomic loci39. Furthermore, we employed the AiDIvA cell system, in which AsiSI expression is induced by doxycycline and 4-hydroxytamoxifen (4-OHT), to trigger the inducible generation of DSBs genome-wide40 (Fig. 2d). CUT&Tag analysis confirmed CRTC2 enrichment proximal to DSB sites relative to undamaged genomic regions (Fig. 2e). Collectively, these data demonstrate that CRTC2 is recruited to DSB sites and acts as an early responder to DNA breaks. Activated PARP1 recruits CRTC2 to DNA breaks. a Images of eGFP-CRTC2-transfected HEK293T cells at 0, 4, 8, 12, 16, and 18 min after laser micro-IR treatment; white line represents the laser path. Scale bar, 20 µm. b Schematic diagram of Shield1- and 4-OHT-induced ER-mCherry expression and DSB generation. c Representative micrographs of eGFP or eGFP-CRTC2 foci formation in FokI-mCherry-LacI cells; white dashed line marks the nuclei. Scale bars, 20 µm. d Schematic diagram of doxycycline (Dox)- and 4-OHT-induced AsiSI expression and DSB generation for CUT&Tag sequencing. e Heatmaps of DSB and non-DSB sites of Dox- and 4-OHT-treated AiDIvA cells in CUT&Tag-seq with anti-CRTC2 and anti-γH2AX. f DNA damage repair proteins detected in LC-MS/MS analysis. g Co-IP assay examining the interaction of CRTC2 with PARP1 in HCCLM3 cells. h Immunofluorescence images of the co-localization of CRTC2 and PARP1 in HCCLM3 cells; white dashed line marks the nuclei. Scale bar, 20 µm. i Immunofluorescence images (left) and quantitative analysis (right) of the co-localization of CRTC2 and γH2AX in control and PARP1-knockdown HCCLM3 cells (si-Control, n = 120 cells; si-PARP1, n = 123 cells); white dashed line marks the nuclei. Scale bars, 20 µm. j Representative micrographs (left) and quantitative analysis (right) of CRTC2 foci formation ratios in the absence or presence of si-PARP1 in FokI-mCherry-LacI cells (n = 3 biologically independent experiments; 100 cells were examined per group); white dashed line marks the nuclei. Scale bars, 20 µm. k Pull-down assay with biotin-PAR and purified CRTC2-TY1. l Co-IP assay examining the interaction of CRTC2 with PARP1-MYC in the absence or presence of Olaparib (10 µM, 24 h) in HCCLM3 cells. m Immunofluorescence images (left) and quantitative analysis (right) of the co-localization of CRTC2 and γH2AX in the absence or presence of Olaparib (10 µM, 24 h) in HCCLM3 cells (n = 116 cells per group); white dashed line marks the nuclei. Scale bars, 20 µm. All data are shown as the mean ± SEM. Comparisons between two groups were analyzed by Student’s t test. All statistical tests were conducted as two-sided. Source Data are provided with this paper. To identify CRTC2 interactors involved in the DDR, we performed mass spectrometry of CRTC2-associated proteins following IR and pulldown using CRTC2-coated beads. CRTC2 co-purified with PARP1 and core NHEJ factors, including KU70, KU80, and DNA-PKcs, ranking among the highest in the interactome profile (Fig. 2f). Given PARP1’s function as a rapid DNA damage sensor (arriving within seconds) and prior reports of PARP-dependent transcription factor recruitment to DNA lesions41, we next investigated whether PARP1 mediates CRTC2 localization to DSB sites. Co-immunoprecipitation (Co-IP) revealed enhanced CRTC2–PARP1 binding post-IR (Fig. 2g), and immunofluorescence confirmed their increased co-localization at damage sites (Fig. 2h). To evaluate the clinical relevance of this interaction, we examined two PDX tissue samples. Proximity ligation assay (PLA) revealed that IR significantly promoted nuclear co-localization of PARP1 and CRTC2 (Supplementary Fig. 4a). Crucially, PARP1 knockdown attenuated CRTC2 accumulation at DSBs, as validated by both the laser micro-irradiation assays and the LacI-LacO system (Fig. 2i, j). Conversely, CRTC2 knockdown did not affect PARP1 recruitment (Supplementary Fig. 4b). Given that PARP1 binds DNA breaks and undergoes auto-PARylation, generating a negatively charged poly(ADP-ribose) (PAR) scaffold that recruits repair factors42. To determine whether a direct interaction exists between CRTC2 and PAR, we performed an in vitro co-incubation assay. Biotin-labeled PAR was incubated with purified CRTC2 protein and captured with streptavidin beads. CRTC2 was detected in the precipitated complexes (Fig. 2k), demonstrating its ability to bind PAR directly in vitro. Consistent with this finding, PARP1 catalytic inhibitor Olaparib substantially attenuated both the IR-induced PAR surge (Supplementary Fig. 4c) and the enhanced CRTC2–PARP1 interaction (Fig. 2l), along with significantly reducing CRTC2 recruitment to DSBs in micro-irradiation assays (Fig. 2m). Collectively, these results indicate that PARP1 and its catalytic activity are essential for the recruitment of CRTC2 to DSB sites. CRTC2 promotes activation of the DNA–PK complex To further elucidate the mechanism by which CRTC2 promotes NHEJ repair, we examined its role in recruiting core NHEJ factors identified as top CRTC2 interactors: the KU70/KU80 heterodimer and DNA-PKcs. Since KU70/KU80 heterodimer formation is constitutive and occurs independently of DNA damage, we focused on KU80 as a representative marker. Strikingly, CRTC2 knockout did not affect KU80 enrichment at DSBs but significantly impaired DNA-PKcs enrichment, as quantified by the LacI-LacO system (Supplementary Fig. 5a, b). Laser micro-irradiation assays further validated CRTC2’s requirement for DNA-PKcs enrichment to DNA lesions (Fig. 3a; Supplementary Fig. 5c). Consistent with these results, Co-IP revealed an enhancement of the interaction of CRTC2 and DNA-PKcs binding post-IR (Fig. 3b). Immunofluorescence further demonstrated that the co-localization between CRTC2 and DNA-PKcs increased at damage sites (Fig. 3c). CRTC2 recruits DNA-PKcs and promotes activation of the DNA-PK holoenzyme. a Immunofluorescence images (left) and quantitative analysis (right) of the co-localization of DNA-PKcs and γH2AX in control and CRTC2-knockdown HCCLM3 cells (n = 120 cells per group). Scale bars, 20 µm. b Co-IP assay examining the interaction of CRTC2 with DNA-PKcs, KU70, and KU80 with or without IR treatment. c Immunofluorescence images of the co-localization of DNA-PKcs and CRTC2 with or without laser micro-IR treatment in HCCLM3 cells. Scale bars, 20 µm. d Co-IP assay examining the interaction of KU80 with DNA-PKcs and KU70 in the absence or presence of si-CRTC2 after IR treatment. e Co-IP assay examining the interaction of KU80 with DNA-PKcs and KU70 with or without CRTC2 overexpression after IR treatment. f Western blot analysis of phosphorylation of DNA-PKcs S2056 in CRTC2-knockout and -overexpressing HCCLM3 cells after IR treatment. g Western blot analysis of phosphorylation of DNA-PKcs S2056 in liver tissues from Crtc2+/+ and Crtc2−/− mice after IR treatment. Change of DNA-PK activity with time in CRTC2-overexpressing (h) and -knockout (i) HCCLM3 cells (n = 3 biologically independent experiments). j Interface region prediction of the CRTC2 TAD domain and DNA-PKcs KD domain by AlphaFold3. k In vitro Co-IP assay examining the interaction of purified KD-Flag and TAD-MYC proteins. l Western blot analysis of phosphorylation of DNA-PKcs S2056 in CRTC2WT- or CRTC2∆TAD-transfected HCCLM3 cells after IR treatment. m In vitro DNA-PK activity detection in CRTC2- or CRTC2∆TAD-transfected HCCLM3 cell lysates (n = 3 biologically independent experiments). n Representative comet assay images (left) and quantitative analysis of tail moment and % of tail DNA (right) in CRTC2- or CRTC2∆TAD-transfected HCCLM3 cells under IR treatment (n = 50 cells per group). Scale bar, 50 µm. All data are shown as the mean ± SEM. Comparisons between two groups were analyzed by Student’s t test. Comparisons between three or more groups with one condition were assessed by one-way ANOVA. Comparisons between three or more groups with two or more conditions were assessed by two-way ANOVA. All statistical tests were conducted as two-sided. Source Data are provided with this paper. Following DNA damage, the KU70/KU80 heterodimer binds DSBs, recruits DNA-PKcs, and forms the DNA-PK holoenzyme, thereby activating DNA-PKcs kinase activity43. This activation facilitates the NHEJ transition from the long-range to short-range synaptic complex, which is prerequisite for efficient repair44. We therefore investigated whether CRTC2 promotes functional DNA-PK holoenzyme assembly. Co-IP analysis demonstrated that CRTC2 modulation (knockdown or overexpression) did not alter KU70/KU80 heterodimerization but changed DNA-PKcs enrichment to the complex (Fig. 3d, e). To assess the functional impact of CRTC2-mediated DNA-PKcs enrichment, we monitored DNA-PKcs auto-phosphorylation at Ser2056, a key activation marker. In both HCCLM3 and SK-HEP1 cell lines, CRTC2 overexpression enhanced Ser2056 phosphorylation (P-S2056), while knockdown suppressed it (Fig. 3f; Supplementary Fig. 5d). Consistent with these, CRTC2 deficiency significantly attenuated DNA-PK activity in irradiated mouse liver tissue (Fig. 3g), highlighting the physiological relevance of this regulation. These results suggest that CRTC2 promotes DNA-PK holoenzyme formation and activation. Additionally, in vitro kinase assays confirmed that CRTC2 overexpression enhanced DNA-PK activity, while CRTC2 knockdown suppressed it (Fig. 3h, i), suggesting a direct interaction between CRTC2 and DNA-PKcs. To map the CRTC2–DNA-PKcs interaction interface, AlphaFold 3 analysis predicted primary contact sites within the CRTC2 transactivation domain (TAD) and the DNA-PKcs kinase domain (KD) (Fig. 3j). Subsequent co-incubation assays using in vitro-purified TAD-MYC and KD-Flag proteins (Supplementary Fig. 5e) confirmed direct binding between these domains (Fig. 3k). Moreover, PLA performed on two PDX tissue samples showed that IR treatment significantly promoted the nuclear co-localization of DNA-PKcs and CRTC2, indicating their interaction in a DNA damage-dependent manner in clinical samples (Supplementary Fig. 5f). Consistent with these results, CRTC2ΔTAD failed to activate DNA-PKcs both in vitro and in vivo (Fig. 3l, m). Following irradiation, CRTC2ΔTAD cells displayed persistent γH2AX foci compared to wild-type CRTC2 cells (Supplementary Fig. 5g). Moreover, consistent with defective repair, comet assays demonstrated significantly elevated DSB levels in irradiated CRTC2ΔTAD cells, as evidenced by increased tail DNA percent and tail moment (Fig. 3n). Collectively, these data indicate that CRTC2 interacts with DNA-PKcs via a specific TAD–catalytic domain binding interface, thereby promoting DNA-PK assembly and activation. CRTC2 Ser433 is phosphorylated by DNA-PK DNA-PK, a serine/threonine kinase of the PIKK family, is well known for its autophosphorylation activity and its critical role in phosphorylating core NHEJ factors, including KU70, KU80, XRCC4, and Artemis, to promote DNA end processing and ligation45. Given this function, we investigated whether DNA-PK also phosphorylates CRTC2 in response to ionizing radiation. To address this, we performed phosphoproteomic analysis in HCCLM3 cells post-IR exposure using HPLC/MS. This analysis revealed seven high-confidence phosphorylation sites within the CRTC2 TAD, ranked