CXCR3+ LEF1low NK cells cause immunopathological hepatic damage in MASH

CXCR3+ LEF1low NK cells cause immunopathological hepatic damage in MASH | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article CXCR3+ LEF1low NK cells cause immunopathological hepatic damage in MASH jia bo, Jie Yang, Chen LI, Jia Zheng, Jianzhong Song This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8654560/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract Metabolic dysfunction-associated steatohepatitis (MASH) is a systemic metabolic disorder associated with obesity that leads to liver disease (such as hepatocyte damage and fibrosis) and hepatocellular carcinoma. During the progression of MASH, the accumulation of metabolites triggers hepatocyte damage and inflammation. however, the mechanisms underlying MASH-related hepatic damage remain incompletely understood. Using several mouse models that simulate critical features of human MASH (hereinafter referred to as MASH mice) and single-cell transcriptomics sequencing, we identify a distinct NK cell subset critically implicated in hepatic immunopathology. We found NK cells with tissue residency characteristics (IL21R, RGS1) and effector functions (GZMK) aggregated in the livers of MASH mice. These CXCR3+ NK cells exhibited reduced activity of the LEF1 transcription factor and were found with elevated abundance in the context of MASH, as observed in both experimental mice and human subjects. The mechanism involves IL-21 sensitizes the responsiveness of hepatic CXCR3⁺ NK cells to metabolic stimuli, such as Acetate and CXCL9, by downregulating LEF1 and upregulating CXCR3, collectively triggering immunopathological hepatic damage. Hepatic CXCR3+LEF1low NK cells from MASH patients and mice exhibit similar transcriptional profiles. During MASH progression, CXCR3+ LEF1lowNK cells were recruited to and retained in the liver, were metabolically reprogrammed in the hepatic microenvironment, and mediated MHC class I-independent auto-destructive killing of hepatocytes. CXCR3+LEF1lowNK cell cytotoxicity fundamentally diverges from canonical anti-tumor NK cell immunity, mechanistically distinguishing auto-destructive versus protective NK cell functional modalities. Biological sciences/Cell biology/Cell signalling/Nutrient signalling Biological sciences/Cell biology/Mechanisms of disease Biological sciences/Cell biology/Cell signalling/Lipid signalling Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Metabolic dysfunction-associated statistic liver disease (MASLD) has a global adult prevalence of approximately 32.4%, making it the most common chronic liver disease 1 . While the majority of MASLD patients do not develop histologic inflammation, approximately 16.7% progress to metabolic dysfunction-associated steatohepatitis (MASH) 2 . A World Health Organization's epidemiological survey shows that more than 115 million adults worldwide are affected by MASH 3 . MASH is pathologically defined by excessive lipid accumulation, lobular inflammation, hepatocellular damage, and progressive fibrosis. This condition can progressively advance to serious complications, including cirrhosis, hepatic failure, and hepatocellular carcinoma (HCC) 4 , 5 . MASH poses a substantial clinical burden, with MASH-related cirrhosis forecasted to be the primary reason for liver transplantation in industrialized nations by 2030 6 . Despite substantial basic research and clinical drug development for MASH treatment, the precise mechanisms underlying hepatic fat accumulation, hepatocellular damage, and fibrosis during MASH pathogenesis remain incompletely elucidated. MASH progression is propelled by hepatic inflammation, orchestrated by a dyschondrosteosis of innate and adaptive immune responses 7 . Within adaptive immunity, CD4⁺ and CD8⁺ T cells are principal players. The latter, particularly cytotoxic T lymphocytes (CTLs), mediate cytotoxicity via the release of effector molecules such as granzyme B (GZMB) and tumor necrosis factor (TNF) 8 . Previous research has implicated hepatic CD8⁺ T cells as pro-inflammatory mediators that propagate hepatic inflammation, hepatocyte injury, and fibrosis during MASH pathogenesis 9 . Previous studies have demonstrated that treatment with the SGLT2 inhibitor empagliflozin significantly reduces the accumulation of autoreactive CD8⁺ T cells and decreases granzyme B levels in the livers of MASH mice. This effect occurs because empagliflozin enhances ketogenesis in CD8⁺ T cells, thereby suppressing their activation and effector functions 10 . Thus, immune mechanisms are profoundly involved in the pathogenesis and progression of MASH. A comprehensive understanding of how immune cells contribute to the pathogenesis of MASH will be instrumental in designing novel therapeutics for this condition. Single-cell sequencing serves as a pivotal tool for investigating metabolic diseases, cancer, and neurodegenerative disorders 11 . Through single-cell sequencing of livers from MASH model mice, we identified the accumulation of CXCR3⁺ NK cells in the liver during MASH progression. Previous studies demonstrated that CXCR3-knockout (KO) mice exhibit enhanced resistance to HFHC diet-induced steatohepatitis. Additionally, the CXCR3-specific inhibitor AMG487 confers significant protection against MASH 12 . Another study demonstrated that Nfil3 ⁻/⁻ mice (also known as E4bp4 ⁻/⁻ , which exhibit defective NK cell development) confer significant resistance to CDHFD diet-induced MASH progression 13 . Collectively, these findings identify NK cells as critically involved in driving the pathogenesis and progression of MASH. Thus, we hypothesize that Cxcr3 + NK cells contribute to the pathogenesis of MASH. However, the phenotype of Cxcr3 + NK cells, how they contribute to the pathogenesis of MASH, and which hepatic cells regulate them are still unclear. In this study, we found NK cells with tissue-resident properties ( IL21R , RGS1 ) and effector functions ( GZMK ) aggregated in the hepatic tissue of MASH mice. The significant feature of these CXCR3 + NK cells is reduced LEF1 transcription factor activity, and they are abundant in the hepatic tissue of both NASH patients and MASH mice. Molecular mechanism studies have shown that IL-21 induces downregulation of LEF1 and upregulation of CXCR3, significantly increasing the hepatic accumulation of CXCR3 + LEF1 low NK cells, when exposed to relevant metabolic stimuli such as Acetate and CXCL9, contributed to auto-destructive hepatocyte killing. Additionally, our results also showed that, the palmitic acid levels are increased in the livers of both NASH patients and MASH mice. Palmitate accumulation promotes hepatic endothelial cells to secrete Cxcl9 by facilitating IRF4 activation, and recruiting and activating Cxcr3 + Lef1 low NK cells in MHC class I-independent. In summary, our research will contribute to a comprehensive understanding of how immune cells contribute to the pathogenesis of MASH, highlighting the potential clinical value of targeting specific NK cell subpopulations to develop innovative drugs or therapeutic approaches for MASH. Methods and Materials 1. MASH mouse model and treatments C57BL/6J mice were obtained from GemPharmatech (Nanjing, China). Both Gimap4 -/- and Irf4 -/- mice on the C57BL/6J background were constructed by Cyagen Biosciences (Suzhou, China) using the CRISPR Cas9 method. All mice were housed under specific pathogen-free (SPF) conditions in compliance with the Association for Research Animal Science guidelines. The mice were raised in 12 hours light/dark cycle environment, the ambient temperature is 24°C, and the humidity is 55%. All mice were fed ad libitum. Mice were fed group-specifically according to experimental plan. Normal diet (ND; Cat. No. D12450J, 70% kcal carbohydrates, 20% kcal protein, 10% kcal fat, Research Diets), choline-deficient high-fat diet (CDHFD; Cat. No. D05010402; 60% kcal fat, 20% kcal carbohydrates, 20% kcal (choline-free casein) protein, Research Diets), high-fat high-cholesterol diet (HFHC diet; Cat. No. D09100301; 40 kcal% fat, 20 kcal% fructose, and 2% cholesterol, Research Diets), western diet (WD; Cat. No. D12079B; 41% kcal fat, 30 kcal% fructose, and no cholesterol, Research Diets), high-fat diet (HFD; Cat. No. D12492; 60% kcal fat, 20% kcal protein, and 20% kcal carbohydrates, Research Diets) were used to establish the mouse model of MASH. After being fed a CDHFD, mice were randomly divided into groups and treated with specified blocking antibody (100 μg/mouse) via intraperitoneal injection: Anti-CD122 antibody (BioXcell, Cat. No. BE0272, clone number 5H4) or Anti-LFA1 antibody (BioXcell, Cat. No. BE0001-1, clone number M17/4). Additional details on monoclonal antibodies are provided in the Supplementary Table. Mice received intraperitoneal injections once every 3 days (4 injections total). Upon experiment termination, mice were euthanized, and tissue and serum samples were collected for subsequent analysis. The Animal Ethics Committee of China Pharmaceutical University approved all mouse experiments (Permit number: SYXK-2021-0011, 25 January 2021). 2. Human sample research A total of 18 participants from the Nanjing University Drum Tower Hospital fatty liver disease prospective observational cohort were included in the present analysis, comprising 12 biopsy-confirmed MASH patients and 6 biopsy-confirmed controls with near-normal hepatic histology. From 2020 to 2025, we collected liver tissue samples from MASH patients and performed pathological sectioning as well as serum biochemical marker and physical parameter testing. A total of 12 MASH patients satisfied the predefined inclusion criteria were: (1) age between 38 and 80 years; (2) biopsy-confirmed MASH diagnosis per current clinical guidelines; (3) absence of other liver diseases and excessive alcohol consumption (≤20 g/day for women, ≤30 g/day for men). Six biopsy-confirmed individuals from the same cohort were enrolled into the non-MASH control group, exhibiting essentially normal liver status. All biopsy results were independently assessed by two hepatopathologists. The NAS Score and fibrosis stage were determined according to the NASH Clinical Research Network scoring system. This study was conducted with the approval of the Ethics Committee of Nanjing University Drum Tower Hospital and with the written informed consent of all participants. 3. I solation of primary mouse lymphocyte Spleen lymphocyte isolation: Following euthanasia, spleens were aseptically excised in a laminar flow hood, placed in cold PBS, and cleared of surrounding adipose tissue and blood. The spleen was placed into a centrifuge tube containing 5 mL of cold PBS and mechanically disrupted. Then, the tissue suspension was transferred into a digestion solution containing collagenase IV and DNase I, followed by incubation at 37°C for 30 minutes with constant agitation. Add serum-containing medium to stop digestion. Filter sequentially through 70 μm and 40 μm cell sieves and collect the filtrate. Centrifuge (4°C, 300 ×g, 5 min) and discard supernatant. Add 3 mL ACK lysis buffer and incubate at room temperature for 2 minutes. Immediately after incubation, add 10 mL cold PBS to stop the reaction and centrifuge (300 ×g, 5 min). Resuspend the cells in 3 mL of cold PBS. Add 3 mL of lymphocyte separation medium to a 15 mL centrifuge tube and slowly add the cell suspension to the upper layer of the separation medium (keep the interface clear). Centrifuge (4°C, 800 ×g, 30 min, acceleration and deceleration set to 0). Transfer the lymphocyte layer to a new tube. Add 10 mL cold PBS to wash the lymphocytes (4°C, 300 ×g, 5 min, repeat twice). Liver immune cell isolation: Following 1-minute perfusion via the portal vein. To obtain a single-cell suspensions, the resulting cells suspension was subsequently passed through the cell strainer (Corning; Cat. No. 352350; 70 μm). After washing, the cells pellet was subjected to digestion in 6 mL of GBSS supplemented with collagenase IV (MedChemExpress; Cat. No. HY-K0012; 1:500 dilution) and digested at 37°C for 15 minutes under constant agitation. The resulting cell suspension was subjected to Percoll density gradient centrifugation (GE Healthcare, Cat. No. 10607095; 40% / 80% layers) using predefined parameters (acceleration rate 7, deceleration rate 0; 1,440 ×g, 25 min) The lymphocyte-enriched interphase was carefully collected and subjected to two washes with 8 mL cold PBS (5 min each at 300 ×g). 4. Isolation of primary mouse hepatocytes Following anesthesia with sodium pentobarbital (50 mg/kg, i.p.), mice were immobilized on a 37°C thermostatic plate. The abdominal skin was disinfected with 75% ethanol. Sequential portal vein perfusion was performed using Pre-perfusion solution (HBSS (Ca²⁺/Mg²⁺-free) + 500 μM EGTA + 15 mM HEPES (pH 7.4), preheated to 37°C) and digestion solution (HBSS (with Ca²⁺/Mg²⁺) + 0.05% type IV collagenase (Sigma, Cat. No.C5138) + 15 mM HEPES + 0.1% BSA, pre-heated to 37°C). Following perfusion, the liver was excised into a pre-chilled culture dish. After removal of Glisson's capsule, hepatocytes were released by gentle teasing with forceps into cold Williams' E medium supplemented with 10% FBS(Gibco, Cat. No. 26140079 ). Hepatocytes were isolated by filtering the suspension through a 70 μm mesh, followed by three washing/centrifugation cycles (50 ×g, 4°C, 3 min). The cells were then purified using 40% Percoll density gradient centrifugation at 100 ×g for 15 min. The high-viability hepatocyte fraction sedimented at the bottom was collected. 5. Isolation of mouse hepatic Cxcr3 + NK cells The same liver perfusion method used for mouse primary hepatocyte isolation was employed, with the digestion solution substituted with 0.02% type IV collagenase (Sigma, Cat. No.C5138). The perfusion time was reduced to 5 minutes to yield a liver cell suspension. Following suspension centrifugation (50 ×g, 3 min, 4°C), the supernatant containing liver non-parenchymal cells (NPCs) was transferred to fresh tubes. Subsequently, the cell supernatant was centrifuged (300 ×g, 15 min, 4°C) to pellet the NPCs. After 1-minute incubation in red blood cell lysis buffer, the cells underwent two washing cycles with cold PBS. The obtained NPCs were subjected to magnetic bead-based negative selection commercial Mouse NK Cell Isolation Kit (Miltenyi Biotec, Cat. No.130-115-818) to isolate mouse hepatic NK cells. Antibody staining was subsequently initiated with Fc receptor blockade using anti-CD16/32 antibody (clone 2.4G2) at 1:100 dilution for 10 minutes. Surface staining: CD49b-APC, CD3ε-FITC (to exclude T cells), Cxcr3-PE. Incubate on ice under light-protected conditions for 20 minutes, then wash with cold PBS. Use the 70 μm nozzle to collect cells at low temperature (4°C) into a collection tube containing RPMI-1640 medium with 50% FBS. The isolated Cxcr3⁺ NK cells were maintained in RPMI-1640 medium with 10% FBS, 10 ng/ml IL-15 and 1000 U/ml IL-2. 6. Isolation of NK cells from human peripheral blood Peripheral blood was collected from donors in sodium heparin tubes and diluted 1:1 with PBS. The diluted blood was carefully layered over Ficoll-Paque™ solution and subjected to density gradient centrifugation at 400 ×g, 25°C for 30 minutes to obtain peripheral blood mononuclear cell (PBMC). The PBMC layer was collected, and NK cells were isolated by negative selection commercial Human NK cell isolation kit (Miltenyi Biotec, Cat. No. 130-092-657) according to the manufacturer's protocol. 7. Flow cytometric analysis Cell surface receptor immunolabeling was performed using fluorescently labeled antibodies in PBS supplemented with 2% FCS (Merck) at 4 °C for 30 minutes. The Transcription Factor Staining Buffer Set (ThermoFisher, Cat. No. 00-5523-00) was utilized for intracellular cytokine staining following the manufacturer's guidelines. Briefly, liver-associate lymphocytes were stimulated in the presence of protein transport inhibitors (Brefeldin A and Monensin, both 1:1,000 from ThermoFisher) for 6 hours. After staining with a fixable viability dye (Thermo Fisher Scientific; eFluor™ 780; 1:3,000) and surface antibodies, cells were processed for fixation, permeabilization, and final intracellular staining against TGF-β and GzmK. For intracellular transcription factor staining, the Foxp3 Transcription Factor Staining Buffer Kit (ThermoFisher, Cat. No. 00-5523-00) was used according to the manufacturer's instructions. For LEF1 staining, incubation with LEF1 primary antibody (CST; Cat. No. 2230; 1:500) was performed overnight (approximately 16 hours) at 4°C, followed by a 2-hour co-incubation at room temperature with an Alexa Fluor 647- (ThermoFisher; Cat. No. A-21244;1:500) or Alexa Fluor 488- (Cell Signaling; Cat. No. 4412s; 1:500) conjugated goat anti-rabbit secondary antibody. Flow cytometry was performed using a CytoFLEX LX standard flow cytometer (Cat. No. A94649AK). Cell sorting was performed using the MoFlo Astrios™ EQ Basic Research Flow Cytometer (Cat. No. A95728AK-CN-TC) with a standard 70 μm nozzle. Data were analyzed using FlowJo software (Tree Star, version.10.8). For antibodies used in flow cytometry analysis, please refer to the attached table. The intracellular calcium ion concentration in NK cells was quantified using the eFluor™ 514 calcium dye (2 mM, ThermoFisher, Cat. No. 65-0859-39) by flow cytometry, and in combination with fluorescently conjugated antibodies. 8. Stimulation of hepatic CXCR3 + NK cells in vitro Mouse hepatic Cxcr3⁺ NK cells were enriched by fluorescence-activated cell sorting (FACS) and resuspended at 1×10⁶ cells/mL. Then, the cells (200 μL per well) were plated into 24-well plates and activated for 2 days in RPMI-1640 medium supplemented with 10% FCS, 100 U/mL penicillin-streptomycin, 0.1 mM β-mercaptoethanol, 2 mM L-glutamine, 10 ng/ml IL-15 and 100 U/mL IL-2. Subsequently, cells were transferred to fresh wells, adjusted to 1×10⁶ cells/mL, and treated with TGF-β1 (5 ng/mL, known to promote NK cell tissue residency [14]) for an additional 2 days. Following density gradient centrifugation using Pancoll (1440 ×g, 25 min), viable cells were collected from the interface. For enrichment of CXCR3⁺LEF1 low NK cells, activated NK cells were treated for 24 h with a combination of IL-21 (10 ng/mL, Thermo Fisher Scientific, Cat. No. PMC0214) and TGF-β1 (5 ng/mL, R&D Systems, Cat. No. 7666-MB). 9. NK cell cytotoxicity assay We used impedance-based xCelligence RTCA MP technology (ACEA Biosciences) to measure the cytotoxic kinetics of NK cells against primary mouse or primary human liver cells over time. In brief, primary hepatocytes were isolated and plated into Xcelligence plates at the indicated density (1×10⁴ cells/well). After 24 hours, when the hepatocytes had adhered properly, 3×10⁴ stimulated Cxcr3 + NK cells were added, and changes in cell impedance were monitored in real time. Blocking antibodies, Acetate, specific inhibitors, or recombinant cytokines were added to the experiment. The vehicle control group (hepatocytes alone) was designated to establish baseline measurements. The electrical impedance was recorded, converted to the cell index value, and normalized by referencing to the time point of NK cell-hepatocyte co-culture initiation. At 24 hours post‑initiation of co‑culture, the cytotoxic efficacy of NK cells against hepatocytes was assessed by quantifying the cell index. The percentage of primary hepatocytes death was determined by normalizing the difference in cell index between experimental and control groups. 10. Adoptive transfer of hepatic auto-destructiveness NK cells into C57BL/6 mice In this study, CXCR3⁺ NK cells isolated from MASH model mice were stimulated in vitro (as described in "Stimulation of hepatic CXCR3⁺ NK cells in vitro") with IL-21 (10 ng/mL) or the LEF1 degrader OP-V1 [15], followed by 24-hour treatment with acetate (15 mM, Merck). The NK cells were then washed with cold PBS and adoptively transferred (1×10⁶ cells/mouse) into the livers of C57BL/6 wild-type (WT) or IRF4 KO mice. Based on experimental groups, mice received intrahepatic injections of anti-LFA1 antibody (100 μg/mouse, BioXcell, clone M17/4, Cat. No. BE0006), anti-FasL antibody (100 μg/mouse, BioXcell, clone MFL3, Cat. No. BE0319), or isotype IgG. Serum alanine aminotransferase (sALT) levels were measured 36 hours post-transfer. 11. Lentiviral transduction and overexpression experiments The pLVX-EF1α-T2A-GFP vector was double-digested with restriction enzymes EcoRI and XhoI, using the following primer sequences: Lef1-F (5’→3’)：**GAATTC**GCCACCATGGCGCCGAGCTGG (Contains EcoRI restriction site + Kozak sequence GCCACC + Lef1 start codon ATG); Lef-R: TTCAGCTCCTCCACCTTGTCGTC **GGTGGTGGTGGT**TCAGCCGCTGC (Contains flexible joint GGTGGTGGT (Gly-Gly-Gly) + removal of stop codon); GFP-F: GGTGGTGGTGGT**GAGCAAGGGCGAGGAG (Flexible joint extension + GFP start sequence, overlapping with Lef1-R); GFP-R: **CTCGAG**TTACTTGTACAGCTCGTCCATGCC (Contains XhoI restriction site + GFP stop codon TTA)，The pLVX-EF1α-LEF1-T2A-GFP plasmid was generated using Gibson Assembly® seamless cloning. Mixed the pLVX-EF1α-LEF1-T2A-GFP plasmid and lentiviral packaging plasmids in the following mass ratio (total DNA 20 μg): pLVX-EF1α-LEF1-T2A-GFP: 10 μg; pMDLg/pRRE: 7.5 μg, pMD2.G: 5 μg, pRSV-Rev: 2.5 μg. The plasmid mixture and Lipo 3000 transfection reagent were diluted in 300 μL Opti-MEM. After incubation at room temperature for 15 minutes, the prepared complex was added to the HEK293T cells. Following collection at 48- and 72-hour time points post-transfection, the supernatant was passed through a 0.45 μm filter. Subsequent concentration of the lentivirus was achieved by ultracentrifugation (70,000 ×g, 2 h, 4°C). Infected HEK293T cells with virus solutions diluted with gradient. After 72 hours of infection, GFP⁺ cells percentage was analyzed by flow cytometry. Virus solution concentrations with a GFP + cells proportion greater than 90% were used to infect NK cells. After infected with lentivirus, NK cells were maintained in culture for 8 hours and then refreshed with RPMI-1640 medium containing IL-2 (300 U/mL) and IL-15 (10 ng/mL). 48 hours later, the expression of CXCR3, LEF1, and GZMK in GFP + NK cells were measured to distinguish NK cells overexpressing LEF1. For NK cell cytotoxicity assays using the Xcelligence device, GFP⁺ NK cells, sorted 24 hours post-lentiviral transduction, were stimulated with 10 ng/ml IL-21 for 24 hours, exposed to 15 mM acetate for 24 hours, then co-cultured with primary mouse hepatocytes at a 3:1 ratio of effectors to targets. NK cell-mediated damage to primary hepatocytes after 24 hours of co-culture was assessed per the protocol detailed in the “NK cell cytotoxicity assay” section. 