ACSS2-KAT5 complex-driven histone crotonylation orchestrates a pro-inflammatory program to promote the transition from MASLD to MASH Authors: Xiao Wen, Keyan Wu, Mengyao Wang, Zihan Ma, Tao Wang, Jing Zhang, Bei Wang, Siyuan Chen, Jingyi Wang, Siyue Chen, Fan Yang, Chuwei Liu, Xianyang Chen, Lu Deng, Yafan Cheng, Qing Robert Miao, Baofa Sun, Xiongzhong Ruan, Kai Li, Yajun Duan, Wenquan Hu Published: 2026 ABSTRACT Inflammation is a pivotal driver of the progression from metabolic dysfunction-associated steatotic liver disease (MASLD) to metabolic dysfunction-associated steatohepatitis (MASH), an aggressive form associated with substantial liver-related mortality. However, the molecular mechanisms underlying the initiation and persistence of liver inflammation remain poorly defined. Here, we demonstrated a previously unrecognized role for hepatic acetyl-CoA synthetase short-chain family member 2 (ACSS2) in MASH, showing that ACSS2 upregulation in patients exacerbates MASH progression by functioning as an epigenetic regulator, independent of its canonical lipogenic role. Mechanistically, ACSS2, in complex with lysine acetyltransferase 5 (KAT5), upregulates allograft inflammatory factor-1 (AIF1) transcription via histone crotonylation, thereby inducing liver inflammation and subsequently resulting in the aberrant accumulation of senescent hepatocytes, which further enhances proinflammatory cytokine production. This ultimately initiates a vicious cycle of chronic inflammation, which directly promotes the progression from simple steatosis to MASH. Thus, our work reveals a mechanistically defined and pivotal role for ACSS2 in promoting the MASLD-to-MASH transition, highlighting its potential as a compelling therapeutic target. Introduction Metabolic dysfunction-associated steatotic liver disease (MASLD) and its more severe subtype, metabolic dysfunction-associated steatohepatitis (MASH)1, have become a global epidemic2. MASH, is characterized by hepatic inflammation, hepatocyte ballooning and fibrosis and can progress to cirrhosis and hepatocellular carcinoma (HCC)3. In recent years, changes in dietary patterns have contributed to an increasing global prevalence of MASH, which increased by approximately 50% between 2005 and 20204. The most pronounced increase has been observed among young adults aged 18–34 years, indicating a shift toward a younger affected demographic5. Notably, inflammation is a central driver in the transition from simple-steatosis MASLD to immune-activated MASH, which occurs in 12–40% of MASLD patients6,7. This transition is associated with various triggers, including oxidative stress, endoplasmic reticulum stress, insulin resistance, and dyslipidemia, indicating that the underlying metabolic dysfunction in the liver is multifactorial8. Inflammation-induced hepatocyte death leads to the release of damage-associated molecular patterns (DAMPs), which in turn recruit immune cells and stimulate the production of hepatotoxic factors, such as tumor necrosis factor (TNF), further exacerbating the disease9. However, the precise mechanisms that initiate the inflammatory cascade and drive the transition from MASLD to MASH remain elusive. Acetyl-CoA synthetase short-chain family member 2 (ACSS2), a crucial member of the ACSS family, is an enzyme that catalyzes the formation of acetyl-CoA from free acetate10. In addition to its central role in cellular energy metabolism, ACSS2 regulates various acetylation processes, such as those involving histones and transcription factors, and has been implicated in diverse pathological processes, including tumorigenesis, aging and the modulation of neuropsychiatric functions and memory10–14. Recent studies have further revealed that the regulatory role of ACSS2 extends beyond acetylation: ACSS2 governs gene expression programs for metabolism, proliferation, and stress responses through novel acylations, including β-hydroxybutyrylation, butyrylation, and lactylation, thereby establishing ACSS2 as a pivotal epigenetic regulator15–17. Previous studies have shown that ACSS2 and ACLY have overlapping roles in lipid metabolism. Wellen et al. found that ACLY deficiency upregulates ACSS2, which uses exogenous acetate to supply acetyl-CoA for fatty acid synthesis and histone acetylation18. Steinberg et al. showed in MASH mice that inhibiting ACLY in hepatocytes reduces steatosis and ballooning, but compensatory ACSS2 upregulation limits this effect; dual inhibition of ACLY and ACSS2 more effectively reduces inflammation and fibrosis19. Thus, ACLY and ACSS2 have a synergistic and compensatory relationship in hepatic lipid metabolism. Beyond transcriptional control, ACSS2 exerts complex and context-dependent effects on inflammation, precluding its classification as uniformly pro- or anti-inflammatory12,20. This plasticity depends on cell type, disease context, activating stimuli, and metabolic status10. ACSS2 also directly regulates fibrosis: Steinberg and colleagues21 demonstrated that ACSS2 controls hepatic stellate cell activation and fibrogenesis. Moreover, owing to its pivotal role in controlling inflammatory and fibrotic responses, especially in amplifying dysregulated inflammation, ACSS2 presents a compelling target for therapeutic intervention20. The risk for developing MASLD and other metabolic dysfunctions varies among individuals but generally increases with age22. This age-related functional decline often stems from cellular senescence and organ aging23. Senescence, a state of irreversible cell cycle arrest, is a key driver of tissue degeneration and organismal aging24. Senescent cells (SCs) are characterized by established hallmarks, including elevated senescence-associated β-galactosidase (SA-β-gal) activity, the formation of senescence-associated heterochromatin (SAHF) foci, and the sustained activation of tumor suppressor pathways25,26. A defining characteristic of SCs is their secretion of diverse inflammatory mediators, collectively termed the senescence-associated secretory phenotype (SASP)27. During MASH progression, lipotoxicity, oxidative stress, and DNA damage promote the accumulation of SCs in the liver28,29, leading to substantial SASP release that amplifies hepatic inflammation30. Although cellular senescence is well established in conditions such as neurodegeneration and cardiovascular diseases31,32, its specific functional contributions to MASH pathogenesis remain largely undefined. Allograft inflammatory factor-1 (AIF1), a cytoplasmic calcium-binding protein, is a proinflammatory regulator and key marker of activated immune cells33,34. It promotes the pathogenesis of inflammatory and immune-related diseases by mediating key processes, including cytokine release, reactive oxygen species (ROS) production, and phagocytosis35–37. A recent study reported that AIF1 mRNA levels in macrophages were higher in women with obesity compared with those without obesity, and that its expression was positively correlated with insulin resistance38. Moreover, treatment with conditioned medium from AIF1-overexpressing macrophages exacerbated lipid accumulation, ROS production, and inflammation in 3T3-L1 adipocytes39,40. Conversely, AIF1 deficiency confers protection against high-fat diet-induced obesity and metabolic disorders40. Despite its broad involvement in inflammatory processes, the specific role of AIF1 in driving hepatocellular lipid accumulation and inflammation during MASH progression remains elusive. In this study, we demonstrated that hepatocyte-specific deletion of ACSS2 protects against MASH progression by alleviating steatosis, inflammation, and fibrosis. Integrated multiomics and RNA sequencing analyses revealed that ACSS2 drives MASH by epigenetically promoting AIF1-mediated inflammation and hepatocellular senescence. Consistently, pharmacological inhibition of ACSS2 ameliorated hepatic steatosis, inflammation, and fibrosis in a MASH model, whereas ACSS2 overexpression exacerbated these pathological features. Mechanistically, ACSS2 deficiency impaired KAT5-mediated histone crotonylation, leading to downregulation of AIF1 expression and consequent suppression of inflammation and hepatocellular senescence. Thus, our study identifies the ACSS2/KAT5–AIF1 axis as a novel regulatory pathway that promotes MASH progression by linking inflammation with hepatocellular senescence, highlighting its potential as a therapeutic target. Results ACSS2expression is upregulated in the livers of both humans and mice with MASH Acetyl-CoA short chain synthetases (ACSSs) have been identified as significant factors in MASLD and MASH progression, with evidence suggesting that their roles extend beyond metabolism to include functions akin to those of transcriptional regulators41–45. To elucidate the specific mechanisms linking the ACSS family to inflammation in MASH, we first performed a systematic analysis of the Human Protein Atlas (HPA) database. Expression analysis revealed a distinct expression pattern: ACSS2 was the predominant isoform in liver tissue, ACSS3 was expressed at markedly lower levels, and ACSS1 was nearly undetectable (Supplementary Fig. 1A). Given the cellular composition of the liver, we used public single-cell RNA sequencing (scRNA-seq) data to evaluate ACSS2 expression. The results revealed that ACSS2 is the predominant isoform in hepatocytes, exceeding the levels of ACSS1 and ACSS3 (Supplementary Fig. 1B, C). Analysis of the GSE89632 dataset revealed that ACSS2 expression, unlike ACSS1 and ACSS3 expression, was significantly positively correlated with lobular inflammation (Fig. 1A). ACSS2 was also positively associated with fibrosis, body mass index, waist circumference, AST, and ALT (Fig. 1B–F). These findings were corroborated in the validation dataset GSE59045, where ACSS2 levels correlated with key inflammatory chemokines, including CXCL1, CCL4, CXCR4, and IL7R (Fig. 1G).Fig. 1ACSS2 is upregulated in the livers of both humans and mice with MASH.A Correlation between ACSS2 mRNA and lobular inflammation in GEO cohort of MASH patients. Correlation of ACSS2 with fibrosis (B), BMI (C), waist (D), AST (E), ALT (F) in the same GEO cohort of MASH patients. G ACSS2 expression shows positive correlation with inflammatory factors levels in the GSE59045 cohort. H Representative images of the immunohistochemistry staining and quantitative analysis of ACSS2 in human liver sections from the healthy controls, (n = 10), MASLD patients (n = 10) and MASH patients (n = 17), Scale bar = 50 µm. I Representative immunohistochemical staining and quantitative analysis of ACSS1, ACSS2, and ACSS3 in the livers of mice fed a normal chow diet (NCD) or high-fat fructose diet (HFFD; 60 kcal% fat with 30% fructose in drinking water, n = 6). Quantification was performed using Image-Pro Plus. Scale bar = 50 µm. J Western blot analysis of relative ACSS1/2/3 protein levels in mouse primary hepatocytes treated with the indicated concentrations of PA, (n = 3 biological replicates per group). K Relative mRNA levels of Il-1, Il-6, Tnfα, and Mcp1 were detected by real-time PCR analysis in PA-treated mouse primary hepatocytes following ACSS2 knockdown by siRNA, (n = 3 biological replicates per group). Data are presented as mean ± SD. Correlations were assessed using Spearman’s rank correlation test (A–G). Statistical significance was determined by two-tailed Student’s t test (I, K) and one-way ANOVA (H, J). Source data are provided as a Source Data file. ACSS2 is upregulated in the livers of both humans and mice with MASH. A Correlation between ACSS2 mRNA and lobular inflammation in GEO cohort of MASH patients. Correlation of ACSS2 with fibrosis (B), BMI (C), waist (D), AST (E), ALT (F) in the same GEO cohort of MASH patients. G ACSS2 expression shows positive correlation with inflammatory factors levels in the GSE59045 cohort. H Representative images of the immunohistochemistry staining and quantitative analysis of ACSS2 in human liver sections from the healthy controls, (n = 10), MASLD patients (n = 10) and MASH patients (n = 17), Scale