by their cross-correlation scores (Xcorr) (Supplementary Fig. 6a). We then applied the GPS 6.0 algorithm (https://gps.biocuckoo.cn/index.php) to predict kinase–substrate relationships, ranking the six TAD sites most likely targeted by DNA-PK (Supplementary Fig. 6b). Integrating the phosphoproteomic data with these in silico predictions pinpointed the evolutionarily conserved serine residue Ser433 of CRTC2 as a potential DNA-PK substrate (Fig. 4a, b). Although PIKK family kinases are generally known for their S/T-Q motif specificity, recent phosphoproteomic evidence suggests that DNA-PK preferentially targets S/T-hydrophobic residues, with notable enrichment at S/T-P and S/T-E motifs46. Significantly, Ser433 resides within an S‑P sequence and lies in the evolutionarily conserved PXSP motif (Fig. 4b), a canonical MAPK recognition site47. To directly test whether DNA-PK phosphorylates CRTC2 at Ser433, we first performed in vitro kinase assays using recombinant DNA‑PK complex and purified wild‑type (TADWT) or S433A‑mutant (TADS433A) CRTC2 TAD fragments (Supplementary Fig. 6c). Immunoblotting with a phospho‑specific antibody against the PXSP motif (P‑PXSP) showed that DNA‑PK phosphorylated the wild‑type fragment, but not the mutant, confirming Ser433 as a direct phosphorylation target of DNA‑PK (Supplementary Fig. 6d). To strengthen these findings, we successfully generated a site‑specific phospho‑antibody targeting CRTC2 Ser433 residue, designated as P-Ser433 (Supplementary Fig. 6e). Using this antibody, we independently verified in vitro that DNA-PK directly phosphorylates CRTC2 at Ser433 (Fig. 4c). DNA-PKcs phosphorylates CRTC2 at Ser433. a LC-MS/MS spectrum of CRTC2 peptides containing phosphorylated Ser433. b Alignment of CRTC2 Ser433 peptide sequences in Homo sapiens, Pan troglodytes, Mus musculus, Rattus norvegicus, Macaca mulatta, and Canis lupus familiaris. c Western blot analysis of CRTC2 Ser433 phosphorylation level by in vitro kinase reaction of DNA-PKcs with purified TAD or TADS433A. d Western blot analysis of Ser433 phosphorylation level in CRTC2-TY1-, S433A-TY1-, and S433E-TY1-transfected HEK293T cells with or without IR treatment (6 Gy). e Western blot analysis of Ser433 phosphorylation level in M3814 (10 µM) treated CRTC2-TY1-overexpressing HEK293T cells with or without IR treatment (6 Gy). Western blot analysis of CRTC2 Ser433 phosphorylation level in the absence or presence of DNA-PKcs overexpression (f) or knockdown (g) in CRTC2-TY1-overexpressing HEK293T cells with or without IR (6 Gy) treatment. DNA-PKcs abundance was normalized to β-actin (g). h Correlation analysis of DNA-PK activity and CRTC2 Ser433 phosphorylation level in patients based on the proteomic database OEP000321 (n = 78). i Western blot analysis of CRTC2 protein abundance, Ser433 phosphorylation level, and DNA-PKcs S2056 phosphorylation level of tumor tissues in patients from Tianjin Medical University Cancer Hospital (TJMUCH) (n = 20). j Correlation analysis of CRTC2 and DNA-PKcs P-S2056 abundance in Fig.4i. k Correlation analysis of CRTC2 P-Ser433 and DNA-PKcs P-S2056 abundance Fig.4i. l Western blot analysis of CRTC2 protein abundance, Ser433 phosphorylation level, and DNA-PKcs S2056 phosphorylation level of tumors in PDX models. The correlation analysis was performed by GraphPad Prism version 9.0 using simple linear regression. Source Data are provided with this paper. To further validate this phosphorylation event in cells, we examined CRTC2 phosphorylation at Ser433 following IR treatment. Both the P-PXSP and P‑Ser433 antibodies detected increased phosphorylation at this site (Supplementary Fig. 6f; Fig. 4d). This phosphorylation was suppressed by a DNA‑PK inhibitor, but not MAPK inhibitor, enhanced by DNA‑PKcs overexpression, and attenuated upon DNA‑PKcs knockdown (Supplementary Fig. 6g-i; Fig. 4e-g). These cellular findings are fully consistent with the in vitro kinase assays, together confirming that DNA‑PK is the kinase responsible for IR‑induced phosphorylation of CRTC2 at Ser433. To validate the clinical relevance of CRTC2 Ser433 phosphorylation and DNA-PK activity in human liver cancer, we analyzed the published OEP000321 proteomic/phosphoproteomic dataset32, as detailed in Supplementary Fig. 2f. Both phosphorylation signals were reliably detected in the non‑AA patient subset. Spearman correlation analysis showed a significant positive correlation between CRTC2 Ser433 phosphorylation levels and DNA-PK activity (Fig. 4h). Furthermore, western blotting of 20 human liver cancer samples revealed a positive correlation between CRTC2 expression levels and DNA-PKcs activation (Fig. 4i, j). Moreover, the degree of DNA-PK activation was significantly associated with the phosphorylation level of CRTC2 at Ser433 (P-Ser433) (Fig. 4i, k). Building on this, we further found that IR treatment also increased DNA-PK activation and the phosphorylation of CRTC2 at Ser433 in PDX models (Fig. 4l). Together, these findings indicate that in clinical tumors, CRTC2 expression is linked to DNA-PK activity, which in turn correlates with CRTC2 Ser433 phosphorylation. In summary, these results demonstrate that DNA-PK complex directly phosphorylates CRTC2 at Ser433 in response to DNA damage, and this phosphorylation event clinically correlates with DNA-PK activity in human liver cancer. CRTC2 Ser433 phosphorylation mediates relocalization from transcription to DNA repair complexes To investigate the biological significance of CRTC2 Ser433 phosphorylation on DNA damage repair, we overexpressed wild-type CRTC2, the non-phosphorylatable S433A mutant, and the phospho-mimetic S433E variant in EJ5-GFP reporter cells and assessed NHEJ efficiency. As shown in Fig. 5a, compared to wild-type CRTC2, S433A expression significantly impaired repair, while S433E substantially enhanced it, indicating that Ser433 phosphorylation is critical for efficient NHEJ. Given CRTC2’s canonical function as a CREB coactivator, we next measured its transcriptional activity using a CRE-luciferase reporter. Strikingly, in contrast to its effect on DNA repair, the S433A mutant exhibited luciferase activity comparable to wild-type CRTC2 without IR, whereas S433E suppressed the activity (Fig. 5b, left), indicating that CRTC2-Ser433 phosphorylation inhibits its transcriptional coactivator function. Furthermore, IR significantly reduced CRE-luciferase activity in wild-type CRTC2-expressing cells, whereas S433A showed minimal response and S433E maintained constitutively low activity (Fig. 5b, right). The above results indicate that CRTC2 Ser433 phosphorylation acts as a molecular switch that represses transcriptional coactivation while simultaneously enabling DNA repair-promoting activity. These results led us to hypothesize that IR-induced phosphorylation of CRTC2 at Ser433 promotes dissociation from transcriptional complexes and facilitates recruitment to DNA repair complexes. To test this, we first examined IR-induced dissociation of CRTC2 from its transcriptional partners CREB and P-TEFb (CyclinT1/CDK9 heterodimer). As expected, IR treatment markedly attenuated CRTC2 interactions with these factors (Supplementary Fig. 7a) while enhancing associations with DNA repair components PARP1 and DNA-PKcs (Figs. 2g, 3b). To determine whether Ser433 phosphorylation governs this dynamic exchange, we performed co-immunoprecipitation assays using CRTC2 mutants. As shown in Fig. 5c, The S433A mutant abolished IR-induced dissociation from transcriptional complexes, maintaining stable binding to CREB and Cyclin T1 (Fig. 5c). Conversely, the S433E phosphomimetic exhibited constitutive dissociation from these partners even without IR (Fig. 5d). Parallel experiments revealed that IR failed to enhance S433A binding to PARP1/DNA-PKcs (Fig. 5c), while S433E maintained strong constitutive interactions with both repair factors (Fig. 5d). Furthermore, PLA assay demonstrated impaired S433A accumulation at DNA break sites post-IR versus wild-type, whereas S433E showed marked enrichment at damage sites (Fig. 5e). We next compared the recruitment kinetics of CRTC2 and its mutants (S433A, S433E). As shown in Supplementary Fig. 7b, wild-type CRTC2 accumulated gradually at DNA damage sites and peaked at 3 min. The S433E mutant exhibited rapid recruitment, reaching peak levels within 1 min and subsequently maintaining a high occupancy (Supplementary Fig. 7b). Conversely, the S433A mutant failed to exhibit significant recruitment (Supplementary Fig. 7b). Collectively, these findings suggest that Ser433 phosphorylation drives CRTC2 redistribution from transcriptional to repair complexes, i.e., from transcriptional coactivator to DNA damage responder. CRTC2 Ser433 phosphorylation mediates the transition from transcription to DNA repair complexes. a NHEJ repair efficiency detection in response to CRTC2WT, CRTC2S433A, and CRTC2S433E overexpression and western blot analysis of I-SceI and CRTC2 expression in EJ5-GFP U2OS cells (n = 3 biologically independent experiments). b Analysis of CRE-mediated transcription using a luciferase reporter in HEK293T cells overexpressing the TY1-tagged constructs CRTC2, S433A, or S433E with or without IR (n = 3 biologically independent experiments). c Co-IP of CRTC2-TY1 with binding partners in HEK293T cells overexpressing wild-type (CRTC2-TY1) or mutant (S433A-TY1) protein, with or without IR (6 Gy) treatment. d Co-IP of CRTC2-TY1 with binding partners in HEK293T cells overexpressing wild-type (CRTC2-TY1) or mutant (S433E-TY1) protein, with or without IR (6 Gy) treatment. e Representative images (left) of Proximity Ligation Assay (PLA) and quantitative analysis (right) of CRTC2 and γH2AX in in CRTC2-TY1, S433A-TY1 and S433E-TY1 overexpressing HEK293T cells (n = 119 cells per group). Scale bars, 20 µm. f The proportion of downregulated genes that were bound by CRTC2 in DSB-induced AiDIvA cell. g Coverage profile plot representing the RPM of CRTC2-bound gene transcripts in AiDIvA cells with or without Dox and 4-OHT induction for each gene body region; TES: transcription end site; TSS: transcription start site. Bold lines represent mean values, whereas semi-transparent shades around the mean curves represent the SEM across the regions. h Heatmap of CRTC2-bound genes in AiDIvA cells treated with or without Dox and 4-OHT in CRTC2 CUT&Tag-seq. i Schematic diagram of the molecular mechanism. All data are shown as the mean ± SEM. Comparisons between three or more groups with one condition were assessed by one-way ANOVA. Comparisons between three or more groups with two or more conditions were assessed by two-way ANOVA. All statistical tests were conducted as two-sided. Source Data are provided with this paper. The above results further indicate that upon DSB induction, CRTC2 undergoes an identity switch that mediates transcriptional silencing of its downstream target genes. To more precisely map where DSBs induce CRTC2-mediated transcriptional repression, we isolated nuclei from AiDIvA cells before and after 4-OHT treatment and profiled nascent transcripts via RNA-seq. Consistent with previous reports, DSBs did not cause genome-wide transcriptional inhibition48 (Supplementary Fig. 7c). Differential