12. shRNA interference experiment For shRNA target design and vector construction, used the pLKO.1-TRC-Puro vector (Addgene, Cat. No.10878), three targets were designed for STAT1 and IRF4. Specifically, S1: CCACAGATGTGCAAGTCCAAT; S2: GCAGATTTCTCCAGACTTATT; S3: GCTGGAGAATTACCTGGAAAT targeting STAT1. I1: GCTGAAGACCTGATCCAGAAT; I2: GCAGATCCTCAACTTCAAGAA; I3: GCCATTACAGACCTGATTCTT targeting IRF4. shRNA oligos containing AgeI/EcoRI restriction sites were synthesized, annealed into double-stranded DNA, and ligated into AgeI/EcoRI-linearized pLKO.1-TRC-Puro vector, following transformation of Stbl3 competent cells with the ligation products, positive clones were subjected to selection and subsequent sequencing validation. Lentivirus was packaged as described in the “Lentivirus infection and overexpression experiment.” After infecting primary liver endothelial cells with lentivirus for 24 hours, treaded with 0.5 μg/mL puromycin (ThermoFisher, Cat. No. A1113803) and screen for 5 days. Then proceed with subsequent experiments. 13. Confocal microscopy of primary cells Primary hepatocytes (3×10⁴ cells) were plated on collagen-coated chamber slides (Ibidi, Cat. No. 81201). Prior to co-culture with 1×10⁵ CXCR3⁺ NK cells, the complete medium was supplemented with TGF-β1 (5 ng/mL), IL-21 (10 ng/mL), and sodium acetate (15 mM). After 24 hours, cells underwent fixation in 4% paraformaldehyde. washed with PBS, and stained using anti-ICAM1-Alexa Fluor® 488 (1:200, Abcam, Cat. No. ab 2210), anti-LFA1-PE (1:200, Abcam, Cat. No. ab 33477), and DAPI with antifade mounting medium (1 μg/mL, Beyotime, Cat. No. P0131). Stained samples were mounted in neutral resin and imaged using a Zeiss LSM 800 Airyscan confocal microscope with 20× and 40× objectives. 14. Histology, immunohistochemistry, scanning, and automated analysis Mouse liver tissue was processed using standardized procedures, including fixation, dehydration, embedding, and sectioning (Leica RM2235 rotary microtome). Histological staining was performed according to the manufacturers' protocols using the following kits: Hematoxylin and Eosin (H&E) Staining Kit (Beyotime, Cat. No. C0105M); Masson's Trichrome Staining Kit (Solarbio, Cat. No. G1340). Immunohistochemistry was performed according to the method provided on the Thermo Fisher Scientific website (https://www.thermofisher.com). Stained sections were mounted with neutral balsam and scanned using a NanoZoomer S60 (Hamamatsu, Cat. No. C13210-01, 20× objective, 0.46 μm/pixel resolution). 15. Hepatic acetate level measurement Mouse and human liver tissues were weighed, washed with cold physiological saline, and homogenized in 50 μL cold PBS at 4°C for 60 seconds using a tissue homogenizer. The homogenates were subjected to centrifugation at 300 ×g for 10 minutes. Acetate concentrations in the collected supernatants were measured according to the instructions of the commercial assay kit (Acetate Assay Kit, Merck, Cat. No. MAK086). Acetate concentrations in liver were normalized to tissue weight. 16. Serum ALT level measurement Blood samples were allowed to clot at room temperature for 30 minutes and were subsequently centrifuged at 2000 ×g for 12 minutes at 4°C to obtain serum. The ThermoFisher Alanine Aminotransferase (ALT/GPT) Activity Kit (ThermoFisher, Cat. No. MAK052) was then used to test and calculate the sALT levels of each sample according to the experimental procedure described in the manufacturer’s instructions. ALT activity (IU/L) = [a× (OD sample -OD control )2+b×(OD sample -ODc ontrol )+c]×0.482 IU/L×f. a, b, c: the constant of standard curve. f: Dilution factor of sample before tested. 17. Serum Cxcl9 level measurement In this study, serum Cxcl9 levels in mice were quantified using an R&D Systems Mouse Cxcl9 ELISA Kit (Cat. No. DY492). Procedures strictly followed the manufacturer’s protocols. 18. Measurement of IL-21 levels In this study, serum IL-21 levels in MASH model mice and human samples were quantified using enzyme-linked immunosorbent assay (ELISA) kits. Mouse serum IL-21 levels was measured with the R&D Systems Mouse IL-21 ELISA Kit (Cat. No. DY594), and human serum IL-21 levels was assessed using the Thermo Fisher Scientific Human IL-21 ELISA Kit (Cat. No. BMS221). All procedures strictly followed the manufacturer’s protocols. 19. Single-cell transcriptomics sequencing of mouse livers Single-cell RNA sequencing was performed on hepatic tissues from both healthy and MASH mice using the 10x Genomics Chromium platform. Sequencing libraries were processed with CellRanger (v4.0.5) for demultiplexing, read alignment to the GRCm38 reference genome, and initial feature-barcode matrix generation. Subsequent processing and quality control were conducted in R (v4.0.5) using the Seurat package (v5.0.4). Low-quality cells were filtered out based on the following criteria: nFeature_RNA between 300 and 7000, nCount_RNA > 1000 (excluding the top 3% of cells with highest UMI counts), mitochondrial gene percentage (mt_percent) < 10%, and hemoglobin gene percentage (HB_percent) < 3%. 20. UMAP dimension reduction and cell annotation Data integration and dimensionality reduction were conducted with Seurat (version 5.0.4). The selection of features for this process was carried out using the package's FindVariableFeatures and SelectIntegrationFeatures functions. Then, we used the FindIntegrationAnchors function to identify anchor points in the data files for data integration. Finally, we used the IntegrateData function to integrate the four sets of data files. We used the NormalizeData and ScaleData functions to standardize each data matrix file. We performed dimensionality reduction on the integrated data and identified cell subpopulations using the FindClusters function. We used the Unitary Manifold Approximation and Projection (UMAP) method as a visualization method for cell clustering. By identifying cell subpopulations, we identified a total of nine cell subpopulations. Using Single R (version. 2.2.0) in combination with the annotation method from the Panglao DB database 16 , we annotated 7 distinct cell types (B cells, T/NK cells, Endothelial cells, Hepatocytes, Fibroblasts, Macrophages, Neutrophils). The FindAllMarkers function was used for dual purposes: to define conserved gene markers characteristic of each cell subpopulation and to identify differentially expressed genes (DEGs) through comparative analysis across groups or subpopulations. 21. Single-cell sequencing intercellular communication analysis The R package CellChat (version 1.7.2) was employed to infer and analyze the intercellular communication network among defined cell subpopulations. First, we extracted data from each group based on orig.ident, created CellChat objects, performed receptor analysis of highly expressed genes based on CellChatDB.mouse (the pathway database provided by CellChat), and projected the genes onto the PPI (protein-protein interaction) database. We filtered out intercellular communication networks with fewer than 50 cells. Then, a standardized analytical pipeline with consistent filtering criteria was applied to interrogate cell-cell communication networks among subpopulations within each individual sample. 22. Single-cell sequencing pseudotime analysis The preprocessed scRNA-seq expression matrix (rows = genes, columns = cells) and associated metadata (cell type, sample groups) of liver endothelial cell subpopulations were imported into Monocle2 to construct a CellDataSet object. Sequencing depth variation was corrected using the negative binomial distribution model (negbinomial. size ()), and gene expression dispersion was estimated (estimateDispersions ()). The top 2,000 highly variable genes were selected for trajectory construction. Nonlinear dimensionality reduction was performed using the DDRTree algorithm (reduceDimension ()). Pseudotime trajectories were rooted at the ND12-month-CDHFD group with State 1 designated as the developmental origin. A 2D trajectory plot of cellular development was generated. 23. GSEA We retrieved gene sets from the GEO database, including the NK cell effector gene set, the GPCR signaling pathway gene set, the key transcription factor signature gene set for NK cell activation, and the apoptosis signature gene set, among others. Among these, GSE54215 (BATF knockout group vs. wild-type group (day 3)), GSE122931 (TBX21 knockout group vs. wild-type group), GSE215188 (Lef1 knockout group vs. wild-type group), and Genes were considered differentially expressed if they showed |log2FC| > 1 and an adjusted p-value (FDR) < 0.05, as identified by the DESeq2 package in R. These gene sets were used to perform GSEA analysis (using GSEA v.3.0 software) on the Log2-transformed fold changes obtained from RNA-seq data of differentially expressed genes (MASH mouse liver CXCR3 + NK cells vs. normal diet mouse liver CXCR3 + NK cells, and MASH mouse liver CXCR3 + NK cells vs. CXCR6 - NK cells). To assess the normalized enrichment score (NES), gene set fold change data were submitted to the PreRanked analysis module in GSEA v3.0 software. Significance of enrichment was assessed based on an FDR threshold of q ≤ 0.25. 24. Transcription factors identification and network analysis The transcription factors regulating CXCR3 + NK cells activation was analyzed using differentially expressed genes from RNA-seq datasets of CXCR3 + vs. CXCR3 - NK cells from MASH mouse livers. To infer upstream transcriptional regulators of the differentially expressed genes, analysis was performed with the “Transcription Regulation Binding Analysis” module of BART (version 1.1) using its default parameters. Our analysis revealed 25 key transcription factors from the MASH mouse liver RNA-seq data (CXCR3⁺ vs. CXCR3⁻ NK cells), applying a significance threshold of P < 0.01. First, we downloaded the promoter sequences (approximately 2 kb) of significantly differentially expressed genes from the eukaryotic promoter database. Second, we retrieved transcription factor binding sites from the JASPAR core and Hocomoco databases. Then, we used a custom Python script to scan the promoter sequences (-2 kb promoters) and transcription factors at the binding sites of differentially expressed genes. We constructed a transcription factor (TF) interaction network using Cytoscape software (version 3.7.1). To assess the hierarchical structure of the transcription factor network, we calculated the weight of each transcription factor's role performed utilizing the Igraph R package (version.1.2.6). 25. Quantitative real-time PCR Total mRNA was extracted from cells or liver tissue samples using Trizol reagent. Subsequently, 1 μg of total mRNA was reverse-transcribed into cDNA using the PrimeScript™ RT reagent Kit with gDNA Eraser (Perfect Real Time) (Takara, Cat. No. RR047A/B). For cDNA amplification, SYBR Green (Vazyme, Cat. No. Q711-02) and the QuantStudio™ 1 real-time PCR system (Thermo Fisher Scientific, Cat. No. A43179) were employed to detect the total product amount after each PCR cycle and to calculate the CT values. Mouse and human 18S rRNA or HPRT were used as housekeeping genes in the experiments. The complete list of primer sequences for all genes is provided in Data S4. 26. RNA-seq analysis RNA sequencing of MASH mouse liver Cxcr3 + NK cells vs . normal diet mouse liver Cxcr3 + NK cells, and MASH mouse liver Cxcr3 + NK cells vs . CXCR3 - NK cells were performed using the Illumina platform. The raw sequencing reads were processed through FastQC for quality control and aligned using Trinity. Subsequently, differential expression analysis was performed with DESeq2, considering genes with |log2FC| > 1 and FDR < 0.05 as statistically significant. All other RNA-seq analyses in this study followed identical procedures for quality control, alignment, and differential expression analysis. The specific grouping information for these additional analyses is detailed in the Results section. 27. Western blot Total protein was extracted using RIPA buffer. Following lysis, samples were denatured in 6× loading buffer by boiling at 100°C for 8 min and resolved on 12% SDS-PAGE gels. Proteins were then electroblotted onto PVDF membranes (Millipore, IPVH00010). After blocking with 5% non-fat milk (1 h, RT), membranes were probed with primary antibodies overnight at 4°C, followed by incubation with species-matched secondary antibodies (1 h, RT). Signal detection was performed with an ECL substrate (A+B reagents) and darkroom development. Band intensities were quantified via grayscale analysis in ImageJ (NIH) and statistically analyzed with GraphPad Prism. 28. Co-immunoprecipitation For Gimap4 and PLC-γ1 interaction analysis, transfected HEK293T cells with Lipo3000 transfection reagent, co-transfecting the pcDNA3.1-Gimap4-Flag plasmid and the pcDNA3.1-PLCγ1-Myc plasmid (5 µg each), After 48 hours, lyse the cells using IP buffer (Beyotime Biotechnology, Cat. No. P0013), centrifuge at 12,000 rpm for 10 minutes, then add 20 μL of magnetic beads to the supernatant and pre-clear for 2 hours. Divide the mixture into two equal portions, add 2 µg of Flag antibody and isotype IgG to each portion, and incubate at 4°C overnight (16 hours). The next day. Then, 30 µL of Protein A/G magnetic beads (Thermo Fisher, Cat. No.88802) that had been pre-equilibrated were added to the mixture. Incubate at 4°C with gentle rotation for 6 hours. Then, perform Western blot analysis to detect Flag and Myc tags. To detect the direct interaction between Gimap4 and p-PLC-γ1 in Cxcr3 + NK cells, we lysed the cells using IP buffer, then performed IP using the Gimap4 antibody and Western blotting to recognize the p-PLC-γ1 protein. 29. Statistical analysis Data are presented as the mean ± SEM from at least three independent experiments. Statistical analyses were performed using GraphPad Prism software (version 9.0). Significance between two groups was assessed by unpaired two-tailed Student’s t-test; comparisons among multiple groups were analyzed by one-way or two-way ANOVA followed by Tukey’s or Sidak’s multiple comparisons test, as appropriate. A P value of less than 0.05 was considered statistically significant ( *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant). Results 1. Significant alterations in hepatic cell populations in the development of MASH Over 70% of MASH patients are associated with moderate or severe obesity. To investigate the novel mechanisms of obesity-associated MASH, C57BL/6J mice were subjected to a CDHFD for 4 months, 8 months and 12 months, respectively, to establish early-, mid-, and late-stage MASH mouse models (Fig1A). Histopathological examination confirmed that hepatic lipid accumulation and the areas of fibrosis in MASH mice live significantly increased with the duration of CDHFD feeding (Fig 1B). Compared with health mice, MASH mice exhibited significantly increased serum ALT levels, MASH activity scores, hepatic TG content, and liver α-SMA expression (Fig 1C-F). We then performed single-cell RNA sequencing (scRNA-seq) on liver tissues from ND- and CDHFD-fed mice, followed by UMAP dimensionality reduction and clustering analysis. we classified mouse liver cells into seven subpopulations by identifying characteristic genes: B cells, Endothelial cells, Hepatocytes, Macrophages, Fibroblasts, Neutrophils, T/NK cells (Fig 1G, Fig S1). Furthermore, we analyzed changes in the cellular composition across the different groups (Fig 1H). During MASH progression, T/NK cells accumulated significantly in the liver, whereas hepatocytes and endothelial cells declined markedly (Fig 1I). We propose that the accumulated T/NK cells in the liver may participate in critical cell-cell communication. Therefore, we employed the CellChat R package to analyze intercellular crosstalk in the liver. The results revealed a significantly enhanced crosstalk between hepatic endothelial cells and T/NK cells with prolonged CDHFD feeding duration (Fig 1J). The Cxcl9-Cxcr3 receptor pair and the App-CD74 receptor pair primarily mediate enhanced cross-talk between hepatic endothelial cell populations and T/NK cell populations (Fig 1K-L, Fig S2). Cxcl9 is highly expressed specifically in the hepatic endothelial cell population, while Cxcr3 is highly expressed specifically in T/NK cell populations (Fig 1M, Fig S1G-H). 2. Cxcr3 + GZMK high NK cells are accumulate in MASH livers Flow cytometric analysis revealed a marked expansion of hepatic CD8⁺ T cell and NK cell populations during MASH progression (Fig S4A). The frequency of Cxcr3 + cells was significantly elevated within the hepatic CD45 + cell population, while no significant change occurred in the kidneys (Fig 2A-C). Therefore, we further investigated the compositional changes in CD8 + T cells, NK cells, CD4 + T cells and NKT cells within hepatic Cxcr3 + CD45 + cells after 8 months of CDHFD feeding. We found that only Cxcr3 + NK cells were significantly accumulated in the hepatic tissue following the CDHFD diet (Fig 2D). Genome-wide transcriptomic profiling revealed distinct transcriptomic profiles between hepatic Cxcr3⁺ and Cxcr3⁻ NK cells from MASH mice (Fig 2E). In Cxcr3⁺ NK cells isolated from the livers of MASH mice, we observed upregulated expression of genes associated with effector function (Gzmk, Bhlhe40), exhaustion markers (Pdcd1, Tox), as well as key signature genes related to NK cell tissue residency (IL21R, Klf2, and Rgs1) (Fig. 1E). Genetic set enrichment analysis (GSEA) further confirmed that Cxcr3 + NK cells enriched genes associated with NK cell effector functions, activation, and tissue residency, but did not enrich Th17 cells or IL-17-dependent genes (Fig. 2F)-The latter has been previously demonstrated to be associated with immunopathology in infection[17]. Flow cytometry also revealed that the number of Gzmk + , IFN-γ + , IL21R + , Ki67 + , and Bcl2 + cells among Cxcr3 + NK cells in the livers of MASH mice were significantly increased (Fig 2G). Collectively, the hepatic Cxcr3⁺ NK cells in MASH were characterized as long-term liver residential effector cells that simultaneously integrate activation and modulatory signals Next, we investigated the relationship between Cxcr3 + NK cells and MASH development. The frequency of hepatic Cxcr3⁺ NK cells showed a significant positive correlation with serum ALT levels in each stage of MASH mice (Fig 2H). Analysis also revealed a significant positive correlation between hepatic Cxcl9 levels and the Cxcr3⁺ NK cell proportion (Fig 2I). To bridge the gap between preclinical research and human pathophysiology, we collected 18 clinical NASH liver samples. Similarly, the same phenotype was observed in liver samples from NASH patients, and the frequency of hepatic CXCR3⁺ NK cells also was positively associated with serum ALT levels (Fig 2J). Additionally, NASH patients exhibited a marked increase in hepatic CXCL9 expression levels (Fig 2K). CD122 serves as an essential hub for IL-2 /IL-15 signaling axis, centrally orchestrating NK cell development, survival, and homeostatic maintenance 18 . Following treatment with an anti-CD122 blocking monoclonal antibody, the number of Cxcr3 and Gzmk positive cells within the CD45 + cell population decreased (Fig 2L). Hepatic CXCR3-positive Gzmk-high NK cells (but not NKT cells, Cxcr3⁺ CD8⁺ T cells and CD4⁺ T cells (Fig 2Q-S)) were drained (Fig 2M, 2O). Moreover, this treatment was accompanied by improvements in hepatic damage and hepatic lipid accumulation (Fig 2N, 2P). These findings underscore Cxcr3 + NK cells as a pivotal driver in the pathogenesis and progression of MASH IL-21 induces hepatic Cxcr3⁺ NK cells accumulation by downregulating Lef1 and upregulating Cxcr3 expression Transcriptomic analysis of CXCR3⁺ NK cells from MASH mouse livers did not reveal significant upregulation of key transcription factors (T-bet, EOMES, and c-Rel) associated with canonical NK cell cytotoxic activity (Fig 2E). Through hierarchical clustering and transcription factor regulatory network analysis, we identified Lef1 as a key orchestrator of the phenotypic characteristics in hepatic Cxcr3⁺ NK cells (Fig 3A). GSEA further confirmed a reduction in the expression of Lef1-dependent genes (Fig 3B). Flow cytometric analysis also indicated that Lef1 expression was downregulated in hepatic Cxcr3⁺ NK cells from MASH mice (Fig 3C-D). To investigate the regulatory mechanisms of Lef1 in Cxcr3 + NK cells, we next performed gene loss-of-function and gain-of-function studies. The experimental data revealed that inhibition of Lef1 in hepatic Cxcr3 neg NK cells from MASH mice upregulates Cxcr3 expression (Fig 3E), whereas overexpression of Lef1 in Cxcr3⁺ NK cells reduce Cxcr3 expression (Fig 3F). In the livers of MASH mice, a marked inverse correlation was observed between the expression levels of Lef1 and Cxcr3 (Fig 3G). Furthermore, analysis revealed that Lef1 expression levels and hepatic damage severity in MASH were reciprocally related (as measured by serum ALT level and NAS scores) (Fig 3H, J). In clinical samples from NASH patients, we also observed decreased LEF1 expression levels (Fig 3K) and increased CXCR3 expression levels (Fig 3M). The expression level of LEF1 in CD56 + NK cells showed a significant negative correlation with serum ALT levels of NASH patients (Fig 3L). We further identified potential mediators of Lef1 downregulation in Cxcr3 + NK cells through cytokine screening. Significantly increased mRNA levels of IL-21 , IL27 , Tnf , and Cxcl9 were observed in MASH mice liver tissue (Fig 3I). Only IL-21 could upregulate Cxcr3 expression in hepatic CD49b + Cxcr3 neg NK cells (Fig 3N). IL-21 significantly induced the upregulation of Cxcr3 and Gzmk expression, while downregulating Lef1 expression in hepatic Cxcr3 + NK cells (Fig 3O). Serum IL-21 levels in MASH mice exhibited a significant positively correlated with serum ALT (Fig 3P) and, conversely, negatively associated with LEF1 expression in hepatic CXCR3⁺ NK cells (Fig 3Q). In clinical liver samples from NASH patients, we also detected elevated IL-21 expression levels (Fig 3R). Furthermore, IL-21 expression levels in NASH patient livers showed a positive correlation with CXCR3 expression levels (Fig 3S). Acetate induces metabolic reprogramming in Cxcr3 + NK cells and drives their attack on hepatocytes . IL-21 induces Cxcr3 + NK cells to express high levels of GzmK, but not GzmA or GzmB (Fig 3O, S5D). However, exposure to IL-21 did not enhance the capacity of Cxcr3 + NK cells to induce hepatocyte damage (Fig 4A). It is well established that dysregulated lipid metabolism and aberrant free fatty acid levels occur in the livers of NASH patients and MASH mice 19 . Therefore, we further explored the impact of free fatty acids on Cxcr3⁺ NK cells function. We found that acetate significantly upregulated GZMK expression in IL-21 stimulated Cxcr3 + NK cells (Fig 4B), and this upregulation exhibited a dose-dependent increase in response to acetate concentration (Fig 4C). Simultaneously, the hepatic acetate levels showed a marked increase in MASH mice relative to healthy controls. which exhibited a positive correlation with disease progression (Fig 4D). Additionally, NASH patients also exhibited significantly elevated acetate levels in liver tissue (Fig 4E). We hypothesize that acetate may mediate adhesion between Cxcr3 + NK cells and hepatocytes. Therefore, we performed an in vitro CRISPR screen targeting receptors for NK cell-derived cytokines on the surface of hepatocytes and identified that knockout of the TGF-β1 receptor most significantly