bar = 50 µm. I Representative immunohistochemical staining and quantitative analysis of ACSS1, ACSS2, and ACSS3 in the livers of mice fed a normal chow diet (NCD) or high-fat fructose diet (HFFD; 60 kcal% fat with 30% fructose in drinking water, n = 6). Quantification was performed using Image-Pro Plus. Scale bar = 50 µm. J Western blot analysis of relative ACSS1/2/3 protein levels in mouse primary hepatocytes treated with the indicated concentrations of PA, (n = 3 biological replicates per group). K Relative mRNA levels of Il-1, Il-6, Tnfα, and Mcp1 were detected by real-time PCR analysis in PA-treated mouse primary hepatocytes following ACSS2 knockdown by siRNA, (n = 3 biological replicates per group). Data are presented as mean ± SD. Correlations were assessed using Spearman’s rank correlation test (A–G). Statistical significance was determined by two-tailed Student’s t test (I, K) and one-way ANOVA (H, J). Source data are provided as a Source Data file. To substantiate these findings at the protein level, we performed IHC staining for ACSS2 protein expression in 37 liver tissue sections from normal controls (n = 10), MASLD patients (n = 10), and MASH patients (n = 17). Statistical analysis revealed a pronounced upregulation of ACSS2, specifically in MASH patients, with levels in MASLD patients remaining comparable to those in normal controls (Fig. 1H). We also examined the expression of ACSS1, ACSS2, and ACSS3 in a MASH mouse model. IHC staining and western blotting revealed that ACSS2 protein levels were significantly elevated in mice with HFFD-induced MASH compared with those in the NCD group, whereas ACSS1 and ACSS3 levels did not significantly change (Fig. 1I, Supplementary Fig. 1D). Furthermore, in vitro experiments revealed that palmitic acid (PA) treatment upregulated ACSS2 in a dose- and time-dependent manner across multiple hepatocyte models, with no corresponding changes in ACSS1 or ACSS3 (Fig. 1J, Supplementary Fig. 1E–I). Previous studies have shown that SREBP1 can directly bind to the SRE element (−475 to −483 bp) in the ACSS2 promoter to drive its transcription46; and that ChREBP activates the target gene Acss2 to regulate hepatic acetate metabolism41. Our experiments confirmed this finding, showing that HFFD-fed mice had significantly higher hepatic mRNA levels of both SREBP1 and ChREBP than NCD controls (Supplementary Fig 1J–K). Functionally, while ACSS2 deficiency had a negligible effect under basal conditions, it significantly attenuated the PA-induced expression of key inflammatory cytokines (IL-1, IL-6, TNFα, and MCP1; Fig. 1K, Supplementary Fig. 1L–N). On the basis of these findings, we hypothesize that ACSS2 promotes the progression from MASLD to MASH by driving hepatic inflammation. Hepatocyte-specificACSS2deficiency ameliorated HFFD/MCD/CCL4-induced inflammation and suppressed MASH development To investigate the role of hepatocyte ACSS2, we generated hepatocyte-specific ACSS2-knockout mice (Acss2-HKO, Supplementary Fig. 2A, B). Compared with Acss2-Flox control mice, Acss2-HKO mice presented significantly lower ACSS2 protein and mRNA levels in both liver tissue and primary hepatocytes (Supplementary Fig. 2C–E, Supplementary Fig. 3A). These mice were subjected to a HFFD or NCD for 4 months (Fig. 2A). Although no phenotypic differences were observed under NCD conditions (Supplementary Fig. 3B–E), Acss2-HKO mice fed the HFFD presented reduced hepatomegaly and significantly lower liver-to-body weight ratios than Acss2-Flox controls did (Fig. 2B–D, Supplementary Fig. 3F). Acss2-HKO mice also presented an improved metabolic profile characterized by decreased fasting blood glucose and increased glucose and insulin tolerance (Fig. 2E–G). Furthermore, the serum levels of alanine aminotransferase (ALT) and aspartate aminotransferase (AST), as well as triglyceride (TG) accumulation in both the liver and serum, were markedly attenuated in Acss2-HKO mice (Fig. 2H–K). Histological analysis via H&E staining and Sirius Red staining revealed reduced liver hepatic steatosis and fibrosis in Acss2-HKO mice than in Acss2-Flox mice (Fig. 2L–M). Correspondingly, ACSS2 deficiency significantly attenuated the hepatic inflammatory response, macrophage infiltration and lipid accumulation. (Fig. 2N–Q, Supplementary Fig. 3G–M). Surprisingly, despite hepatic stellate cells (HSCs) being the primary fibrogenic cells, in vitro PA treatment did not elevate ACSS2 in LX2 cells. Instead, hepatocytic ACSS2 influenced inflammatory cytokine expression, leading to macrophage recruitment and secretion of pro-inflammatory factor, which in turn modulated HSCs activation (Supplementary Fig. 4).Fig. 2Hepatocyte-specific deficiency of ACSS2 ameliorates HFFD-induced MASH.A Experimental timeline: 8–10 weeks Acss2-Flox and Acss2-HKO mice were fed a normal chow diet (NCD) or a high-fat fructose diet (HFFD) for 16 weeks. B Representative gross morphology of livers from Acss2-Flox and Acss2-HKO mice fed with NCD or HFFD. Scale bar: 1 cm. C Liver-to-body weight ratio of Acss2-Flox and Acss2-HKO mice fed with NCD (n = 6) or HFFD (n = 8). Changes in body weight (D) and blood glucose levels (E) in the indicated groups, (n = 8 biological replicates per group). Glucose tolerance test (GTT, F) and insulin tolerance test (ITT, G) in the indicated groups, (n = 8 biological replicates per group). Serum ALT (H), AST (I), serum TG (J) and hepatic TG (K) in the indicated groups, (n = 8 biological replicates per group). Representative images of H&E staining with quantification of relative vacuolar area (L), Sirius Red staining with quantification of stained area (M) in the liver sections from the indicated groups, (NCD, n = 6 biological replicates per group, HFFD, n = 8 biological replicates per group). Scale bar: 50 μm. N Flow cytometry analysis of percentage of CD11b high Ly6G+ cells, (n = 5 biological replicates per group). Serum IL-6 (O) and TNFα (P) levels determined by ELISA, (NCD, n = 6 biological replicates per group, HFFD, n = 8 biological replicates per group). Q Relative mRNA levels of pro-inflammatory and chemokine molecules in the livers of HFFD-fed Acss2-Flox and Acss2-HKO mice analyzed by real-time PCR, (n = 8 biological replicates per group). Data are presented as mean ± SD. Statistical significance was determined by two-tailed Student’s t test (D–G, N, Q) and two-way ANOVA (C, H–M, O–P). Source data are provided as a Source Data file. Hepatocyte-specific deficiency of ACSS2 ameliorates HFFD-induced MASH. A Experimental timeline: 8–10 weeks Acss2-Flox and Acss2-HKO mice were fed a normal chow diet (NCD) or a high-fat fructose diet (HFFD) for 16 weeks. B Representative gross morphology of livers from Acss2-Flox and Acss2-HKO mice fed with NCD or HFFD. Scale bar: 1 cm. C Liver-to-body weight ratio of Acss2-Flox and Acss2-HKO mice fed with NCD (n = 6) or HFFD (n = 8). Changes in body weight (D) and blood glucose levels (E) in the indicated groups, (n = 8 biological replicates per group). Glucose tolerance test (GTT, F) and insulin tolerance test (ITT, G) in the indicated groups, (n = 8 biological replicates per group). Serum ALT (H), AST (I), serum TG (J) and hepatic TG (K) in the indicated groups, (n = 8 biological replicates per group). Representative images of H&E staining with quantification of relative vacuolar area (L), Sirius Red staining with quantification of stained area (M) in the liver sections from the indicated groups, (NCD, n = 6 biological replicates per group, HFFD, n = 8 biological replicates per group). Scale bar: 50 μm. N Flow cytometry analysis of percentage of CD11b high Ly6G+ cells, (n = 5 biological replicates per group). Serum IL-6 (O) and TNFα (P) levels determined by ELISA, (NCD, n = 6 biological replicates per group, HFFD, n = 8 biological replicates per group). Q Relative mRNA levels of pro-inflammatory and chemokine molecules in the livers of HFFD-fed Acss2-Flox and Acss2-HKO mice analyzed by real-time PCR, (n = 8 biological replicates per group). Data are presented as mean ± SD. Statistical significance was determined by two-tailed Student’s t test (D–G, N, Q) and two-way ANOVA (C, H–M, O–P). Source data are provided as a Source Data file. Wellen and colleagues showed that ACSS2 and ACLY act synergistically, prompting us to test whether Acss2-HKO affects ACLY18. As shown in Supplementary Fig. 5, ACSS2 knockdown did not change ACLY expression or activity, revealing functional dissociation in MASH. ACLY supports basal lipid metabolism and is elevated in MASH, yet it is not regulated by ACSS2. Hence, ACSS2 deletion alone blocks MASH progression, positioning ACSS2 an irreplaceable driver. We propose that ACSS2 promotes MASLD-to-MASH transition by driving liver inflammation. Given the recognized sexual dimorphism in metabolic responses47,48, we further assessed the effects of ACSS2 knockout in female mice. In line with the observed metabolic improvements, ACSS2 deletion in HFFD-fed females significantly reduced body weight and liver-to-body weight ratios and improved glucose and insulin tolerance (Supplementary Fig. 6A–F). Moreover, histological assessment via H&E staining and Sirius Red staining, together with body fat imaging, demonstrated that ACSS2 deficiency also significantly attenuated liver steatosis, fibrosis, and lipid accumulation in females (Supplementary Fig. 6G–L). Owing to the established increased susceptibility of males to diet-induced MASH, all subsequent mechanistic studies were conducted in male mice. Consistent with HFFD results, Acss2-HKO mice fed a methionine-choline deficient (MCD) diet for 4 weeks exhibited multiple protective effects compared to Acss2-Flox controls: reduced hepatomegaly (Supplementary Fig. 7A, B), lower serum ALT/AST and triglycerides (Supplementary Fig. 7C–F), and decreased hepatic steatosis, fibrosis (Supplementary Fig. 7G–I), and inflammation (Supplementary Fig. 7J, K). Collectively, these data demonstrate that hepatocyte-specific ACSS2 knockout alleviates MCD-induced liver inflammation and injury. To further validate the role of ACSS2 in inflammation, we employed a CCL4-induced liver injury model. In this model, CCL4-derived free radicals induce lipid peroxidation of hepatocyte membranes, leading to inflammatory responses45. Interestingly, hepatocyte-specific knockout of ACSS2 significantly alleviated CCl4-induced inflammatory cell infiltration and liver injury (Supplementary Fig. 8), paralleling the protective effects observed in the HFFD-induced MASH model. Therefore, these findings suggest that ACSS2 regulates MASH mainly through inflammatory modulation rather than lipid metabolism. ACSS2overexpression exacerbated HFFD-induced MASH progression Given the upregulation of hepatic ACSS2 in HFFD-fed mice and the protective effect of ACSS2 knockout, we hypothesized that restoring ACSS2 would exacerbate MASH progression. To test this hypothesis, we generated an adeno-associated virus 8 (AAV8) vector encoding the Acss2 gene to restore hepatic ACSS2 expression (Supplementary Fig. 9A). The transfection efficiency of AAV8-Acss2 was determined by western blot, real-time PCR and immunohistochemistry (IHC) (Supplementary Fig. 9B, C, 10A), while hepatocyte‑specific AAV8-Acss2 adenovirus significantly boosted ACSS2 expression in hepatocytes, it did not change ACSS2 expression in HSCs (Supplementary Fig. 10B). We then evaluated the impact of ACSS2 overexpression on hepatic steatosis, fibrosis, and inflammation during HFFD-induced MASH progression. As shown in supplementary Fig. 9D–I, ACSS2 gain-of-function significantly increased the serum ALT and AST levels and increased triglyceride (TG) accumulation in both the liver and serum. Furthermore, ACSS2 overexpression promoted substantial weight gain and impaired systemic glucose homeostasis, as indicated by elevated random blood glucose, aggravated glucose intolerance, and exacerbated insulin resistance (Supplementary Fig. 9J–M). Concurrently, it worsened HFFD-induced hepatic steatosis, fibrosis, inflammatory infiltration, and lipid deposition (Supplementary Fig. 9N–Q, 