gene expression analysis revealed 284 genes that were down-regulated upon DSB induction (Supplementary Fig. 7d). Among these, 38 genes (12.7%) showed CRTC2 binding at their promoters and gene bodies when integrated with CRTC2 CUT&Tag data (Fig. 5f, g), and this binding markedly decreased following DSB induction (Fig. 5h). Collectively, these findings suggest that DSBs repurpose CRTC2, converting it from a transcriptional coactivator into a DNA repair cofactor that directs the silencing of its target genes. In summary, our mechanistic investigations revealed that upon DSB, PARP1-dependent recruitment of CRTC2 to DNA damage sites promotes activation of the DNA-PK complex, enabling DNA-PK to phosphorylate CRTC2 at Ser433. This phosphorylation facilitates CRTC2’s localization to DNA repair complexes, where its physical association with DNA-PKcs further enhances DNA-PK holoenzyme kinase activity, thereby establishing a positive feedback loop. This loop ensures rapid amplification of the NHEJ repair while inhibiting CRTC2’s transcriptional activity and locking it in a repair-competent state until damage resolution is achieved (Fig. 5i). CRTC2 knockout enhances radiosensitivity in liver cancer Emerging evidence indicates that hyperactivated DNA repair pathways contribute to radio-chemoresistance in malignancies49,50. To evaluate the clinical relevance of CRTC2, we analyzed pretreatment needle biopsies from 26 liver cancer patients undergoing radiotherapy. Stratification by CRTC2 immunohistochemistry (IHC) staining intensity revealed significantly improved therapeutic responses in the CRTC2-low cohort versus the CRTC2-high group (P < 0.05), indicating that CRTC2 high expression is significantly associated with radioresistance in liver cancer patients (Fig. 6a, b). To further validate this association, we employed CRTC2-knockout and -overexpressing cell lines. CRTC2 knockout significantly enhanced radiosensitivity and amplified radiation-induced apoptosis post-irradiation (6 Gy) (Supplementary Fig. 8a, b, e), whereas CRTC2 overexpression conferred radioresistance (Supplementary Fig. 8c, d, f). These phenotypes were consistently replicated in HCCLM3 (Supplementary Fig. 8g-i). Given that our previous work showed CRTC2 knockdown inhibits HCC progression29, we sought to determine whether this effect reflects true radiosensitization or merely additive tumor growth inhibition. We further analyzed the interaction between siCRTC2 and IR using the Zero Interaction Potency (ZIP) model, an advanced algorithm that quantifies drug combinations and differentiates additive from synergistic effects51,52. By precisely modulating CRTC2 expression through graded concentrations of siCRTC2 in combination with varying doses of IR, we observed a synergy score with a Z-value of 13.08 (Supplementary Fig. 8j, k), exceeding the established threshold of 10.00. This indicates that the combination of CRTC2 knockdown and IR treatment produces a clear synergistic effect, rather than an additive one. Specifically, CRTC2 knockdown effectively sensitizes cells to IR. CRTC2 ablation enhances radiosensitivity in liver cancer. a Representative image of CRTC2 immunohistochemistry staining and computed tomography scans in patients with liver cancer treated with radiotherapy, scale bar, 5 mm. b Correlation analysis between CRTC2 expression and radiotherapy response (n = 26 patients) (PR: partial remission; SD: stable disease; PD: progressed disease; nPD, non-progressed disease). Schematic of the development of the Keap1KO/MYCOE- liver cancer model (c) and representative liver images and H&E staining (d) of Crtc2+/+ and Crtc2−/− mice with or without IR (6 Gy). Scale bar, 25 µm; white dashed line marks the tumor edge. Liver versus body weight ratio (LBR) (e) and tumor burden (f) of Crtc2+/+ and Crtc2−/− mice in the Keap1KO/MYCOE-liver cancer model with or without IR (Control-Crtc2+/+ n = 5, Control-Crtc2−/− n = 6, IR-Crtc2+/+ n = 6, IR-Crtc2−/− n = 6). IR inhibitory effect below the bar graph (f). Tumor images (g), tumor growth curves (h), tumor weight (i) and IR inhibitory effect (j) of PDX-1 with indicated treatment (n = 4 tumors). Tumor images (k), tumor growth curves (l), tumor weight (m) and IR inhibitory effect (n) of PDX-2 with indicated treatment (n = 4 tumors). All data are shown as the mean ± SEM. Comparisons between two groups were analyzed by Student’s t test. Comparisons between three or more groups with two or more conditions were assessed by two-way ANOVA. All statistical tests were conducted as two-sided. Source Data are provided with this paper. In subcutaneous xenograft models, irradiated sgCRTC2 tumors exhibited accelerated regression, characterized by reduced volume, decreased weight, and increased IR inhibitory effect following irradiation (Supplementary Fig. 9a-d). Conversely, CRTC2-overexpressing tumors displayed diminished regression, marked by increased volume and weight post-irradiation, along with reduced IR inhibitory effect (Supplementary Fig. 9e-h). To further explore CRTC2’s role in radioresistance, we utilized Crtc2 deletion (Crtc2−/−) mice. An oncogenic liver cancer model was established via hydrodynamic tail vein (HDTV) injection of plasmids encoding Keap1 knockout and c-MYC overexpression (Keap1KO/MYCOE) into wild-type (Wt, Crtc2+/+) and Crtc2−/− mice. Spontaneous liver cancer developed 3–6 weeks post-injection, and irradiation was administered 3 weeks after HDTV injection (Fig. 6c). As shown in Fig. 6d-f, Crtc2−/− mice exhibited significantly attenuated tumor progression post-irradiation, as evidenced by reduced liver-to-body weight ratios (LBR) and tumor burden. Collectively, these results confirm CRTC2’s functional role in radioresistance across experimental models. Furthermore, we utilized two clinically relevant liver cancer PDX models (previously shown in Fig. 4l). In these models, CRTC2 expression was disrupted using a liver-specific adeno-associated virus system, AAV8.sgCRTC2, which carries a sgRNA as previously described29. As shown in Fig. 6g-n, CRTC2 deficiency significantly enhanced the sensitivity of both tumors to IR, as evidenced by reduced tumor weight and volume, along with a greater inhibitory effect of irradiation. To further elucidate the role of CRTC2 Ser433 phosphorylation in radiotherapy resistance, we first analyzed the OEP000321 dataset. Among non‑AA patients, high CRTC2 S433 phosphorylation levels significantly correlated with poor overall survival (Supplementary Fig. 10a). To validate this finding, we performed P-Ser433 IHC on the same 26 patient specimens from Fig. 6a. High P-Ser433 levels correlated with radiotherapy resistance, whereas low levels correlated with better response (Supplementary Fig. 10b). Collectively, these results indicate that both CRTC2 expression and its S433 phosphorylation correlate with radiotherapy response in liver cancer. CRTC2 suppresses radiotherapy-induced cGAS-STING activation and the abscopal effect Radiotherapy-generated cytosolic DNA fragments function as damage-associated molecular patterns that potently activate the cGAS-STING axis53,54. To investigate CRTC2’s regulatory role in this process, we first quantified cytoplasmic double-stranded DNA (dsDNA) accumulation. As shown in Supplementary Fig. 11a, after IR treatment, CRTC2 knockdown significantly enhanced this accumulation, whereas CRTC2 overexpression attenuated it. In parallel, CRTC2 knockout led to a marked increase in micronucleus formation compared with control cells (Supplementary Fig. 11b). Following recognition of cytoplasmic dsDNA, cGAS produces the second messenger cGAMP to activate STING. Activated STING translocates to the Golgi apparatus, recruiting kinases TBK1. TBK1 then phosphorylates IRF3, inducing the expression of type I interferons (IFN-I) and inflammatory cytokines55,56. Consistent with these findings, western blot analysis demonstrated upon IR treatment, CRTC2 knockout substantially potentiated cGAS-STING pathway activation and upregulated downstream targets (Supplementary Fig. 11c, d, left), whereas CRTC2 overexpression attenuated these responses (Supplementary Fig. 11c, d, right). In subcutaneous xenograft models, CRTC2 knockout consistently amplified cGAS-STING signaling, whereas CRTC2 overexpression suppressed pathway activity (Supplementary Fig. 12a, b). Additionally, in Keap1KO/MYCOE-driven liver cancer mice, Crtc2−/− consistently enhanced cGAS-STING signaling (Supplementary Fig. 12c). Clinical validation using human liver cancer specimens revealed that an inverse correlation between CRTC2 and P-STING expression (Supplementary Fig. 12d). Furthermore, in PDX models, IR combined with CRTC2 targeting decreased DNA-PK kinase activity and significantly activated the cGAS-STING pathway (Supplementary Fig. 12e). Altogether, these data indicate that CRTC2 suppresses radiotherapy-triggered cGAS-STING activation by regulating cytoplasmic dsDNA retention. cGAS-STING signaling triggers an IFN-I response that drives the maturation of dendritic cells (DCs) and cross-presentation of antigens, ultimately activating tumor-specific CD8a + T cells. This process converts localized irradiation into systemic abscopal immunity57. To investigate CRTC2’s role in this radiotherapy-induced immune microenvironment remodeling, we performed single-cell RNA sequencing on irradiated primary liver cancer tissues from mice with NASH-driven liver cancer (Supplementary Fig. 13a, b). Notably, NASH-driven liver cancer exhibits significantly lower responsiveness to immunotherapy compared to other forms of liver cancer, a difference that is closely related to the impaired functions of DCs and CD8a+ T cells in the NASH microenvironment58. Strikingly, CRTC2-deficient (Crtc2−/−) tumors exhibited enhanced DCs and CD8a + T cell infiltration, specifically by cytotoxic subpopulations including Granzyme B + CD8a + T cells and IFN-γ + CD8 + T cells (Supplementary Fig. 13c; Fig. 7a, b). T cell-mediated killing assays confirmed that CRTC2 ablation enhanced irradiation-induced PBMC cytotoxicity, whereas CRTC2 overexpression reduced this effect (Fig. 7c). To further validate these findings, we generated a CRTC2 knockout in the mouse hepatoma cell line Hepa1-6 (Supplementary Fig. 13d). Consistent with the above phenotype, IR significantly enhanced activation of the cGAS-STING signaling pathway in CRTC2-knockout cells compared to controls (Supplementary Fig. 13d). Furthermore, subcutaneous xenograft experiments demonstrated that CRTC2 knockout significantly enhanced tumor radiosensitivity (Supplementary Fig. 13e-h). Corroborating the enhanced anti-tumor immunity, flow cytometry analysis of tumor-infiltrating lymphocytes from these xenografts revealed that CRTC2 knockout significantly increased CD8a + T cell infiltration, as well as the frequencies of cytotoxic Granzyme B + CD8a + T cells and IFN-γ + CD8a + T cells (Fig. 7d, e). These results suggest that CRTC2 negatively regulates the cGAS-STING pathway, thereby modulating antitumor immunity. CRTC2 suppresses radiotherapy-induced abscopal immunity. a Total immune cell cluster analysis with UMAP visualization of murine tumor tissue scRNA-seq in the DEN-HFD liver cancer model. b UMAP visualization