protected hepatocytes. Cxcr3 + NK cells exhibited a significant increase in TGF-β1 expression (but not TNF) after IL-21 stimulation and acetate exposure (Fig 4F-G). Overexpression of Lef1 suppressed the acetate induced upregulation of Gzmk expression (Fig 4H). Using the impedance-based xCelligence RTCA MP system, we found that IL-21 stimulation and subsequent acetate exposure enabled Cxcr3⁺ Lef1 low NK cells to induce hepatocyte damage (Fig 4I). We termed this non-specific cytotoxic activity of Cxcr3⁺ Lef1 low NK cells "Hepatic Auto-destructiveness" . Conversely, Lef1 overexpression was found to block this auto- destructiveness phenotype (Fig 4J). In adoptive transfer experiments of Cxcr3 + NK cells from MASH mouse livers, we also observed that IL-21-induced and acetate-exposed NK cells or Lef1 inhibitor-treated and acetate-exposed NK cells caused liver injury in recipient mice (Fig 4K-L). Blocking TGF-β1 protects hepatocytes from injury both in vivo and in vitro (Fig 4M-N). Similarly, TGF-β1 expression levels in the livers of NASH patients showed a significant positive correlation with serum ALT levels (Fig 4O). Further research showed that TGF-β1 promoted a robust induction of the adhesion molecule ICAM-1 in primary hepatocytes (Fig 4P). Similarly, Cxcr3⁺ NK cells stimulated with IL-21 could kill TGF-β1-induced hepatocytes, even in the absence of acetate exposure, while LFA-1 blockade conferred protection against Cxcr3⁺ NK cell-mediated hepatocyte injury both in vivo and in vitro (Fig 4Q-R). Additionally, we observed direct contact between Cxcr3 + NK cells and primary hepatocytes following acetate exposure using laser confocal microscopy (Fig 4S). Cxcl9 recruits and activates hepatic Cxcr3⁺ Lef1 low NK cells through Gimap4 The preliminary findings prompted us to investigate whether the cytotoxic activity of hepatic Cxcr3⁺ NK cells from MASH mice against normal hepatocytes is analogous to their targeting of tumor/infected cells. We found that blockade of MHC I on hepatocytes failed to induce Cxcr3⁺ NK cell cytotoxicity, and that inhibiting their primary triggering receptors (NKG2D and NKp30) did not protect hepatocytes from acetate-exposed Cxcr3⁺ NK cell-mediated killing (Fig 5A). Furthermore, in contrast to the null effect of blocking the PLC and MAPK cascades, the auto-destructive function of Cxcr3⁺ NK cells were strictly dependent on intact PI3K and calcium signaling (Fig 5B). This prompted us to identify novel activating stimuli for Cxcr3⁺ NK cells. We referenced the laboratory's earlier MASH mouse metabolomics data to screen several signaling molecules that showed significantly elevated levels in the livers of MASH mice. We found that Cxcl9 (rather than Cxcl10 or Cxcl11) not only recruits Cxcr3 + Lef1 low NK cells but also significantly activates them (Fig 5C). Proteomics revealed that Cxcl9 may activate Cxcr3 + NK cells by inducing the expression of GTPase of the immunity-associated protein family member 4 (Gimap4) (Fig 5D). Fluo-4AM calcium fluorescent probes revealed that both Cxcl9 and overexpression Gimap4 upregulate calcium levels in NK cells, whereas Cxcl9 fails to elevate calcium levels in Gimap4 KO NK cells (Fig 5E). Additionally, Gimap4 knockout does not affect MICA-Fc (NKG2D ligand, activating NKG2D) activation of NK cells (Fig 5E). We predicted downstream molecules of Gimap4 using AlphaFold 3 (Fig S5A-B). Concurrently, CO-IP experiments demonstrated that Gimap4 directly binds to PLC-γ1 in both HEK293T cells and hepatic Cxcr3 + Lef1 low NK cells (Fig 5F-G) and mediates PLC-γ1 phosphorylation (Fig 5H), further activated the hepatic Cxcr3 + Lef1 low NK cells. The Asparagine at site 647 and the Aspartic Acid at site 560 of PLC-γ1 is essential for its interaction with Gimap4 (Fig S5C). In vivo experiments demonstrated that Gimap4 knockout protected against CDHFD-induced weight gain in mice (Fig. 5I) Pathological analysis revealed that Gimap4 knockout protected against CDHFD-induced hepatic lipid accumulation, fibrosis, and increased Gzmk levels (Fig 5J). Concurrently, it reduced serum ALT and IFN-γ levels in MASH mice (Fig 5K-L) as well as hepatic TG and α-SMA expression levels (Fig 5M-N). Flow cytometric analysis revealed that Cxcr3⁺ NK cells from Gimap4 knockout mice fed a CDHFD diet failed to exhibit increased calcium levels upon Cxcl9 stimulation (Fig 5O). In vitro experiments and NK cell adoptive transfer experiments, Gimap4 knockout protected against Cxcr3⁺NK cell-mediated hepatocyte killing (Fig 5P-Q), while it did not affect NK cell anti-infective or anti-tumor immunity (Fig 5R). Hepatic NK cells in MASH patients employ similar mechanisms to mediate hepatocyte damage. To further prove our findings, we validated the function of hepatic Cxcr3 + Lef1 low NK cells in MASH model mice with other obesity backgrounds and observed a conserved role consistent with that found in CDHFD-induced MASH mice (Fig S3). To establish the clinical relevance of our findings, we next investigated whether hepatic NK cells in MASH patients are involved in a similar mechanism of hepatocyte damage. IL-21-induced, acetate-stimulated CXCR3 + IL21R + NK cells from NASH patients exhibit cytotoxic activity against both allogeneic and autologous primary hepatocytes (Fig S7A). Hepatic CXCR3 + IL21R + NK cells in NASH patients also express characteristic genes of auto-destructive NK cells in MASH mice (Fig S7B). PCA and correlation analysis revealed that CXCR3 + IL21R + NK cells in the livers of NASH patients share similar transcriptional profiles with Cxcr3 + Lef1 low NK cells in MASH mice (Fig S7C-D). Similarly, IL21R⁺ NK cells derived from healthy human PBMCs required the presence of CXCL9 to induce hepatocyte damage (Fig. S7E). Furthermore, the cytotoxicity of these reprogrammed NK cells against hepatocytes was dependent on both IL-21 concentration and the effector-to-target ratio (Fig S7F-G). Consistent with the findings in MASH mice, blockade of NKG2D failed to rescue hepatocytes from NK cell-mediated damage, whereas inhibition of calcium signaling significantly suppressed NK cell-induced hepatocyte damage (Fig S7H). In addition to mediating hepatocyte death through GZMK, CXCR3⁺IL21R⁺ NK cells in NASH patient livers also exhibited CXCL9-induced rapid upregulation of Fas ligand (FasL), but not TRAIL (Fig S7I-J). Furthermore, hepatocyte death was found to be dependent on apoptosis rather than necroptosis (Fig S7K). Scd1 suppression mediated palmitic acid accumulation promotes CXCL9 secretion in hepatic endothelial cells. Single-cell RNA sequencing revealed that CXCL9 is predominantly secreted by hepatic endothelial cells (HECs). Therefore, we performed clustering analysis on HECs subsets from MASH mice and identified three distinct subpopulations based on their signature gene expression profiles: Scd1 + HECs, Cxcl9 + HECs and Fbln + HECs (Fig 6A-B). Pseudotime analysis revealed that HECs gradually differentiate into a Cxcl9-high-expressing subpopulation from Scd1 + HECs during MASH progression (Fig 6C-E). We observed no significant changes in hepatic Stearoyl-CoA desaturase 1 (SCD1) expression between normal and MASH mice. A modest downregulation was only observed after 12 months of CDHFD feeding (Fig 6F). In contrast, a significant decrease in Scd1 expression was detected specifically in HECs isolated from the livers of Normal and MASH mice (Fig 6F). SCD1 is a pivotal regulator in maintaining the homeostasis between saturated (SFAs) and monounsaturated fatty acids (MUFAs). Inhibition of SCD1 activity or downregulation of its gene expression leads to intracellular accumulation of SFAs (e.g., palmitate) and a concomitant reduction in MUFAs (e.g., oleate) [20]. Consistently, we observed palmitic acid accumulation in both liver and HECs during MASH progression (Fig 6G-H). To assess gene expression differences between Cxcl9+ HECs and Scd1+ HECs at deeper sequencing depths, we performed RNA sequencing on Cxcl9 + HECs and Scd1+ HECs isolated from the livers of MASH mice fed an 8-month CDHFD diet (Fig 6I). KEGG enrichment analysis and transcription factor screening revealed that palmitate accumulation mediated by Scd1 expression suppression may promote CXCL9 expression through JAK-STAT signaling and IRF4 signaling (Fig 6J-K). This lipid metabolic reprogramming drives CXCL9 secretion in HECs. Western Blot results also showed that JANK-STAT signaling pathway and IRF4 signaling were significantly upregulated in Cxcl9 + HECs (Fig 6L-M). Next, we performed gene knockout and overexpression experiments. Overexpressing STAT1 alone in HECs failed to induce upregulation of CXCL9 expression (Fig 6N). Conversely, knocking down STAT1 and IRF4 both blocked IFN-γ-induced upregulation of CXCL9 expression (Fig 6O). These results indicate that the JAK2-STAT1 signaling pathway regulates CXCL9 expression by modulating IRF4. Furthermore, IL21R + NK cells from healthy mice were stimulated with IL-21 and exposed to acetate to generate Cxcr3 + Lef1 low NK cells. Subsequently, WT and IRF4-KO mice injected with AAV2-Cdh5pro-gRNA-EGFP (gRNA specifically targeting Scd1) and AAV8-TBG-spCas9 were adoptively transferred with Cxcr3 + Lef1 low NK cells and Cxcr3 - NK cells (Fig 6P). Adoptive transfer of Cxcr3⁺Lef1 low NK cells significantly induced hepatocyte injury in wild-type mice. In contrast, Irf4 knockout mice were protected against this liver damage(Fig 6Q-U). Furthermore, Irf4 knockout mice exhibited significantly lower serum Cxcl9 levels (Fig 6W) and reduced hepatic Gimap4 expression compared to wild-type mice (Fig 6V). Collectively, our results reveal that suppression of Scd1 is a critical driver of palmitic acid accumulation in HECs. Palmitic acid accumulation-mediated lipid metabolic reprogramming promotes CXCL9 secretion in HECs, which in turn activates Cxcr3⁺Lef1 low NK cells, ultimately leading to liver damage. Discussion The development and progression of metabolic dysfunction-associated steatohepatitis (MASH) is closely linked to chronic hepatic immune dysregulation and subsequent inflammatory responses 21 . In this study, we observed an accumulation of Cxcr3⁺ NK cells exhibiting tissue-resident features ( IL21R, RGS1 ) and effector function ( GZMK ) in the livers of MASH mice. Cxcr3⁺ NK cells exhibited diminished LEF1 activity and were present with elevated abundance in the livers of both MASH mice and NASH patients. IL-21 reprograms liver-resident NK cells into a sensitized state via LEF1 downregulation and CXCR3 upregulation, enabling robust responses to acetate and CXCL9. Acetate stimulation prompted Cxcr3 + Lef1 low NK cells to secrete TGF-β1 and express high levels of GZMK. TGF-β1 facilitates the interaction between Cxcr3 + Lef1 low NK cells and hepatocytes by inducing ICAM-1 expression on hepatocytes, thereby collectively triggering an immunopathological damage response against hepatocytes. SCD1 suppression in hepatic endothelial cells drives palmitic acid accumulation and subsequent CXCL9 production, ultimately mediating the MHC-I-independent recruitment and activation of Cxcr3⁺Lef1low NK cells. Previous studies have demonstrated that interleukin 21 receptor (IL21R) and interleukin 18 receptor (IL18R) are highly expressed in trNK and cNK cells, respectively, and serve as key markers distinguishing tissue-resident NK cells from their circulating counterparts 22 . This indicates the decisive effect of the cytokine microenvironment in determining NK cells differential activation. Consistently, we observed that IL21R was significantly high expressed in Cxcr3 + Lef1 low NK cells in the liver of MASH mice, while IL18R was significantly downregulated compared with Cxcr3 - NK cells. This also demonstrates the critical regulatory role of IL-21 signaling in Cxcr3 + Lef1 low NK cells. Intestinal dysbiosis and disrupted hepatic fatty acid metabolism are hallmark features and driving forces in the development and progression of MASH 23 , 24 . We propose that the elevated acetate levels in the livers of MASH mice and patients with NASH may result from gut dysbiosis and disrupted fatty acid metabolism. As a complex liver metabolic disease, there is currently no animal model that can completely simulate MASH 25 . Therefore, in this study, we validated our findings in several MASH mouse models in an obese background. Specifically, these included: the HFHC diet-induced MASH model 26 , the HFD diet-induced MASH model 27 , and the WD diet-induced MASH model 28 . These MASH model mice also exhibit similar phenotypes to CDHFD induce MASH model (Fig. S3). Previous reports have shown that NK cell development is severely impaired in Nfil3 KO (also known as E4bp4) mice, while the number and function of T cells, NKT cells, and B cells are unaffected by Nfil3 deficiency 29 . Similarly, our study also showed that Nfil3 -/- mice were significantly resistant to MASH diet-induced hepatic lipid accumulation, inflammation, and liver damage (Fig. S4). In the CDHFD diet-induced MASH mouse model, Nfil3 KO also protects against CDHFD diet-induced NASH development, with significant attenuation of the chemokine signaling pathway and cytokine-cytokine receptor pathway 30 , 31 . These cumulative findings establish NK cells as critical contributors to the development and progression of MASH. The GTPase of Immunity-Associated Protein 4 (Gimap4), alternatively known as IAN4, belongs to the GIMAP protein family 32 . The Gimap family is a class of GTP-binding proteins primarily that is expressed in the immune system (especially lymphocytes), and it resides in the inner mitochondrial and endoplasmic reticulum membranes, orchestrating key processes in immune cell development, functional regulation, and signal transduction 33 . Our research shows that Gimap4 directly binds to Plc-γ1 and regulates calcium signaling in NK cells. Knockout of Gimap4 directly inhibits the activation of Cxcr3 + Lef1 low NK cells. We observed that Gimap4 deletion not only markedly ameliorated liver damage and hepatic inflammation, but also alleviated CDHFD-induced hepatic lipid accumulation and elevated triglyceride levels. We speculate that the reason may be that Cxcr3 + Lef1 low NK cells attack hepatocytes, damaging their ability to regulate lipid metabolism. It is worth noting that Nfil3 -/- mice also exhibited a protective effect against lipid accumulation compared to wild-type mice 30 , which is consistent with our results. MASH represents a significant risk factor and a major etiology for HCC, with patients exhibiting a significantly elevated risk of developing HCC 34 . Previous studies have shown that liver cirrhosis caused by liver fat accumulation, inflammation, and hepatocyte damage associated with MASH is a potential factor in inducing HCC 35 . Although cirrhosis is the predominant risk factor for MASH-related HCC, a substantial proportion (36.6%-50%) of cases arise in a non-cirrhotic background 36 , implicating alternative underlying mechanisms in MASH-driven hepatocarcinogenesis. It is worth noting that our research found that Cxcr3 + Lef1 low NK cells not only damage normal liver cells but also possess the antitumor function as normal NK cells. Cxcr3 + Lef1 low NK cells also damage hepatic endothelial cells, with a significant reduction in MASH late-stage hepatic endothelial cells, leading to a decrease in the activation and recruitment of NK cells, thereby enabling cancerous cells to evade immune surveillance. Collectively, this study provides evidence that hepatic Cxcr3 + Lef1 low cells of MASH mice and MASH patients are activated by Cxcl9 after IL-21 stimulation and acetate exposure, attacking normal hepatocytes. Unlike NK cell activation in anti-infection and anti-tumor responses, the Cxcl9-induced activation of Cxcr3⁺Lef1 low NK cells diverges from this conventional pathway, being uniquely orchestrated by Gimap4. Knocking out Gimap4 to suppress Cxcr3 + Lef1 low NK cell activation can significantly improve MASH. In summary, our findings unveil a previously unappreciated mechanism of a novel NK cell subset contribute to MASH progression. Simultaneously, results obtained from experimental MASH models and human clinical samples demonstrated that targeted inhibition of Cxcr3 + Lef1 low NK cell activation and Gimap4 signaling may represent a promising therapeutic strategy for MASH. Limitations of the study: Cxcr3 + Lef1 low NK cells on normal hepatocytes depends on LAF1-ICAM1-mediated cell adhesion. TGF-β1 is a key cytokine that induces ICAM1 expression in hepatocytes, and knocking out the TGF-β1 receptor on the surface of hepatocytes significantly protects against damage caused by Cxcr3 + Lef1 low NK cells to normal hepatocytes. Notably, during MASH progression, while liver Cxcr3 + Lef1 low NK cells exhibit high TGF-β1 expression, other cells (such as Kupffer cells or hepatic stellate cells) are activated from a quiescent state and produce large amounts of TGF-β1 in response to inflammatory and damage signals 37 , 38 . In this study, we did not investigate the potential interplay between Cxcr3⁺Lef1 low NK cells and other hepatic cell types, such as Kupffer cells or hepatic stellate cells. Declarations Resource availability Lead contact Further information and requests for resources and reagents should be directly to and will be fulfilled by the lead contact, Jiaqiang Bo ( [email protected] ). Materials availability All materials used in this study were commercially available or obtained from existing sources; no novel reagents were developed. Data and code availability Please refer to Data S1 for all raw data of Western blotting. Processed RNA sequencing data was provided in Data S2 and Data S6. This paper does not report original code. Single-cell RNA-sequencing data will be used for future research and are currently available only upon request to the corresponding author. Any additional information required to reanalyze the data reported in this paper is avail-able from the lead contact upon request. Acknowledgments This work was financially supported by grants from the Natural Science Foundation of Xinjiang Uygur Autonomous Region (82560821), and the Project of State Key Laboratory of Natural Medicines, China Pharmaceutical University (SKLNMZZ202214). Author contributions Jiaqiang Bo designed and supervised the experiments. Jiaqiang Bo, Jie Yang, Chen Li analyzed the results and drafted the manuscript. Jiaqiang Bo, Jie Yang, Chen Li, Jia Zheng carried out most of the experiments. Chen Li, Jia Zheng helped conduct animal experiments and methodology. Jiaqiang Bo helped to analyze scRNA-seq data. Jianzhong Song helped collect and analyze the human samples. Jianzhong Song acquired funding and provided funding support. Declaration of interests The authors declare no competing interests. References Riazi K, Azhari H, Charette JH, Underwood FE, King JA, Afshar EE, Swain MG, Congly SE, Kaplan GG, Shaheen AA. The prevalence and incidence of NAFLD worldwide: a systematic review and meta-analysis. Lancet Gastroenterol Hepatol. 2022 Sep;7(9):851-861. doi: 10.1016/S2468-1253(22)00165-0. Ma C, Wang S, Dong B, Tian Y. Metabolic reprograming of immune cells in MASH. Hepatology. 2025 May 5. doi: 10.1097/HEP.0000000000001371. Younossi ZM, Golabi P, Paik JM, Henry A, Van Dongen C, Henry L. The global epidemiology of nonalcoholic fatty liver disease (NAFLD) and nonalcoholic steatohepatitis (NASH): a systematic review. Hepatology. 2023 Apr 1;77(4):1335-1347. doi: 10.1097/HEP.0000000000000004. Luo X, Wang K, Jiang C. Gut microbial enzymes and metabolic dysfunction-associated steatohepatitis: Function, mechanism, and therapeutic prospects. Cell Host Microbe. 2025 May 21:S1931-3128(25)00153-2. doi: 10.1016/j.chom.2025.04.020. Crane H, Eslick GD, Gofton C, Shaikh A, Cholankeril G, Cheah M, Zhong JH, Svegliati-Baroni G, Vitale A, Kim BK, Ahn SH, Kim MN, Strasser SI, George J. Global prevalence of metabolic dysfunction-associated fatty liver disease-related hepatocellular carcinoma: A systematic review and meta-analysis. Clin Mol Hepatol. 2024 Jul;30(3):436-448. doi: 10.3350/cmh.2024.0109. Younossi ZM, Alqahtani SA, Alswat K, Yilmaz Y, Keklikkiran C, Funuyet-Salas J, et al . Global NASH Council. Global survey of stigma among physicians and patients with nonalcoholic fatty liver disease. J Hepatol. 2024 Mar;80(3):419-430. doi: 10.1016/j.jhep.2023.11.004. Tilg H, Adolph TE, Dudek M, Knolle P. Non-alcoholic fatty liver disease: the interplay between metabolism, microbes and immunity. Nat Metab. 2021 Dec;3(12):1596-1607. doi: 10.1038/s42255-021-00501-9. Huby T, Gautier EL. Immune cell-mediated features of non-alcoholic steatohepatitis. Nat Rev Immunol. 2022 Jul;22(7):429-443. doi: 10.1038/s41577-021-00639-3. Dudek M, Pfister D, Donakonda S, Filpe P, Schneider A, Laschinger M, Hartmann D, et al . Auto-aggressive CXCR6+ CD8 T cells cause liver immune pathology in NASH. Nature. 2021 Apr;592(7854):444-449. doi: 10.1038/s41586-021-03233-8. Liu W, You D, Lin J, Zou H, Zhang L, Luo S, et al . SGLT2 inhibitor promotes ketogenesis to improve MASH by suppressing CD8 + T cell activation. Cell Metab. 2024 Nov 5;36(11):2489. doi: 10.1016/j.cmet.2024.10.009. Epub 2024 Oct 16. Erratum for: Cell Metab. 2024 Oct 1;36(10):2245-2261.e6. doi: 10.1016/j.cmet.2024.08.005. Bennett HM, Stephenson W, Rose CM, Darmanis S. Single-cell proteomics enabled by next-generation sequencing or mass spectrometry. Nat Methods. 2023 Mar;20(3):363-374. doi: 10.1038/s41592-023-01791-5. Zhang X, Han J, Man K, Li X, Du J, Chu ES, Go MY, Sung JJ, Yu J. CXC chemokine receptor 3 promotes steatohepatitis in mice through mediating inflammatory cytokines, macrophages and autophagy. J Hepatol. 2016 Jan;64(1):160-70. doi: 10.1016/j.jhep.2015.09.005. Wang F, Zhang X, Liu W, Zhou Y, Wei W, Liu D, Wong CC, Sung JJY, Yu J. Activated Natural Killer Cell Promotes Nonalcoholic Steatohepatitis Through Mediating JAK/STAT Pathway. Cell Mol Gastroenterol Hepatol. 2022;13(1):257-274. doi: 10.1016/j.jcmgh.2021.08.019. Sparano C, Solís-Sayago D, Zangger NS, Rindlisbacher L, Van Hove H, Vermeer M, Westermann F, Mussak C, et al . Autocrine TGF-β1 drives tissue-specific differentiation and function of resident NK cells. J Exp Med. 2025 Mar 3;222(3):e20240930. doi: 10.1084/jem.20240930. Shao J, Yan Y, Ding D, Wang D, He Y, Pan Y, Yan W, Kharbanda A, Li HY, Huang H. Destruction of DNA-Binding Proteins by Programmable Oligonucleotide PROTAC (O'PROTAC): Effective Targeting of LEF1 and ERG. Adv Sci (Weinh). 2021 Oct;8(20):e2102555. doi: 10.1002/advs.202102555. Franzén O, Gan LM, Björkegren JLM. PanglaoDB: a web server for exploration of mouse and human single-cell RNA sequencing data. Database (Oxford). 2019 Jan 1;2019:baz046. doi: 10.1093/database/baz046. Veldhoen M. Interleukin 17 is a chief orchestrator of immunity. Nat Immunol. 2017 May 18;18(6):612-621. doi: 10.1038/ni.3742. PMID: 28518156. Kennedy PR, Arvindam US, Phung SK, Ettestad B, Feng X, Li Y, et al . Metabolic programs drive function of therapeutic NK cells in hypoxic tumor environments. Sci Adv. 2024 Nov;10(44):eadn1849. doi: 10.1126/sciadv.adn1849. Tincopa MA, Anstee QM, Loomba R. New and emerging treatments for metabolic dysfunction-associated steatohepatitis. Cell Metab. 