10C–I). Notably, approximately 15% of AAV8-Acss2-treated Acss2-Flox mice developed cirrhosis, an end-stage manifestation of MASH, indicating accelerated disease progression upon ACSS2 reconstitution (Supplementary Fig. 10J). Consistently, AAV8-mediated ACSS2 reconstitution in Acss2-HKO mice abrogated the protective effect of genetic deletion and reinstated key disease features, including steatosis, fibrosis, inflammatory infiltration, and the upregulation of inflammatory and lipogenic genes (Supplementary Figs. 9, 10). Hepatocyte-specificACSS2deficiency suppressed hepatocellular senescence To elucidate the mechanism by which hepatocyte ACSS2 drives MASH progression, we performed RNA sequencing (RNA-seq) to examine the transcriptome alterations in HFFD-fed Acss2-HKO versus Acss2-Flox mice (Fig. 3A). While KEGG enrichment analysis revealed alterations in multiple pathways (Supplementary Fig. 11A), gene set enrichment analysis (GSEA) and heatmap analyses revealed significant downregulation of cellular senescence and coordinated suppression of senescence-associated genes in Acss2-HKO livers (Fig. 3B, C). Cellular senescence is a state triggered by various stimuli and is characterized primarily by stable cell cycle arrest and a complex senescence-associated secretory phenotype (SASP)49. Consistent with its age-associated upregulation, genetic ablation of ACSS2 led to the downregulation of key pathways involved in cellular senescence, including those related to chemokines, the cell cycle, and NF-κB signaling (Supplementary Fig. 11B–H). This transcriptional signature was confirmed at the protein and histological levels. Specifically, IHC staining, real-time PCR, and immunofluorescence revealed significant reductions in the expression of senescence markers (P16, P21, P53, and γH2AX) and β-gal in knockout livers (Fig. 3D–K). In vitro, ACSS2 knockdown significantly attenuated the upregulation of senescence-associated genes and positive β-gal staining induced by PA (Fig. 3L, M, Supplementary Fig. 11I–J). Taken together, these results suggest that ACSS2 deficiency alleviates MASH, at least partially, by attenuating hepatocellular senescence.Fig. 3Hepatocyte-specific deficiency of ACSS2 attenuates hepatocellular senescence.A The experimental design and workflow for performing RNA-seq analysis on microdissected liver tissues. Gene Set Enrichment Analysis (GSEA) plot (B) and heatmap (C) showing the expression of hepatocellular senescence pathway-related genes in livers of HFFD-fed Acss2-Flox and Acss2-HKO mice. Representative images of immunohistochemical staining and quantitative analysis of P16 (D), P21 (E), and γH2AX (F) in liver sections from the indicated groups of mice, (n = 6). Scale bar: 50 μm. Real-time PCR analysis of the relative mRNA levels of P16 (G), P21 (H), and P53 (I) in the livers of Acss2-Flox and Acss2-HKO mice, (n = 6). J Representative images of immunofluorescence and quantitative analysis of P21 (green) in the liver tissues of the indicated groups of mice. nuclei were counterstained with DAPI, (n = 6), Scale bar: 20 μm. K Representative images of β-gal staining and quantitative analysis of hepatocellular senescence in the liver sections of the indicated groups of mice (n = 6). Scale bar: 50 μm. L Western blot analysis of relative ACSS2, P53, P21, and γH2AX protein levels in control (Si-Control) and ACSS2-knockdown (Si-Acss2) MIHA cells treated with PA, (n = 3 biological replicates per group). M Representative images of β-gal staining and quantitative analysis of hepatocellular senescence in control (Si-Control) and ACSS2-knockdown (Si-Acss2) primary hepatocytes treated with PA, (n = 3 biological replicates per group), Scale bar: 50 μm. Data are presented as the mean ± SD. The p values were calculated by two-sided test and adjusted for p values by the Benjamini-Hochberg method to control for false positives due to multiple comparisons (B). The p values were calculated by two-tailed Student’s t test (D–F, J, L, M), two-way ANOVA (G–I, K). Source data are provided as a Source Data file. Hepatocyte-specific deficiency of ACSS2 attenuates hepatocellular senescence. A The experimental design and workflow for performing RNA-seq analysis on microdissected liver tissues. Gene Set Enrichment Analysis (GSEA) plot (B) and heatmap (C) showing the expression of hepatocellular senescence pathway-related genes in livers of HFFD-fed Acss2-Flox and Acss2-HKO mice. Representative images of immunohistochemical staining and quantitative analysis of P16 (D), P21 (E), and γH2AX (F) in liver sections from the indicated groups of mice, (n = 6). Scale bar: 50 μm. Real-time PCR analysis of the relative mRNA levels of P16 (G), P21 (H), and P53 (I) in the livers of Acss2-Flox and Acss2-HKO mice, (n = 6). J Representative images of immunofluorescence and quantitative analysis of P21 (green) in the liver tissues of the indicated groups of mice. nuclei were counterstained with DAPI, (n = 6), Scale bar: 20 μm. K Representative images of β-gal staining and quantitative analysis of hepatocellular senescence in the liver sections of the indicated groups of mice (n = 6). Scale bar: 50 μm. L Western blot analysis of relative ACSS2, P53, P21, and γH2AX protein levels in control (Si-Control) and ACSS2-knockdown (Si-Acss2) MIHA cells treated with PA, (n = 3 biological replicates per group). M Representative images of β-gal staining and quantitative analysis of hepatocellular senescence in control (Si-Control) and ACSS2-knockdown (Si-Acss2) primary hepatocytes treated with PA, (n = 3 biological replicates per group), Scale bar: 50 μm. Data are presented as the mean ± SD. The p values were calculated by two-sided test and adjusted for p values by the Benjamini-Hochberg method to control for false positives due to multiple comparisons (B). The p values were calculated by two-tailed Student’s t test (D–F, J, L, M), two-way ANOVA (G–I, K). Source data are provided as a Source Data file. ACSS2promotes hepatocellular senescence by controllingAIF1expression To further investigate the underlying mechanisms by which ACSS2 regulates hepatocellular senescence, we performed Gene Ontology (GO) enrichment analysis of key biological processes (BP). The results revealed that ACSS2-mediated MASH was associated primarily with biological processes, including phagocytosis, the inflammatory response, fatty acid metabolic processes, cell cycle regulation, and the regulation of immune effector processes (Fig. 4A). To identify key mediators, we intersected the top two significantly enriched biological processes with the 100 most downregulated DEGs, which revealed four overlapping genes: AIF1, CLEC7A, PLCG2, and CD36. Among these genes, only the expression of AIF1, a calcium-binding protein involved in phagocytosis and inflammation50, was positively correlated with ACSS2 levels (Fig. 4B, C). This transcriptomic alteration reflected a broader shift toward anti-inflammatory and lipid metabolic pathways, along with the suppression of proliferative, inflammatory, and lipogenic genes (Fig. 4D). Additionally, IHC staining of human liver tissue sections revealed that AIF1 expression was significantly upregulated in MASH patients, which is consistent with the ACSS2 expression pattern. In contrast, AIF1 levels in patients with simple steatosis (MASLD) did not differ from those in healthy controls (Fig. 4E). In vivo studies revealed a significant reduction in AIF1 expression in Acss2-HKO mice, as demonstrated by western blot, real-time PCR, and IHC analyses of liver tissues (Fig. 4F–H). This decrease was also observed in primary hepatocytes isolated from HFFD-fed mice, with AIF1 levels in Acss2-HKO cells being lower than those in Acss2-Flox control cells but still higher than those in NCD-fed mice (Fig. 4I). We extended these observations in vitro via gain- and loss-of-function experiments: in PA-treated MIHA cells and primary hepatocytes, ACSS2 knockdown reduced AIF1 protein levels, whereas ACSS2 overexpression increased AIF1 protein levels (Fig. 4J, Supplementary Fig. 12A, B). Furthermore, AIF1 knockdown recapitulated the effects of ACSS2 knockdown by reducing the levels of inflammatory cytokines (IL-6, TNFα, Fig. 4K) and attenuated hepatocellular senescence (Fig. 4L, M), whereas its overexpression had the opposite effect (Supplementary Fig. 12C). Similarly, in cells overexpressing ACSS2, AIF1 knockdown suppressed the elevated inflammatory cytokines (TNFα, IL-1, etc.) and senescence markers (p53, p21, etc.) (Supplementary Fig. 12D–J). Collectively, these lines of evidence demonstrate that AIF1 is a key downstream effector of ACSS2.Fig. 4ACSS2 modulates hepatocellular senescence by regulating AIF1 expression.A Gene Ontology (GO) enrichment analysis of biological processes (BP) in the livers of HFFD-fed Acss2-Flox and Acss2-HKO mice. B Venn diagram showing the intersection of the two most significantly enriched BP pathways with the top 100 downregulated differentially expressed genes (DEGs). C Correlation analysis between the Venn diagram results (Aif1, Clec7a, Plcg2, and Cd36) and Acss2. D Volcano plot displaying upregulated (red) or downregulated (blue) DEGs between HFFD-fed Acss2-Flox and Acss2-HKO mice. E Representative images of the immunohistochemistry staining and quantitative analysis of AIF1 in human liver sections from the healthy controls (n = 10), MASLD patients (n = 10) and MASH patients (n = 17). Scale bar: 50 µm. F Western blot analysis of relative AIF1 protein levels in the liver tissues from indicated groups, (n = 6 biological replicates per group). G Representative images of immunohistochemical staining and quantitative analysis of AIF1 in liver sections of the indicated groups, (NCD, n = 6 biological replicates per group, HFFD, n = 8 biological replicates per group). Scale bar: 50 µm. H Real-time PCR analysis of relative Aif1 mRNA levels in the livers of HFFD-fed Acss2-Flox and Acss2-HKO mice, (n = 8). I Western blot analysis of relative AIF1 protein levels in primary hepatocytes of indicated groups, (n = 3 biological replicates per group). J Western blot analysis of relative AIF1 and ACSS2 protein levels in control (Si-Control) and ACSS2-knockdown (Si-Acss2) MIHA cells treated with PA, (n = 3 biological replicates per group). K Real-time PCR analysis of relative mRNA levels of inflammation- and cellular senescence-related genes in control (Si-Control) and AIF1-knockdown (Si-Aif1) MIHA cells treated with PA, (n = 3 biological replicates per group). L Western blot analysis of protein levels of AIF1, P53, P21, and γH2AX in control (Si-Control) and AIF1-knockdown (Si-Aif1) MIHA cells treated with PA, (n = 3 biological replicates per group). M Representative images of β-galactosidase staining and quantitative analysis of hepatocellular senescence in control (Si-Control) and AIF1-knockdown (Si-Aif1) MIHA cells treated with PA, (n = 3 biological replicates per group). Scale bar: 50 µm. Data are presented as the mean ± SD. The p values were calculated by two-sided test and adjusted for p values by the Benjamini-Hochberg method to control for false positives due to multiple comparisons (D). The p values were calculated by two-tailed Student’s t test (F, H, J–M), one-way ANOVA (E, I), two-way ANOVA (G). Source data are provided as a Source Data file. ACSS2 modulates hepatocellular senescence by regulating AIF1 expression. A Gene Ontology (GO) enrichment analysis of biological processes (BP) in the livers of HFFD-fed Acss2-Flox and Acss2-HKO mice. B Venn diagram showing the intersection of the two most significantly enriched BP pathways with the top 100 downregulated differentially expressed genes (DEGs). C Correlation analysis between the Venn diagram results (Aif1, Clec7a, Plcg2, and Cd36) and Acss2. D Volcano plot displaying upregulated (red) or downregulated (blue) DEGs between HFFD-fed Acss2-Flox and Acss2-HKO mice. E Representative images of the immunohistochemistry staining and quantitative analysis of AIF1 in human liver sections from the healthy controls (n = 10), MASLD patients (n = 10) and MASH patients (n = 17). Scale bar: 50 µm. F Western blot analysis of relative AIF1 protein levels in the liver tissues from indicated groups, (n = 6 biological replicates per group). G Representative images of immunohistochemical staining and quantitative analysis of AIF1 in liver sections of the indicated groups, (NCD, n = 6 biological replicates per group, HFFD, n = 8 biological replicates per group). Scale bar: 50 µm. H Real-time PCR analysis of relative Aif1 mRNA levels in the livers of HFFD-fed Acss2-Flox and Acss2-HKO mice, (n = 8). I Western blot analysis of relative AIF1 protein levels in primary hepatocytes of indicated groups, (n = 3 biological replicates per group). J Western blot analysis of relative AIF1 and ACSS2 protein levels in control (Si-Control) and ACSS2-knockdown (Si-Acss2) MIHA cells treated with PA, (n = 3 biological replicates per group). K Real-time PCR analysis of relative mRNA levels of inflammation- and cellular senescence-related genes in control (Si-Control) and AIF1-knockdown (Si-Aif1) MIHA cells treated with PA, (n = 3 biological replicates per group). L Western blot analysis of protein levels of AIF1, P53, P21, and γH2AX in control (Si-Control) and AIF1-knockdown (Si-Aif1) MIHA cells treated with PA, (n = 3 biological replicates per group). M Representative images of β-galactosidase staining and quantitative analysis of hepatocellular senescence in control (Si-Control) and AIF1-knockdown (Si-Aif1) MIHA cells treated with PA, (n = 3 biological replicates per group). Scale bar: 50 µm. Data are presented as the mean ± SD. The p values were calculated by two-sided test and adjusted for p values by the Benjamini-Hochberg method to control for false positives due to multiple comparisons (D). The p values were calculated by two-tailed Student’s t test (F, H, J–M), one-way ANOVA (E, I), two-way ANOVA (G). Source data are provided as a Source Data file. ACSS2overexpression promoted HFFD-induced hepatocellular senescence To define the contribution of ACSS2 to HFFD-induced hepatocellular senescence, we examined key markers of hepatocellular senescence. ACSS2 gain-of-function was sufficient to upregulate AIF1 and, in turn, induce hepatocellular senescence, as evidenced by elevated AIF1 expression at both the protein and mRNA levels and elevated expression of senescence markers (P16, P21, γH2AX, P53) via IHC staining, real-time PCR, western blotting and immunofluorescence (Supplementary Fig. 13A–H). Furthermore, AAV8-Acss2 treatment increased β-gal-positive staining in both genotypes, although the increase was less pronounced in Acss2-HKO mice (Supplementary Fig. 13I). Together, these data substantiate the role of ACSS2 in promoting MASH through the induction of hepatocellular senescence. AIF1overexpression in hepatocytes reversed the suppression of inflammation, hepatocellular senescence, and liver injury inAcss2-HKO mice Given that ACSS2 exacerbates hepatocellular senescence via AIF1, we hypothesized that AIF1 is the key downstream effector in ACSS2-driven MASH. To test this hypothesis, we administered an AAV8 vector carrying the Aif1 gene to both Acss2-Flox and Acss2-HKO mice, and the transfection efficiency was confirmed by real-time PCR, western blotting, and IHC staining (Fig. 5A, Supplementary Fig. 14A–C). Compared with those in AAV8-Control-treated Acss2-HKO mice, AIF1 overexpression abrogated the protective phenotype in Acss2-HKO mice, as shown by increased body weight, liver-to-body weight ratio (Fig. 5B, C, Supplementary Fig. 14F), and elevated markers of liver injury (serum ALT, AST) and triglycerides (hepatic and serum) (Fig. 5D–G). It also impaired glucose homeostasis and reversed protection against steatosis, fibrosis, and inflammation (Fig. 5H–N, Supplementary Fig. 14D). Moreover, AIF1 overexpression promoted lipogenesis (Supplementary Fig. 14E–G) and enhanced hepatocellular senescence (Supplementary Fig. 15), recapitulating the pro-MASH effects of ACSS2 overexpression. Collectively, our data demonstrate that ACSS2 deficiency alleviates MASH progression through the suppression of AIF1-mediated hepatocellular senescence.Fig. 5AIF1 overexpression counteracts the protective effects of ACSS2 deficiency on HFFD-induced liver injury, inflammation, and senescence.A Experimental timeline: 8–10 weeks male Acss2-Flox and Acss2-HKO mice were injected via the tail vein with AAV8-Control or AAV8-Aif1, followed by HFFD feeding for 16 weeks. B Body weight changes in AAV8-Control and AAV8-Aif1 mice, (n = 6). C Gross liver morphology and liver-to-body weight ratios of AAV8-Control and AAV8-Aif1 mice, (n = 6). Scale bar: 1 cm. Serum levels of AST (D), ALT (E), and hepatic TG (F) and serum TG (G) in the indicated groups, (n = 6 biological replicates per group). H Blood glucose level changes in the indicated groups, (n = 6 biological replicates per group). Glucose tolerance test (GTT, I) and insulin tolerance test (ITT, J) in the indicated groups, (n = 6 biological replicates per group). Representative images of H&E staining with quantitative analysis of the relative vacuolar area (K), Sirius Red staining with quantitation of the relative stained area (L) in liver sections from the indicated groups, (n = 6 biological replicates per group). Scale bar: 50 μm. M Representative images of immunofluorescence and quantitative analysis of F4/80 (red) in the liver sections from indicated groups. nuclei were counterstained with DAPI, (n = 6 biological replicates per group). Scale bar: 20 μm. N Real-time PCR analysis of the relative mRNA levels of pro-inflammatory cytokines and chemokines in the livers of AAV8-Control-HKO and AAV8-Aif1-HKO mice, (n = 6). Data are presented as mean ± SD. Statistical significance was determined by two-tailed Student’s t test (N) and two-way ANOVA (C–M). Source data are provided as a Source Data file. AIF1 overexpression counteracts the protective effects of ACSS2 deficiency on HFFD-induced liver injury, inflammation, and senescence. A Experimental timeline: 8–10 weeks male Acss2-Flox and Acss2-HKO mice were injected via the tail vein with AAV8-Control or AAV8-Aif1, followed by HFFD feeding for 16 weeks. B Body weight changes in AAV8-Control and AAV8-Aif1 mice, (n = 6). C Gross liver morphology and liver-to-body weight ratios of AAV8-Control and AAV8-Aif1 mice, (n = 6). Scale bar: 1 cm. Serum levels of AST (D), ALT (E), and hepatic TG (F) and serum TG (G) in the indicated groups, (n = 6 biological replicates per group). H Blood glucose level changes in the indicated groups, (n = 6 biological replicates per group). Glucose tolerance test (GTT, I) and insulin tolerance test (ITT, J) in the indicated groups, (n = 6 biological replicates per group). Representative images of H&E staining with quantitative analysis of the relative vacuolar area (K), Sirius Red staining with quantitation of the relative stained area (L) in liver sections from the indicated groups, (n = 6 biological replicates per group). Scale bar: 50 μm. M Representative images of immunofluorescence and quantitative analysis of F4/80 (red) in the liver sections from indicated groups. nuclei were counterstained with DAPI, (n = 6 biological replicates per group). Scale bar: 20 μm. N Real-time PCR analysis of the relative mRNA levels of pro-inflammatory cytokines and chemokines in the livers of AAV8-Control-HKO and AAV8-Aif1-HKO mice, (n = 6). Data are presented as mean ± SD. Statistical significance was determined by two-tailed Student’s t test (N) and two-way ANOVA (C–M). Source data are provided as a Source Data file. ACSS2governsAIF1expression via histone crotonylation ACSS2 has a compartmentalized function: in addition to its role in cytoplasmic lipid synthesis, it also regulates nuclear histone acylation51. In HFFD-fed Acss2-HKO mice, both hepatic tissue and isolated hepatocytes presented reduced global and histone crotonylation without significant changes in other types of acylation (Supplementary Fig. 16A, 17A–I). RNA-seq and real-time PCR results indicated that this reduction was associated with the downregulation of crotonylation-related enzymes (Supplementary Fig. 16B, Supplementary Fig. 18). Importantly, a comparative analysis of four GEO datasets indicated that, among these enzymes, ACSS2 was consistently upregulated in the livers of MASH patients (Supplementary Fig. 16C). The essential role of ACSS2 was validated in vitro. Its knockdown attenuated PA- and crotonate-induced crotonylation and specific histone marks (H3K18cr, H3K27cr, H4K12cr), whereas its overexpression increased global crotonylation (Supplementary Figs. 16D–H, 17J–K). Notably, in the PA-treated MASH cell model, ACSS2 knockdown specifically reduced histone crotonylation (H3K18cr, H4K12cr) without significantly effecting histone acetylation (H3K9ac, H4K12ac, Supplementary Fig. 19A). We next examined whether ACSS2 regulates AIF1 expression through a histone crotonylation-mediated epigenetic mechanism. Crotonate treatment increased AIF1 protein levels in a concentration-dependent manner, and this effect was abolished by ACSS2 knockdown (Supplementary Fig. 16I–J). Moreover, knockdown of either ACSS2 or AIF1 attenuated crotonate-induced inflammatory gene expression and hepatocellular senescence (Supplementary Fig. 16K, 19B, C), and importantly, AIF1 overexpression abolished the anti-senescence effect of ACSS2 knockdown (Supplementary Fig. 19D). In summary, our findings demonstrate that ACSS2 promotes inflammation and hepatocellular senescence by regulating AIF1 expression via histone crotonylation. ACSS2regulates the transcription ofAIF1 via KAT5-mediated histone crotonylation ACSS2 participates in multiple acylations. For example, it regulates histone acetylation by recruiting CBP and activating its nuclear histone acetyltransferase (HAT) activity12,52. ACSS2 also acts as a lactoyl-CoA synthase and interacts with KAT2A to catalyze histone lactylation16. To elucidate the underlying mechanism by which ACSS2 regulates histone crotonylation, we performed liquid chromatography‒mass spectrometry (LC‒MS) in MIHA cells expressing empty FLAG or ACSS2-FLAG plasmids (Fig. 6A). We identified a physical interaction between ACSS2 and KAT5, which we validated with coimmunoprecipitation (co-IP) assays in 293T and MIHA cells for both endogenous and exogenous proteins (Fig. 6B–D). This interaction was further confirmed by proximity ligation assays (PLA), immunofluorescence assays and molecular docking in MIHA cells (Fig. 6E–G). Next, our results showed that nuclear ACSS2 was elevated in the MASH model (Fig. 1H–I). Lu et al. showed that EGFR-ERK phosphorylates ACSS2 at Ser267, promoting its nuclear translocation16. Consistent with this, p‑ERK was higher in HFFD‑fed mouse livers and PA‑treated hepatocytes (Supplementary Fig. 3M, 20A), accompanied by greater nuclear ACSS2. Moreover, ERK inhibition (U0126) blocked this nuclear accumulation (Supplementary Fig. 20B, C). To test functional role of S267 phosphorylation, we generated S267A (phospho‑deficient) and S267D (phospho‑mimetic) mutants (Supplementary Fig. 20D). S267A did not bind KAT5 (Supplementary Fig. 20E) nor enhance histone crotonylation (Supplementary Fig. 20F). In contrast, S267D exacerbated the senescence phenotype, whereas S267A did not (Supplementary Fig. 20G). Accordingly, U0126 also lowered histone crotonylation (Fig. 6H). Moreover, we found that nuclear ACSS2 retains its ability to convert acetate to acetyl‑CoA, although the cytoplasmic acetyl‑CoA concentration was approximately eight‑fold higher than that in the nucleus (Supplementary Fig. 20H–I). However, our Western blot analysis revealed that exogenous acetate supplementation did not significantly alter the protein expression level of AIF1. By contrast, exogenous crotonate supplementation markedly increased AIF1 expression (Supplementary Fig. 20J, K). collectively, these findings confirm that ACSS2 regulates AIF1 transcription through KAT5-mediated histone crotonylation rather than histone acetylation.Fig. 6ACSS2 governs AIF1 transcription by modulating KAT5-mediated histone crotonylation.A LC-MS