of CD8a +, CD8a + Granzyme B + and CD8a + IFNγ + T cell ratios in total T cell cluster. c T cell killing efficiency detection with CRTC2 knockout or overexpression under IR treatment (n = 3 biologically independent experiments). CD8a + (d), CD8a + Granzyme B + and CD8a + IFNγ+ (e) T cell ratios of immune cells in tumor tissue from control and CRTC2-knockout xenograft model (n = 6 tumors). f Schematic of the development of the xenograft model with control or CRTC2-knockout Hepa1-6 cells in C57BL/6 mice and IR strategy. g Tumor growth curves after subcutaneous injection of control and CRTC2-knockout Hepa1-6 cells and IR for left side from day 10 to day 22 (sgControl-L, sgControl+IR-L, sgCRTC2-L, sgCRTC2+IR-L, sgControl-R, sgControl+IR-R, sgCRTC2-R, n = 6; sgCRTC2-+IR-R, n = 4). Tumors were imaged (h) and weighed (i) after mice were sacrificed (sgControl-L, sgControl+IR-L, sgCRTC2-L, sgCRTC2+IR-L, sgControl-R, sgControl+IR-R, sgCRTC2-R, n = 6; sgCRTC2-+IR-R, n = 4). Abscopal efficiency below the bar graph (i). Immunofluorescence images (left) and quantitative analysis (right) of CD8a, Granzyme B and IFNγ in the abscopal sites of sgControl (j) and sgCRTC2 (k) tumors with or without IR, scale bar, 50 µm (sgControl-R, sgControl+IR-R, sgCRTC2-R, n = 6; sgCRTC2-+IR-R, n = 4). All data are shown as the mean ± SEM. Comparisons between two groups were analyzed by Student’s t test. Comparisons between three or more groups with two or more conditions were assessed by two-way ANOVA. All statistical tests were conducted as two-sided. Source Data are provided with this paper. To further investigate the role of CRTC2 in modulating the abscopal effect of radiotherapy, we utilized immunocompetent C57BL/6 mice bearing bilateral Hepa1-6 flank tumors. The left flank tumor (primary tumor, irradiated site) was irradiated as described previously59, while the right flank tumor (secondary tumor, abscopal site) remained unirradiated (Fig. 7f). Strikingly, irradiation of the primary tumor failed to elicit a significant abscopal effect on the secondary tumor in the control group (sgControl-IR-R vs. sgControl-R; Fig. 7g-i). Conversely, CRTC2 knockout potentiated the abscopal effect (sgCRTC2-IR-R vs. sgCRTC2-R; Fig. 7g-i). Accordingly, immunofluorescence staining revealed no significant difference in the infiltration of CD8a + T cells or cytotoxic T cells between sgControl-IR-R and sgControl-R at the abscopal site (sgControl-IR-R vs. sgControl-R; Fig. 7j). In contrast, sgCRTC2-IR-R exhibited a marked increase in the infiltration of CD8a + T cells or cytotoxic T cells compared to sgCRTC2-R (sgCRTC2-IR-R vs. sgCRTC2-R; Fig. 7k). These results indicate that CRTC2 depletion potentiates the abscopal effect of radiotherapy by enhancing antitumor immunity. Furthermore, CRTC2 knockout significantly delayed growth of non-irradiated primary tumors (sgCRTC2-L vs. sgControl-L) and enhanced radiosensitivity in irradiated primary tumors (sgCRTC2-IR-L vs. sgControl-IR-L; Fig. 7g-i), consistent with prior findings (Supplementary Fig. 13e-h). In summary, these results demonstrate that CRTC2 functions as a suppressor of radiotherapy-triggered cGAS-STING activation, thereby modulating the immune microenvironment and consequently antagonizing the abscopal effect. Targeting CRTC2 enhances the efficacy of radioimmunotherapy Recent clinical trials have demonstrated that the abscopal effect induced by radiotherapy arises from localized irradiation-triggered systemic anti-tumor immunity. Adding immune-checkpoint inhibitors or other immunotherapies markedly enhances this effect and improves overall efficacy, leading to the emergence of radioimmunotherapy (RIT) as a more potent treatment paradigm60–63. Based on our findings, we propose that targeting CRTC2 could enhance RIT efficacy in liver cancer. We first evaluated the therapeutic efficacy of AAV8.sgCrtc2 in a subcutaneous xenograft model. Mice received AAV8.sgCrtc2 via tail vein injection 10 days post-tumor inoculation, followed by irradiation and anti-PD-1 combination therapy (Fig. 8a, upper). Western blot analysis of CRTC2 in tumor tissues and confirmed that AAV8.sgCrtc2 can successfully inhibit the expression of Crtc2 in tumors (Supplementary Fig. 14a). As shown in Fig. 8a-c, both AAV8.sgCrtc2 and RIT monotherapy significantly inhibited tumor growth, as assessed by tumor weight and volume. Notably, the combined treatment group (AAV8.sgCrtc2 + RIT) exhibited an enhanced inhibitory effect (Fig. 8a-c). Furthermore, RIT inhibitory effect was markedly enhanced in the AAV8.sgCrtc2 group (Fig. 8c), indicating that CRTC2 ablation potentiates the antitumor efficacy of radioimmunotherapy. Flow cytometric analysis confirmed a significant increase in the infiltration of CD8a + T cells, Granzyme + CD8a + T cells, and IFN-γ + CD8a + T cells in tumors from the AAV8.sgCrtc2 and RIT combination-treated group (Fig. 8d). We next assessed AAV8.sgCrtc2’s potential to enhance RIT in the Keap1KO/MYCOE-driven liver cancer mouse model (Fig. 8e). Compared to controls, RIT efficacy was significantly enhanced in mice receiving AAV8.sgCrtc2, as assessed by maximum tumor area, and tumor burden (Fig. 8f, g). Furthermore, serum ALT and AST levels were substantially reduced in the AAV8.sgCrtc2 group, indicating ameliorated liver dysfunction (Fig. 8h). Further examination revealed that tumors from AAV8.sgCrtc2-treated mice exhibited inhibited DNA-PK activity alongside activated cGAS-STING signaling (Fig. 8i; Supplementary Fig. 14b, c). Collectively, these findings demonstrate that CRTC2 targeting significantly enhances the efficacy of RIT in liver cancer models. CRTC2 ablation enhances the efficacy of radioimmunotherapy. a Schematic of the development of the xenograft model with Hepa1-6 cells in C57BL/6 mice in the absence and presence of AAV8.sgCrtc2 under IR (6 Gy) combined with anti-PD-1 therapy (upper) and tumor images (lower) after mice were sacrificed. Tumor growth curves (b) after subcutaneous injection from day 7 to day 22 and tumor weight (c) after mice were sacrificed (n = 6). IR inhibitory effect below the bar graph (c). d CD8a + (top), CD8a + Granzyme B + (middle) and CD8a + IFNγ + (bottom) T cell ratio of immune cells in tumor tissue from C57BL/6 mice in the absence and presence of AAV8.sgCrtc2 under IR combined with anti-PD-1 therapy (n = 6). e Schematic of the development of the Keap1KO/MYCOE- liver cancer model in the absence and presence of AAV8.sgCrtc2 with IR (6 Gy) combined with anti-PD-1 therapy. Representative liver images and H&E staining (f), tumor max area (left) and tumor burden (right) (g), ALT (left) and AST level (right) (h) of the Keap1KO/MYCOE- liver cancer model in the absence and presence of AAV8.sgCrtc2 under IR combined with anti-PD-1 therapy; white dashed line marks the tumor edge. (AAV8.sgControl n = 7, AAV8.sgCrtc2 n = 7). Scale bar, 25 µm. I DNA-PK activity detection of tumor tissue from the xenograft model with Hepa1-6 cells in the absence and presence of AAV8.sgCrtc2 under IR combined with anti-PD-1 therapy (n = 6). All data are shown as the mean ± SEM. Comparisons between two groups were analyzed by Student’s t test. Comparisons between three or more groups with two or more conditions were assessed by two-way ANOVA. All statistical tests were conducted as two-sided. Source Data are provided with this paper. Discussion DSBs represent the most lethal form of cellular damage and trigger radiotherapy-induced cancer cell death, but aberrant repair drives therapeutic resistance64. In this study, we identify CRTC2 as the pivotal regulator of this “repair–resistance” axis. Mechanistically, upon DSB induction, PARP1 activation recruits CRTC2 to lesions, where CRTC2 in turn facilitates DNA-PKcs enrichment and accelerates active DNA-PK holoenzyme assembly. Subsequently, DNA-PK phosphorylates CRTC2 at Ser433, disrupting its association with the CREB/P-TEFb transcription complex while enhancing its binding to the PARP1/DNA-PKcs repair platform. This establishes a CRTC2 phosphorylation-driven positive feedback loop that maximizes NHEJ efficiency and concomitantly represses transcription of downstream target genes. Functionally, CRTC2 high expression is correlated with radioresistance, which is consistently confirmed in vitro and in vivo. CRTC2 deletion potentiates irradiation-induced cGAS–STING activation, promotes cytotoxic CD8a+ T cell infiltration, enhances antitumor immunity, as well as the abscopal effect. Furthermore, AAV8-mediated targeting of CRTC2 sensitizes tumors to radioimmunotherapy. Collectively, these findings demonstrate that CRTC2 is a therapeutically targetable regulator at the nexus of DNA repair, radioresistance, and anti-tumor immunity. We and others have observed that DSB-induced transcriptional silencing occurs locally near the break or spreads distally, even reaching nucleolar regions3,6,65–67. Traditionally, this “local” vs. “distal” dichotomy has been attributed to heterogeneity in DNA-damage models, including variations in break density, chromatin landscape, or cell-cycle stage3. However, the possibility of a unified molecular mechanism governing the phenomena has been largely overlooked. Our study reveals that DNA-PK-mediated reprogramming of CRTC2 links the distal silencing of CRTC2-regulated genes to DNA repair activation, thereby uncovering a damage-induced transcriptional control mechanism. CRTC2 undergoes spatial redistribution and functional conversion, rapidly switching its target genes from transcriptional activation to repression. The core significance of this dynamic regulatory process lies in its ability to divert resources from the transcriptional machinery to the repair machinery, prioritizing DSB repair both spatially and temporally, and thereby providing a crucial safeguard for maintaining genomic stability. Building on evidence that numerous transcription activators are recruited to DNA damage sites to facilitate repair21–23, our study further indicates that DSB signaling triggers an immediate identity switch in these factors—from “drivers of gene expression” to “guardians of the genome”. This switch suppresses transcription while prioritizing DNA repair, equipping cells with a rapid, reversible, and spatially precise stress response pathway. Early studies established that PIKK kinases (ATR, ATM, DNA-PK) primarily recognize the canonical S/T-Q motif68,69. However, accumulating evidence over the past decade has revealed that DNA-PK’s substrate specificity extends far beyond this consensus70,71. Recent phosphoproteomics and in silico motif analyses reveal that DNA-PK preferentially phosphorylates serine/threonine residues followed by hydrophobic residues, with prominent enrichment at S/T-P and S/T-E motifs46. Significantly, Güllülü et al. showed that survivin binding to DNA-PK’s kinase domain both enhances kinase activation and directs substrate selection toward S/T-hydrophobic motifs, particularly S/T-P sites, which comprise one third of identified DNA-PK targets72,73. In this study, we reveal that CRTC2 physically interacts with the kinase domain of DNA-PK, driving its assembly and activation. Although no unbiased global specificity screen was performed, targeted phosphoproteomics