2024 May 7;36(5):912-926. doi: 10.1016/j.cmet.2024.03.011. Epub 2024 Apr 11. Erratum in: Cell Metab. 2024 Jun 4;36(6):1430. doi: 10.1016/j.cmet.2024.04.016. Sen U, Coleman C, Sen T. Stearoyl coenzyme A desaturase-1: multitasker in cancer, metabolism, and ferroptosis. Trends Cancer. 2023 Jun;9(6):480-489. doi: 10.1016/j.trecan.2023.03.003. Huby T, Gautier EL. Immune cell-mediated features of non-alcoholic steatohepatitis. Nat Rev Immunol. 2022 Jul;22(7):429-443. doi: 10.1038/s41577-021-00639-3. Epub 2021 Nov 5. Hu L, Han M, Deng Y, Gong J, Hou Z, Zeng Y, Zhang Y, He J, Zhong C. Genetic distinction between functional tissue-resident and conventional natural killer cells. iScience. 2023 Jun 20;26(7):107187. doi: 10.1016/j.isci.2023.107187. Nie Q, Luo X, Wang K, Ding Y, Jia S, Zhao Q, Li M, Zhang J, Zhuo Y, Lin J, Guo C, Zhang Z, Liu H, Zeng G, You J, Sun L, Lu H, Ma M, Jia Y, Zheng MH, Pang Y, Qiao J, Jiang C. Gut symbionts alleviate MASH through a secondary bile acid biosynthetic pathway. Cell. 2024 May 23;187(11):2717-2734.e33. doi: 10.1016/j.cell.2024.03.034. Horn P, Tacke F. Metabolic reprogramming in liver fibrosis. Cell Metab. 2024 Jul 2;36(7):1439-1455. doi: 10.1016/j.cmet.2024.05.003. Epub 2024 May 31. Gallage S, Avila JEB, Ramadori P, Focaccia E, Rahbari M, Ali A, Malek NP, Anstee QM, Heikenwalder M. A researcher's guide to preclinical mouse NASH models. Nat Metab. 2022 Dec;4(12):1632-1649. doi: 10.1038/s42255-022-00700-y. Zhang X, Fan L, Wu J, Xu H, Leung WY, Fu K, Wu J, Liu K, Man K, Yang X, Han J, Ren J, Yu J. Macrophage p38α promotes nutritional steatohepatitis through M1 polarization. J Hepatol. 2019 Jul;71(1):163-174. doi: 10.1016/j.jhep.2019.03.014. Zhang Z, Yuan Y, Hu L, Tang J, Meng Z, Dai L, et al . ANGPTL8 accelerates liver fibrosis mediated by HFD-induced inflammatory activity via LILRB2/ERK signaling pathways. J Adv Res. 2023 May;47:41-56. doi: 10.1016/j.jare.2022.08.006. Gallage S, Ali A, Barragan Avila JE, Seymen N, Ramadori P, Joerke V, Zizmare L, et al . A 5:2 intermittent fasting regimen ameliorates NASH and fibrosis and blunts HCC development via hepatic PPARα and PCK1. Cell Metab. 2024 Jun 4;36(6):1371-1393.e7. doi: 10.1016/j.cmet.2024.04.015. Kamizono S, Duncan GS, Seidel MG, Morimoto A, Hamada K, Grosveld G, Akashi K, Lind EF, Haight JP, Ohashi PS, Look AT, Mak TW. Nfil3/E4bp4 is required for the development and maturation of NK cells in vivo. J Exp Med. 2009 Dec 21;206(13):2977-86. doi: 10.1084/jem.20092176. Wang F, Zhang X, Liu W, Zhou Y, Wei W, Liu D, Wong CC, Sung JJY, Yu J. Activated Natural Killer Cell Promotes Nonalcoholic Steatohepatitis Through Mediating JAK/STAT Pathway. Cell Mol Gastroenterol Hepatol. 2022;13(1):257-274. doi: 10.1016/j.jcmgh.2021.08.019. Wang S, Gao J, Yang M, Zhang G, Yin L, Tong X. OPN-Mediated Crosstalk Between Hepatocyte E4BP4 and Hepatic Stellate Cells Promotes MASH-Associated Liver Fibrosis. Adv Sci (Weinh). 2024 Dec;11(47):e2405678. doi: 10.1002/advs.202405678. Schnell S, Démollière C, van den Berk P, Jacobs H. Gimap4 accelerates T-cell death. Blood. 2006 Jul 15;108(2):591-9. doi: 10.1182/blood-2005-11-4616. Heinonen MT, Laine AP, Söderhäll C, Gruzieva O, Rautio S, Melén E, Pershagen G, Lähdesmäki HJ, Knip M, Ilonen J, Henttinen TA, Kere J, Lahesmaa R; Finnish Pediatric Diabetes Registry. GIMAP GTPase family genes: potential modifiers in autoimmune diabetes, asthma, and allergy. J Immunol. 2015 Jun 15;194(12):5885-94. doi: 10.4049/jimmunol.1500016. Huang DQ, El-Serag HB, Loomba R. Global epidemiology of NAFLD-related HCC: trends, predictions, risk factors and prevention. Nat Rev Gastroenterol Hepatol. 2021 Apr;18(4):223-238. doi: 10.1038/s41575-020-00381-6. Povero D, Chen Y, Johnson SM, McMahon CE, Pan M, Bao H, et al . HILPDA promotes NASH-driven HCC development by restraining intracellular fatty acid flux in hypoxia. J Hepatol. 2023 Aug;79(2):378-393. doi: 10.1016/j.jhep.2023.03.041. Epub 2023 Apr 13. Huang Y, Xie Y, Zhang Y, Liu Z, Jiang W, Ye Y, et al . Single-cell transcriptome reveals the reprogramming of immune microenvironment during the transition from MASH to HCC. Mol Cancer. 2025 Jun 11;24(1):177. doi: 10.1186/s12943-025-02370-2. Trivedi P, Wang S, Friedman SL. The Power of Plasticity-Metabolic Regulation of Hepatic Stellate Cells. Cell Metab. 2021 Feb 2;33(2):242-257. doi: 10.1016/j.cmet.2020.10.026. Epub 2020 Nov 23. Massagué J, Sheppard D. TGF-β signaling in health and disease. Cell. 2023 Sep 14;186(19):4007-4037. doi: 10.1016/j.cell.2023.07.036. Additional Declarations There is NO Competing Interest. Supplementary Files GraphicalAbstract.jpg Graphical abstract DataS3Cxcl9transcriptionfactor.xlsx Cxcl9 transcription factor Declarationofinterestsform.docx Declaration of interests form F2.tif Figure2 F4.tif Figure4 F3.tif Figure3 DataS5Cxcl9HECsvsScd1HECsRNAseqrawdata.xlsx Cxcl9+HECs_vs_Scd1+HECs RNA-seq raw data F1.tif Figure1 SF3.tif Figure S3 DataS2CXCR3NKcellsRNAseqrowdata.xlsx S2-CXCR3+NK cells RNA-seq row data SF2.tif Figure S2 supplementarydatafile.docx supplementary data information SF7.tif Figure S7 DataS4Primer.xlsx Primers for RT-qPCR SF1.tif Figure S1 DataS1Westernblotrawdata.tif western blot raw data SF4.tif Figure S4 SF6.tif Figure S6 SF5.tif Figure S5 F6.tif Figure6 F5.tif Figure5 Cite Share Download PDF Status: Under Review Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8654560","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":584573770,"identity":"5e306b04-871a-4ea6-8c18-ca9216cc0d01","order_by":0,"name":"jia bo","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA2klEQVRIiWNgGAWjYBACxmYgkQBiSTAwPkioqCFNC7PBgzPHSLFPgoFN8mELM2GFzO3MD288qLlj1yDdfKwisYGNgb+9O4GAw9iMLRKOPUtukDmWdiNxhwyDxJmzGwhoYTCTSGA7nMwgkWN2I/EMG4OBRC4hLezfJBL+QbQUJLYxE6OFx0wise2wHUgLA7Faii0S+w4nMMgcS5ZIOHOMh6BfDPuPb7z549thewbp5oMff1TUyPG39xLQ0gCKEQaGxP0HIAI8eJWDgDwDRIs9QZWjYBSMglEwcgEALkZH20pDKf4AAAAASUVORK5CYII=","orcid":"","institution":"China Pharmaceutical University","correspondingAuthor":true,"prefix":"","firstName":"jia","middleName":"","lastName":"bo","suffix":""},{"id":584573771,"identity":"01aa2f37-095a-42ab-8367-9dae22b39ae8","order_by":1,"name":"Jie Yang","email":"","orcid":"","institution":"China Pharmaceutical University","correspondingAuthor":false,"prefix":"","firstName":"Jie","middleName":"","lastName":"Yang","suffix":""},{"id":584573772,"identity":"c61c4f33-25db-4356-a695-ebfaea743285","order_by":2,"name":"Chen LI","email":"","orcid":"","institution":"Southeast University","correspondingAuthor":false,"prefix":"","firstName":"Chen","middleName":"","lastName":"LI","suffix":""},{"id":584573773,"identity":"8b3686fe-5b37-4451-b245-edcfd9c476ad","order_by":3,"name":"Jia Zheng","email":"","orcid":"","institution":"China Pharmaceutical University","correspondingAuthor":false,"prefix":"","firstName":"Jia","middleName":"","lastName":"Zheng","suffix":""},{"id":584573774,"identity":"49cf9174-50dd-432b-be27-d217251a370d","order_by":4,"name":"Jianzhong Song","email":"","orcid":"","institution":"Tumor Hospital Affiliated to Xinjiang Medical University","correspondingAuthor":false,"prefix":"","firstName":"Jianzhong","middleName":"","lastName":"Song","suffix":""}],"badges":[],"createdAt":"2026-01-21 03:22:21","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8654560/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8654560/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":102377471,"identity":"bba3e2c7-5aae-41ed-b677-1c5b90e1c5d3","added_by":"auto","created_at":"2026-02-11 05:48:46","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":6551849,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCXCR9-CXCR3 signaling axis mediates cross-talk between hepatic endothelial cell and T/NK cell.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eA\u003c/strong\u003e. Experimental design for inducing the MASH model in C57BL/6 mice using the CDHFD diet, n = 6. \u003cstrong\u003eB\u003c/strong\u003e. H\u0026amp;E and Masson's trichrome staining of representative liver sections from ND and CDHFD-fed mice. Scale bar = 50 mm. \u003cstrong\u003eC\u003c/strong\u003e. Serum ALT levels. \u003cstrong\u003eD\u003c/strong\u003e. NAS, NASH score. \u003cstrong\u003eE\u003c/strong\u003e. Hepatic triglycerides levels. \u003cstrong\u003eF\u003c/strong\u003e. Hepatic α-smooth muscle actin (α-SMA) mRNA expression levels in ND and CDHFD group mice. \u003cstrong\u003eG\u003c/strong\u003e. Phenograph-based analysis of liver cell clusters (\u003cstrong\u003eG\u003c/strong\u003e). \u003cstrong\u003eH-I\u003c/strong\u003e. The changes in hepatic cell populations (\u003cstrong\u003eH\u003c/strong\u003e) and the proportional abundance of specific cell types (\u003cstrong\u003eI\u003c/strong\u003e) within each experimental group. \u003cstrong\u003eJ\u003c/strong\u003e. Network diagram depicts cell communication weights and strength among cell subtypes of different groups using the CellChat R package (version: 1.5.0). \u003cstrong\u003eK\u003c/strong\u003e. Heatmap depicting intercellular communication strength across cell populations in the ND and CDHFD groups. \u003cstrong\u003eL\u003c/strong\u003e. Venn diagram displaying the Key molecules mediating intercellular communication between endothelial cell and T/NK cell populations. \u003cstrong\u003eM\u003c/strong\u003e. UMAP visualization illustrates the expression levels of App, Cd74, Cxcl9, and Cxcr3 across hepatic cell populations. Statistical significance was determined by one-way ANOVA followed by Tukey's multiple comparison test (\u003cstrong\u003eC-F\u003c/strong\u003e). Data represent the mean ± SEM. \u003cem\u003e*P\u003c/em\u003e \u0026lt; 0.05, \u003cem\u003e**P\u003c/em\u003e \u0026lt; 0.01, \u003cem\u003e***P\u003c/em\u003e \u0026lt; 0.001.\u003c/p\u003e","description":"","filename":"image2.png","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/1a3b790ace087cca0775d277.png"},{"id":102377473,"identity":"1b95cae4-9756-44b3-b367-7e9d05073ae2","added_by":"auto","created_at":"2026-02-11 05:48:46","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":4516700,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eThe population of Cxcr3⁺Gzmk⁺NK cells was significantly increased in the livers of MASH mice.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eA\u003c/strong\u003e. Analysis of Cxcr3 expression level on Liver (left) and Splenic (right) NK cells from ND and MASH mice, assessed by flow cytometry, n=6. \u003cstrong\u003eB-C\u003c/strong\u003e. Proportion of Cxcr3⁺ cells among lymphocytes in the liver (B) and spleen (C) of ND and MASH mice, n=6. \u003cstrong\u003eD\u003c/strong\u003e. Hepatic Cxcr3⁺ lymphocyte frequencies in ND and MASH mice (fed with CDHFD for 8 months), n=6. \u003cstrong\u003eE-F\u003c/strong\u003e. Volcano plot (E) and GSEA (F) analysis of differentially expressed genes from hepatic Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells compared to Cxcr3\u003csup\u003e neg\u003c/sup\u003e NK cells sorted from the livers of MASH mice, n=5 or 6. \u003cstrong\u003eG\u003c/strong\u003e. The number of hepatic Gzmk\u003csup\u003e+\u003c/sup\u003e, IFNg\u003csup\u003e+\u003c/sup\u003e, TNF\u003csup\u003e+\u003c/sup\u003e, Ki67\u003csup\u003e+\u003c/sup\u003e, Bcl2\u003csup\u003e+ \u003c/sup\u003eNK cells of normal diet (ND)-fed and MASH mice, n=6. \u003cstrong\u003eH-I\u003c/strong\u003e. Correlation of sALT and Hepatic Cxcl9 levers (\u003cstrong\u003eI\u003c/strong\u003e) with frequency of hepatic Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells from normal diet (ND)-fed and MASH mice, n=6. \u003cstrong\u003eJ\u003c/strong\u003e. Correlation between serum ALT levels\u0026nbsp; (sALT) and hepatic Cxcr3⁺ NK cell frequency in healthy controls (n=6) versus NASH patients (n=12). \u003cstrong\u003eK\u003c/strong\u003e. CXCL9 mRNA expression in NASH patient liver tissue was assessed by RT-qPCR. \u003cstrong\u003eL\u003c/strong\u003e. Frequencies and cell numbers (\u003cstrong\u003eO\u003c/strong\u003e) of Cxcr3⁺Gzmk⁺ NK cells in MASH mice (8-month CDHFD fed) following 2-week treatment: CDHFD + anti-CD122 or CDHFD + control Ig, n=6. \u003cstrong\u003eM-N\u003c/strong\u003e. CD49b\u003csup\u003e+ \u003c/sup\u003eNK cell count (\u003cstrong\u003eM\u003c/strong\u003e) and sALT levels (\u003cstrong\u003eN\u003c/strong\u003e) after 2 weeks anti-CD122 therapy in MASH mice. \u003cstrong\u003eP\u003c/strong\u003e. Representative photomicrographs of liver H\u0026amp;E staining.\u003cstrong\u003e Q-S\u003c/strong\u003e. Numbers of NKT cells (\u003cstrong\u003eQ\u003c/strong\u003e), CD44\u003csup\u003e+ \u003c/sup\u003eCD4\u003csup\u003e+ \u003c/sup\u003eT cells (\u003cstrong\u003eR\u003c/strong\u003e) and Cxcr3\u003csup\u003e+ \u003c/sup\u003eCD8\u003csup\u003e+ \u003c/sup\u003eT cells (\u003cstrong\u003eS\u003c/strong\u003e) in the liver after anti-CD122 antibody application from mice in \u003cstrong\u003eL\u003c/strong\u003e. \u003cem\u003e*P\u003c/em\u003e \u0026lt; 0.05, \u003cem\u003e**P\u003c/em\u003e \u0026lt; 0.01,\u003cem\u003e ***P \u003c/em\u003e\u0026lt; 0.001 with all data presented as mean ± SEM. Statistical significance was determined using one-way ANOVA with Tukey’s test (panels A-B, M-O, Q-S) or two-way ANOVA with Sidak’s (D) or Tukey’s (G) test, as indicated. Pearson’s correlation was used to assess the coefficient of determination (R²) and \u003cem\u003eP\u003c/em\u003e values (H-J). unpaired two-tailed t-test (C, K).\u003c/p\u003e","description":"","filename":"image3.png","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/644bb35b79b2694ad16922a5.png"},{"id":102404168,"identity":"1a750ca0-73a1-4b81-914a-96ff224f2826","added_by":"auto","created_at":"2026-02-11 11:01:59","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":1860467,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIdentification and analysis of key transcription factor of hepatic CXCR3\u003c/strong\u003e\u003csup\u003e\u003cstrong\u003e+\u003c/strong\u003e\u003c/sup\u003e\u003cstrong\u003e NK cells in NASH mice.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eA\u003c/strong\u003e. Transcription-factor network analysis performed using RNA-seq data from DEGs in hepatic Cxcr3⁺ versus Cxcr3⁻ NK cells. \u003cstrong\u003eB\u003c/strong\u003e. GSEA of transcription-factor-related signatures from the RNA-seq profiling of hepatic Cxcr3⁺ versus Cxcr3⁻ NK cells. \u003cstrong\u003eC\u003c/strong\u003e. Assessment of LEF1 expression by flow cytometry in hepatic CXCR3⁺ and CXCR3⁻ NK cells. \u003cstrong\u003eD\u003c/strong\u003e. Expression of Lef1 in hepatic Cxcr3\u003csup\u003e+\u003c/sup\u003e and Cxcr3\u003csup\u003eneg\u003c/sup\u003e NK cells in normal diet mice and MASH mice. \u003cstrong\u003eE\u003c/strong\u003e. Cxcr3 expression levels in sorted hepatic CXCR3\u003csup\u003e neg\u003c/sup\u003e NK cells from MASH mouse after Lef1 overexpression (transfected pLVX-EF1α-T2A-GFP or pLVX-EF1α-LEF1-T2A-GFP), n=6. \u003cstrong\u003eF\u003c/strong\u003e. Cxcr3 expression levels after treatment with the Lef1 inhibitor (OP-V1) 24 hours in sorted CXCR3\u003csup\u003e+\u003c/sup\u003e NK cells from MASH mouse, n=6.\u003cstrong\u003e G-H\u003c/strong\u003e. Correlation of frequency of hepatic Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells and serum ALT levels (\u003cstrong\u003eH\u003c/strong\u003e) with Lef1 expression levels in sorted hepatic CD49b\u003csup\u003e+\u003c/sup\u003e Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells from ND and MASH mice, n=6. \u003cstrong\u003eI\u003c/strong\u003e. Analysis revealed differential hepatic cytokine expression MASH mice (CDHFD) compared with ND mice. \u003cstrong\u003eJ\u003c/strong\u003e. Correlation of NAS score with Lef1 expression levels in sorted hepatic CD49b\u003csup\u003e+\u003c/sup\u003e Cxcr3\u003csup\u003e+\u003c/sup\u003eNK cells from mice fed a normal diet and MASH mice, n=6. \u003cstrong\u003eK\u003c/strong\u003e. LEF1 expression quantified by flow cytometry in liver tissue from MASH patients. \u003cstrong\u003eL\u003c/strong\u003e. Correlation of serum ALT levels with Lef1 expression levels in sorted hepatic CD56\u003csup\u003e+\u003c/sup\u003e CXCR3\u003csup\u003e+\u003c/sup\u003eNK cells from patients with MASH, n=12. \u003cstrong\u003eM\u003c/strong\u003e. CXCR3 expression quantified by flow cytometry in liver tissue from MASH patients. \u003cstrong\u003eN\u003c/strong\u003e. Flow cytometric assessment of Cxcr3 induction after 24-hour cytokines stimulation in sorted hepatic CD49b⁺ Cxcr3⁻ NK cells. \u003cstrong\u003eO\u003c/strong\u003e. Expression of Cxcr3, Gzmk, and Lef1 in hepatic Cxcr3⁺ NK cells following 24 hours IL-21 treatment. \u003cstrong\u003eP-Q\u003c/strong\u003e. Correlation of serum ALT level and Lef1 MFI in CD49b\u003csup\u003e+\u003c/sup\u003e NK cells with serum IL-21 levels in normal diet (ND)-fed and MASH mice, n=6. \u003cstrong\u003eR\u003c/strong\u003e. LEF1 expression quantified by flow cytometry in sorted hepatic CXCR3\u003csup\u003e+ \u003c/sup\u003eNK cells from MASH mouse. \u003cstrong\u003eS\u003c/strong\u003e. Correlation of hepatic CXCR3 expression levels with IL-21 expression levels in patients with MASH, n=12. \u003cem\u003e*P\u003c/em\u003e \u0026lt; 0.05, \u003cem\u003e**P\u003c/em\u003e \u0026lt; 0.01, \u003cem\u003e***P\u003c/em\u003e \u0026lt; 0.001. Statistical significance was determined using two-way ANOVA with Tukey's test (D) or unpaired two-tailed t-test (E-F, K, M, O, R). All data are shown as mean ± SEM. Pearson’s correlation was used to assess the coefficient of determination (R²) and \u003cem\u003eP\u003c/em\u003e values (G, H, J, L, P, Q, S).\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e","description":"","filename":"image4.png","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/bf831205e6dc8c27e4ee5408.png"},{"id":102398224,"identity":"49db24ff-c780-4557-a8df-d929b8d50114","added_by":"auto","created_at":"2026-02-11 10:21:49","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":2983900,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAcetate exposure confers hepatic auto-destructive phenotype on IL-21-stimulated CXCR3⁺ NK cells.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eA\u003c/strong\u003e. Auto-destructiveness activity of IL-21 stimulated hepatic CXCR3\u003csup\u003e+\u003c/sup\u003e Lef1\u003csup\u003e low\u003c/sup\u003e NK cells sorted from MASH mice against primary mouse hepatocytes, n=5. \u003cstrong\u003eB\u003c/strong\u003e. Gzmk expression after metabolites exposure and IL-21 treated in hepatic CXCR3\u003csup\u003e+\u003c/sup\u003e Lef1\u003csup\u003e low\u003c/sup\u003e NK cells, n=3. \u003cstrong\u003eC\u003c/strong\u003e. Gzmk expression in IL-21-treated hepatic CXCR3⁺ NK cells following a 24-hour exposure to concentration gradient Acetate, n=5. \u003cstrong\u003eD\u003c/strong\u003e. Hepatic Acetate levels in normal diet (ND) and MASH mice \u003cstrong\u003eE\u003c/strong\u003e. Hepatic acetate levels in NASH patients. \u003cstrong\u003eF-G\u003c/strong\u003e. Quantification of TNF and TGF-β1 expression in CXCR3\u003csup\u003e-\u003c/sup\u003e and CXCR3\u003csup\u003e+\u003c/sup\u003e NK cells after IL-21 stimulation and acetate (15 mM) exposure by RT-qPCR., n=5. \u003cstrong\u003eH\u003c/strong\u003e. Gzmk expression in acetate-exposed CXCR3⁺LEF1\u003csup\u003elow\u003c/sup\u003e NK cells following retroviral transduction with either pLVX-EF1α-T2A-GFP (control) or pLVX-EF1α-LEF1-T2A-GFP, n = 5. \u003cstrong\u003eI-J\u003c/strong\u003e. The auto-destructiveness activity of IL-21-treated CXCR3⁺ NK cells against hepatocytes was assessed following exposure to 15 mM Acetate or LEF1-overexpression, n=3. \u003cstrong\u003eK-L\u003c/strong\u003e. Serum ALT (sALT) levels at day 3 following the adoptive transfer of Acetate-exposed Cxcr3⁺ NK cells that were either IL-21-stimulated (\u003cstrong\u003eK\u003c/strong\u003e) or LEF1-inhibited (\u003cstrong\u003eL\u003c/strong\u003e), n=3. \u003cstrong\u003eM\u003c/strong\u003e. Auto-destructiveness activity of IL-21-treated, Acetate-exposed CXCR3⁺ NK cells following blockade with anti-TNFR1/2, anti-TGF-β1, with anti-TβRII (each at 10 μg/ml) or not, n=3. \u003cstrong\u003eN\u003c/strong\u003e. Serum ALT levels in NASH mice after administration of an anti-TGF-β1 mAb, n = 6. \u003cstrong\u003eO\u003c/strong\u003e. Correlation of serum ALT levels with hepatic TGF-β1 expression levels in patients with NASH, n=12. \u003cstrong\u003eP\u003c/strong\u003e. The expression of ICAM-1 in hepatocytes after 24 hours treatment with TGF-β1, n=6. \u003cstrong\u003eQ\u003c/strong\u003e. Auto-destructiveness activity of IL-21-treated CXCR3⁺ NK cells, with or not anti-LFA1 antibody therapy, in response to 15 mM acetate or 5 ng/ml TGF-β1, n=3. \u003cstrong\u003eR\u003c/strong\u003e. Serum ALT levels at day 3 following the adoptive transfer of Cxcr3⁺ Lef1\u003csup\u003elow\u003c/sup\u003e NK cells and concomitant administration of anti-LFA1 or control Ig (each 100 μg/mouse). \u003cstrong\u003eS\u003c/strong\u003e. Representative confocal images of CXCR3⁺ NK cells co-cultured with hepatocytes following IL-21 stimulation and acetate exposure. Scale bars: 10 μm (Right), 50 μm (Left). Pearson's correlation was applied to determine the coefficient of determination (R²) and corresponding \u003cem\u003ep\u003c/em\u003e-values (O). Data are presented as mean ± SEM.\u003cem\u003e *P\u003c/em\u003e \u0026lt; 0.05, \u003cem\u003e**P\u003c/em\u003e \u0026lt; 0.01,\u003cem\u003e ***P\u003c/em\u003e \u0026lt; 0.001. Statistical significance was determined using one-way ANOVA with Tukey’s test (C-D, I-N, Q-R); two-way ANOVA with Sidak’s (B) or Tukey’s test (F-H); or unpaired two-tailed t-test (A, E, P).\u003c/p\u003e","description":"","filename":"image5.png","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/32deb55a2eed9b6e64126a30.png"},{"id":102397488,"identity":"05d2804c-35ee-4fd0-906e-f2296a994f6f","added_by":"auto","created_at":"2026-02-11 10:17:22","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":6071693,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCxcl9 activation of Cxcr3⁺ NK cells elicit their hepatic auto-destructive function\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eA\u003c/strong\u003e. Auto-destructiveness of IL-21-treated Cxcr3⁺ NK cells against hepatocytes, assessed under conditions of acetate presence or absence and antibody (10 μg/ml) blockade of MHC-I, NKG2D, or NKp30, n=3. \u003cstrong\u003eB\u003c/strong\u003e. Auto-destructiveness of Cxcr3⁺ NK cells following inhibition of specific signaling pathways, n=3. \u003cstrong\u003eC\u003c/strong\u003e. Intracellular calcium influx was measured in IL-21-primed CXCR3⁺ and CXCR3⁻ NK cells following stimulation with TLR ligands, metabolites, chemokines or cytokines, n=3. \u003cstrong\u003eD\u003c/strong\u003e. Volcano plot analysis of differentially expressed proteins from hepatic CXCR3⁺ NK cells with or without Cxcl9 treatment, identified by quantitative proteomics, n=3. \u003cstrong\u003eE\u003c/strong\u003e. Intracellular calcium fluorescence, cells are stained with Fluo-4AM. Scale bar, 25 μm. Time, 24 hours.