analysis of MIHA cells transfected with Flag-ACSS2 overexpression plasmid or control vector. B Western blot analysis of co-immunoprecipitation (Co-IP) assays in 293 T cells. The left panel shows the interaction between endogenous ACSS2 and endogenous KAT5, while the right panel shows the interaction between overexpressed ACSS2-Flag and endogenous KAT5. C Western blot analysis of Co-IP between endogenous ACSS2 and endogenous KAT5 in MIHA cells. D Western blot analysis of Co-IP between KAT5-Myc and ACSS2-Flag in MIHA cells co-transfected with respective plasmids. Each experiment was independently repeated three times with similar results. E In situ proximity ligation assay (PLA) of ACSS2 and KAT5 interaction in MIHA cells. Scale bar: 10 μm. F Representative images of immunofluorescence and quantitative analysis of co-localization between KAT5 (red) and ACSS2 (green) in the liver sections from indicated groups. Nuclei were counterstained with DAPI. Scale bar: 10 μm (left panel) and 1 μm (right panel). G Representative model of molecular docking between KAT5 and ACSS2. H Western blot analysis of histone crotonylation in MIHA cells treated with Control, PA, or PA + U0126, (n = 3). I Western blot analysis of global protein crotonylation in MIHA cells transfected with Si-Kat5 or Si-Control and subsequent treated with crotonate, (n = 3 biological replicates per group). J Schematic diagram of the ChIP-qPCR assay procedure. K ChIP-qPCR analysis of Kat5, H3k18cr and H3k27cr enrichment at the Aif1 promoter in MIHA cells transfected with Si-Acss2 or Si-Control, (n = 3 biological replicates per group). IgG was used as a negative control. Data are presented as mean ± SD. Statistical significance was determined by two-tailed Student’s t test (K) and one-way ANOVA (H). Source data are provided as a Source Data file. ACSS2 governs AIF1 transcription by modulating KAT5-mediated histone crotonylation. A LC-MS analysis of MIHA cells transfected with Flag-ACSS2 overexpression plasmid or control vector. B Western blot analysis of co-immunoprecipitation (Co-IP) assays in 293 T cells. The left panel shows the interaction between endogenous ACSS2 and endogenous KAT5, while the right panel shows the interaction between overexpressed ACSS2-Flag and endogenous KAT5. C Western blot analysis of Co-IP between endogenous ACSS2 and endogenous KAT5 in MIHA cells. D Western blot analysis of Co-IP between KAT5-Myc and ACSS2-Flag in MIHA cells co-transfected with respective plasmids. Each experiment was independently repeated three times with similar results. E In situ proximity ligation assay (PLA) of ACSS2 and KAT5 interaction in MIHA cells. Scale bar: 10 μm. F Representative images of immunofluorescence and quantitative analysis of co-localization between KAT5 (red) and ACSS2 (green) in the liver sections from indicated groups. Nuclei were counterstained with DAPI. Scale bar: 10 μm (left panel) and 1 μm (right panel). G Representative model of molecular docking between KAT5 and ACSS2. H Western blot analysis of histone crotonylation in MIHA cells treated with Control, PA, or PA + U0126, (n = 3). I Western blot analysis of global protein crotonylation in MIHA cells transfected with Si-Kat5 or Si-Control and subsequent treated with crotonate, (n = 3 biological replicates per group). J Schematic diagram of the ChIP-qPCR assay procedure. K ChIP-qPCR analysis of Kat5, H3k18cr and H3k27cr enrichment at the Aif1 promoter in MIHA cells transfected with Si-Acss2 or Si-Control, (n = 3 biological replicates per group). IgG was used as a negative control. Data are presented as mean ± SD. Statistical significance was determined by two-tailed Student’s t test (K) and one-way ANOVA (H). Source data are provided as a Source Data file. To explore the potential mechanism by which the ACSS2-KAT5 complex regulates AIF1 expression, we performed ChIP‒qPCR using specific ACSS2/KAT5 antibodies to determine the binding of the ACSS2-KAT5 complex to the promoter region of AIF1 (Fig. 6J). KAT5 bound to the region between −1199 and −1100 of the AIF1 promoter, and this binding was dramatically reduced upon ACSS2 knockdown in MIHA cells (Fig. 6K). Consistent with the role of KAT5 in mediating crotonylation and gene expression, ChIP‒qPCR analysis revealed that ACSS2 knockdown significantly diminished the enrichment of H3K18cr and H3K27cr on the AIF1 promoter (Fig. 6K). Furthermore, electrophoresis mobility shift assay (EMSA) demonstrated that the ACSS2/KAT5 complex is present at the promoter of AIF1. (Supplementary Fig. 20L). Together, these results suggest that ACSS2 regulates AIF1 expression via KAT5-mediated crotonylation of H3K18 and H3K27. Pharmacological inhibition of ACSS2 significantly suppresses HFFD-induced MASH Having established that hepatocyte-specific ACSS2 deficiency mitigates MASH progression, we evaluated the therapeutic potential of an ACSS2 inhibitor (ACSS2i) in an HFFD-induced MASH model (Fig. 7A). ACSS2i treatment significantly reduced body weight, liver-to-body weight ratios (Fig. 7B–D, Supplementary Fig. 21A), and levels of liver injury markers (AST, ALT) and triglycerides in both the serum and liver (Fig. 7E–H). Furthermore, ACSS2i administration improved systemic insulin resistance and glucose homeostasis (Fig. 7I–K) and attenuated hepatic steatosis, inflammation, and fibrosis. This effect was accompanied by suppressed lipogenesis (Fig. 7L, Supplementary Fig. 21B–E), reduced lipid droplet accumulation (Supplementary Fig. 21F), downregulated expression of inflammatory and fibrogenic genes (Fig. 7M, N, Supplementary Fig. 21G–I), and decreased expression of AIF1 (Supplementary Fig. 22A–C), which collectively contributed to the alleviation of hepatocellular senescence (Supplementary Fig. 22D–I). In conclusion, our study demonstrated that hepatic ACSS2 deficiency alleviates HFFD-induced MASH by downregulating AIF1-mediated hepatocellular senescence. This work identifies the ACSS2-AIF1 axis as a key pathogenic driver of MASH and proposes a promising therapeutic strategy.Fig. 7Pharmacological inhibition of ACSS2 significantly ameliorates HFFD-induced MASH development.A Experimental timeline: 8–10 weeks male wild-type (WT) mice were fed an HFFD for 8 weeks and then randomized to receive intraperitoneal injections of either an ACSS2 inhibitor (ACSS2i; 25 mg/kg) or vehicle control three times per week for another 8 weeks on the HFFD. B Body weight changes in the indicated groups of mice, (n = 6). Gross liver morphology (C) and liver-to-body weight ratios (D) of mice from the indicated groups, (NCD, n = 5, HFFD, n = 8). Serum levels of ALT (E), AST (F), and hepatic TG (G) and serum TG (H) in the indicated groups of mice, (n = 6). Glucose tolerance test (GTT, I) and insulin tolerance test (ITT, J) in the indicated groups, (n = 6 biological replicates per group). K Blood glucose level changes in the indicated groups, (n = 6 biological replicates per group). Representative images of H&E staining and quantitation of the relative vacuolar area (L), Sirius Red staining and quantitation of the relative stained area (M) in liver sections from the indicated groups, (n = 6 biological replicates per group). Scale bar: 50 μm. N Representative images of immunofluorescence and quantitative analysis of F4/80 (red) in the livers of the indicated groups. Nuclei were counterstained with DAPI, (n = 6 biological replicates per group). Scale bar: 20 μm. Data are presented as mean ± SD. Statistical significance was determined by two-tailed Student’s t test (N) and one-way ANOVA (B, D–M). Source data are provided as a Source Data file. Pharmacological inhibition of ACSS2 significantly ameliorates HFFD-induced MASH development. A Experimental timeline: 8–10 weeks male wild-type (WT) mice were fed an HFFD for 8 weeks and then randomized to receive intraperitoneal injections of either an ACSS2 inhibitor (ACSS2i; 25 mg/kg) or vehicle control three times per week for another 8 weeks on the HFFD. B Body weight changes in the indicated groups of mice, (n = 6). Gross liver morphology (C) and liver-to-body weight ratios (D) of mice from the indicated groups, (NCD, n = 5, HFFD, n = 8). Serum levels of ALT (E), AST (F), and hepatic TG (G) and serum TG (H) in the indicated groups of mice, (n = 6). Glucose tolerance test (GTT, I) and insulin tolerance test (ITT, J) in the indicated groups, (n = 6 biological replicates per group). K Blood glucose level changes in the indicated groups, (n = 6 biological replicates per group). Representative images of H&E staining and quantitation of the relative vacuolar area (L), Sirius Red staining and quantitation of the relative stained area (M) in liver sections from the indicated groups, (n = 6 biological replicates per group). Scale bar: 50 μm. N Representative images of immunofluorescence and quantitative analysis of F4/80 (red) in the livers of the indicated groups. Nuclei were counterstained with DAPI, (n = 6 biological replicates per group). Scale bar: 20 μm. Data are presented as mean ± SD. Statistical significance was determined by two-tailed Student’s t test (N) and one-way ANOVA (B, D–M). Source data are provided as a Source Data file. Pharmacological clearance of senescent hepatocytes by Navitoclax reverses ACSS2 overexpression‑induced MASH progression To determine whether hepatocyte senescence drives inflammation and MASH progression, we selectively eliminated senescent hepatocytes using Navitoclax. Acss2-Flox mice received AAV8-Acss2 adenovirus to overexpress ACSS2 in hepatocytes, followed by Navitoclax treatment (Fig. 8A). The results showed that ACSS2 overexpression exacerbated MCD‑induced liver injury, steatosis, inflammation, and hepatocyte senescence, whereas Navitoclax reversed all these phenotypes (Fig. 8B–J).Fig. 8Pharmacological clearance of senescent hepatocytes by Navitoclax reverses ACSS2 overexpression-induced MASH progression.A Experimental timeline: Acss2-Flox mice were injected with AAV8-control or AAV8-Acss2 to achieve hepatocyte-specific overexpression of ACSS2, followed by feeding with a methionine-choline deficient (MCD) diet for 4 weeks. Navitoclax (50 mg/kg) was administered orally every other day during the final 2 weeks of MCD feeding. B Representative gross morphology of livers from the indicated groups. Scale bar: 1 cm. C, D Serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels, (n = 6 biological replicates per group). E, F Hepatic triglyceride (TG) content and serum TG levels, (n = 6 biological replicates per group). G Representative images of H&E staining, Sirius Red staining, and immunohistochemical staining for ACSS2, AIF1, and γH2AX in liver sections from the indicated groups. Scale bar: 50 μm, (n = 6). H Western blot analysis of ACSS2, AIF1, and α-SMA protein levels in liver tissues from the indicated groups, (AAV8-Control, n = 4 biological replicates per group, AAV8-Acss2, n = 5 biological replicates per group). I Relative mRNA levels of pro-inflammatory cytokines in liver tissues determined by real-time PCR, (n = 6 biological replicates per group). J Relative mRNA levels of senescence-associated genes (p16, p21, p53) in liver tissues determined by real-time PCR (n = 6 biological replicates per group). Data are presented as mean ± SD. Statistical significance was determined by two-tailed Student’s t test (I–J) and one-way ANOVA (C–H). Source data are provided as a Source Data file. Pharmacological clearance of senescent hepatocytes by Navitoclax reverses ACSS2 overexpression-induced MASH progression. A Experimental timeline: Acss2-Flox mice were injected with AAV8-control or AAV8-Acss2 to achieve hepatocyte-specific overexpression of ACSS2, followed by feeding with a methionine-choline deficient (MCD) diet for 4 weeks. Navitoclax (50 mg/kg) was administered orally every other day during the final 2 weeks of MCD feeding. B Representative gross morphology of livers from the indicated groups. Scale bar: 1 cm. C, D Serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels, (n = 6 biological replicates per group). E, F Hepatic triglyceride (TG) content and