and in vitro kinase assays conclusively demonstrate DNA-PK-mediated phosphorylation of CRTC2 specifically at Ser433. Strikingly, Ser433 is within an S-P sequence. More importantly, Ser433 is embedded within the evolutionarily conserved PXSP motif, which is a canonical MAPK recognition site47. Using a phospho-specific antibody against the PXSP motif, we readily detected DNA-PK-catalyzed phosphorylation at Ser433. Importantly, MAPK inhibitors failed to reduce IR-induced Ser433 phosphorylation, confirming DNA-PK dependence under genotoxic stress. This kinase “cross-talk” at shared sites is not unprecedented. Over two decades ago, multiple groups reported that both DNA-PK and MAPK phosphorylate p53 at Ser4674–76, with Bulavin and Komiyama providing key demonstrations. Our findings add CRTC2 to the growing list of substrates phosphorylated at identical sites by distinct kinase families. This highlights a fundamental mechanism whereby single residues integrate signals from disparate pathways to fine-tune cellular stress responses. CRTC2 functions as an adaptive transcriptional co-activator, enabling cellular energy conservation and survival under adaptive stress through dynamic regulation by multiple signaling pathways77. This adaptability allows cells and tissues to withstand fluctuating environmental pressures. Under basal conditions, CRTC2 is phosphorylated by salt-inducible kinases (SIKs) and sequestered in the cytoplasm via 14-3-3 binding24. During fasting, convergent cAMP and calcium signaling promote CRTC2 dephosphorylation through SIK inhibition and calcineurin activation, respectively, triggering CRTC2 release from 14-3-3 proteins and nuclear translocation. Within the nucleus, CRTC2 forms complexes with CREB to upregulate key gluconeogenic genes (PEPCK, G6Pase) to sustain metabolic adaptation24. Unfortunately, malignancies co-opt this adaptive machinery. In LKB1-mutant non-small cell lung cancer, constitutively active CRTC2 upregulates ID1 to fuel tumor progression78. LKB1 constrains inflammatory potential by suppressing CRTC2-dependent histone acetylation. In gastrointestinal polyps, aberrant CRTC2 signaling, driven by LKB1 loss, orchestrates pathogenic TH17 programs28,79. Beyond transcription, CRTC2 exhibits non-transcriptional functions. mTORC1 phosphorylates CRTC2 at Ser136, relieving COPII complex inhibition to facilitate SREBP1 Golgi transport and lipogenesis80. Our prior work revealed that CRTC2 forms co-condensates with YTHDF2 that enhance translation of m⁶A-modified mRNAs29. Here, we identify a DNA repair function: CRTC2 senses double-strand breaks and promotes NHEJ repair, with Ser433 phosphorylation serving as the molecular switch for stress detection. Our findings indicate that Ser433 phosphorylation reduces CRTC2 binding to P-TEFb and CREB while promoting its association with PARP1 and DNA-PKcs (Fig. 5). Based on these and published data, we propose that this phosphorylation mediates the dissociation of CRTC2 from the transcriptional complex through the following potential mechanisms: 1. Phosphorylation at Ser433 may induce CRTC2 conformational changes and render the CREB-binding domain into a “closed” state, thereby directly impairing CREB binding. 2. Phosphorylation of CRTC2 at Ser433 may alter the biophysical properties of condensates (e.g., their dynamics, charge, or organization), thereby impairing CRTC2‑dependent P‑TEFb activation and disrupting functional co‑condensate formation. 3. Phosphorylation at Ser433 may act as a molecular switch that triggers or promotes the removal of other modifications (e.g., deacetylation, demethylation, or deglycosylation) within the regulatory domain. These coordinated changes could induce conformational rearrangements, resulting in its dissociation from the transcriptional complex. 4. Ser433 is located within the TAD region, which is required for interaction with DNA-PKcs (Fig. 3j, k). Phosphorylation at this site significantly strengthens CRTC2’s binding to DNA-PKcs. This enhanced interaction may compete with CREB for the CBD interface on CRTC2, or indirectly interfere with CBD-CREB interaction through steric hindrance, thereby promoting the displacement of CRTC2 from the transcriptional complex. Our analysis of the recruitment kinetics reveals that CRTC2 is recruited to DSB sites within seconds of damage induction, reaching its peak at approximately 180 s (Supplementary Fig. 7b). However, when compared with the recruitment kinetics of DNA-PKcs, CRTC2 exhibits notably weaker and slower accumulation at damage sites (Supplementary Fig. 5c). Despite this relatively modest recruitment, CRTC2 plays a functional role in enhancing DNA-PKcs accumulation, as supported by the observation that CRTC2 knockdown impairs DNA-PKcs accumulation at damage sites (Supplementary Fig. 5c). This seemingly paradoxical observation can be explained by the existence of a positive feedback loop between CRTC2 and DNA-PKcs, in which mutual reinforcement promotes sustained repair complex assembly. Such mutual reinforcement mechanisms are well recognized in DNA damage signaling and allow factors with modest individual recruitment to exert significant functional impact81,82. Thus, although CRTC2 does not exhibit the rapid or robust recruitment characteristic of core repair factors, its presence at damage sites is sufficient to facilitate DNA-PKcs accumulation and activation, thereby contributing to efficient NHEJ. Consistently, the recruitment kinetics of CRTC2 further support its role as a regulatory node rather than a primary damage sensor. Radiation therapy, a cornerstone clinical modality, noninvasively eradicates tumors by inducing DNA damage while sparing normal tissues64. However, therapeutic efficacy is often limited by radioresistance, driven primarily by excessive DNA repair capacity50. We now provide convergent evidence identifying CRTC2 as a critical driver of this resistance. Specifically, CRTC2 overexpression correlates with reduced radiosensitivity in both preclinical liver cancer models and patient specimens. Conversely, CRTC2 knockout significantly enhances tumor radiosensitivity in vitro and in vivo, indicating CRTC2 as a tractable target for radiosensitization in liver cancer. Beyond improving local control, our findings hold relevance for systemic anti-tumor responses, particularly the radiation-induced abscopal effect. First reported by Robin Moles in 1953, this phenomenon has gained renewed interest in the immuno-oncology era57. Consequently, researchers have pursued various strategies to amplify this effect, including tumor immunogenicity enhancement83–85, priming T cells in tumor-draining lymph nodes86,87, delivering low-dose radiation to distant lesions88,89, and systemically activating anti-tumor macrophages90,91. Despite these efforts, clinical response rates remain modest, suggesting persistent immunosuppressive barriers. Notably, a recent study revealed that Sfrp2 knockout in cancer-associated fibroblasts enhances the abscopal effect by remodeling the vascular–immune microenvironment to promote CD8a + T cell recruitment to unirradiated tumors59. Here, we identify CRTC2 as a suppressor of the abscopal responses. CRTC2 deletion attenuates NHEJ-mediated DNA repair efficiency, leading to increased cytosolic DNA accumulation and amplified cGAS–STING signaling. This cascade enhances anti-tumor CD8a + T cell infiltration, synergistically suppressing growth of non-irradiated tumors. The radiotherapy-induced abscopal effect serves as potent catalyst for systemic immunity, eradicating distant lesions. With clinical trials establishing radiotherapy–immunotherapy as the leading therapeutic paradigm92, we demonstrate that CRTC2 targeting via AAV8.sgCrtc2 significantly enhances synergy between radiotherapy and immune checkpoint blockade. In summary, targeting CRTC2 not only disrupts radioresistance but also potentiates systemic anti-tumor immunity. Future studies should optimize sgCRTC2-specific delivery systems or develop anti-CRTC2 small molecules to accelerate clinical translation. Methods Cell lines and cell culture The SK-HEP1 and HEK293T cell lines employed in this investigation were procured from the American Type Culture Collection (ATCC) within the United States. The HCCLM3 cell line was provided by Dr. Ke Li (Peking Union Medical College, Beijing, CN)93. FokI-mCherry-LacI U2OS39 and ER-AsiSI nuclease expression HeLa cells (AiDIvA cells)40 were kindly donated by Dr. Lei Shi from Tianjin Medical University. All cells have been authenticated via short tandem repeat (STR) analysis and verified to be devoid of mycoplasma contamination. All cells were grown and sustained in DMEM (Gibco) enriched with 10% fetal bovine serum (FBS) (Gibco) and 1% penicillin-streptomycin solution (Gibco), incubated at a temperature of 37 °C in an atmosphere containing 5% CO2. For FokI nuclease expression and DSB generation, FokI-mCherry-LacI cells were treated with Shield-1 (1 μM) and 4-OHT (1 μM) for 4 h. For AsiSI-induced DSB generation, AiDIvA cells were treated with doxycycline (1 ng/ml) for 24 h and 4-OHT (0.5 μM) for an additional 4 h to stabilize ER-AsiSI and promote its nuclear translocation. Cells were routinely tested for mycoplasma contamination and maintained under mycoplasma-free conditions. CRTC2- knockout and CRTC2-TY1- overexpressed HCCLM3 and SK-HEP1 stable expressed cells were described as previously reported29. In vivo animal studies C57BL/6J mice (6–8-week-old, male) and BALB/c nude mice (5-week-old, female) were obtained from SPF (Beijing) Biotechnology Co. Ltd (Beijing, CN). NCG (ShiLtJGpt-Prkdcem26Cd52 Il2rgem26Cd22/Gpt) mice were obtained from GemPharmatech (Nanjing, CN). Crtc2−/− mice were generated as previously described27. The results reported in this study are not influenced by the sex of mice. All mice were maintained in the animal facility at Tianjin Medical University Cancer Institute and Hospital under specific pathogen-free conditions. All animal studies were sanctioned by the Animal Experiment Ethics Committee of the Institute of Tumor Research and Tianjin Medical University Cancer Hospital, and all procedures were carried out in compliance with the guidelines of the Institutional Animal Care and Use Committee of the National Institutes of Health. For Keap1KO/MYCOE- liver cancer model94, 8-week-old male C57BL/6J mice were injected with a mixture of plasmids diluted in Ringer’s solution through hydrodynamic tail vein (HDTV) injection. The plasmids mix for per mouse consists of 30 μg pX330-Cas9-sgKeap1, 10 μg of pT3-EF1α-MYC and 2 μg of sleeping beauty transposase vector (SB). Mice were sacrificed 4–6 weeks after injection. The following sgRNA sgKeap1 (GGTTCGGTTACCGTCCTGCGA) was used to target Keap1 locus. To evaluate the effect of combination treatment with AAV8.sgCRTC2 and PD-1 mAb, mice were separately treated with saline, AAV8.sgCrtc2, PD-1 mAb, AAV8.sgCrtc2 + PD-1 mAb. AAV8.sgCrtc2 (1 × 1011 vg per mouse) was injected via the tail vein at 10 weeks. PD-1 mAb (10 mg/kg) was injected intraperitoneally every 3 days, six times. Mice were sacrificed at 16 weeks. Serum and liver tissues were collected and stored at −80 °C for subsequent analysis. Tumor