\u003cstrong\u003e F\u003c/strong\u003e. Co-immunoprecipitation of Gimap4 and Plc-γ1. HEK 293T cells were transfected with plasmids over-expressing Flag-Gimap4 and Myc-Plc-γ1.\u003cstrong\u003e G\u003c/strong\u003e. Co-immunoprecipitation of Gimap4 and p-Plc-γ1 in Hepatic Cxcr3\u003csup\u003e+ \u003c/sup\u003eLef1\u003csup\u003elow\u003c/sup\u003e NK cells from MASH mouse. \u003cstrong\u003eH\u003c/strong\u003e. Plc-γ1 phosphorylation assay in vitro. \u003cstrong\u003eI.\u003c/strong\u003e Body weight gain of wild-type (WT) and Gimap4 knockout (GKO) mice maintained on either a ND or CDHFD, n=8. \u003cstrong\u003eJ.\u003c/strong\u003e H\u0026amp;E staining, Masson staining and GZMK IHC of WT and GKO mice liver after 8 months ND or CDHFD fed. \u003cstrong\u003eK\u003c/strong\u003e. sALT of WT and GKO mice after 8 months ND or CDHFD fed, n=8. \u003cstrong\u003eL-N\u003c/strong\u003e. Serum IFN-γ levels, liver α-SMA expression levels and Hepatic hydroxyproline levels of WT and GKO mice after 8 months fed a ND or CDHFD, n=8. \u003cstrong\u003eO\u003c/strong\u003e. Dynamic Ca²⁺ mobilization in CXCR3⁺ and CXCR3⁻ NK cells from WT and GKO MASH mice, measured upon Cxcl9 exposure following IL-21 stimulated, n=5. \u003cstrong\u003eP\u003c/strong\u003e. Auto-destructiveness activity of IL-21 treated CXCR3\u003csup\u003e+\u003c/sup\u003e NK cells and GKO Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells with 15 mM acetate against hepatocytes, n=3. \u003cstrong\u003eQ\u003c/strong\u003e. Serum ALT (sALT) levels at day 5 following the adoptive transfer of IL-21-stimulated, acetate-exposed Cxcr3⁺ NK cells from either WT or GKO mice, n=5. \u003cstrong\u003eR\u003c/strong\u003e. The killing activity of NK cells or Gimap4\u003csup\u003e-/-\u003c/sup\u003e NK cells against B16F10 mouse melanoma cells at Cxcr3\u003csup\u003e+\u003c/sup\u003eNK cells: B16F10 cells ratios of 1:3, 1:1, and 3:1. Data are presented as mean ± SEM. \u003cem\u003e*P\u003c/em\u003e \u0026lt; 0.05,\u003cem\u003e **P\u003c/em\u003e \u0026lt; 0.01, \u003cem\u003e***P\u003c/em\u003e \u0026lt; 0.001. Statistical analyses were performed using one-way ANOVA with Tukey’s test (A-B, P-Q) or two-way ANOVA followed by Sidak’s (C, I, K-N, R) or Tukey’s (O) test for multiple comparisons.\u003c/p\u003e","description":"","filename":"image6.png","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/2794bdd98d8d874e4c253920.png"},{"id":102377480,"identity":"a8fa8a3d-d701-41b0-9a92-f4f91c2b3a3f","added_by":"auto","created_at":"2026-02-11 05:48:47","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":3911480,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eScd1 inhibition drives CXCL9 secretion in HECs by inducing palmitate accumulation.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eA\u003c/strong\u003e. Marker Genes for Liver Endothelial Cell Subsets. \u003cstrong\u003eB\u003c/strong\u003e. Phenograph-based analysis of hepatic endothelial cells (HECs), n = 6. \u003cstrong\u003eC-E\u003c/strong\u003e. Pseudotime analysis was performed on HECs subsets, showing the ordering of cells by State (\u003cstrong\u003eC\u003c/strong\u003e), by Pseudotime (\u003cstrong\u003eD\u003c/strong\u003e), and by cell type (\u003cstrong\u003eE\u003c/strong\u003e). F. Analysis of Scd1 expression in total liver lysates and isolated HECs of mice fed a CDHFD. \u003cstrong\u003eG-H\u003c/strong\u003e. Palmitic acid levels in the liver tissue (\u003cstrong\u003eG\u003c/strong\u003e) and liver-isolated endothelial cells (\u003cstrong\u003eH\u003c/strong\u003e) of Normal and MASH mice. \u003cstrong\u003eI-J\u003c/strong\u003e. Volcano plot (\u003cstrong\u003eI\u003c/strong\u003e) and KEGG enrichment analysis (\u003cstrong\u003eJ\u003c/strong\u003e) of differentially expressed genes between Cxcl9⁺ and Scd1⁺ HECs, n = 6. \u003cstrong\u003eK\u003c/strong\u003e. Venn diagram analysis of differentially expressed CXCL9-related transcription factors in Cxcl9⁺ HECs compared to Scd1⁺ HECs. \u003cstrong\u003eL-M\u003c/strong\u003e. Quantification of JAK-STAT and IRF4 signaling pathway activation via Western blot, n=3. \u003cstrong\u003eN\u003c/strong\u003e. Luciferase activity analysis of CXCL9 transcriptional activity in HEK293 cells with or without STAT1 or IRF4 overexpression, n=6. \u003cstrong\u003eO\u003c/strong\u003e. CXCL9 mRNA expression quantified by RT-qPCR in primary hepatic endothelial cells after knockdown of IRF4 or STAT1, n=6. \u003cstrong\u003eP\u003c/strong\u003e. Schematic Diagram of Adoptive Transfer of NK Cells. \u003cstrong\u003eQ-R\u003c/strong\u003e. Liver H\u0026amp;E and TUNEL staining at day 3 after adoptive transfer of acetate-exposed, IL-21-stimulated Cxcr3⁺ NK cells or Cxcr3\u003csup\u003e-\u003c/sup\u003eNK cells, after 21 days of injection with AAV2-Cdh5pro-gRNA-EGFP (Scd1-targeting) and AAV8-TBG-spCas9. \u003cstrong\u003eS-T\u003c/strong\u003e. Serum ALT and AST levels at day 3 after adoptive transfer of acetate-exposed IL-21-stimulated Cxcr3⁺ NK cells, n = 6. \u003cstrong\u003eU-W\u003c/strong\u003e. Hepatic caspase 3/7 activity (\u003cstrong\u003eU\u003c/strong\u003e), serum Cxcl9 level (\u003cstrong\u003eV\u003c/strong\u003e) and hepatic Gimap4 mRNA expression level (\u003cstrong\u003eW\u003c/strong\u003e) at day 3 after adoptive transfer of acetate-exposed IL-21-stimulated Cxcr3⁺ NK cells, n = 6. Data are presented as mean ± SEM. \u003cem\u003e*P \u003c/em\u003e\u0026lt; 0.05, \u003cem\u003e**P\u003c/em\u003e \u0026lt; 0.01, \u003cem\u003e***P\u003c/em\u003e \u0026lt; 0.001. Statistical analyses were performed using one-way ANOVA with Tukey’s test (F-H, N-O, R-W) or two-tailed t-test (M).\u003c/p\u003e","description":"","filename":"image7.png","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/da1f34c8a8f29dfae91ee77f.png"},{"id":103056309,"identity":"4d0b07c6-d7a2-43e1-9a61-c4c8b2568d39","added_by":"auto","created_at":"2026-02-20 09:05:47","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":27467652,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/13903b82-7a5f-4914-83a1-4d181340425a.pdf"},{"id":102377468,"identity":"7aa19bd8-750b-4a22-a7a1-df660b5a4754","added_by":"auto","created_at":"2026-02-11 05:48:46","extension":"jpg","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":1740372,"visible":true,"origin":"","legend":"Graphical abstract","description":"","filename":"GraphicalAbstract.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/3012704cbe25bf1117fda74a.jpg"},{"id":102377470,"identity":"0af0c900-673f-4c9b-8885-11154d111729","added_by":"auto","created_at":"2026-02-11 05:48:46","extension":"xlsx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":12442,"visible":true,"origin":"","legend":"Cxcl9 transcription factor","description":"","filename":"DataS3Cxcl9transcriptionfactor.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/b6d544c45a7eb3d2d0dc1206.xlsx"},{"id":102398428,"identity":"8092b685-55b1-4bf5-8f64-cbbfa6f6d243","added_by":"auto","created_at":"2026-02-11 10:22:38","extension":"docx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":13735,"visible":true,"origin":"","legend":"Declaration of interests form","description":"","filename":"Declarationofinterestsform.docx","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/313ccfcc1be516e5db11655e.docx"},{"id":102398273,"identity":"2d18fa91-a0bd-4776-945f-6e4200998923","added_by":"auto","created_at":"2026-02-11 10:22:01","extension":"tif","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":15176820,"visible":true,"origin":"","legend":"Figure2","description":"","filename":"F2.tif","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/b5241ae5ba9c2434abecd210.tif"},{"id":102377474,"identity":"b5d1a910-43d1-4a82-8bd3-924e1de0a583","added_by":"auto","created_at":"2026-02-11 05:48:46","extension":"tif","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":577455,"visible":true,"origin":"","legend":"Figure4","description":"","filename":"F4.tif","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/5d8126125d37db77705eccb1.tif"},{"id":102377476,"identity":"04d16627-0e50-4c2a-b50d-0a783316aeaf","added_by":"auto","created_at":"2026-02-11 05:48:46","extension":"tif","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":345021,"visible":true,"origin":"","legend":"Figure3","description":"","filename":"F3.tif","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/87a3c96b513ebd027e3a1b82.tif"},{"id":102398252,"identity":"70e83589-0a36-4ff2-96fe-0ba9ff785f37","added_by":"auto","created_at":"2026-02-11 10:21:56","extension":"xlsx","order_by":7,"title":"","display":"","copyAsset":false,"role":"supplement","size":5038866,"visible":true,"origin":"","legend":"Cxcl9+HECs_vs_Scd1+HECs RNA-seq raw data","description":"","filename":"DataS5Cxcl9HECsvsScd1HECsRNAseqrawdata.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/90c74482b054a9ffb0f67286.xlsx"},{"id":102377488,"identity":"6603a402-2745-4cb6-afe6-f1f01d0dfff4","added_by":"auto","created_at":"2026-02-11 05:48:47","extension":"tif","order_by":8,"title":"","display":"","copyAsset":false,"role":"supplement","size":4222690,"visible":true,"origin":"","legend":"Figure1","description":"","filename":"F1.tif","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/d6e722c9091f00953c5fcff4.tif"},{"id":102397757,"identity":"af82ca86-746a-4c7d-926a-6f698be1804e","added_by":"auto","created_at":"2026-02-11 10:19:34","extension":"tif","order_by":9,"title":"","display":"","copyAsset":false,"role":"supplement","size":7477799,"visible":true,"origin":"","legend":"Figure S3","description":"","filename":"SF3.tif","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/d5e75cb6cd8b87f506ee14af.tif"},{"id":102377478,"identity":"6862f321-6387-466b-b050-2463e3f8d85b","added_by":"auto","created_at":"2026-02-11 05:48:47","extension":"xlsx","order_by":10,"title":"","display":"","copyAsset":false,"role":"supplement","size":2112252,"visible":true,"origin":"","legend":"S2-CXCR3+NK cells RNA-seq row data","description":"","filename":"DataS2CXCR3NKcellsRNAseqrowdata.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/ffa1d4d95dac80a6ddf2fe06.xlsx"},{"id":102377495,"identity":"46a9e7da-d2c0-4d54-9078-a8e028e52076","added_by":"auto","created_at":"2026-02-11 05:48:48","extension":"tif","order_by":11,"title":"","display":"","copyAsset":false,"role":"supplement","size":4001199,"visible":true,"origin":"","legend":"Figure S2","description":"","filename":"SF2.tif","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/8b9c0c8bd2f8db4f8be9cf52.tif"},{"id":102377490,"identity":"9f37c08a-acfa-4074-8ee7-0e9e5cb7b19a","added_by":"auto","created_at":"2026-02-11 05:48:47","extension":"docx","order_by":12,"title":"","display":"","copyAsset":false,"role":"supplement","size":19514674,"visible":true,"origin":"","legend":"supplementary data information","description":"","filename":"supplementarydatafile.docx","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/6868db9b436c45c927e8a9f0.docx"},{"id":102377481,"identity":"ec6928eb-c653-443b-a970-7c71953cfe02","added_by":"auto","created_at":"2026-02-11 05:48:47","extension":"tif","order_by":13,"title":"","display":"","copyAsset":false,"role":"supplement","size":2972189,"visible":true,"origin":"","legend":"Figure S7","description":"","filename":"SF7.tif","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/300f13ce231bd07bd185d8f3.tif"},{"id":102377486,"identity":"b86ac5ae-b72b-4d08-a833-fa449e9c8c96","added_by":"auto","created_at":"2026-02-11 05:48:47","extension":"xlsx","order_by":14,"title":"","display":"","copyAsset":false,"role":"supplement","size":7362,"visible":true,"origin":"","legend":"Primers for RT-qPCR","description":"","filename":"DataS4Primer.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/860b375cdf6527cb39544df8.xlsx"},{"id":102398416,"identity":"8c1c7ed8-7939-4529-bf65-122de7077002","added_by":"auto","created_at":"2026-02-11 10:22:34","extension":"tif","order_by":15,"title":"","display":"","copyAsset":false,"role":"supplement","size":4284089,"visible":true,"origin":"","legend":"Figure S1","description":"","filename":"SF1.tif","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/aa3a68e3c7f8b23705cb3c37.tif"},{"id":102377491,"identity":"93cd4a74-9365-458b-a5e8-0008ed29ce40","added_by":"auto","created_at":"2026-02-11 05:48:47","extension":"tif","order_by":16,"title":"","display":"","copyAsset":false,"role":"supplement","size":16928940,"visible":true,"origin":"","legend":"western blot raw data","description":"","filename":"DataS1Westernblotrawdata.tif","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/c1a1fcc801b6615fcc4c93c2.tif"},{"id":102377493,"identity":"41fd1bfb-d84a-40ed-ab8f-1ec5121b79e6","added_by":"auto","created_at":"2026-02-11 05:48:48","extension":"tif","order_by":17,"title":"","display":"","copyAsset":false,"role":"supplement","size":2694166,"visible":true,"origin":"","legend":"Figure S4","description":"","filename":"SF4.tif","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/52c7772ed61e5f2597f19b7c.tif"},{"id":102377492,"identity":"901dd06c-8f4d-4e41-bc59-6b2a26e383e3","added_by":"auto","created_at":"2026-02-11 05:48:48","extension":"tif","order_by":18,"title":"","display":"","copyAsset":false,"role":"supplement","size":7073372,"visible":true,"origin":"","legend":"Figure S6","description":"","filename":"SF6.tif","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/fbb7e0c71386b856f2b69841.tif"},{"id":102377484,"identity":"2bbd8316-5e4a-4387-9f34-3a88d8bd0fd6","added_by":"auto","created_at":"2026-02-11 05:48:47","extension":"tif","order_by":19,"title":"","display":"","copyAsset":false,"role":"supplement","size":3954712,"visible":true,"origin":"","legend":"Figure S5","description":"","filename":"SF5.tif","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/458906a59009f96d2e366e71.tif"},{"id":102377489,"identity":"b89daca6-7039-4578-adcf-885d0f08e2a3","added_by":"auto","created_at":"2026-02-11 05:48:47","extension":"tif","order_by":20,"title":"","display":"","copyAsset":false,"role":"supplement","size":6710552,"visible":true,"origin":"","legend":"Figure6","description":"","filename":"F6.tif","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/d7c996600186cacff57a56e0.tif"},{"id":102377494,"identity":"3e4ee9ec-24ed-4e79-aaef-c027e16cb5a4","added_by":"auto","created_at":"2026-02-11 05:48:48","extension":"tif","order_by":21,"title":"","display":"","copyAsset":false,"role":"supplement","size":129169108,"visible":true,"origin":"","legend":"Figure5","description":"","filename":"F5.tif","url":"https://assets-eu.researchsquare.com/files/rs-8654560/v1/6e5fecdc32015812b55b2b62.tif"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"CXCR3+ LEF1low NK cells cause immunopathological hepatic damage in MASH","fulltext":[{"header":"Introduction","content":"\u003cp\u003eMetabolic dysfunction-associated statistic liver disease (MASLD) has a global adult prevalence of approximately 32.4%, making it the most common chronic liver disease\u0026nbsp;\u003csup\u003e1\u003c/sup\u003e. While the majority of MASLD patients do not develop histologic inflammation, approximately 16.7% progress to metabolic dysfunction-associated steatohepatitis (MASH)\u0026nbsp;\u003csup\u003e2\u003c/sup\u003e. A World Health Organization's epidemiological survey shows that more than 115 million adults worldwide are affected by MASH\u0026nbsp;\u003csup\u003e3\u003c/sup\u003e. MASH is pathologically defined by excessive lipid accumulation, lobular inflammation, hepatocellular damage, and progressive fibrosis. This condition can progressively advance to serious complications, including cirrhosis, hepatic failure, and hepatocellular carcinoma (HCC)\u0026nbsp;\u003csup\u003e4\u003c/sup\u003e\u003csup\u003e,\u003c/sup\u003e\u003csup\u003e5\u003c/sup\u003e. MASH poses a substantial clinical burden, with MASH-related cirrhosis forecasted to be the primary reason for liver transplantation in industrialized nations by 2030\u0026nbsp;\u003csup\u003e6\u003c/sup\u003e. Despite substantial basic research and clinical drug development for MASH treatment, the precise mechanisms underlying hepatic fat accumulation, hepatocellular damage, and fibrosis during MASH pathogenesis remain incompletely elucidated.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp;\u0026nbsp;MASH progression is propelled by hepatic inflammation, orchestrated by a dyschondrosteosis of innate and adaptive immune responses\u003csup\u003e7\u003c/sup\u003e. Within adaptive immunity, CD4⁺ and CD8⁺ T cells are principal players. The latter, particularly cytotoxic T lymphocytes (CTLs), mediate cytotoxicity via the release of effector molecules such as granzyme B (GZMB) and tumor necrosis factor (TNF)\u003csup\u003e8\u003c/sup\u003e. Previous research has implicated hepatic CD8⁺ T cells as pro-inflammatory mediators that propagate hepatic inflammation, hepatocyte injury, and fibrosis during MASH pathogenesis\u0026nbsp;\u003csup\u003e9\u003c/sup\u003e. Previous studies have demonstrated that treatment with the SGLT2 inhibitor empagliflozin significantly reduces the accumulation of autoreactive CD8⁺ T cells and decreases granzyme B levels in the livers of MASH mice. This effect occurs because empagliflozin enhances ketogenesis in CD8⁺ T cells, thereby suppressing their activation and effector functions\u0026nbsp;\u003csup\u003e10\u003c/sup\u003e. Thus, immune mechanisms are profoundly involved in the pathogenesis and progression of MASH.\u003c/p\u003e\n\u003cp\u003eA comprehensive understanding of how immune cells contribute to the pathogenesis of MASH will be instrumental in designing novel therapeutics for this condition. Single-cell sequencing serves as a pivotal tool for investigating metabolic diseases, cancer, and neurodegenerative disorders\u0026nbsp;\u003csup\u003e11\u003c/sup\u003e. Through single-cell sequencing of livers from MASH model mice, we identified the accumulation of CXCR3⁺ NK cells in the liver during MASH progression. Previous studies demonstrated that CXCR3-knockout (KO) mice exhibit enhanced resistance to HFHC diet-induced steatohepatitis. Additionally, the CXCR3-specific inhibitor AMG487 confers significant protection against MASH\u0026nbsp;\u003csup\u003e12\u003c/sup\u003e. Another study demonstrated that \u003cem\u003eNfil3\u003c/em\u003e\u003csup\u003e⁻/⁻\u0026nbsp;\u003c/sup\u003emice (also known as \u003cem\u003eE4bp4\u003c/em\u003e\u003csup\u003e⁻/⁻\u003c/sup\u003e, which exhibit defective NK cell development) confer significant resistance to CDHFD diet-induced MASH progression\u0026nbsp;\u003csup\u003e13\u003c/sup\u003e. Collectively, these findings identify NK cells as critically involved in driving the pathogenesis and progression of MASH. Thus, we hypothesize that Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells contribute to the pathogenesis of MASH. However, the phenotype of Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells, how they contribute to the pathogenesis of MASH, and which hepatic cells regulate them are still unclear.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; In this study, we found NK cells with tissue-resident properties (\u003cem\u003eIL21R\u003c/em\u003e, \u003cem\u003eRGS1\u003c/em\u003e) and effector functions (\u003cem\u003eGZMK\u003c/em\u003e) aggregated in the hepatic tissue of MASH mice. The significant feature of these CXCR3\u003csup\u003e+\u0026nbsp;\u003c/sup\u003eNK cells is reduced LEF1 transcription factor activity, and they are abundant in the hepatic tissue of both NASH patients and MASH mice. Molecular mechanism studies have shown that IL-21 induces downregulation of LEF1 and upregulation of CXCR3, significantly increasing the hepatic accumulation of CXCR3\u003csup\u003e+\u003c/sup\u003e LEF1\u003csup\u003elow\u003c/sup\u003e NK cells, when exposed to relevant metabolic stimuli such as Acetate and CXCL9, contributed to auto-destructive hepatocyte killing. Additionally, our results also showed that, the palmitic acid levels are increased in the livers of both NASH patients and MASH mice. Palmitate accumulation promotes hepatic endothelial cells to secrete Cxcl9 by facilitating IRF4 activation, and recruiting and activating Cxcr3\u003csup\u003e+\u0026nbsp;\u003c/sup\u003eLef1\u003csup\u003elow\u003c/sup\u003e NK cells in MHC class I-independent. In summary, our research will contribute to a comprehensive understanding of how immune cells contribute to the pathogenesis of MASH, highlighting the potential clinical value of targeting specific NK cell subpopulations to develop innovative drugs or therapeutic approaches for MASH.\u003c/p\u003e"},{"header":"Methods and Materials","content":"\u003cp\u003e\u003cstrong\u003e1. MASH mouse model and treatments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eC57BL/6J mice were obtained from GemPharmatech (Nanjing, China). Both Gimap4\u003csup\u003e-/-\u003c/sup\u003e and Irf4\u003csup\u003e-/-\u003c/sup\u003e mice on the C57BL/6J background were constructed by Cyagen Biosciences (Suzhou, China) using the CRISPR Cas9 method. All mice were housed under specific pathogen-free (SPF) conditions in compliance with the Association for Research Animal Science guidelines. The mice were raised in 12 hours light/dark cycle environment, the ambient temperature is 24°C, and the humidity is 55%. All mice were fed ad libitum. Mice were fed group-specifically according to experimental plan. Normal diet (ND; Cat. No. D12450J, 70% kcal carbohydrates, 20% kcal protein, 10% kcal fat, Research Diets), choline-deficient high-fat diet (CDHFD; Cat. No. D05010402; 60% kcal fat, 20% kcal carbohydrates, 20% kcal (choline-free casein) protein, Research Diets), high-fat high-cholesterol diet (HFHC diet; Cat. No. D09100301; 40 kcal% fat, 20 kcal% fructose, and 2% cholesterol, Research Diets), western diet (WD; Cat. No. D12079B; 41% kcal fat, 30 kcal% fructose, and no cholesterol, Research Diets), high-fat diet (HFD; Cat. No. D12492; 60% kcal fat, 20% kcal protein, and 20% kcal carbohydrates, Research Diets) were used to establish the mouse model of MASH. After being fed a CDHFD, mice were randomly divided into groups and treated with specified blocking antibody (100 μg/mouse) via intraperitoneal injection: Anti-CD122 antibody (BioXcell, Cat. No.\u0026nbsp;BE0272, clone number 5H4) or Anti-LFA1 antibody (BioXcell,\u0026nbsp;Cat. No.\u0026nbsp;BE0001-1, clone number M17/4). Additional details on monoclonal antibodies are provided in the Supplementary Table. Mice received intraperitoneal injections once every 3 days (4 injections total). Upon experiment termination, mice were euthanized, and tissue and serum samples were collected for subsequent analysis. The Animal Ethics Committee of China Pharmaceutical University approved all mouse experiments (Permit number: SYXK-2021-0011, 25 January 2021).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.