serum TG levels, (n = 6 biological replicates per group). G Representative images of H&E staining, Sirius Red staining, and immunohistochemical staining for ACSS2, AIF1, and γH2AX in liver sections from the indicated groups. Scale bar: 50 μm, (n = 6). H Western blot analysis of ACSS2, AIF1, and α-SMA protein levels in liver tissues from the indicated groups, (AAV8-Control, n = 4 biological replicates per group, AAV8-Acss2, n = 5 biological replicates per group). I Relative mRNA levels of pro-inflammatory cytokines in liver tissues determined by real-time PCR, (n = 6 biological replicates per group). J Relative mRNA levels of senescence-associated genes (p16, p21, p53) in liver tissues determined by real-time PCR (n = 6 biological replicates per group). Data are presented as mean ± SD. Statistical significance was determined by two-tailed Student’s t test (I–J) and one-way ANOVA (C–H). Source data are provided as a Source Data file. Together, these findings demonstrate that ACSS2 accelerates MASH progression by inducing hepatocyte senescence, and that Navitoclax clears senescent cells to recapitulate the benefit of ACSS2 loss. Thus, hepatocyte senescence is a functional consequence of ACSS2 activity, and senolytic strategies may offer a novel approach for treating MASH. Discussion The global increase in MASH incidence, particularly its earlier onset in younger individuals5, underscores an urgent public health challenge. Despite extensive research, the intricate pathogenesis of MASH remains a major obstacle, necessitating innovative treatment approaches. Our study revealed that ACSS2, which is upregulated in both HFFD-induced MASH mice and human MASH livers, acts as a positive regulator of inflammation in MASH, extending beyond its canonical role in lipogenesis53. Owing to its predominant expression in hepatocytes12,54, we focused on its liver-specific functions. We demonstrated that either genetic or pharmacological inhibition of ACSS2 alleviated HFFD-induced liver inflammation and injury by reducing AIF1 expression, whereas ACSS2 overexpression abrogated this protective effect. Mechanistically, ACSS2 facilitates KAT5-mediated histone crotonylation to promote AIF1 transcription, which subsequently triggers inflammation and hepatocellular senescence, thereby accelerating MASH progression(Fig. 9). Together, these results reveal the ACSS2/KAT5-AIF1 axis as a previously unrecognized link between histone crotonylation and the core pathologies of MASH, offering fundamental insights into disease development.Fig. 9Schematic diagram of the mechanism by which ACSS2 regulates inflammation and promotes MASH.ACSS2 interacts with KAT5 at the AIF1 promoter to promote histone crotonylation and drive AIF1 transcription. In MASH, upregulation of hepatic ACSS2 enhances this pathway, leading to increased AIF1 expression. Consequently, intensified hepatic inflammation promotes hepatocellular senescence and the secretion of the SASP. The SASP, in turn, establishes a positive feedback loop by further recruiting immune cells and exacerbating the inflammatory response. This figure was created by MedPeer (medpeer.cn). Schematic diagram of the mechanism by which ACSS2 regulates inflammation and promotes MASH. ACSS2 interacts with KAT5 at the AIF1 promoter to promote histone crotonylation and drive AIF1 transcription. In MASH, upregulation of hepatic ACSS2 enhances this pathway, leading to increased AIF1 expression. Consequently, intensified hepatic inflammation promotes hepatocellular senescence and the secretion of the SASP. The SASP, in turn, establishes a positive feedback loop by further recruiting immune cells and exacerbating the inflammatory response. This figure was created by MedPeer (medpeer.cn). Inflammation is widely recognized as a critical driver of disease progression from MASLD to MASH55,56, increasing the risk of cirrhosis and hepatocellular carcinoma in susceptible individuals57. Identifying key inflammatory triggers is therefore essential for deciphering the MASH pathogenesis of the transition from MASLD. In this study, we identified the ACSS2-AIF1 axis as a key contributor to hepatic inflammation50,58. As a well-characterized interferon (IFN)-g-responsive gene, AIF1 (also known as IBA1) is a key mediator of proinflammatory activity and is implicated in regulating inflammatory responses in rheumatoid arthritis, inflammatory skin diseases, and systemic sclerosis50,59,60. Emerging evidence also links AIF1 to metabolic regulation40,61. For example, serum AIF1 levels correlate with HbA1c and waist circumference in a Japanese population, and sequence variations near its locus were associated with obesity and diabetes in a Greek cohort62. Nevertheless, the underlying mechanisms remain unclear. The well-established bidirectional link between cellular senescence and inflammation constitutes a core biological axis that underpins processes from aging and tissue homeostasis to diverse pathologies, such as arthritis and cancer63,64. In the liver, inflammatory accumulation promotes hepatocellular senescence and the SASP, thereby driving multiple pathologies65,66. We demonstrated that ACSS2 in hepatocytes exacerbates metabolic liver injury by promoting inflammation and hepatocellular senescence through KAT5-dependent histone crotonylation at the AIF1 locus. This finding extends the role of ACSS2 beyond lipogenesis to include inflammation modulation, clarifying its functional significance and therapeutic potential. Unlike direct lipogenesis regulation, targeting ACSS2 represents an upstream intervention that concurrently restrains lipogenesis and attenuates inflammatory signaling, suggesting a novel therapeutic strategy for MASH treatment. Epigenetics, particularly through environmental factor-induced protein posttranslational modifications (PTMs), plays a crucial role in various diseases, including MASH67,68. Acylations, a major class of PTMs encompassing more than ten types, such as acetylation, crotonylation, lactylation, and butyrylation, are implicated in MASH progression when dysregulated in hepatocytes69. Among these PTMs, lysine crotonylation (Kcr) is an emerging PTM that is highly enriched at active enhancers and promoters. While Kcr contributes to various diseases, such as diabetic kidney disease and inflammatory bowel disease15,70, its role in MASH remains largely unexplored. ACSS2 regulates diverse biological processes through distinct PTMs, including butyrylation in Treg cells, lactylation in tumorigenesis, and benzoylation in inflammation16,71,72. Notably, our study revealed that hepatocyte-specific ACSS2 knockout in an MASH model led to the most significant reduction in crotonylation among all acylations. Despite this specificity, the precise function of ACSS2-mediated crotonylation in MASH remains unclear, as most PTMs operate through shared and highly similar enzyme systems15. This ambiguity underscores the necessity for further studies to delineate the individual contributions of specific PTMs. While ACSS2 has a well-established lipogenic role, emerging evidence indicates that it also acts as a transcriptional regulator by modulating the epigenetic landscape12,73. Our previous work demonstrated that ACSS2 protects against alcohol-induced liver injury by regulating HAMP1/2 to maintain iron homeostasis, thereby reducing hepatic steatosis and inflammation12. In contrast, the results of the present study revealed a detrimental role for ACSS2 in MASH, indicating that ACSS2 performs context-specific functions in liver pathophysiology. Here, we demonstrated that genetic deficiency of ACSS2 alleviates HFFD-induced inflammation, hepatic steatosis and liver injury, potentially by attenuating inflammatory secretion and hepatocellular senescence (Graphical Abstract). While the role of ACSS2 in lipogenesis has been established, recent research by Zhang et al.45 suggested that inhibiting ACSS2 alone may not sufficiently reverse MASLD progression. In contrast, dual inhibition of ACLY and ACSS2 not only suppresses de novo lipogenesis but also enhances fatty acid oxidation through PPAR-α activation, representing a promising and synergistic therapeutic strategy for MASLD treatment. We propose that seemingly contradictory functions in liver diseases are determined by distinct subcellular localizations, which govern context-dependent mechanistic pathways. This model is strongly supported by our observation of significant nuclear accumulation of ACSS2 in hepatocytes from both MASH patients and murine models compared with controls (Fig. 1H–I). The nuclear accumulation of ACSS2 facilitates KAT5-mediated histone crotonylation to regulate AIF1 expression, thus establishing the necessity of its nuclear localization for this epigenetic mechanism. Under standard feeding conditions, hepatocyte-specific deletion of ACSS2 had minimal metabolic impact, presumably due to systemic compensation. However, in vitro crotonate supplementation elevated global histone crotonylation and upregulated AIF1 protein expression in a dose-dependent manner, suggesting that AIF1 is regulated by histone crotonylation. Crucially, adenovirus-mediated overexpression of AIF1 abolished the beneficial phenotype in ACSS2 hepatocyte-specific knockout mice, exacerbating inflammation, hepatocellular senescence, and liver injury. Together, these results underscore the therapeutic potential of targeting ACSS2 to counteract inflammation, hepatocellular senescence and liver injury driven by chronic nutrient overload. In conclusion, our study demonstrated that ACSS2 promotes inflammation and hepatocellular senescence by facilitating the histone crotonylation-mediated activation of AIF1. This work reveals how the ACSS2-KAT5 complex drives MASH progression via this epigenetic mechanism, thus positioning ACSS2 inhibition as a promising therapeutic strategy to counter the progression from MASLD to MASH. Methods Human tissue Human liver tissue sections from patients diagnosed with MASH, including both MASLD, MASH and adjacent non-MASH tissues, were obtained from Beijing China-Japan Friendship Hospital. The specific classification followed the diagnostic criteria of Beijing China-Japan Friendship Hospital. Briefly, the diagnosis of MASLD was based on the following three criteria: (1) imaging-based diagnosis of fatty liver and/or liver biopsy revealing ≥5% hepatic macrovesicular steatosis; (2) the presence of one or more components of metabolic syndrome (MetS); and (3) the exclusion of other causes of fatty liver, such as excessive alcohol consumption, malnutrition, or Wilson’s disease. The diagnosis of MASH was based on the following two criteria: (1) meeting the diagnostic criteria for MASLD and (2) the coexistence of ≥5% hepatic macrovesicular steatosis with ballooning degeneration and lobular inflammation and/or portal inflammation. Histological scoring was performed according to the NASH Clinical Research Network criteria, with NAS (steatosis 0–3, lobular inflammation 0–3, and ballooning 0–2) and fibrosis stage (F0–F4). NAS ≥ 5 was considered definite MASH, and NAS ≤ 4 as non‑MASH steatotic liver disease. Patient information is provided in Supplementary Table 1. Immunohistochemical staining was performed on the slides with a monoclonal ACSS2 antibody (Cat# 16087-1-AP; Proteintech). The slides were digitally analyzed via a 3DHISTECH Kft scanning system with CaseViewer 2.4.0 software (3DHISTECH, Budapest, Hungary). Statistical analysis was conducted via Image-Pro Plus 6.0. All procedures involving the collection and use of human specimens were approved by the Medical Ethics Committee of the Second Affiliated Hospital of Chongqing Medical University (approval number 2025(37)) and were performed in accordance with the principles of the Declaration of Helsinki. Written informed consent was obtained from all human participants or their legal guardians. The clinical characteristics of the participants are summarized in Supplementary Table 1. Mice, cell lines, and reagents All the animal procedures employed in the project were approved