numbers were determined by the number of visible tumors. The largest tumor size was measured with a caliper. For xenograft mouse model, the bilateral subcutaneous tumor model was performed as described95. sgControl and sgCRTC2 Hepa1-6 cells were injected into the left hind leg region (6 × 106 cells) and right hind leg region (2 × 106 cells) of C57BL/6J mice in 100 μL PBS. Subcutaneous tumors were treated with irradiation (IR) (10 Gy) at day 13 after injection. For subcutaneous tumorigenesis model, 5-week-old female BALB/c nude mice were used to develop subcutaneous tumorigenesis model. Cells (5 × 106) suspended in 100 μL PBS were subcutaneously injected into the left or double side inner thigh of mice. Tumor volume was measured every 3 days. Mice were sacrificed at the indicated days after injection. The tumors were weighed and their volumes were calculated as length × width2 × 0.5. For patient-derived xenograft (PDX) mouse model, liver cancer tissue from patients was cut into about 0.5 cm3 and subcutaneously injected into double-side inner thigh of NCG mice. Tumor volume was measured every 3 days. Mice were sacrificed at the indicated days after injection. The tumors were weighed and their volumes were calculated as length × width2 × 0.5. For the diethylnitrosamine/high-fat diet (DEN/HFD)- liver cancer model, 2-week-old male C57BL/6 mice were injected intraperitoneally with DEN (25 mg/kg body weight). Mice were switched to HFD (Research Diets, Cat# D09100301) at 8 weeks of age and sacrificed at 26 weeks. IR Inhibitory Effect (%) = (1 − IR-treated Tumor Weight/non-IR-treated Tumor Weight) × 100%; Abscopal Efficiency (%) = (1 − weight of IR-treated distant tumor/weight of non-IR-treated distant tumor) × 100%; RIT Inhibitory Effect (%) = (1 − RIT-treated Tumor Weight/non-RIT-treated Tumor Weight) × 100%. Patient tissue samples Frozen liver tumor tissue samples from 20 patients were collected and lysed for Western blot. Fixed sections of liver tumor tissue from 26 patients were used for immunohistochemistry staining. All patients enrolled in this study were pathologically diagnosed with liver cancer. The radiotherapy efficacy of patients in these cases was classified according to the MRI reports. CRTC2 expression level was determined by immunohistochemistry and evaluated by H-Score. All tissue samples were collected from Tianjin Medical University Cancer Institute and Hospital. All participants signed informed consent in accordance with the Helsinki Declaration. Mouse embryonic fibroblasts (MEFs) isolation and metaphase spread Embryos were harvested from pregnant Crtc2+/+ and Crtc2−/− mice on day 12.5–13.5 post-fertilization. The head and invisible internal organs of each embryo were removed. The remaining embryonic tissue was cut into small pieces and transferred to a 15 ml tube containing 0.25% trypsin-EDTA solution and incubated at 37 °C for 30 min. Subsequently, twice the volume of complete medium was added to inactivate the trypsin and obtain a single-cell suspension. Cells were pipetted up and down 15–20 times and seeded into a 0.2% gelatin-coated 15 cm dish, and incubated for 3 h to adhere. Then the suspension was removed and washed with warmed PBS for three times. Crtc2+/+ and Crtc2−/− MEFs were treated with colcemid (0.2 mg/ml, MCE, Cat# HY-N0282) for 4 h to arrest cells in mitosis. Cells were then incubated in pre-warmed 75 mM KCl at 37 °C for 20 min and in Carnoy’s buffer (3:1 methanol: acetic acid) at room temperature for 15 min. After incubation, the cells were collected and washed twice with Carnoy’s buffer. Cell droplets were spread onto glass slides and dried for 15 min. Slides were stained with Giemsa solution (Solarbio, Cat# G1015). Genome instability was investigated by counting metaphases with chromosomal breaks. X-ray irradiation and laser micro-irradiation IR was performed using an X-ray generator (RS2000 PRO, Radsource Corporation) under the condition of 1 Gy/min (160 kV, 25 mA). After IR, the cells were cultured at 37 °C for the indicated times before collection. Laser micro-IR on CRTC2-GFP-tagged live cells was performed using a microscope equipped with a 37 °C heating chamber and a 365-nm laser diode (Micropoint, Andor Technology). A detectable laser path was performed using a laser setting of 60% of full power under a 60× objective lens. For laser micro-IR on non-fluorescent cells, cells were grown on LabTek II chamber slides (Thermo Scientific, Cat# 154453) and DNA damage was induced by a UV-A laser (355 nm, 40% of full power) using a Zeiss microscope with a PALM MicroBeam laser microdissection workstation under a 40× objective lens. To induce focal damage, HEK293T cells were transfected with PARP1-GFP, DNA-PKsc-GFP, CRTC2-GFP, S433A-GFP or S433E-GFP plasmids. Cells were incubated with DMEM containing Hoechst 33342 (Beyotime, Cat# C1027) at 37 °C for 10 min. Micro-IR was performed using a Zeiss LSM 900 microscope at full power laser bleaching (405 nm, 6 scans, speed 2). Time-lapse images were captured with a 40× objective under 488 nm light. HR and NHEJ reporter assays Reporter assays were performed to assess the efficiency of HR and NHEJ. Briefly, HR-GFP and EJ5-GFP cells were kindly donated by Dr. Lei Shi from Tianjin Medical University. DSBs were induced by transfecting cells with HA-tagged I-SceI for 48 h. I-SceI site cleavage within these genome-integrated constructs resulted in restoration of GFP gene expression, either through gene conversion directed-HR using a 3’ N-terminal GFP fragment as an HR template or through NHEJ-mediated elimination of the non-sense frameshift. The percentage of GFP-positive cells was determined by flow cytometry using an LSRFortessa (BD Biosciences) flow cytometer. A minimum of 10,000 cells were detected and analyzed with FlowJo for each group. Comet assay The Comet Assay kit (Beyotime, Cat# C2041) was used to detect damaged DNA according to the manufacturer’s instructions. Briefly, cells were collected and resuspended in PBS at a concentration of 1 × 105 cells/ml. Cells were mixed with pre-warmed low-melting agarose and then spread onto the comet slides. Slides were then maintained at 4 °C for 60 min. Slides were then maintained in lysis buffer at 4 °C for 60 min. Slides were transferred to an electrophoresis tank containing a neutral electrophoresis buffer, and the system was run at 1 V/cm for 30 min at 4 °C. The slides were immersed twice in neutral electrophoresis buffer and stained with propidium iodide solution for 20 min. The length of comet tails was determined by ImageJ. Mouse serum immunoglobulin detection Serum samples were collected from 8–10-week-old Crtc2+/+ and Crtc2−/− mice. Serum was diluted in assay buffer, and the concentrations of IgM, IgA, IgG1, IgG2a, IgG2b, or IgG3 were determined with the mouse Ig isotyping ELISA kit (Invitrogen, Cat# 88-50630-88) according to the manufacturer’s instructions. The concentrations for each immunoglobulin were calculated and normalized to the Crtc2+/+ group. Class switch recombination (CSR) assay Mouse pan B cells were isolated from the spleen of 10-week-old mice by magnetic negative selection using MojoSort™ Mouse Pan B Cell Isolation Kit (BioLegend, Cat# 480051) according to the manufacturer’s instructions. The enriched cells were cultured in RPMI 1640 supplemented with 10% FBS. To induce class switching in mouse B cells, they were cultured for 96 h in medium supplemented with LPS (37.5 µg/ml, Solarbio, Cat# 8880) or a mixture of IL4 (10 ng/ml, MCE, Cat# HY-P7080) and LPS 4, 8. CSR was assessed by flow cytometry after live cell staining with FITC-labeled anti-CD19 antibodies and PE-labeled anti-IgG3 and anti-IgE. The CSR efficiency was then examined by labeling cells with FITC-conjugated anti-CD19 and PE-conjugated anti-IgG3 or anti-IgE antibodies. All flow cytometry assays were detected on LSRFortessa (BD Biosciences) and analyzed with FlowJo. LC−MS/MS analysis Irradiated HCCLM3 cells were collected and subjected to pulldown by anti-CRTC2 and Dynabeads (Thermo Fisher, 10006D). Then, the lysis buffer was reduced, alkylated, and digested by trypsin overnight. After that, the tryptic digest was redissolved in HPLC solvent A (0.1% (v/v) formic acid in water) and injected into a nano-LC system (EASY-nLC 1200, Thermo Fisher Scientific). Each sample was separated by a C18 column (75 μm inner-diameter × 15 cm, 3 μm) with a 130 min HPLC gradient at a flow rate of 300 nl/min. The HPLC gradient was as follows: 5% to 7% solvent B (0.1%(v/v) formic acid in acetonitrile) in 2 min, 7% to 22% solvent B in 80 min, 22% to 38% solvent B in 38 min, 38% to 100% solvent B in 3 min and hold for 7 min at 100% solvent B. The HPLC elute was electrosprayed into an Orbitrap Eclipse mass spectrometer (Thermo Fisher Scientific). The source was performed at 2.2 kV. The mass spectrometric analysis was performed in a data-dependent mode. For MS1 survey scan, automatic gain control (AGC) target was 4 × 105 and the resolution is 60,000. The MS2 spectra were obtained with 15,000 resolution. Raw data were analyzed with Thermo Proteome Discoverer 3.0 with an overall false discovery rate (FDR) for peptides of less than 1%. Peptide sequences were identified using trypsin specificity and allowing a maximum of two missed cleavages. Carbamidomethylating on cysteine was specified as fixed modification. Oxidation of methionine and acetylation on N-terminal were set as variable modifications. Mass tolerances for precursor ions were set at ±10 ppm for precursor ions and ±0.02 Da for MS/MS. Plasmids pCBA-I-SceI-HA was used for HR-GFP and EJ5-GFP reporter assay. pAAV2/8, pAdDeltaF6 and pAAV- U6-sgCrtc2-TBG-Nme2Cas9 were used for AAV package. pX330-Cas9-sgKeap1, pT3-EF1α-MYC and sleeping beauty transposase vector were used for Keap1KO/MYCOE- liver cancer model development. All plasmids were constructed by the Gibson assembly approach. TY1-tagged CRTC2 (1- 693 aa) and TY1-tagged CRTC2∆TAD (1- 430 aa) were cloned into pLVX vector. CRTC2 (1- 693 aa) was cloned into pLVX-EGFP-C1 vector. TAD (431-693 aa)- MYC, TADS433A (431-693 aa)- MYC and KD (3677- 4128 aa)- Flag were cloned into pGEX-6P-1. For CRTC2 mutation, S433 of TY1-tagged CRTC2 was mutated to A or E using Mut Express II Fast Mutagenesis Kit V2 (Vazyme, #C214) according to the manufacturer’s instructions. GFP tag was cloned into CRTC2-TY1, PARP1-MYC, S433A-TY1, S433E-TY1 and DNA-PKcs plasmids. Protein expression and purification The TY1-tagged CRTC2, TAD and S433A mutant TAD domains of CRTC2 linked with MYC tag (TAD-MYC and TADS433A-MYC) and the kinase domain of DNA-PKcs linked with Flag tag (KD-Flag) were cloned and constructed on pGEX-6P-1 vector containing GST tag at N-terminal. TAD-MYC, TADS433A-MYC and KD-Flag were expressed in BL21 (DE3) chemically competent cells (Tsingke, Cat# TSC-E01). Escherichia coli cells expressing TAD-MYC, TADS433A-MYC and KD-Flag were induced by 0.5 mM IPTG at 16 °C for 16 h, after which they were collected and resuspended in lysis buffer (50 mM NaH2PO4, 300 mM NaCl, 20 mM imidazole, EDTA-free protease inhibitor cocktail) and maintain ultrasonic for 15 min (5 s run and 5 s stop) on ice. TAD-MYC, TADS433A-MYC and KD-Flag proteins were pulldown by GST-tag protein A/G