\u003c/strong\u003e \u003cstrong\u003eHuman sample research\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA total of 18 participants from the Nanjing University Drum Tower Hospital fatty liver disease prospective observational cohort were included in the present analysis, comprising 12 biopsy-confirmed MASH patients and 6 biopsy-confirmed controls with near-normal hepatic histology. From 2020 to 2025, we collected liver tissue samples from MASH patients and performed pathological sectioning as well as serum biochemical marker and physical parameter testing. A total of 12 MASH patients satisfied the predefined inclusion criteria were: (1) age between 38 and 80 years; (2) biopsy-confirmed MASH diagnosis per current clinical guidelines; (3) absence of other liver diseases and excessive alcohol consumption (≤20 g/day for women, ≤30 g/day for men). Six biopsy-confirmed individuals from the same cohort were enrolled into the non-MASH control group, exhibiting essentially normal liver status. All biopsy results were independently assessed by two hepatopathologists. The NAS Score and fibrosis stage were determined according to the NASH Clinical Research Network scoring system. This study was conducted with the approval of the Ethics Committee of Nanjing University Drum Tower Hospital and with the written informed consent of all participants.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eI\u003c/strong\u003e\u003cstrong\u003esolation of primary mouse lymphocyte\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSpleen lymphocyte isolation:\u0026nbsp;\u003c/strong\u003eFollowing euthanasia, spleens were aseptically excised in a laminar flow hood, placed in cold PBS, and cleared of surrounding adipose tissue and blood. The spleen was placed into a centrifuge tube containing 5 mL of cold PBS and mechanically disrupted. Then, the tissue suspension was transferred into a digestion solution containing collagenase IV and DNase I, followed by incubation at 37°C for 30 minutes with constant agitation. Add serum-containing medium to stop digestion. Filter sequentially through 70 μm and 40 μm cell sieves and collect the filtrate. Centrifuge (4°C, 300 ×g, 5 min) and discard supernatant. Add 3 mL ACK lysis buffer and incubate at room temperature for 2 minutes. Immediately after incubation, add 10 mL cold PBS to stop the reaction and centrifuge (300 ×g, 5 min). Resuspend the cells in 3 mL of cold PBS. Add 3 mL of lymphocyte separation medium to a 15 mL centrifuge tube and slowly add the cell suspension to the upper layer of the separation medium (keep the interface clear). Centrifuge (4°C, 800 ×g, 30 min, acceleration and deceleration set to 0). Transfer the lymphocyte layer to a new tube. Add 10 mL cold PBS to wash the lymphocytes (4°C, 300 ×g, 5 min, repeat twice).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLiver immune cell isolation:\u0026nbsp;\u003c/strong\u003eFollowing 1-minute perfusion via the portal vein. To obtain a single-cell suspensions, the resulting cells suspension was subsequently passed through the cell strainer (Corning;\u0026nbsp;Cat. No. 352350; 70 μm). After washing, the cells pellet was subjected to digestion in 6 mL of GBSS supplemented with collagenase IV (MedChemExpress; Cat. No. HY-K0012; 1:500 dilution) and digested at 37°C for 15 minutes under constant agitation. The resulting cell suspension was subjected to Percoll density gradient centrifugation (GE Healthcare,\u0026nbsp;Cat. No.\u0026nbsp;10607095; 40%\u003cstrong\u003e/\u003c/strong\u003e80% layers) using predefined parameters (acceleration rate 7, deceleration rate 0; 1,440 ×g, 25 min) The lymphocyte-enriched interphase was carefully collected and subjected to two washes with 8 mL cold PBS (5 min each at 300 ×g).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4. Isolation of primary mouse hepatocytes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFollowing anesthesia with sodium pentobarbital (50 mg/kg, i.p.), mice were immobilized on a 37°C thermostatic plate. The abdominal skin was disinfected with 75% ethanol. Sequential portal vein perfusion was performed using Pre-perfusion solution (HBSS (Ca²⁺/Mg²⁺-free) + 500 μM EGTA + 15 mM HEPES (pH 7.4), preheated to 37°C) and digestion solution (HBSS (with Ca²⁺/Mg²⁺) + 0.05% type IV collagenase (Sigma,\u0026nbsp;Cat. No.C5138) + 15 mM HEPES + 0.1% BSA, pre-heated to 37°C). Following perfusion, the liver was excised into a pre-chilled culture dish. After removal of Glisson's capsule, hepatocytes were released by gentle teasing with forceps into cold Williams' E medium supplemented with 10% FBS(Gibco, Cat. No. 26140079 ). Hepatocytes were isolated by filtering the suspension through a 70 μm mesh, followed by three washing/centrifugation cycles (50 ×g, 4°C, 3 min). The cells were then purified using 40% Percoll density gradient centrifugation at 100 ×g for 15 min. The high-viability hepatocyte fraction sedimented at the bottom was collected.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e5. Isolation of mouse hepatic Cxcr3\u003csup\u003e+\u0026nbsp;\u003c/sup\u003eNK cells\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe same liver perfusion method used for mouse primary hepatocyte isolation was employed, with the digestion solution substituted with 0.02% type IV collagenase (Sigma,\u0026nbsp;Cat. No.C5138). The perfusion time was reduced to 5 minutes to yield a liver cell suspension. Following suspension centrifugation (50 ×g, 3 min, 4°C), the supernatant containing liver non-parenchymal cells (NPCs) was transferred to fresh tubes. Subsequently, the cell supernatant was centrifuged (300 ×g, 15 min, 4°C) to pellet the NPCs. After 1-minute incubation in red blood cell lysis buffer, the cells underwent two washing cycles with cold PBS. The obtained NPCs were subjected to magnetic bead-based negative selection commercial Mouse NK Cell Isolation Kit (Miltenyi Biotec,\u0026nbsp;Cat. No.130-115-818) to isolate mouse hepatic NK cells. Antibody staining was subsequently initiated with Fc receptor blockade using anti-CD16/32 antibody (clone 2.4G2) at 1:100 dilution for 10 minutes. Surface staining: CD49b-APC, CD3ε-FITC (to exclude T cells), Cxcr3-PE. Incubate on ice under light-protected conditions for 20 minutes, then wash with cold PBS. Use the 70 μm nozzle to collect cells at low temperature (4°C) into a collection tube containing RPMI-1640 medium with 50% FBS. The isolated Cxcr3⁺ NK cells were maintained in RPMI-1640 medium with 10% FBS, 10 ng/ml IL-15 and 1000 U/ml IL-2.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e6. Isolation of NK cells from human peripheral blood\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePeripheral blood was collected from donors in sodium heparin tubes and diluted 1:1 with PBS. The diluted blood was carefully layered over Ficoll-Paque™ solution and subjected to density gradient centrifugation at 400 ×g, 25°C for 30 minutes to obtain peripheral blood mononuclear cell (PBMC). The PBMC layer was collected, and NK cells were isolated by negative selection commercial Human NK cell isolation kit (Miltenyi Biotec, Cat. No. 130-092-657) according to the manufacturer's protocol.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e7. Flow cytometric analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCell surface receptor immunolabeling was performed using fluorescently labeled antibodies in PBS supplemented with 2% FCS (Merck) at 4 °C for 30 minutes. The Transcription Factor Staining Buffer Set (ThermoFisher, Cat. No. 00-5523-00) was utilized for intracellular cytokine staining following the manufacturer's guidelines. Briefly, liver-associate lymphocytes were stimulated in the presence of protein transport inhibitors (Brefeldin A and Monensin, both 1:1,000 from ThermoFisher) for 6 hours. After staining with a fixable viability dye (Thermo Fisher Scientific; eFluor™ 780; 1:3,000) and surface antibodies, cells were processed for fixation, permeabilization, and final intracellular staining against TGF-β and GzmK. For intracellular transcription factor staining, the Foxp3 Transcription Factor Staining Buffer Kit (ThermoFisher, Cat. No. 00-5523-00) was used according to the manufacturer's instructions.\u0026nbsp;For LEF1 staining, incubation with LEF1 primary antibody (CST; Cat. No. 2230; 1:500) was performed overnight (approximately 16 hours) at 4°C, followed by a 2-hour co-incubation at room temperature with an Alexa Fluor 647- (ThermoFisher; Cat. No. A-21244;1:500) or Alexa Fluor 488- (Cell Signaling; Cat. No. 4412s; 1:500) conjugated goat anti-rabbit secondary antibody. Flow cytometry was performed using a CytoFLEX LX standard flow cytometer (Cat. No.\u0026nbsp;A94649AK). Cell sorting was performed using the MoFlo Astrios™ EQ Basic Research Flow Cytometer (Cat. No. A95728AK-CN-TC) with a standard 70 μm nozzle. Data were analyzed using FlowJo software (Tree Star, version.10.8). For antibodies used in flow cytometry analysis, please refer to the attached table.\u0026nbsp;The intracellular calcium ion concentration in NK cells was quantified using the eFluor™ 514 calcium dye (2 mM, ThermoFisher, Cat. No. 65-0859-39) by flow cytometry, and in combination with fluorescently conjugated antibodies.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e8. Stimulation of hepatic CXCR3\u003csup\u003e+\u003c/sup\u003e NK cells in vitro\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMouse hepatic Cxcr3⁺ NK cells were enriched by fluorescence-activated cell sorting (FACS) and resuspended at 1×10⁶ cells/mL. Then, the cells (200 μL per well) were plated into 24-well plates and activated for 2 days in RPMI-1640 medium supplemented with 10% FCS, 100 U/mL penicillin-streptomycin, 0.1 mM β-mercaptoethanol, 2 mM L-glutamine, 10 ng/ml IL-15 and 100 U/mL IL-2. Subsequently,\u0026nbsp;cells were transferred to fresh wells, adjusted to 1×10⁶ cells/mL, and treated with TGF-β1 (5 ng/mL, known to promote NK cell tissue residency [14]) for an additional 2 days. Following density gradient centrifugation using Pancoll (1440 ×g, 25 min), viable cells were collected from the interface. For enrichment of CXCR3⁺LEF1\u003csup\u003elow\u003c/sup\u003e NK cells, activated NK cells were treated for 24 h with a combination of IL-21 (10 ng/mL, Thermo Fisher Scientific, Cat. No. PMC0214) and TGF-β1 (5 ng/mL, R\u0026amp;D Systems, Cat. No. 7666-MB).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e9. NK cell cytotoxicity assay\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe used impedance-based xCelligence RTCA MP technology (ACEA Biosciences) to measure the cytotoxic kinetics of NK cells against primary mouse or primary human liver cells over time. In brief, primary hepatocytes were isolated and plated into Xcelligence plates at the indicated density (1×10⁴ cells/well). After 24 hours, when the hepatocytes had adhered properly, 3×10⁴ stimulated Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells were added, and changes in cell impedance were monitored in real time. Blocking antibodies, Acetate, specific inhibitors, or recombinant cytokines were added to the experiment. The vehicle control group (hepatocytes alone) was designated to establish baseline measurements. The electrical\u0026nbsp;impedance was recorded, converted to the cell index value, and normalized by referencing to the time point of NK cell-hepatocyte co-culture initiation.\u0026nbsp;At 24 hours post‑initiation of co‑culture, the cytotoxic efficacy of NK cells against hepatocytes was assessed by quantifying the cell index. The percentage of primary hepatocytes death was determined by normalizing the difference in cell index between experimental and control groups.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e10. Adoptive transfer of hepatic auto-destructiveness NK cells into C57BL/6 mice\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn this study, CXCR3⁺ NK cells isolated from MASH model mice were stimulated in vitro (as described in \"Stimulation of hepatic CXCR3⁺ NK cells in vitro\") with IL-21 (10 ng/mL) or the LEF1 degrader OP-V1 [15], followed by 24-hour treatment with acetate (15 mM, Merck). The NK cells were then washed with cold PBS and adoptively transferred (1×10⁶ cells/mouse) into the livers of C57BL/6 wild-type (WT) or IRF4 KO mice. Based on experimental groups, mice received intrahepatic injections of anti-LFA1 antibody (100 μg/mouse, BioXcell, clone M17/4,\u0026nbsp;Cat. No. BE0006), anti-FasL antibody (100 μg/mouse, BioXcell, clone MFL3, Cat. No. BE0319), or isotype IgG. Serum alanine aminotransferase (sALT) levels were measured 36 hours post-transfer.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e11. Lentiviral transduction and overexpression experiments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe\u0026nbsp;pLVX-EF1α-T2A-GFP vector was double-digested with restriction enzymes EcoRI and XhoI, using the following primer sequences: Lef1-F (5’→3’)：**GAATTC**GCCACCATGGCGCCGAGCTGG (Contains EcoRI restriction site + Kozak sequence GCCACC + Lef1 start codon ATG); Lef-R: TTCAGCTCCTCCACCTTGTCGTC **GGTGGTGGTGGT**TCAGCCGCTGC (Contains flexible joint GGTGGTGGT (Gly-Gly-Gly) + removal of stop codon); GFP-F: GGTGGTGGTGGT**GAGCAAGGGCGAGGAG (Flexible joint extension + GFP start sequence, overlapping with Lef1-R); GFP-R: **CTCGAG**TTACTTGTACAGCTCGTCCATGCC (Contains XhoI restriction site + GFP stop codon TTA)，The pLVX-EF1α-LEF1-T2A-GFP plasmid was generated using Gibson Assembly® seamless cloning. Mixed the pLVX-EF1α-LEF1-T2A-GFP plasmid and lentiviral packaging plasmids in the following mass ratio (total DNA 20 μg): pLVX-EF1α-LEF1-T2A-GFP: 10 μg; pMDLg/pRRE: 7.5 μg, pMD2.G: 5 μg, pRSV-Rev: 2.5 μg. The plasmid mixture and Lipo 3000 transfection reagent were diluted in 300 μL Opti-MEM. After incubation at room temperature for 15 minutes, the prepared complex was added to the HEK293T cells. Following collection at 48- and 72-hour time points post-transfection, the supernatant was passed through a 0.45 μm filter. Subsequent concentration of the lentivirus was achieved by ultracentrifugation (70,000 ×g, 2 h, 4°C). Infected HEK293T cells with virus solutions diluted with gradient. After 72 hours of infection, GFP⁺ cells percentage was analyzed by flow cytometry. Virus solution concentrations with a GFP\u003csup\u003e+\u0026nbsp;\u003c/sup\u003ecells proportion greater than 90% were used to infect NK cells.\u003c/p\u003e\n\u003cp\u003eAfter infected with lentivirus, NK cells were maintained in culture for 8 hours and then refreshed with RPMI-1640 medium containing IL-2 (300 U/mL) and IL-15\u0026nbsp;(10 ng/mL). 48 hours later, the expression of CXCR3, LEF1, and GZMK in GFP\u003csup\u003e+\u003c/sup\u003e NK cells were measured to distinguish NK cells overexpressing LEF1. For NK cell cytotoxicity assays using the Xcelligence device, GFP⁺ NK cells, sorted 24 hours post-lentiviral transduction, were stimulated\u0026nbsp;with 10 ng/ml IL-21 for 24 hours, exposed to 15 mM acetate for 24 hours, then co-cultured with primary mouse hepatocytes at a 3:1 ratio of effectors to targets. NK cell-mediated damage to primary hepatocytes after 24 hours of co-culture was assessed per the protocol detailed in the “NK cell cytotoxicity assay” section.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e12. shRNA interference experiment\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFor shRNA target design and vector construction, used the pLKO.1-TRC-Puro vector (Addgene,\u0026nbsp;Cat. No.10878), three targets were designed for STAT1 and IRF4. Specifically, S1: CCACAGATGTGCAAGTCCAAT; S2: GCAGATTTCTCCAGACTTATT; S3: GCTGGAGAATTACCTGGAAAT targeting STAT1. I1: GCTGAAGACCTGATCCAGAAT; I2: GCAGATCCTCAACTTCAAGAA; I3: GCCATTACAGACCTGATTCTT targeting IRF4. shRNA oligos containing AgeI/EcoRI restriction sites were synthesized, annealed into double-stranded DNA, and ligated into AgeI/EcoRI-linearized pLKO.1-TRC-Puro vector, following transformation of Stbl3 competent cells with the ligation products, positive clones were subjected to selection and subsequent sequencing validation. Lentivirus was packaged as described in the “Lentivirus infection and overexpression experiment.” After infecting primary liver endothelial cells with lentivirus for 24 hours, treaded with 0.5 μg/mL puromycin (ThermoFisher, Cat. No. A1113803) and screen for 5 days. Then proceed with subsequent experiments.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e13. Confocal microscopy of primary cells\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePrimary hepatocytes (3×10⁴ cells) were plated on collagen-coated chamber slides (Ibidi, Cat. No. 81201). Prior to co-culture with 1×10⁵ CXCR3⁺ NK cells, the complete medium was supplemented with TGF-β1 (5 ng/mL), IL-21 (10 ng/mL), and sodium acetate (15 mM). After 24 hours, cells underwent fixation in 4% paraformaldehyde. washed with PBS, and stained using anti-ICAM1-Alexa Fluor® 488 (1:200, Abcam,\u0026nbsp;Cat. No. ab 2210), anti-LFA1-PE (1:200, Abcam, Cat. No. ab 33477), and DAPI with antifade mounting medium (1 μg/mL, Beyotime, Cat. No. P0131). Stained samples were mounted in neutral resin and imaged using a Zeiss LSM 800 Airyscan confocal microscope with 20× and 40× objectives.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e14. Histology, immunohistochemistry, scanning, and automated analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMouse liver tissue was processed using standardized procedures, including fixation, dehydration, embedding, and sectioning (Leica RM2235 rotary microtome). Histological staining was performed according to the manufacturers' protocols using the following kits: Hematoxylin and Eosin (H\u0026amp;E) Staining Kit (Beyotime,\u0026nbsp;Cat. No. C0105M); Masson's Trichrome Staining Kit (Solarbio, Cat. No. G1340). Immunohistochemistry was performed according to the method provided on the Thermo Fisher Scientific website (https://www.thermofisher.com). Stained sections were mounted with neutral balsam and scanned using a NanoZoomer S60 (Hamamatsu, Cat. No. C13210-01, 20× objective, 0.46 μm/pixel resolution).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e15. Hepatic acetate level measurement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMouse and human liver tissues were weighed, washed with cold physiological saline, and homogenized in 50 μL cold PBS at 4°C for 60 seconds using a tissue homogenizer. The homogenates were subjected to centrifugation at 300 ×g for 10 minutes. Acetate concentrations in the collected supernatants were measured according to the instructions of the commercial assay kit (Acetate Assay Kit, Merck, Cat. No. MAK086). Acetate concentrations in liver were normalized to tissue weight.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e16. Serum ALT level measurement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBlood samples were allowed to clot at room temperature for 30 minutes and were subsequently centrifuged at 2000 ×g for 12 minutes at 4°C to obtain serum. The ThermoFisher Alanine Aminotransferase (ALT/GPT) Activity Kit (ThermoFisher,\u0026nbsp;Cat. No. MAK052) was then used to test and calculate the sALT levels of each sample according to the experimental procedure described in the manufacturer’s instructions. ALT activity (IU/L) = [a× (OD\u003csub\u003esample\u003c/sub\u003e-OD\u003csub\u003econtrol\u003c/sub\u003e)2+b×(OD\u003csub\u003esample\u003c/sub\u003e-ODc\u003csub\u003eontrol\u003c/sub\u003e)+c]×0.482 IU/L×f. a, b, c: the constant of standard curve. f: Dilution factor of sample before tested.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e17. Serum Cxcl9 level measurement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn this study, serum Cxcl9 levels in mice were quantified using an R\u0026amp;D Systems Mouse Cxcl9 ELISA Kit (Cat. No. DY492). Procedures strictly followed the manufacturer’s protocols.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e18. Measurement of IL-21 levels\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn this study, serum IL-21 levels in MASH model mice and human samples were quantified using enzyme-linked immunosorbent assay (ELISA) kits. Mouse serum IL-21 levels was measured with the R\u0026amp;D Systems Mouse IL-21 ELISA Kit (Cat. No. DY594), and human serum IL-21 levels was assessed using the Thermo Fisher Scientific Human IL-21 ELISA Kit (Cat. No. BMS221). All procedures strictly followed the manufacturer’s protocols.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e19. Single-cell transcriptomics sequencing of mouse livers\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSingle-cell RNA sequencing was performed on hepatic tissues from both healthy and MASH mice using the 10x Genomics Chromium platform. Sequencing libraries were processed with CellRanger (v4.0.5) for demultiplexing, read alignment to the GRCm38 reference genome, and initial feature-barcode matrix generation. Subsequent processing and quality control were conducted in R (v4.0.5) using the Seurat package (v5.0.4). Low-quality cells were filtered out based on the following criteria: nFeature_RNA between 300 and 7000, nCount_RNA \u0026gt; 1000 (excluding the top 3% of cells with highest UMI counts), mitochondrial gene percentage (mt_percent) \u0026lt; 10%, and hemoglobin gene percentage (HB_percent) \u0026lt; 3%.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e20. UMAP dimension reduction and cell annotation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData integration and dimensionality reduction were conducted with Seurat (version 5.0.4). The selection of features for this process was carried out using the package's FindVariableFeatures and SelectIntegrationFeatures functions. Then, we used the FindIntegrationAnchors function to identify anchor points in the data files for data integration. Finally, we used the IntegrateData function to integrate the four sets of data files. We used the NormalizeData and ScaleData functions to standardize each data matrix file. We performed dimensionality reduction on the integrated data and identified cell subpopulations using the FindClusters function. We used the Unitary Manifold Approximation and Projection (UMAP) method as a visualization method for cell clustering. By identifying cell subpopulations, we identified a total of nine cell subpopulations. Using Single R (version. 2.2.0) in combination with the annotation method from the Panglao DB\u0026nbsp;database\u0026nbsp;\u003csup\u003e16\u003c/sup\u003e, we annotated 7 distinct cell types (B cells, T/NK cells, Endothelial cells, Hepatocytes, Fibroblasts, Macrophages, Neutrophils). The FindAllMarkers function was used for dual purposes: to define conserved gene markers characteristic of each cell subpopulation and to identify differentially expressed genes (DEGs) through comparative analysis across groups or subpopulations.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e21. Single-cell sequencing intercellular communication analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe R package CellChat (version 1.7.2) was employed to infer and analyze the intercellular communication network among defined cell subpopulations. First, we extracted data from each group based on orig.ident, created CellChat objects, performed receptor analysis of highly expressed genes based on CellChatDB.mouse (the pathway database provided by CellChat), and projected the genes onto the PPI (protein-protein interaction) database. We filtered out intercellular communication networks with fewer than 50 cells. Then, a standardized analytical pipeline with consistent filtering criteria was applied to interrogate cell-cell communication networks among subpopulations within each individual sample.