by the Institutional Animal Care and Use Committee of Chongqing Medical University (Approval number: IACUC-CQMU-2024-04025). All studies were performed in accordance with institutional guidelines. All experiments were conducted using 8‑ to 10‑week‑old male and female mice. Mice were housed in a specific pathogen‑free (SPF) environment at 22 ± 2 °C under a 12‑h light/dark cycle with ad libitum access to food and water. Hepatocyte-specific ACSS2 knockout (Acss2-HKO) mice were generated by crossing Acss2 flox/flox (Acss2-Flox) mice with Alb-Cre mice (Jackson Laboratory, USA). All animal protocols were approved by the Animal Ethics Committee of Chongqing Medical University. MIHA cells were obtained from Shanghai Yaji Biotechnology Co., Ltd. (Cat# YS4004C). HEK293T cells were obtained from ATCC (Cat# CRL-3216). AML12 cells were obtained from Procell (Cat# CL-0602). All the cell lines were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Cat# C11965500BT, Gibco), except for AML12 cells, which were cultured in DMEM/F12(Cat# 11330032, Gibco); supplemented with 10% fetal bovine serum (FBS, Cat# 04-010-1 A, Biological Industries) and 100 U/mL penicillin‒streptomycin (Cat# BL505A, Biosharp). A high-fat diet (60% fat, Cat# D12492) was purchased from Research Diets. Palmitic acid (Cat# P9767) was obtained from Sigma‒Aldrich. Bovine serum albumin and fatty-free albumin (BSA, Cat# 9048-46-8) were obtained from BBI Co., Ltd. The ACSS2 inhibitor (Cat# S8588) was obtained from Selleck Chemicals. MASH model MASH was induced in 8–10-week-old male and female mice as previously described74. The mice were fed either a normal chow diet (NCD,10 kcal% fat, 20 kcal% protein, and 70 kcal% carbohydrate; D12450J, Research Diets, USA) or a high-fat/fructose diet (HFFD) for 16 weeks. The HFFD consisted of a high‑fat diet (60 kcal% fat, 20 kcal% protein, and 20 kcal% carbohydrate; D12492, Research Diets, USA), supplemented with 30% fructose in the drinking water. For the CCL4-induced liver injury model, Acss2-Flox and Acss2-HKO mice were intraperitoneally injected with a 20% CCL4 solution in olive oil at a dosage of 2 ml/kg body weight. Injections were administered every other day (three times per week) for a total of 8 weeks75. Mice in the vehicle control group received the same volume of olive oil on an identical schedule. For the AAV8 intervention study, 8–10-week-old Acss2-Flox and Acss2-HKO mice received a tail vein injection of AAV8-control, AAV8-Acss2, or AAV8-Aif1 virus, each mouse received 5 × 10¹¹ viral genomes (VG) of AAV8 virus via tail vein injection, with an injection volume of 100 μL, followed by HFFD feeding for an additional 16 weeks. The virus used in this study was obtained from Shandong WZ Biosciences Co., Ltd. For the ACSS2 inhibitor study, 8–10-week-old male C57BL/6 J mice were divided into three groups: the NCD, HFFD-Control, and HFFD+ACSS2i groups. After 8 weeks on their respective diets (NCD or HFFD), the HFFD+ACSS2i group received intraperitoneal injections of the ACSS2 inhibitor (25 mg/kg body weight) three times per week for another 8 weeks76, whereas the control groups received an equal volume of the vehicle. Throughout the studies, body weight and 12-h fasting blood glucose levels were monitored biweekly. All the mice were housed under specific pathogen-free conditions with a 12-h light/dark cycle. At the endpoint, the mice were euthanized under avertin anaesthesia. Blood was collected via the orbital sinus, allowed to clot at room temperature for 2 h, and centrifuged at 13,000 × g for 10 min to obtain the plasma supernatant, which was stored at −80 °C73. Livers were harvested and processed for paraffin embedding, OCT embedding, or snap-freezing in liquid nitrogen at −80 °C. Navitoclax administration in MCD-fed mice Eight‑ to 10‑week‑old male Acss2‑Flox mice were administered adeno‑associated virus 8 (AAV8) encoding ACSS2 or control vector via tail vein injection (5 × 10¹¹ viral genomes per mouse). At 1 month after injection, mice were fed either a methionine‑choline deficient (MCD) diet (21% fat, 17% protein, 63% carbohydrate, 0% methionine, 0% choline; A02082002B, Research Diets, USA) or a normal chow diet for 4 weeks. Navitoclax (ABT‑263; A276560, Aladdin, China) was suspended in 0.5% carboxymethylcellulose sodium (CMC‑Na; STL007-100g, Shanghai Winherb Medical Science Co., Ltd., China) and administered by oral gavage at 50 mg/kg body weight every other day during the final 2 weeks of MCD feeding. The control group received an equal volume of vehicle (0.5% CMC‑Na) on the same schedule. At the end of the treatment, all mice were sacrificed under anesthesia, and serum, liver tissues, and other relevant samples were collected for further analysis. Isolation of mouse primary hepatocytes and nonparenchymal cells Primary mouse hepatocytes and nonparenchymal cells (NPCs) were isolated via a modified two-step collagenase perfusion technique. Briefly, anaesthetized mice were subjected to laparotomy to expose the inferior vena cava (IVC) and hepatic portal vein. The liver was initially perfused via the IVC with Buffer A (calcium- and magnesium-free Hank’s balanced salt solution containing 0.2 μM EGTA) with outflow through the severed portal vein until the tissue turned pale yellow. The perfusion mixture was subsequently switched to Hank’s balanced salt solution with calcium and magnesium (Buffer B) containing 0.6 g/L collagenase IV (Cat# BS165-1 g, Biosharp) until a honeycomb-like structure appeared. The liver was then excised, transferred into 10 mL of Buffer B, and gently dissociated. The cell suspension was filtered through a 70-μm strainer and centrifuged at 100 × g for 5 min at 4 °C. The supernatant was carefully collected and retained for subsequent NPC isolation. The pellet, containing the crude hepatocyte fraction, was washed twice and further purified by resuspension in 36% Percoll (Cat# BS909, Biosharp) and centrifugation at 2000 × g for 20 min at 4 °C. The purified hepatocytes were resuspended in DMEM supplemented with 10% FBS and 1% penicillin‒streptomycin and then cultured in a 37 °C incubator. For NPC isolation, the supernatant was centrifuged at 600 × g. The cell pellet was resuspended in 5 mL of 25% Percoll and carefully layered over 5 mL of 50% Percoll to form a density gradient. After centrifugation at 2000 × g for 20 min, the NPCs were enriched at the interface between the 25% and 50% Percoll layers. Plasmid and siRNA transfection For transient siRNA transfection, cells were seeded in 6- or 12-well plates and transfected 24 h later, upon adherence, using Lipofectamine™ RNAiMAX (Cat# 13778150, Thermo Fisher Scientific) in Opti-MEM™ I Reduced-Serum Medium. The siRNA targets included Si-Acss2:5′-CAGGATTGATGACATGCTCAATT-3′, Si-Aif1: 5′-GGAUGAGA-UCAACAAGGUAGA-3′, and Si-Kat5: 5′- GAGAUUGAUGGACGUAAGATT-3′. The transfection mixture was replaced with complete culture medium after 24 h. For plasmid transfection, the cells were transfected with Lip2000 Transfection Reagent (Cat# BL623D, Biosharp) in Opti-MEM™ I Reduced-Serum Medium. The ACSS2-FLAG plasmid was a gift from Dr. Zhimin Lu’s laboratory at the University of Texas MD Anderson Cancer Center), whereas the AIF1-HA and KAT5-MYC plasmids were purchased from ZhuHai GeneRulor Medical Technology Co., Ltd. (China). The MolPure® Endo‑free Plasmid Maxi Kit was purchased from Yeasen Biotechnology (Shanghai) Co., Ltd. (Cat#19036ES10). Following transfection, the transfection mixture was replaced with fresh complete medium after 6 h. For the rescue experiments, plasmid DNA transfection was performed 24 h after the initial siRNA transfection (48 h after seeding). Histological analysis Fresh liver tissues were fixed in 4% paraformaldehyde for at least 24 h and subsequently processed through dehydration, clearing, and paraffin infiltration for embedding. The resulting paraffin blocks were sectioned for hematoxylin and eosin (H&E) and immunohistochemical (IHC) staining. Additionally, separate fresh liver tissue samples were embedded in optimal cutting temperature (OCT) compound and rapidly frozen for cryosectioning. These cryosections were subjected to immunofluorescence (IF) staining and Oil Red O staining. Immunohistochemical (IHC) and Sirius Red staining IHC staining was performed on paraffin-embedded sections. Briefly, the sections were deparaffinized in xylene, rehydrated through a graded ethanol series, and subjected to antigen retrieval via sodium citrate buffer or Tris-EDTA buffer. Endogenous peroxidase activity was quenched by incubation with peroxidase blocking buffer (Reagent 1, PV-9001/PV-9002, ZSGB-BIO, China) for 20 min at room temperature. The sections were subsequently incubated with the primary antibody overnight at 4 °C. After being washed with PBS, the sections were sequentially treated with an enhancer solution (Reagent 2) and a horseradish peroxidase (HRP)-conjugated secondary antibody (both from PV-9001/PV-9002, ZSGB-BIO, China) for 30 min at 37 °C. The signal was developed via a DAB substrate kit (Cat# ZLI-9019; ZSGB-BIO, China), and the nuclei were counterstained with hematoxylin for 8–10 seconds, followed by rinsing under running tap water for bluing for 5 min. Finally, the sections were dehydrated through a graded ethanol series, cleared in xylene, and mounted with neutral resin. For histological assessment of collagen deposition, liver sections were stained with the Modified Sirius Red Stain Kit (No Picric Acid; Solarbio, Cat# G1472, China) according to the manufacturer's instructions. All the slides were examined and imaged via a Zeiss Scope A1 microscope (Carl Zeiss AG, Oberkochen, Germany). Immunofluorescence (IF) staining IF staining was performed on frozen tissue sections and cultured cells. Briefly, frozen sections were air-dried for 15 min at room temperature. Both sections and cells were fixed with 4% paraformaldehyde for 10 min at room temperature, permeabilized with 0.3% Triton X-100 for 10 min (extended to 15 min for tissue sections), and blocked with 5% bovine serum albumin (BSA) for 2 h at room temperature. The samples were then incubated with primary antibodies overnight at 4 °C. After being washed with PBS, the samples were incubated with fluorochrome-conjugated secondary antibodies for 2 h at RT in the dark. Nuclei were counterstained with DAPI, and samples were mounted with antifade mounting medium. Images were acquired via a confocal laser scanning microscope. Oil Red O staining The frozen tissue sections were incubated at room temperature for 15 min and then washed with PBS. The sections were fixed with 4% paraformaldehyde for 15 min at room temperature, followed by incubation in 60% isopropanol for 3‒5 min. After a brief wash with PBS to remove residual isopropanol, the sections were stained with an Oil Red O working solution for 30 min at room temperature. The staining solution was then removed, and images were captured via a standard light microscope. Flow cytometry Liver tissues were minced and homogenized in cold PBS. The resulting single-cell suspension was filtered through a 70 μm cell strainer. The filtrate was initially centrifuged at 50 × g for 5 min at 4 °C to remove hepatocytes and large debris. The supernatant was collected and centrifuged at 1000 × g for 5 min at 4 °C. The pellet was resuspended in 1 mL of red blood cell (RBC) lysis buffer and incubated for 5–10 min at room temperature. After lysis, the cells were washed by centrifugation at 1000 × g for 5 min at 4 °C. The cell pellet was then resuspended in an appropriate volume of FBS-containing stain buffer (Cat# 554656, BD Pharmingen) and counted. The cells were incubated with an Fc receptor blocking agent (Cat# 553141; BD Pharmingen) on ice for 10 min. Without washing, fluorochrome-conjugated antibodies were added directly, and the cells were incubated on ice for 30 min in the dark. After two washes with stain buffer, 7-amino-actinomycin D (7-AAD, Cat# 559925, BD Pharmingen) was added 5 min prior to acquisition to discriminate nonviable cells. Samples were acquired on a flow cytometer and analyzed via CytExpert 2.5 and FlowJo 10.0.7 software. The gating strategy was applied as follows: the single-cell suspension was first gated on a forward scatter area (FSC-A) versus side scatter area (SSC-A) plot to exclude debris. Single cells were then selected via FSC-A versus forward scatter height (FSC-H) to exclude doublets. 