agarose beads and the GST-tag was removed by PreScission Protease (Beyotime, Cat# P2303) according to the manufacturer’s instructions. Western blot RIPA buffer (150 mM NaCl, 1% NP-40, 50 mM Tris-HCl, 0.1% SDS, 0.5% sodium deoxycholate) supplemented with phosphatase and protease inhibitor cocktail was used for cell lysis. The concentration of proteins was measured using the BCA protein assay kit (Thermo Fisher Scientific, Cat# 23225). Following this, the proteins were separated by SDS-PAGE and then transferred to a PVDF membrane (Millipore, Cat# IPVH00010). The membranes were blocked with 5% skim milk in TBST at room temperature and incubated with the primary antibody at 4 °C overnight. After washing with TBST, the membrane was incubated with the antibody conjugated to horseradish peroxidase (HRP) at room temperature for 1 h. Protein signals were captured by GelView 6000Plus imaging system (Biolight, CN) and the densitometric analysis was performed with ImageJ software. The following antibodies were employed: anti-CRTC2 (Proteintech, 12497-1-AP, 1:2000), anti-β-actin (Proteintech, 66009-1-Ig, 1:5000), anti-GFP (Proteintech, 50430-2-AP, 1:2000), anti-TY1 (Genscript, A01004, 1:2000). Anti-γ-H2AX (Sigma-Aldrich, 05-636, 1:5000), anti-PARP1 (Proteintech, 66520-1-lg, 1: 2000), anti-KU70 (CST, 4588, 1:1000), anti-KU80 (CST, 2180, 1: 1000), anti-DNA-PKcs (Abclonal, A20837, 1:2000), anti- P-DNA-PKcs-S2056 (Abclonal, AP0621, 1:2000), anti-MYC (Proteintech, 60003-2-lg, 1:5000), anti-Flag (Proteintech, 20543-1-AP, 1:5000), anti-Phospho-MAPK/CDK Substrates (PXS*P or S*PXR/K) (CST, 2325, 1:1000), anti-CREB (CST, 9197, 1:1000), anti-Cyclin T1 (CST, 81464, 1:1000), anti-IRF3 (CST, 4302, 1: 1000), anti-TBK1 (CST, 3504, 1:1000), anti-STING (CST, 13647, 1:1000), anti-Phospho-IRF3 (Ser386) (Affinity, AF3438, 1:2000), anti-Phospho-TBK1 (Ser172) (Affinity, AF8190, 1:2000), anti-Phospho-STING (Ser366) (Affinity, AF7416, 1:2000), anti-Poly/Mono ADP Ribose (CST, 89190, 1:1000). A custom rabbit polyclonal antibody against CRTC2 p-Ser433 was generated by HUABIO (Hangzhou, CN). ELISA confirmed its specific binding to the phospho-antigen (Supplementary Fig. 6e), and it was used at a 1:10,000 dilution. Co-immunoprecipitation (Co-IP) assay Cells were lysed with NP-40 lysis buffer (150 mM NaCl, 1% NP-40, 50 mM) containing protease and phosphatase inhibitor cocktail (Protease Inhibitor Cocktail Tablets, 04693132001, Roche) and incubated on ice for 1 h. The supernatant was mixed with the indicated primary antibody and incubated at 4 °C overnight. Dynabeads™ Protein G (Thermo Scientific, 10004D, USA) was added to the supernatant after for additional 2 h. Subsequently, the supernatant was washed with Co-IP washing buffer (20 mM Hepes, 50 mM NaCl, 1 mM EDTA, 1 mM EGTA, 0.1% NP-40, 2% glycerol) three times and resuspended with 1× loading buffer for western blotting detection. Primary antibodies as following: anti-Flag (Proteintech, 20543-1-AP, 2 μg/mg), anti-MYC (Proteintech, 60003-2-Ig, 2 μg/mg), anti-TY1 (Genscript, A01004, 1.5 μg/mg), anti-CRTC2 (Proteintech, 12497-1-AP, 2 μg/mg), anti-KU80 (CST, 2180, 2 μg/mg). PAR pull-down assay Purified CRTC2-TY1 protein was incubated with Biotin-PAR (R&D Systems, Ca# 4336-100-02) in reaction buffer (100 mM Tris, pH 8.0, 10 mM MgCl2, 10% glycerol, 1.5 mM DTT). Biotin-PAR was pulled down using streptavidin beads (Invitrogen, 65002). The PARylation of CRTC2 was detected by western blot using PAR antibody (CST, 89190). DNA-PK kinase activity assay DNA-PK kinase activity was detected using the ADP-GloTM Kinase Assay (Promega, V6930) and DNA-Dependent Protein Kinase Peptide Substrate (Promega, V5671) according to the manufacturer’s instructions. HCCLM3 cells were collected at 1, 2, and 3 h after IR and incubated with DNA-Dependent Protein Kinase Peptide Substrate in DNA-PK reaction buffer. The relative DNA-PK kinase activity was measured by luminescence signal. For the in vitro DNA-PK reaction, purified TAD and TADS433A proteins were mixed with purified DNA-PK kinase (50 U) (Promega, V5811) in DNA-PK reaction buffer (50 mM HEPES (pH 7.5), 1 mM DTT, 0.1 mM EDTA, 0.2 mM EGTA, 10 mM MgCl2, 0.1 M KCl, 1.14 mM DNA-PK Peptide Substrate, 80 µg/ml BSA, 0.2 mM ATP) according to the assay described. Immunohistochemistry and H&E staining Liver tissues were fixed in 4% paraformaldehyde (PFA) for at least 24 h, paraffin-embedded and sectioned into 5-μm slices. H&E (Solarbio, Cat# G1120) staining was performed according to the manufacturer’s instructions. Immunohistochemical staining was performed according to the established protocol. In summary, the tissue samples were treated with xylene to remove paraffin and then rehydrated through a series of graded alcohols followed by distilled water. Following antigen retrieval and blocking, the tissue sections were incubated with the indicated primary antibody at 4 °C overnight and then with the corresponding secondary antibody at room temperature for 1 h. The signals were developed using a DAB substrate solution. The slides underwent counterstaining with hematoxylin, followed by dehydration and the application of coverslips for mounting. Images were obtained using an optical microscope (Zeiss, Germany). IHC results were evaluated using a double-blind method and given a score between 0 and 12. The scores were obtained by multiplying the intensity score (0, negative; 1, weak; 2, moderate; and 3, high) by the percentage of cells (0, no staining; 1+, 1–25% positive cells; 2+, 26–50% positive cells; 3+, 51–75% positive cells; and 4+, 76–100% positive cells). The primary antibodies as follows: CRTC2 (Proteintech, 12497-1-AP, 1:150), anti-Phospho-STING (Ser366) (Affinity, AF7416, 1:200). Immunofluorescence staining To assess cell immunofluorescence of protein co-localization, cells on glass coverslips were washed with PBS and fixed using 4% PFA. Following this, cells were incubated in blocking solution (3% FBS, which included 0.3% Triton X-100) at room temperature for 30 min. The cells were incubated with indicated primary antibody at 4 °C overnight, then incubated with the secondary antibody for 1 h at room temperature. Glass coverslips were put on slides and mounted using FluoroshieldTM containing DAPI (Sigma-Aldrich, F6057, USA). Images were obtained through a confocal microscope (ZIESS, Germany). The primary antibodies as follows: CRTC2 (Proteintech, 12497-1-AP, 1:100), Anti-γH2AX (Sigma-Aldrich, 05-636, 1:300), anti-PARP1 (Proteintech, 66520-1-lg, 1:100), anti-KU80 (CST, 2180, 1:100), anti-DNA-PKcs (Abclonal, A20837, 1:200), anti-53BP1 (BD Biosciences, 612522, 1:150). For γH2AX and 53BP1 foci quantification, only spots with well-defined edges and uniform fluorescence intensity were counted. RNA interference siRNAs were synthesized by GENECFPS (Jiangsu, CN) and transfected into the indicated cells using Lipofectamine RNAiMAX (Thermo Fisher, Cat# 13778075). The siRNA sequences as following: si-CRTC2-1 (5’- CAUACACAAGGAGCUCUC ATT-3’); si-CRTC2-2 (5’- −3’); si-CRTC2-2 (5’- CUGCUAUUGAGGAGAACUUTT-3’); si-KU80 (5’- CAAGGAUGAGAUUGCUUUAGU-3’); si-PARP1 (5’- AGAA AGUGUGUUCAACUAAUG-3’); si-DNA-PKcs (5’- GCUCUAGAAGAAUGGUCA ATT-3’) Proximity ligation assay (PLA) Cells were seeded on glass coverslips and transfected with CRTC2-TY1, S433A-TY1 and S433E-TY1. After 48 h, they were washed with cold PBS and fixed with 4% paraformaldehyde for 10 min on ice, then blocked with the blocking buffer supplied in Duolink In Situ Detection Reagents at 37 °C for 1 h. Cells were then incubated with the indicated primary antibodies at 37 °C for 1 h. PLA assay was performed with Duolink In Situ Detection Reagents (Sigma-Aldrich, Cat# DUO92101) according to the manufacturer’s instructions. Coverslips were mounted on slides with Fluoroshield containing DAPI. Images were captured using a Zeiss LSM 900 microscope and quantified by ImageJ. Cell proliferation, colony formation and apoptosis assays Cells were seeded into 96-well plates at a density of 2000 cells per well for the CCK-8 assay. Cell viability was measured using CCK-8 solution (Beyotime, Cat# C0038) after 1, 2, 3, or 5-days culture. Cells were seeded into 96-well plates at 500, 1000, 1500, and 2000 cells per well and treated without or with IR at 2, 4, or 6 Gy for the colony formation assay. After 14-day culture, cells were fixed and stained with 0.1% crystal violet, and then counted. For apoptosis assay, Cells were subjected to IR at a dose of 6 Gy and cultures for 48 h and 72 h. Cells were collected and washed with PBS, and then suspended in binding buffer (2.5 μL of FITC and 2.5 μL of PI) and incubated at 4 °C for 15–20 min. On completion of staining, 200 μL of binding buffer was added to each tube and mixed for flow cytometer detection. Micronucleus assay HCCLM3 cells were treated with IR (6 Gy) and cultured in medium containing Cytochalasin B (2 µM, MCE, Cat# HY-16928) for 48 h. Cells were then incubated in pre-warmed 75 mM KCl at 37 °C for 20 min and in Carnoy’s buffer (3:1 methanol: acetic acid) at room temperature for 15 min. After incubation, the cells were collected and washed twice with Carnoy’s buffer. Cell droplets were spread onto glass slides and dried for 15 min. Slides were stained with DAPI. The micronucleus frequency was calculated by micronucleus number/cell number. Real-time qPCR Total RNA was extracted using TRIzol (Invitrogen, Cat# 15596026CN). The RNA concentration was measured with a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific). cDNA synthesis was performed with PrimeScript™ RT reagent Kit with gDNA Eraser (Takara, Cat# RR047A). mRNA level was detected with the 2× TSINGKE® Master qPCR Mix Kit (Tsingke, Cat# TSE201). The sequences of the PCR primers are presented in Supplementary Table 1. Gene expression levels were determined using the 2-ΔΔCt approach. T cell killing assay PBMCs were first separated from whole blood from healthy humans. Anti-CD3 antibody (1 µg/mL) was diluted with sterile PBS for a 6-well plate precoating. The culture plate was placed in a 5% CO2 incubator and incubated at 37 °C for 2 h. After which, the anti-CD3 was removed, but the culture plate was not washed. PBMC cells isolated from whole blood or lymphoid tissue were resuspended in complete RPMI 1640 medium (10% FBS) to a concentration of 3 × 106 cells/ml. Anti-CD28 (1 µg/mL) and IL-2 (30 ng/ml) antibodies were added to the cell solution. Subsequently, cells were placed in CD3 pre-coated 6-well plate and cultured for 2 days for T cell activation. Activated T cells were mixed with pre-plated HCCLM3 cells and culture for additional 24 h. The cells were harvested and stained with crystal violet. Images were captured and T cell killing efficiency was analyzed by ImageJ. Flow cytometry For flow cytometry experiments to detect immune cells in tumors, upon the sacrifice of the mice, the tumors were extracted and minced fully in a 12-well plate. Subsequently, 1640 culture medium containing collagenase IV (Solarbio, Cat# C8160) was added for complete digestion. The sample was placed on a shaker at 37 °C for 30 min of digestion and filtered twice through a 70 μm filter, and ultimately centrifuged at 1000 × g for 5 min. After centrifugation, stain buffer was added to