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e22. Single-cell sequencing pseudotime analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe preprocessed scRNA-seq expression matrix (rows = genes, columns = cells) and associated metadata (cell type, sample groups) of liver endothelial cell subpopulations were imported into Monocle2 to construct a CellDataSet object. Sequencing depth variation was corrected using the negative binomial distribution model (negbinomial. size ()), and gene expression dispersion was estimated (estimateDispersions ()). The top 2,000 highly variable genes were selected for trajectory construction. Nonlinear dimensionality reduction was performed using the DDRTree algorithm (reduceDimension ()). Pseudotime trajectories were rooted at the ND12-month-CDHFD group with State 1 designated as the developmental origin. A 2D trajectory plot of cellular development was generated.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e23. GSEA\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe retrieved gene sets from the GEO database, including the NK cell effector gene set, the GPCR signaling pathway gene set, the key transcription factor signature gene set for NK cell activation, and the apoptosis signature gene set, among others. Among these, GSE54215 (BATF knockout group vs. wild-type group (day 3)), GSE122931 (TBX21 knockout group vs. wild-type group), GSE215188 (Lef1 knockout group vs. wild-type group), and Genes were considered differentially expressed if they showed |log2FC| \u0026gt; 1 and an adjusted p-value (FDR) \u0026lt; 0.05, as identified by the DESeq2 package in R. These gene sets were used to perform GSEA analysis (using GSEA v.3.0 software) on the Log2-transformed fold changes obtained from RNA-seq data of differentially expressed genes (MASH mouse liver CXCR3\u003csup\u003e+\u0026nbsp;\u003c/sup\u003eNK cells vs. normal diet mouse liver CXCR3\u003csup\u003e+\u003c/sup\u003e NK cells, and MASH mouse liver CXCR3\u003csup\u003e+\u003c/sup\u003e NK cells vs. CXCR6\u003csup\u003e-\u003c/sup\u003e NK cells). To assess the normalized enrichment score (NES), gene set fold change data were submitted to the PreRanked analysis module in GSEA v3.0 software. Significance of enrichment was assessed based on an FDR threshold of q ≤ 0.25.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e24. Transcription factors identification and network analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe transcription factors regulating CXCR3\u003csup\u003e+\u0026nbsp;\u003c/sup\u003eNK cells activation was analyzed using differentially expressed genes from RNA-seq datasets of CXCR3\u003csup\u003e+\u0026nbsp;\u003c/sup\u003e\u003cem\u003evs.\u003c/em\u003e CXCR3\u003csup\u003e-\u0026nbsp;\u003c/sup\u003eNK cells from MASH mouse livers. To infer upstream transcriptional regulators of the differentially expressed genes, analysis was performed with the “Transcription Regulation Binding Analysis” module of BART (version 1.1) using its default parameters. Our analysis revealed 25 key transcription factors from the MASH mouse liver RNA-seq data (CXCR3⁺ vs. CXCR3⁻ NK cells), applying a significance threshold of P \u0026lt; 0.01. First, we downloaded the promoter sequences (approximately 2 kb) of significantly differentially expressed genes from the eukaryotic promoter database. Second, we retrieved transcription factor binding sites from the JASPAR core and Hocomoco databases. Then, we used a custom Python script to scan the promoter sequences (-2 kb promoters) and transcription factors at the binding sites of differentially expressed genes. We constructed a transcription factor (TF) interaction network using Cytoscape software (version 3.7.1). To assess the hierarchical structure of the transcription factor network, we calculated the weight of each transcription factor's role performed utilizing the Igraph R package (version.1.2.6).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e25. Quantitative real-time PCR\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTotal mRNA was extracted from cells or liver tissue samples using Trizol reagent. Subsequently, 1 μg of total mRNA was reverse-transcribed into cDNA using the PrimeScript™ RT reagent Kit with gDNA Eraser (Perfect Real Time) (Takara,\u0026nbsp;Cat. No. RR047A/B). For cDNA amplification, SYBR Green (Vazyme, Cat. No. Q711-02) and the QuantStudio™ 1 real-time PCR system (Thermo Fisher Scientific, Cat. No. A43179) were employed to detect the total product amount after each PCR cycle and to calculate the CT values. Mouse and human 18S rRNA or HPRT were used as housekeeping genes in the experiments. The complete list of primer sequences for all genes is provided in Data S4.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e26. RNA-seq analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRNA sequencing of MASH mouse liver Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells \u003cem\u003evs\u003c/em\u003e. normal diet mouse liver Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells, and MASH mouse liver Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells \u003cem\u003evs\u003c/em\u003e. CXCR3\u003csup\u003e-\u003c/sup\u003e NK cells were performed using the Illumina platform. The raw sequencing reads were processed through FastQC for quality control and aligned using Trinity. Subsequently, differential expression analysis was performed with DESeq2, considering genes with |log2FC| \u0026gt; 1 and FDR \u0026lt; 0.05 as statistically significant. All other RNA-seq analyses in this study followed identical procedures for quality control, alignment, and differential expression analysis. The specific grouping information for these additional analyses is detailed in the Results section.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e27. Western blot\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTotal protein was extracted using RIPA buffer. Following lysis, samples were denatured in 6× loading buffer by boiling at 100°C for 8 min and resolved on 12% SDS-PAGE gels. Proteins were then electroblotted onto PVDF membranes (Millipore, IPVH00010). After blocking with 5% non-fat milk (1 h, RT), membranes were probed with primary antibodies overnight at 4°C, followed by incubation with species-matched secondary antibodies (1 h, RT). Signal detection was performed with an ECL substrate (A+B reagents) and darkroom development. Band intensities were quantified via grayscale analysis in ImageJ (NIH) and statistically analyzed with GraphPad Prism.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e28. Co-immunoprecipitation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFor Gimap4 and PLC-γ1 interaction analysis, transfected HEK293T cells with Lipo3000 transfection reagent, co-transfecting the pcDNA3.1-Gimap4-Flag plasmid and the pcDNA3.1-PLCγ1-Myc plasmid (5 µg each), After 48 hours, lyse the cells using IP buffer (Beyotime Biotechnology,\u0026nbsp;Cat. No. P0013), centrifuge at 12,000 rpm for 10 minutes, then add 20 μL of magnetic beads to the supernatant and pre-clear for 2 hours. Divide the mixture into two equal portions, add 2 µg of Flag antibody and isotype IgG to each portion, and incubate at 4°C overnight (16 hours). The next day. Then, 30 µL of Protein A/G magnetic beads (Thermo Fisher, Cat. No.88802) that had been pre-equilibrated were added to the mixture. Incubate at 4°C with gentle rotation for 6 hours. Then, perform Western blot analysis to detect Flag and Myc tags. To detect the direct interaction between Gimap4 and p-PLC-γ1 in Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells, we lysed the cells using IP buffer, then performed IP using the Gimap4 antibody and Western blotting to recognize the p-PLC-γ1 protein.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e29. Statistical analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData are presented as the mean ± SEM from at least three independent experiments. Statistical analyses were performed using GraphPad Prism software (version 9.0). Significance between two groups was assessed by unpaired two-tailed Student’s t-test; comparisons among multiple groups were analyzed by one-way or two-way ANOVA followed by Tukey’s or Sidak’s multiple comparisons test, as appropriate. A \u003cem\u003eP\u003c/em\u003e value of less than 0.05 was considered statistically significant (\u003cem\u003e*P\u003c/em\u003e \u0026lt; 0.05, \u003cem\u003e**P\u003c/em\u003e \u0026lt; 0.01, \u003cem\u003e***P\u003c/em\u003e \u0026lt; 0.001; ns, not significant).\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003e1. Significant alterations in hepatic cell populations in the development of MASH\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOver 70% of MASH patients are associated with moderate or severe obesity. To investigate the novel mechanisms of obesity-associated MASH, C57BL/6J mice were subjected to a CDHFD for 4 months, 8 months and 12 months, respectively, to establish early-, mid-, and late-stage MASH mouse models (Fig1A). Histopathological examination confirmed that hepatic lipid accumulation and the areas of fibrosis in MASH mice live significantly increased with the duration of CDHFD feeding (Fig 1B). Compared with health mice, MASH mice exhibited significantly increased serum ALT levels, MASH activity scores, hepatic TG content, and liver \u0026alpha;-SMA expression (Fig 1C-F). We then performed single-cell RNA sequencing (scRNA-seq) on liver tissues from ND- and CDHFD-fed mice, followed by UMAP dimensionality reduction and clustering analysis. we classified mouse liver cells into seven subpopulations by identifying characteristic genes: B cells, Endothelial cells, Hepatocytes, Macrophages, Fibroblasts, Neutrophils, T/NK cells (Fig 1G, Fig S1). Furthermore, we analyzed changes in the cellular composition across the different groups (Fig 1H). During MASH progression, T/NK cells accumulated significantly in the liver, whereas hepatocytes and endothelial cells declined markedly (Fig 1I). We propose that the accumulated T/NK cells in the liver may participate in critical cell-cell communication. Therefore, we employed the CellChat R package to analyze intercellular crosstalk in the liver. The results revealed a significantly enhanced crosstalk between hepatic endothelial cells and T/NK cells with prolonged CDHFD feeding duration (Fig 1J). The Cxcl9-Cxcr3 receptor pair and the App-CD74 receptor pair primarily mediate enhanced cross-talk between hepatic endothelial cell populations and T/NK cell populations (Fig 1K-L, Fig S2). Cxcl9 is highly expressed specifically in the hepatic endothelial cell population, while Cxcr3 is highly expressed specifically in T/NK cell populations (Fig 1M, Fig S1G-H).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.\u003c/strong\u003e \u003cstrong\u003eCxcr3\u003csup\u003e+\u003c/sup\u003eGZMK\u003csup\u003ehigh\u003c/sup\u003e NK cells are accumulate in MASH livers\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFlow cytometric analysis revealed a marked expansion of hepatic CD8⁺ T cell and NK cell populations during MASH progression (Fig S4A). The frequency of Cxcr3\u003csup\u003e+\u003c/sup\u003e cells was significantly elevated within the hepatic CD45\u003csup\u003e+\u003c/sup\u003e cell population, while no significant change occurred in the kidneys (Fig 2A-C). Therefore, we further investigated the compositional changes in CD8\u003csup\u003e+\u003c/sup\u003e T cells, NK cells, CD4\u003csup\u003e+\u003c/sup\u003e T cells and NKT cells within hepatic Cxcr3\u003csup\u003e+\u0026nbsp;\u003c/sup\u003eCD45\u003csup\u003e+\u003c/sup\u003e cells after 8 months of CDHFD feeding. We found that only Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells were significantly accumulated in the hepatic tissue following the CDHFD diet (Fig 2D). Genome-wide transcriptomic profiling revealed distinct transcriptomic profiles between hepatic Cxcr3⁺ and Cxcr3⁻ NK cells from MASH mice (Fig 2E). In Cxcr3⁺ NK cells isolated from the livers of MASH mice, we observed upregulated expression of genes associated with effector function (Gzmk, Bhlhe40), exhaustion markers (Pdcd1, Tox), as well as key signature genes related to NK cell tissue residency (IL21R, Klf2, and Rgs1) (Fig. 1E). Genetic set enrichment analysis (GSEA) further confirmed that Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells enriched genes associated with NK cell effector functions, activation, and tissue residency, but did not enrich Th17 cells or IL-17-dependent genes (Fig. 2F)-The latter has been previously demonstrated to be associated with immunopathology in infection[17]. Flow cytometry also revealed that the number of Gzmk\u003csup\u003e+\u003c/sup\u003e, IFN-\u0026gamma;\u003csup\u003e+\u003c/sup\u003e, IL21R\u003csup\u003e+\u003c/sup\u003e, Ki67\u003csup\u003e+\u003c/sup\u003e, and Bcl2\u003csup\u003e+\u003c/sup\u003e cells among Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells in the livers of MASH mice were significantly increased (Fig 2G). Collectively, the hepatic Cxcr3⁺ NK cells in MASH were characterized as long-term liver residential effector cells that simultaneously integrate activation and modulatory signals\u003c/p\u003e\n\u003cp\u003eNext, we investigated the relationship between Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells and MASH development. The frequency of hepatic Cxcr3⁺ NK cells showed a significant positive correlation with serum ALT levels in each stage of MASH mice (Fig 2H). Analysis also revealed a significant positive correlation between hepatic Cxcl9 levels and the Cxcr3⁺ NK cell proportion (Fig 2I). To bridge the gap between preclinical research and human pathophysiology, we collected 18 clinical NASH liver samples. Similarly, the same phenotype was observed in liver samples from NASH patients, and the frequency of hepatic CXCR3⁺ NK cells also was positively associated with serum ALT levels (Fig 2J). Additionally, NASH patients exhibited a marked increase in hepatic CXCL9 expression levels (Fig 2K).\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; CD122 serves as an essential hub for IL-2 /IL-15 signaling axis, centrally orchestrating NK cell development, survival, and homeostatic maintenance\u003csup\u003e18\u003c/sup\u003e. Following treatment with an anti-CD122 blocking monoclonal antibody, the number of Cxcr3 and Gzmk positive cells within the CD45\u003csup\u003e+\u003c/sup\u003e cell population decreased (Fig 2L). Hepatic CXCR3-positive Gzmk-high NK cells (but not NKT cells, Cxcr3⁺ CD8⁺ T cells and CD4⁺ T cells (Fig 2Q-S)) were drained (Fig 2M, 2O). Moreover, this treatment was accompanied by improvements in hepatic damage and hepatic lipid accumulation (Fig 2N, 2P). These findings underscore Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells as a pivotal driver in the pathogenesis and progression of MASH\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eIL-21 induces hepatic Cxcr3⁺ NK cells accumulation by downregulating Lef1 and upregulating Cxcr3 expression\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTranscriptomic analysis of CXCR3⁺ NK cells from MASH mouse livers did not reveal significant upregulation of key transcription factors (T-bet, EOMES, and c-Rel) associated with canonical NK cell cytotoxic activity (Fig 2E). Through hierarchical clustering and transcription factor regulatory network analysis, we identified Lef1 as a key orchestrator of the phenotypic characteristics in hepatic Cxcr3⁺ NK cells (Fig 3A). GSEA further confirmed a reduction in the expression of Lef1-dependent genes (Fig 3B). Flow cytometric analysis also indicated that Lef1 expression was downregulated in hepatic Cxcr3⁺ NK cells from MASH mice (Fig 3C-D). To investigate the regulatory mechanisms of Lef1 in Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells, we next performed gene loss-of-function and gain-of-function studies. The experimental data revealed that inhibition of Lef1 in hepatic Cxcr3\u003csup\u003eneg\u003c/sup\u003e NK cells from MASH mice upregulates Cxcr3 expression (Fig 3E), whereas overexpression of Lef1 in Cxcr3⁺ NK cells reduce Cxcr3 expression (Fig 3F). In the livers of MASH mice, a marked inverse correlation was observed between the expression levels of Lef1 and Cxcr3 (Fig 3G). Furthermore, analysis revealed that Lef1 expression levels and hepatic damage severity in MASH were reciprocally related (as measured by serum ALT level and NAS scores) (Fig 3H, J). In clinical samples from NASH patients, we also observed decreased LEF1 expression levels (Fig 3K) and increased CXCR3 expression levels (Fig 3M). The expression level of LEF1 in CD56\u003csup\u003e+\u003c/sup\u003e NK cells showed a significant negative correlation with serum ALT levels of NASH patients (Fig 3L).\u003c/p\u003e\n\u003cp\u003eWe further identified potential mediators of Lef1 downregulation in Cxcr3\u003csup\u003e+\u0026nbsp;\u003c/sup\u003eNK cells through cytokine screening. Significantly increased mRNA levels of\u003cem\u003e\u0026nbsp;IL-21\u003c/em\u003e, \u003cem\u003eIL27\u003c/em\u003e, \u003cem\u003eTnf\u003c/em\u003e, and\u003cem\u003e\u0026nbsp;Cxcl9\u003c/em\u003e were observed in MASH mice liver tissue (Fig 3I). Only IL-21 could upregulate Cxcr3 expression in hepatic CD49b\u003csup\u003e+\u003c/sup\u003e Cxcr3\u003csup\u003eneg\u003c/sup\u003e NK cells (Fig 3N). IL-21 significantly induced the upregulation of Cxcr3 and Gzmk expression, while downregulating Lef1 expression in hepatic Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells (Fig 3O). Serum IL-21 levels in MASH mice exhibited a significant positively correlated with serum ALT (Fig 3P) and, conversely, negatively associated with LEF1 expression in hepatic CXCR3⁺ NK cells (Fig 3Q). In clinical liver samples from NASH patients, we also detected elevated IL-21 expression levels (Fig 3R). Furthermore, IL-21 expression levels in NASH patient livers showed a positive correlation with CXCR3 expression levels (Fig 3S).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcetate induces metabolic reprogramming in Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells and drives their attack on\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003ehepatocytes\u003c/strong\u003e\u003cstrong\u003e.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIL-21 induces Cxcr3\u003csup\u003e+\u0026nbsp;\u003c/sup\u003eNK cells to express high levels of GzmK, but not GzmA or GzmB (Fig 3O, S5D). However, exposure to IL-21 did not enhance the capacity of Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells to induce hepatocyte damage (Fig 4A). It is well established that dysregulated lipid metabolism and aberrant free fatty acid levels occur in the livers of NASH patients and MASH mice\u003csup\u003e19\u003c/sup\u003e. Therefore, we further explored the impact of free fatty acids on Cxcr3⁺ NK cells function. We found that acetate significantly upregulated GZMK expression in IL-21 stimulated Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells (Fig 4B), and this upregulation exhibited a dose-dependent increase in response to acetate concentration (Fig 4C). Simultaneously, the hepatic acetate levels showed a marked increase in MASH mice relative to healthy controls. which exhibited a positive correlation with disease progression (Fig 4D). Additionally, NASH patients also exhibited significantly elevated acetate levels in liver tissue (Fig 4E). We hypothesize that acetate may mediate adhesion between Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells and hepatocytes. Therefore, we performed an in vitro CRISPR screen targeting receptors for NK cell-derived cytokines on the surface of hepatocytes and identified that knockout of the TGF-\u0026beta;1 receptor most significantly protected hepatocytes. Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells exhibited a significant increase in TGF-\u0026beta;1 expression (but not TNF) after IL-21 stimulation and acetate exposure (Fig 4F-G). Overexpression of Lef1 suppressed the acetate induced upregulation of Gzmk expression (Fig 4H). Using the impedance-based xCelligence RTCA MP system, we found that IL-21 stimulation and subsequent acetate exposure enabled Cxcr3⁺ Lef1\u003csup\u003elow\u003c/sup\u003e NK cells to induce hepatocyte damage (Fig 4I). We termed this non-specific cytotoxic activity of Cxcr3⁺ Lef1\u003csup\u003elow\u003c/sup\u003e NK cells \u0026quot;Hepatic Auto-destructiveness\u0026quot; . Conversely, Lef1 overexpression was found to block this auto- destructiveness phenotype (Fig 4J). In adoptive transfer experiments of Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells from MASH mouse livers, we also observed that IL-21-induced and acetate-exposed NK cells or Lef1 inhibitor-treated and acetate-exposed NK cells caused liver injury in recipient mice (Fig 4K-L). Blocking TGF-\u0026beta;1 protects hepatocytes from injury both in vivo and in vitro (Fig 4M-N). Similarly, TGF-\u0026beta;1 expression levels in the livers of NASH patients showed a significant positive correlation with serum ALT levels (Fig 4O). Further research showed that TGF-\u0026beta;1 promoted a robust induction of the adhesion molecule ICAM-1 in primary hepatocytes (Fig 4P). Similarly, Cxcr3⁺ NK cells stimulated with IL-21 could kill TGF-\u0026beta;1-induced hepatocytes, even in the absence of acetate exposure, while LFA-1 blockade conferred protection against Cxcr3⁺ NK cell-mediated hepatocyte injury both in vivo and in vitro (Fig 4Q-R). Additionally, we observed direct contact between Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells and primary hepatocytes following acetate exposure using laser confocal microscopy (Fig 4S).