7-AAD-negative cells were gated to identify the viable cell population. From the viable single-cell gate, CD45-positive cells (Cat# 557659, BD Pharmingen) were selected to identify the total leukocyte population. Mature neutrophils were identified as CD11b + Ly6G+ cells (Cat# 562950/551460, BD Pharmingen). Serum biochemical and cytokine analyses Serum levels of aspartate aminotransferase (AST, Cat# C010-2-1), alanine aminotransferase (ALT, Cat# C009-2-1), and triglycerides (TG, Cat# A110-1-1) were measured via commercial assay kits purchased from Nanjing Jiancheng Bioengineering Institute (Nanjing, China) according to the manufacturer’s protocols. A microplate reader was used for the measurements. The concentrations of serum TNFα (Cat# CSB-E04741m, CUSABIO) and IL-6 (Cat# SYPM1543, Ruixinbio, Quanzhou, China) were quantified via ELISA kits. following the manufacturer’s instructions. RNA extraction and quantitative real-time PCR Total RNA was isolated from cells or tissues via TRIzol reagent in accordance with the manufacturer’s instructions. The concentration and purity of the isolated RNA were assessed via spectrophotometry. First-strand cDNA was reverse transcribed from 0.5–1 µg of total RNA via the PrimeScript RT Reagent Kit (Cat# RR037A, Takara). Quantitative real-time PCR (qRT‒PCR) assays were carried out with 2 × SP qPCR Mix (Cat# BG0014, Bioground) on a LightCycler 480 System (Roche). The expression of genes was normalized to that of the classic housekeeping gene b-actin, and the relative fold changes were determined via the 2^(-ΔΔCt) method. All sequences of primers used are listed in Supplementary Table 2. Western blot Proteins were extracted from tissues and cells via RIPA lysis buffer supplemented with phosphatase and protease inhibitors. The protein concentration was determined via a BCA assay. The proteins were subsequently denatured by boiling at 100 °C for 10 min in 1 × SDS loading buffer via a metal bath. The denatured proteins were separated by SDS‒polyacrylamide gel electrophoresis (SDS‒PAGE) and transferred onto a PVDF membrane via the wet transfer method. The membrane was blocked with 5% nonfat milk for 1.5 h at room temperature and incubated with primary antibodies overnight at 4 °C. After washing, the membranes were incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies for 1.5 h at room temperature. The protein bands were visualized via an enhanced chemiluminescence (ECL) detection system and captured via a chemiluminescence imaging system. The intensity of each band was quantified using ImageJ software (V1.51). Briefly, the images were converted to 8‑bit grayscale, and the background signal was subtracted using a rolling ball method. The integrated density of each protein band was measured and normalized to the corresponding loading control: β‑actin or α‑Tubulin for whole‑cell lysates and cytoplasmic fractions, and Lamin B1 for nuclear fractions. All data were expressed as relative protein expression levels. Antibody information is provided in Supplementary Table 3. Electrophoretic mobility shift assay (EMSA) Nuclear proteins were extracted from treated MIHA cells using a nuclear extraction kit (Cat# KGB5302-50, Jiangsu KeyGEN Bio TECH, China), and protein concentrations were determined by the BCA assay. Aliquots were stored at −80 °C until use. Biotin-labeled double-stranded DNA probes were prepared by annealing equimolar amounts of 3′-end biotinylated sense and unlabeled antisense oligonucleotides in annealing buffer at 95 °C for 5 min, followed by slow cooling to room temperature. For competition assays, an excess of unlabeled wild-type or mutant probe was added to the reaction mixture; for supershift assays, 1 μg of anti-ACSS2 antibody was added. The reaction products were separated on a 6% native polyacrylamide gel in 0.5X TBE buffer at 4 °C under constant voltage (100 V) for approximately 60 min. After separation, the protein-DNA complexes were then transferred onto a positively charged nylon membrane by wet electroblotting and crosslinked by UV irradiation. The biotin signal was detected using a chemiluminescent EMSA kit (Cat# GS009, Beyotime, China), and images were acquired using a chemiluminescence imaging system. Coimmunoprecipitation (Co-IP) Cells grown in a 10 cm culture dish were washed with ice-cold PBS and lysed with 2 mL of RIPA lysis buffer on ice for 10 min. The cells were then scraped and transferred to a 2 mL microcentrifuge tube, followed by lysis with continuous agitation in an ice‒water mixture for 30 min. After centrifugation at 12,000 × g for 15 min at 4 °C, the supernatant was collected, and the protein concentration was determined. For immunoprecipitation (IP) with Protein A/G beads, 1 mg of total protein was incubated with the corresponding primary antibody (ACSS2, KAT5, or MYC-tag) under gentle rotation overnight at 4 °C. Then, 50 μL of Protein A/G PLUS-Agarose beads (Cat# sc-2003, Santa Cruz) were added, and the mixture was incubated overnight at 4 °C. The beads were collected by centrifugation at 2000 × g for 5 min at 4 °C and washed three times with ice-cold Co-IP/RIPA buffer with gentle rotation. Bound proteins were eluted by boiling the beads in 60 μL of 1 × SDS loading buffer at 100 °C for 10 min and detected by western blotting. For IP with FLAG-tagged proteins, 1 mg of total protein was directly incubated with 30 μL of anti-DYKDDDDK magnetic beads (Cat# L00835-1, GenScript) overnight at 4 °C. The magnetic beads were captured via a magnetic stand and washed three times with ice-cold Co-IP/RIPA buffer. Proteins bound to the beads were eluted by boiling in 60 μL of 1 × SDS loading buffer at 100 °C for 10 min and subsequently analyzed by western blotting. Chromatin immunoprecipitation (ChIP) assay Chromatin immunoprecipitation (ChIP) assays were performed in MIHA cells via the SimpleChIP® Enzymatic Chromatin IP Kit (Cat# 9005; Cell Signaling Technology, USA) following the manufacturer’s instructions. Briefly, cells grown in a 10 cm dish were cross-linked by adding 243 µL of 37% formaldehyde (final concentration of 1%) directly to 9 mL of culture medium for 10 min at room temperature. The cross-linking was quenched with glycine. After cell lysis, the chromatin was enzymatically digested to obtain DNA fragments of 200–500 bp. The chromatin lysates were immunoprecipitated overnight at 4 °C with control IgG or specific antibodies against H3, KAT5, H3K8cr, and H3K27cr. Protein A/G magnetic beads were then added to each sample, followed by incubation with rotation for 2 h at 4 °C to capture the antibody‒chromatin complexes. After a series of washes, the bound chromatin complexes were eluted, and the cross-linking was reversed by incubation with NaCl and proteinase K at 65 °C for 2 h. DNA was purified via the spin columns provided in the kit and analyzed via quantitative real-time PCR with the primers listed in Supplementary Table 4. RNA sequencing and data processing Total RNA was extracted via a commercial RNA isolation kit according to the manufacturer’s instructions. RNA integrity and quantity were assessed via an Agilent 2100 Bioanalyzer (Agilent Technologies, CA, USA). Sequencing libraries were constructed following either standard NEB protocols or strand-specific methods. The libraries were subjected to the Illumina platform by Novogene Co., Ltd. (Beijing, China). The reads were mapped to the reference genome, and the number of reads mapped to each gene was quantified via FeatureCounts (v1.5.0-p3). Gene expression levels were estimated as fragments per kilobase of transcript per million mapped reads (FPKM) values. Differential expression analysis between two comparison groups was performed via DESeq2 software (v1.20.0). Genes with an adjusted P value ≤ 0.05 and a fold change >1 were identified as differentially expressed genes (DEGs). Functional enrichment analysis of the DEGs, including Gene Ontology (GO) terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways, was conducted via R software (V.4.4.1). Terms or pathways with a p value < 0.05 were considered statistically significant. Metabolic flux analysis MIHA cells were seeded in 10 cm dishes and. Upon reaching 60%–70% confluence, they were transfected with Si-Control or Si-Acss2, then treated with 400 μM palmitic acid (PA) for 24 h. Subsequently, the cells were switched to glucose-free DMEM containing 25 mM [U-¹³C]glucose (without unlabeled glucose) and incubated for an additional 8–10 h for metabolic labeling. After labeling, the medium was discarded, and the cells were washed twice with ice-cold PBS, then detached using ice-cold trypsin digestion solution and counted. The cell lysates were sonicated and centrifuged, and the supernatants were collected, dried, and derivatized. The mass isotopologue distributions of tricarboxylic acid cycle intermediates (including citrate, fumarate, malate, and succinate) were measured by liquid chromatography-tandem mass spectrometry (LC-MS/MS), with technical services provided by Shanghai Biotree Biomedical Technology Co., Ltd. Finally, M + 2 enrichment was calculated to assess the flux of glucose-derived carbon into the TCA cycle. Proteomics sample preparation and LC-MS/MS analysis MIHA cells overexpressing FLAG‑ACSS2 or FLAG-empty were lysed, and the protein complexes were pulled down using anti‑FLAG magnetic beads. After thorough washing, the bead‑bound proteins were subjected to on‑bead trypsin digestion. The resulting peptides were desalted using C18 SPE columns, separated on a NanoElute UHPLC system, and analyzed on a timsTOF Pro mass spectrometer (Bruker Daltonics) operating in dia‑PASEF mode. The DIA data were processed using DIA‑NN (v1.8) against the UniProt Homo sapiens database (20,429 sequences). Trypsin/P was specified as the cleavage enzyme, allowing up to one missed cleavage. N‑terminal methionine excision and carbamidomethylation of cysteine were set as fixed modifications. The precursor FDR was set to <1%. Protein quantitation was performed using normalized intensity values. Differentially expressed proteins were defined based on fold change (FC) > 1.5 or <1/1.5 and P < 0.05 (Student’s t test). Technical services were provided by Jingjie PTM BioLab (Hangzhou) Co. Inc. In vitro ACSS2 enzymatic activity assay Briefly, nuclear proteins were extracted and their concentrations were determined using a BCA assay, ACSS2 protein was then immunoprecipitated. The purified protein samples were incubated in a reaction mixture containing ATP (Cat# D7378, Beyotime, China), coenzyme A (CoA, Cat# S35669, Shanghai yuanye Bio-Technology Co., Ltd, China), and sodium acetate (Cat# S8750, Sigma). A negative control was prepared by replacing sodium acetate with an equal volume of double‑distilled water. The reaction mixture was incubated at 37 °C for 100 min, after which the reaction was terminated77. The amount of acetyl‑CoA generated was measured using a commercial Acetyl‑CoA ELISA kit (Cat# JM‑11689M2, JINGMEI BIOTECHNOLOGY), and sample concentrations were calculated based on a standard curve. Statistical analysis Statistical analysis was performed using GraphPad Prism 9. Data are presented as mean ± SD. All results were derived from at least three independent experiments to ensure reproducibility. Comparisons between two groups were analyzed by an unpaired, two-tailed Student’s t test, while Comparisons among multiple groups were performed by one-way or two-way ANOVA with appropriate post-hoc tests. Correlations were assessed using Spearman’s rank correlation analysis. Statistical significance was defined as p < 0.05. 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 Supplementary information Reporting Summary Transparent Peer Review file Source data Source Data Source Data