resuspend the sample, which was then transferred to a flow cytometry tube. Subsequently, cells were blocked on ice and stained with corresponding indicators. Centrifugal washing was conducted at the end of each staining process. When the antibody indicator for staining was located within the nucleus, a membrane disruption operation was necessary. Finally, the samples were resuspended in approximately 300 μl of PBS and subjected to machine-based detection. The antibodies and kit employed for flow cytometry staining were as follows: FITC anti-mouse CD45 (BioLegend, Cat# 157607), PerCP/Cyanine5.5 anti-mouse CD3 (BioLegend, Cat# 100217), APC anti-mouse CD8a Recombinan t(BioLegend, Cat# 155005), PE anti-mouse CD279 (PD-1) (BioLegend, Cat# 114117), Brilliant Violet 421™ anti-human/mouse Granzyme B Recombinant (BioLegend, Cat# 396413), PE/Cyanine7 anti-mouse IFN-γ (BioLegend, Cat# 505825), Cell Activation Cocktail (with Brefeldin A) (BioLegend, Cat# 423303), TruStain FcX™ PLUS (anti-mouse CD16/32) (BioLegend, Cat# 156603), Zombie Aqua™ Fixable Viability Kit (BioLegend, Cat# 423101). All flow cytometry assays were detected on LSRFortessa (BD Biosciences) and analyzed by FlowJo. CUT&Tag sequencing and data processing Cells (1 × 105 cells per sample) were collected for library preparation using the Hyperactive Universal CUT&Tag Assay Kit for Illumina (Vazyme, Cat# TD903) according to the manufacturer’s instructions and sequenced with PE150 (paired-end, 150 bp reads) on Novogene’s NovaSeq high-throughput sequencing platform. CUT&Tag sequencing reads were first processed using fastp (v1.0.1) to trim adaptor and low-quality reads. The quality of all trimmed file was verified by FastQC (v0.12.1) and processed using a classical CUT&Tag pipeline. The trimmed reads were then aligned to the human assembly GRCh38 using bowtie2 (v2.5.4) with the parameters: --end-to-end --very-sensitive, --no-mixed, and --no-discordant and were deduplicated using sambamba (v1.0.1) with the “markdup” command. CUT&Tag target enrichment used MACS3 with a q-value setting of 0.01. Signal tracks for each sample were created using bamCoverage (deepTools v3.5.6) tool and were normalized to Counts Per Million mapped reads (CPM). To identify AsiSI sites (GCGATCGC) with the highest cutting efficiency, the average log2 ratio of γH2AX CUT&Tag in DSB-induced AiDIvA cells versus ER-expressing control cells was calculated in a 1 Mb window around the 1220 predicted AsiSI sites. The top 250 AsiSI sites were then selected for subsequent analysis. The random non-DSB regions across the entire genome (excluding chromosome Y) were generated by R using the following rules:1) at least 1 Mb away from the top cleaved positions, 2) excluding regions with a null CUT&Tag count in a 1 kb window. The primary antibodies as follows: CRTC2 (Proteintech, 12497-1-AP, 2 µL), Anti-γ-H2AX (Sigma-Aldrich, 05-636, 2 µL). RNA sequencing AiDIvA cells treated with doxycycline and in the absence or presence of 4-OHT were collected, centrifuged and washed twice with PBS. Then cells were resuspended with buffer A (10 mM Tris-HCl, 10 mM KCl, 1.5 mM MgCl2, 1× protease inhibitor) and incubated on ice for 10 min. Centrifuge and resuspend the pellets with buffer B (buffer A containing 0.05% NP-40) and incubate lysis on ice for 5 min. Centrifuge the lysis at 14,000 g for 20 min and discard the supernatant. The pellets were resuspended with buffer A for nuclear RNA extraction. Nuclear RNA was isolated by TRIzol (Sigma, USA) and the mRNA was purified from total RNA using poly-T oligo-attached magnetic beads. After fragmentation, first strand cDNA was synthesized using random hexamer primers. And the second strand cDNA was synthesized using dUTP, instead of dTTP. Library preparation and sequencing were performed by Novogene Co., Ltd. (Beijing, CN). The clean reads from RNA-seq experiment were aligned to the GRCh38 assembly human reference genome using the STAR (v2.7.11b) with default parameters. Read count tables were generated with stringtie (v3.0.1). Gene expression normalization and dynamic profiling were evaluated with R package DESeq2 (v1.46.0) using the default parameters. Coverage of RNA-seq reads across genebody regions was normalized and visualized using the latest version of ngs.plot package48. Generation of coverage profile plots was based on the GRCh38 reference genome assembly with gene annotations from Ensembl release 76. Single-cell RNA sequencing Two-week-old Crtc2+/+ and Crtc2−/− mice were intraperitoneally injected with DEN (25 mg/kg). Mice were maintained NASH diet from 8 to 26 weeks and then sacrificed after 26. Tumors from the liver were collected for single-cell RNA sequencing. Tumors cell suspensions were loaded into Chromium microfluidic chips with 3’ chemistry and barcoded with a 10× Chromium Controller (10× Genomics). RNA from the barcoded cells was subsequently reverse-transcribed and sequencing libraries constructed with reagents from a Chromium Single Cell 3’ v2 reagent kit (10× Genomics) according to the manufacturer’s instructions. Sequencing was performed with Illumina (HiSeq 2000) according to the manufacturer’s instructions. Raw reads were demultiplexed and mapped to the reference genome by 10× Genomics Cell Ranger pipeline (https://support.10xgenomics.com/single-cell-gene-expression/software/pipelines/latest/what-is-cell-ranger) using default parameters. All downstream single-cell analyses were performed using Cell Ranger and Seurat96,97. In brief, for each gene and each cell barcode, unique molecule identifiers were counted to construct digital expression matrices. A gene with expression in more than three cells was considered convinced, and each cell was required to have at least 200 expressed genes. AAV production and purification The liver-specific TBG promoter carried Cas9 and U6-sgCrtc2 AAV cassette was used as previously described29. pAAV-U6-sgCrtc2-TBG-Nme2Cas9, pAAV2/8 and pAdDeltaF6 were co-transfected into HEK293T cells using PEI in a ratio of 1:4:2. Refresh the medium every 12 h to keep the cell healthy. Four days after transfection, the supernatant and cells were collected and centrifuged. The supernatant was mixed with 40% PEG8000 and incubated at 4 °C overnight. Cells were resuspended using SAN (Salt active nuclease) buffer (500 mM NaCl, 40 mM Tris-base, 10 mM MgCl2) and 100 U Benzonase (Sigma-Aldrich, Cat# E1014) was added with incubating at 37 °C for 1 h. Then the SAN solution was centrifuged and the pellet was resuspended and mixed with cell lysate as the virus solution. The virus solution was purified by ultracentrifugation in iodixanol density gradient solutions at 350,000 g at 4 °C for 2.5 h and Amicon Ultra-15 centrifugal filter devices (100 kDa) (Millipore, Cat# UFC910024). Titration was performed as previously described98. Liver function detection (AST/ALT) Serum samples in Keap1KO/MYCOE- liver cancer model were collected for liver function detection. The measured biochemical markers in the mouse serum, including ALT and AST, were measured using commercial kits (Nanjing JianCheng Bioengineering Institute) according to the manufacturer’s instructions. Statistics and reproducibility The number of independent experimental replicates, mice for in vivo experiments and in vitro cell-based assays are indicated in the figure legends. Data from experiments were presented as mean ± SEMs or box-plots. ANOVA and Student’s t test analysis were performed with GraphPad Prism version 9.0. All statistical tests were conducted as two-sided. The selection between Pearson’s and Spearman’s correlation coefficients in Prism was dependent on the normality of the data. P values were present in columns or legends. P < 0.05 was considered to be statistically significant. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Reporting Summary Transparent Peer Review file Source data Source Data Acknowledgements The authors sincerely thank Yupeng Chen (Tianjin Medical University, Tianjin, China) for providing Crtc2 knockout mice. This work was supported by the National Natural Science Foundation of China (Grant No. 82472654 and 82170758 to Z.Mi, 82373085 to Z.Wu, 82172674 and 82473240 to Z.Yuan, 82503518 to M.Wang), the Science & Technology Development Fund of Tianjin Education Commission for Higher Education (grant 2023KJ095 to F.Zou), the Natural Science Foundation of Tianjin (S25QN0767 to M.Wang), National key R&D Program of China (2022YFC2304603 to Xinlu Wang), and the Tianjin Key Medical Discipline Construction Project (TJYXZDXK-3-003A). Author contributions L.S., Z.W. and Z.M. conceived, designed, and supervised the project, analyzed data, and wrote the manuscript. F.Z., Z.Y. and X.D. performed biochemistry, cell biology, and animal studies. Z.X. and H.Z. performed bioinformatics analysis and protein structural analysis. Z.Y., Y.S., L.S. and Y.L. provided pathological analysis. K.H. provided expertise on immunofluorescence analysis of DNA-PKcs (phosphorylation or foci formation) following micro-irradiation. M.W. and X.L. provided expertise on animal studies. K.C. provided human liver cancer samples. X.W., J.W. and S.M. analyzed the data. All authors read and approved the final manuscript. Peer review Peer review information Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Data availability The mass spectrometry proteomics data have been deposited at the ProteomeXchange Consortium with the dataset identifier PXD068727. High-throughput sequencing data of CRTC2 CUT&Tag were deposited into NCBI’s Gene Expression Omnibus with the GEO series accession code GSE306488. High-throughput sequencing data of nuclear RNA-seq were deposited into NCBI’s GEO database with the GEO series accession code GSE306599. High-throughput sequencing data of single-cell RNA-seq were deposited into NCBI’s GEO database with the GEO series accession code GSE307028. Any additional information required to reanalyze the data reported in this paper is available from the corresponding authors upon request. Source data are provided with this paper. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Fangdi Zou, Zhiqi Yao, Xiaohan Dong. Contributor Information Lei Shi, Email: shilei@tmu.edu.cn. Zhiqiang Wu, Email: zwu08@tmu.edu.cn. Zeyun Mi, Email: mizeyun@tmu.edu.cn. Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-73228-4. References Associated Data Supplementary Materials Supplementary Information Reporting Summary Transparent Peer Review file Source Data Data Availability Statement The mass spectrometry proteomics data have been deposited at the ProteomeXchange Consortium with the dataset identifier PXD068727. High-throughput sequencing data of CRTC2 CUT&Tag were deposited into NCBI’s Gene Expression Omnibus with the GEO series accession code GSE306488. High-throughput sequencing data of nuclear RNA-seq were deposited into NCBI’s GEO database with the GEO series accession code GSE306599. High-throughput sequencing data of single-cell RNA-seq were deposited into NCBI’s GEO database with the GEO series accession code GSE307028. Any additional information required to reanalyze the data reported in this paper is available from the corresponding authors upon request. Source data are provided with this paper.