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCxcl9 recruits and activates hepatic Cxcr3⁺ Lef1\u003csup\u003elow\u003c/sup\u003e\u003c/strong\u003e \u003cstrong\u003eNK cells through Gimap4\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe preliminary findings prompted us to investigate whether the cytotoxic activity of hepatic Cxcr3⁺ NK cells from MASH mice against normal hepatocytes is analogous to their targeting of tumor/infected cells. We found that blockade of MHC I on hepatocytes failed to induce Cxcr3⁺ NK cell cytotoxicity, and that inhibiting their primary triggering receptors (NKG2D and NKp30) did not protect hepatocytes from acetate-exposed Cxcr3⁺ NK cell-mediated killing (Fig 5A). Furthermore, in contrast to the null effect of blocking the PLC and MAPK cascades, the auto-destructive function of Cxcr3⁺ NK cells were strictly dependent on intact PI3K and calcium signaling (Fig 5B). This prompted us to identify novel activating stimuli for Cxcr3⁺ NK cells. We referenced the laboratory\u0026apos;s earlier MASH mouse metabolomics data to screen several signaling molecules that showed significantly elevated levels in the livers of MASH mice. We found that Cxcl9 (rather than Cxcl10 or Cxcl11) not only recruits Cxcr3\u003csup\u003e+\u0026nbsp;\u003c/sup\u003eLef1\u003csup\u003elow\u003c/sup\u003e NK cells but also significantly activates them (Fig 5C). Proteomics revealed that Cxcl9 may activate Cxcr3\u003csup\u003e+\u003c/sup\u003e NK cells by inducing the expression of GTPase of the immunity-associated protein family member 4 (Gimap4) (Fig 5D). Fluo-4AM calcium fluorescent probes revealed that both Cxcl9 and overexpression Gimap4 upregulate calcium levels in NK cells, whereas Cxcl9 fails to elevate calcium levels in Gimap4 KO NK cells (Fig 5E). Additionally, Gimap4 knockout does not affect MICA-Fc (NKG2D ligand, activating NKG2D) activation of NK cells (Fig 5E). We predicted downstream molecules of Gimap4 using AlphaFold 3 (Fig S5A-B). Concurrently, CO-IP experiments demonstrated that Gimap4 directly binds to PLC-\u0026gamma;1 in both HEK293T cells and hepatic Cxcr3\u003csup\u003e+\u0026nbsp;\u003c/sup\u003eLef1\u003csup\u003elow\u003c/sup\u003e NK cells (Fig 5F-G) and mediates PLC-\u0026gamma;1 phosphorylation (Fig 5H), further activated the hepatic Cxcr3\u003csup\u003e+\u003c/sup\u003e Lef1\u003csup\u003elow\u003c/sup\u003e NK cells. The Asparagine at site 647 and the Aspartic Acid at site 560 of PLC-\u0026gamma;1 is essential for its interaction with Gimap4 (Fig S5C).\u003c/p\u003e\n\u003cp\u003eIn vivo experiments demonstrated that Gimap4 knockout protected against CDHFD-induced weight gain in mice (Fig. 5I) Pathological analysis revealed that Gimap4 knockout protected against CDHFD-induced hepatic lipid accumulation, fibrosis, and increased Gzmk levels (Fig 5J). Concurrently, it reduced serum ALT and IFN-\u0026gamma; levels in MASH mice (Fig 5K-L) as well as hepatic TG and \u0026alpha;-SMA expression levels (Fig 5M-N). Flow cytometric analysis revealed that Cxcr3⁺ NK cells from Gimap4 knockout mice fed a CDHFD diet failed to exhibit increased calcium levels upon Cxcl9 stimulation (Fig 5O). In vitro experiments and NK cell adoptive transfer experiments, Gimap4 knockout protected against Cxcr3⁺NK cell-mediated hepatocyte killing (Fig 5P-Q), while it did not affect NK cell anti-infective or anti-tumor immunity (Fig 5R).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eHepatic NK cells in MASH patients employ similar mechanisms to mediate hepatocyte damage.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo further prove our findings, we validated the function of hepatic Cxcr3\u003csup\u003e+\u003c/sup\u003eLef1\u003csup\u003elow\u003c/sup\u003e NK cells in MASH model mice with other obesity backgrounds and observed a conserved role consistent with that found in CDHFD-induced MASH mice (Fig S3). To establish the clinical relevance of our findings, we next investigated whether hepatic NK cells in MASH patients are involved in a similar mechanism of hepatocyte damage. IL-21-induced, acetate-stimulated CXCR3\u003csup\u003e+\u0026nbsp;\u003c/sup\u003eIL21R\u003csup\u003e+\u003c/sup\u003e NK cells from NASH patients exhibit cytotoxic activity against both allogeneic and autologous primary hepatocytes (Fig S7A). Hepatic CXCR3\u003csup\u003e+\u0026nbsp;\u003c/sup\u003eIL21R\u003csup\u003e+\u003c/sup\u003e NK cells in NASH patients also express characteristic genes of auto-destructive NK cells in MASH mice (Fig S7B). PCA and correlation analysis revealed that CXCR3\u003csup\u003e+\u0026nbsp;\u003c/sup\u003eIL21R\u003csup\u003e+\u003c/sup\u003e NK cells in the livers of NASH patients share similar transcriptional profiles with Cxcr3\u003csup\u003e+\u003c/sup\u003e Lef1\u003csup\u003elow\u003c/sup\u003e NK cells in MASH mice (Fig S7C-D). Similarly, IL21R⁺ NK cells derived from healthy human PBMCs required the presence of CXCL9 to induce hepatocyte damage (Fig. S7E). Furthermore, the cytotoxicity of these reprogrammed NK cells against hepatocytes was dependent on both IL-21 concentration and the effector-to-target ratio (Fig S7F-G). Consistent with the findings in MASH mice, blockade of NKG2D failed to rescue hepatocytes from NK cell-mediated damage, whereas inhibition of calcium signaling significantly suppressed NK cell-induced hepatocyte damage (Fig S7H). In addition to mediating hepatocyte death through GZMK, CXCR3⁺IL21R⁺ NK cells in NASH patient livers also exhibited CXCL9-induced rapid upregulation of Fas ligand (FasL), but not TRAIL (Fig S7I-J). Furthermore, hepatocyte death was found to be dependent on apoptosis rather than necroptosis (Fig S7K).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eScd1 suppression mediated palmitic acid accumulation promotes CXCL9 secretion in hepatic endothelial cells.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSingle-cell RNA sequencing revealed that CXCL9 is predominantly secreted by hepatic endothelial cells (HECs). Therefore, we performed clustering analysis on HECs subsets from MASH mice and identified three distinct subpopulations based on their signature gene expression profiles: Scd1\u003csup\u003e+\u0026nbsp;\u003c/sup\u003eHECs, Cxcl9\u003csup\u003e+\u0026nbsp;\u003c/sup\u003eHECs and Fbln\u003csup\u003e+\u003c/sup\u003e HECs \u0026nbsp; (Fig 6A-B). Pseudotime analysis revealed that HECs gradually differentiate into a Cxcl9-high-expressing subpopulation from Scd1\u003csup\u003e+\u003c/sup\u003eHECs during MASH progression (Fig 6C-E).\u0026nbsp;We observed no significant changes in hepatic Stearoyl-CoA desaturase 1 (SCD1) expression between normal and MASH mice. A modest downregulation was only observed after 12 months of CDHFD feeding (Fig 6F). In contrast, a significant decrease in Scd1 expression was detected specifically in HECs isolated from the livers of\u0026nbsp;Normal\u0026nbsp;and MASH mice (Fig 6F). SCD1 is a pivotal regulator in maintaining the homeostasis between saturated (SFAs) and monounsaturated fatty acids (MUFAs). Inhibition of SCD1 activity or downregulation of its gene expression leads to intracellular accumulation of SFAs (e.g., palmitate) and a concomitant reduction in MUFAs (e.g., oleate) [20]. Consistently, we observed palmitic acid accumulation in both liver and HECs during MASH progression (Fig 6G-H).\u0026nbsp;To assess gene expression differences between Cxcl9+ HECs and Scd1+ HECs at deeper sequencing depths, we performed RNA sequencing on Cxcl9\u003csup\u003e+\u003c/sup\u003e HECs and Scd1+ HECs isolated from the livers of MASH mice fed an 8-month CDHFD diet (Fig 6I). KEGG enrichment analysis and transcription factor screening revealed that palmitate accumulation mediated by Scd1 expression suppression may promote CXCL9 expression through JAK-STAT signaling and IRF4 signaling (Fig 6J-K). This lipid metabolic reprogramming drives CXCL9 secretion in HECs. Western Blot results also showed that JANK-STAT signaling pathway and IRF4 signaling were significantly upregulated in Cxcl9\u003csup\u003e+\u003c/sup\u003eHECs (Fig 6L-M). Next, we performed gene knockout and overexpression experiments. Overexpressing STAT1 alone in HECs failed to induce upregulation of CXCL9 expression (Fig 6N). Conversely, knocking down STAT1 and IRF4 both blocked IFN-\u0026gamma;-induced upregulation of CXCL9 expression (Fig 6O). These results indicate that the JAK2-STAT1 signaling pathway regulates CXCL9 expression by modulating IRF4.\u003c/p\u003e\n\u003cp\u003eFurthermore, IL21R\u003csup\u003e+\u003c/sup\u003e NK cells from healthy mice were stimulated with IL-21 and exposed to acetate to generate Cxcr3\u003csup\u003e+\u003c/sup\u003e Lef1\u003csup\u003elow\u003c/sup\u003e NK cells. Subsequently, WT and IRF4-KO mice injected with AAV2-Cdh5pro-gRNA-EGFP (gRNA specifically targeting Scd1) and AAV8-TBG-spCas9 were adoptively transferred with Cxcr3\u003csup\u003e+\u003c/sup\u003e Lef1\u003csup\u003elow\u003c/sup\u003e NK cells and Cxcr3\u003csup\u003e-\u003c/sup\u003e NK cells (Fig 6P). Adoptive transfer of Cxcr3⁺Lef1\u003csup\u003elow\u003c/sup\u003e NK cells significantly induced hepatocyte injury in wild-type mice. In contrast, Irf4 knockout mice were protected against this liver damage(Fig 6Q-U). Furthermore, Irf4 knockout mice exhibited significantly lower serum Cxcl9 levels (Fig 6W) and reduced hepatic Gimap4 expression compared to wild-type mice (Fig 6V). Collectively, our results reveal that suppression of Scd1 is a critical driver of palmitic acid accumulation in HECs. Palmitic acid accumulation-mediated lipid metabolic reprogramming promotes CXCL9 secretion in HECs, which in turn activates Cxcr3⁺Lef1\u003csup\u003elow\u003c/sup\u003e NK cells, ultimately leading to liver damage.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe development and progression of metabolic dysfunction-associated steatohepatitis (MASH) is closely linked to chronic hepatic immune dysregulation and subsequent inflammatory responses\u003csup\u003e21\u003c/sup\u003e. In this study, we observed an accumulation of Cxcr3⁺ NK cells exhibiting tissue-resident features (\u003cem\u003eIL21R, RGS1\u003c/em\u003e) and effector function (\u003cem\u003eGZMK\u003c/em\u003e) in the livers of MASH mice. Cxcr3⁺ NK cells exhibited diminished LEF1 activity and were present with elevated abundance in the livers of both MASH mice and NASH patients. IL-21 reprograms liver-resident NK cells into a sensitized state via LEF1 downregulation and CXCR3 upregulation, enabling robust responses to acetate and CXCL9. Acetate stimulation prompted Cxcr3\u003csup\u003e+\u003c/sup\u003eLef1\u003csup\u003elow\u003c/sup\u003e NK cells to secrete TGF-β1 and express high levels of GZMK. TGF-β1 facilitates the interaction between Cxcr3\u003csup\u003e+\u003c/sup\u003eLef1\u003csup\u003elow\u003c/sup\u003e NK cells and hepatocytes by inducing ICAM-1 expression on hepatocytes, thereby collectively triggering an immunopathological damage response against hepatocytes. SCD1 suppression in hepatic endothelial cells drives palmitic acid accumulation and subsequent CXCL9 production, ultimately mediating the MHC-I-independent recruitment and activation of Cxcr3⁺Lef1low NK cells.\u003c/p\u003e\n\u003cp\u003ePrevious studies have demonstrated that interleukin 21 receptor (IL21R) and interleukin 18 receptor (IL18R) are highly expressed in trNK and cNK cells, respectively, and serve as key markers distinguishing tissue-resident NK cells from their circulating counterparts\u003csup\u003e22\u003c/sup\u003e. This indicates the decisive effect of the cytokine microenvironment in determining NK cells differential activation. Consistently, we observed that IL21R was significantly high expressed in Cxcr3\u003csup\u003e+\u003c/sup\u003eLef1\u003csup\u003elow\u0026nbsp;\u003c/sup\u003eNK cells in the liver of MASH mice, while IL18R was significantly downregulated compared with Cxcr3\u003csup\u003e-\u003c/sup\u003eNK cells. This also demonstrates the critical regulatory role of IL-21 signaling in Cxcr3\u003csup\u003e+\u003c/sup\u003eLef1\u003csup\u003elow\u003c/sup\u003e NK cells. Intestinal dysbiosis and disrupted hepatic fatty acid metabolism are hallmark features and driving forces in the development and progression of MASH\u003csup\u003e23\u003c/sup\u003e\u003csup\u003e,\u0026nbsp;\u003c/sup\u003e\u003csup\u003e24\u003c/sup\u003e. We propose that the elevated acetate levels in the livers of MASH mice and patients with NASH may result from gut dysbiosis and disrupted fatty acid metabolism.\u003c/p\u003e\n\u003cp\u003eAs a complex liver metabolic disease, there is currently no animal model that can completely simulate MASH\u003csup\u003e25\u003c/sup\u003e. Therefore, in this study, we validated our findings in several MASH mouse models in an obese background. Specifically, these included: the HFHC diet-induced MASH model\u003csup\u003e26\u003c/sup\u003e, the HFD diet-induced MASH model\u003csup\u003e27\u003c/sup\u003e, and the WD diet-induced MASH model\u0026nbsp;\u003csup\u003e28\u003c/sup\u003e. These MASH model mice also exhibit similar phenotypes to CDHFD induce MASH model (Fig. S3). Previous reports have shown that NK cell development is severely impaired in Nfil3 KO (also known as E4bp4) mice, while the number and function of T cells, NKT cells, and B cells are unaffected by Nfil3 deficiency\u003csup\u003e29\u003c/sup\u003e. Similarly, our study also showed that Nfil3\u003csup\u003e-/-\u003c/sup\u003e mice were significantly resistant to MASH diet-induced hepatic lipid accumulation, inflammation, and liver damage (Fig. S4). In the CDHFD diet-induced MASH mouse model, Nfil3 KO also protects against CDHFD diet-induced NASH development, with significant attenuation of the chemokine signaling pathway and cytokine-cytokine receptor pathway\u0026nbsp;\u003csup\u003e30\u003c/sup\u003e\u003csup\u003e,\u0026nbsp;\u003c/sup\u003e\u003csup\u003e31\u003c/sup\u003e. These cumulative findings establish NK cells as critical contributors to the development and progression of MASH.\u003c/p\u003e\n\u003cp\u003eThe GTPase of Immunity-Associated Protein 4 (Gimap4), alternatively known as IAN4, belongs to the GIMAP protein family\u0026nbsp;\u003csup\u003e32\u003c/sup\u003e. The Gimap family is a class of GTP-binding proteins primarily that is expressed in the immune system (especially lymphocytes), and it resides in the inner mitochondrial and endoplasmic reticulum membranes, orchestrating key processes in immune cell development, functional regulation, and signal transduction\u003csup\u003e33\u003c/sup\u003e. Our research shows that Gimap4 directly binds to Plc-γ1 and regulates calcium signaling in NK cells. Knockout of Gimap4 directly inhibits the activation of Cxcr3\u003csup\u003e+\u003c/sup\u003eLef1\u003csup\u003elow\u003c/sup\u003e NK cells. We observed that Gimap4 deletion not only markedly ameliorated liver damage and hepatic inflammation, but also alleviated CDHFD-induced hepatic lipid accumulation and elevated triglyceride levels. We speculate that the reason may be that Cxcr3\u003csup\u003e+\u003c/sup\u003eLef1\u003csup\u003elow\u0026nbsp;\u003c/sup\u003eNK cells attack hepatocytes, damaging their ability to regulate lipid metabolism. It is worth noting that Nfil3\u003csup\u003e-/-\u003c/sup\u003e mice also exhibited a protective effect against lipid accumulation compared to wild-type mice\u003csup\u003e30\u003c/sup\u003e, which is consistent with our results.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eMASH represents a significant risk factor and a major etiology for HCC, with patients exhibiting a significantly elevated risk of developing HCC\u003csup\u003e34\u003c/sup\u003e. Previous studies have shown that liver cirrhosis caused by liver fat accumulation, inflammation, and hepatocyte damage associated with MASH is a potential factor in inducing HCC\u0026nbsp;\u003csup\u003e35\u003c/sup\u003e. Although cirrhosis is the predominant risk factor for MASH-related HCC, a substantial proportion (36.6%-50%) of cases arise in a non-cirrhotic background\u0026nbsp;\u003csup\u003e36\u003c/sup\u003e, implicating alternative underlying mechanisms in MASH-driven hepatocarcinogenesis. It is worth noting that our research found that Cxcr3\u003csup\u003e+\u003c/sup\u003eLef1\u003csup\u003elow\u003c/sup\u003e NK cells not only damage normal liver cells but also possess the antitumor function as normal NK cells. Cxcr3\u003csup\u003e+\u003c/sup\u003eLef1\u003csup\u003elow\u003c/sup\u003e NK cells also damage hepatic endothelial cells, with a significant reduction in MASH late-stage hepatic endothelial cells, leading to a decrease in the activation and recruitment of NK cells, thereby enabling cancerous cells to evade immune surveillance.\u003c/p\u003e\n\u003cp\u003eCollectively, this study provides evidence that hepatic Cxcr3\u003csup\u003e+\u003c/sup\u003eLef1\u003csup\u003elow\u003c/sup\u003e cells of MASH mice and MASH patients are activated by Cxcl9 after IL-21 stimulation and acetate exposure, attacking normal hepatocytes. Unlike NK cell activation in anti-infection and anti-tumor responses, the Cxcl9-induced activation of Cxcr3⁺Lef1\u003csup\u003elow\u003c/sup\u003e NK cells diverges from this conventional pathway, being uniquely orchestrated by Gimap4. Knocking out Gimap4 to suppress Cxcr3\u003csup\u003e+\u003c/sup\u003eLef1\u003csup\u003elow\u003c/sup\u003e NK cell activation can significantly improve MASH. In summary, our findings unveil a previously unappreciated mechanism of a novel NK cell subset contribute to MASH progression. Simultaneously, results obtained from experimental MASH models and human clinical samples demonstrated that targeted inhibition of Cxcr3\u003csup\u003e+\u003c/sup\u003eLef1\u003csup\u003elow\u003c/sup\u003e NK cell activation and Gimap4 signaling may represent a promising therapeutic strategy for MASH.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLimitations of the study:\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCxcr3\u003csup\u003e+\u003c/sup\u003eLef1\u003csup\u003elow\u003c/sup\u003e NK cells on normal hepatocytes depends on LAF1-ICAM1-mediated cell adhesion. TGF-β1 is a key cytokine that induces ICAM1 expression in hepatocytes, and knocking out the TGF-β1 receptor on the surface of hepatocytes significantly protects against damage caused by Cxcr3\u003csup\u003e+\u003c/sup\u003eLef1\u003csup\u003elow\u003c/sup\u003eNK cells to normal hepatocytes. Notably, during MASH progression, while liver Cxcr3\u003csup\u003e+\u003c/sup\u003eLef1\u003csup\u003elow\u003c/sup\u003e NK cells exhibit high TGF-β1 expression, other cells (such as Kupffer cells or hepatic stellate cells) are activated from a quiescent state and produce large amounts of TGF-β1 in response to inflammatory and damage signals\u0026nbsp;\u003csup\u003e37\u003c/sup\u003e\u003csup\u003e,\u0026nbsp;\u003c/sup\u003e\u003csup\u003e38\u003c/sup\u003e. In this study, we did not investigate the potential interplay between Cxcr3⁺Lef1\u003csup\u003elow\u003c/sup\u003e NK cells and other hepatic cell types, such as Kupffer cells or hepatic stellate cells.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eResource availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLead contact\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFurther information and requests for resources and reagents should be directly to and will be fulfilled by the lead contact, Jiaqiang Bo ([email protected]).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMaterials availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll materials used in this study were commercially available or obtained from existing sources; no novel reagents were developed.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData and code availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePlease refer to Data S1 for all raw data of Western blotting. Processed RNA sequencing data was provided in Data S2 and Data S6.\u0026nbsp;This paper does not report original code. Single-cell RNA-sequencing data will be used for future research and are currently available only upon request to the corresponding author. Any additional information required to reanalyze the data reported in this paper is avail-able from the lead contact upon request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was financially supported by grants from the Natural Science Foundation of Xinjiang Uygur Autonomous Region (82560821), and the Project of State Key Laboratory of Natural Medicines, China Pharmaceutical University (SKLNMZZ202214).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eJiaqiang Bo designed and supervised the experiments. Jiaqiang Bo, Jie Yang, Chen Li analyzed the results and drafted the manuscript. Jiaqiang Bo, Jie Yang, Chen Li, Jia Zheng carried out most of the experiments. Chen Li, Jia Zheng helped conduct animal experiments and methodology. Jiaqiang Bo helped to analyze scRNA-seq data. Jianzhong Song helped collect and analyze the human samples. Jianzhong Song acquired funding and provided funding support.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDeclaration of interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eRiazi K, Azhari H, Charette JH, Underwood FE, King JA, Afshar EE, Swain MG, Congly SE, Kaplan GG, Shaheen AA. The prevalence and incidence of NAFLD worldwide: a systematic review and meta-analysis. Lancet Gastroenterol Hepatol. 2022 Sep;7(9):851-861. doi: 10.1016/S2468-1253(22)00165-0.\u003c/li\u003e\n \u003cli\u003eMa C, Wang S, Dong B, Tian Y. Metabolic reprograming of immune cells in MASH. Hepatology. 2025 May 5. doi: 10.1097/HEP.0000000000001371.\u003c/li\u003e\n \u003cli\u003eYounossi ZM, Golabi P, Paik JM, Henry A, Van Dongen C, Henry L. The global epidemiology of nonalcoholic fatty liver disease (NAFLD) and nonalcoholic steatohepatitis (NASH): a systematic review. Hepatology. 2023 Apr 1;77(4):1335-1347. doi: 10.1097/HEP.0000000000000004.\u003c/li\u003e\n \u003cli\u003eLuo X, Wang K, Jiang C. Gut microbial enzymes and metabolic dysfunction-associated steatohepatitis: Function, mechanism, and therapeutic prospects. Cell Host Microbe. 2025 May 21:S1931-3128(25)00153-2. doi: 10.1016/j.chom.2025.04.020.\u003c/li\u003e\n \u003cli\u003eCrane H, Eslick GD, Gofton C, Shaikh A, Cholankeril G, Cheah M, Zhong JH, Svegliati-Baroni G, Vitale A, Kim BK, Ahn SH, Kim MN, Strasser SI, George J. Global prevalence of metabolic dysfunction-associated fatty liver disease-related hepatocellular carcinoma: A systematic review and meta-analysis. Clin Mol Hepatol. 2024 Jul;30(3):436-448. doi: 10.3350/cmh.2024.0109.\u003c/li\u003e\n \u003cli\u003eYounossi ZM, Alqahtani SA, Alswat K, Yilmaz Y, Keklikkiran C, Funuyet-Salas J, \u003cem\u003eet al\u003c/em\u003e. Global NASH Council. Global survey of stigma among physicians and patients with nonalcoholic fatty liver disease. J Hepatol. 2024 Mar;80(3):419-430. doi: 10.1016/j.jhep.2023.11.004.\u003c/li\u003e\n \u003cli\u003eTilg H, Adolph TE, Dudek M, Knolle P. Non-alcoholic fatty liver disease: the interplay between metabolism, microbes and immunity. Nat Metab. 2021 Dec;3(12):1596-1607. doi: 10.1038/s42255-021-00501-9.\u003c/li\u003e\n \u003cli\u003eHuby T, Gautier EL. Immune cell-mediated features of non-alcoholic steatohepatitis. Nat Rev Immunol. 2022 Jul;22(7):429-443. doi: 10.1038/s41577-021-00639-3.\u003c/li\u003e\n \u003cli\u003eDudek M, Pfister D, Donakonda S, Filpe P, Schneider A, Laschinger M, Hartmann D, \u003cem\u003eet al\u003c/em\u003e. Auto-aggressive CXCR6+ CD8 T cells cause liver immune pathology in NASH. Nature. 2021 Apr;592(7854):444-449. doi: 10.1038/s41586-021-03233-8.\u003c/li\u003e\n \u003cli\u003eLiu W, You D, Lin J, Zou H, Zhang L, Luo S,\u003cem\u003e\u0026nbsp;et al\u003c/em\u003e. SGLT2 inhibitor promotes ketogenesis to improve MASH by suppressing CD8\u003csup\u003e+\u003c/sup\u003e T cell activation. Cell Metab. 2024 Nov 5;36(11):2489. doi: 10.1016/j.cmet.2024.10.009. Epub 2024 Oct 16. Erratum for: Cell Metab. 2024 Oct 1;36(10):2245-2261.e6. doi: 10.1016/j.cmet.2024.08.005.\u003c/li\u003e\n \u003cli\u003eBennett HM, Stephenson W, Rose CM, Darmanis S. Single-cell proteomics enabled by next-generation sequencing or mass spectrometry. Nat Methods. 2023 Mar;20(3):363-374. doi: 10.1038/s41592-023-01791-5.\u003c/li\u003e\n \u003cli\u003eZhang X, Han J, Man K, Li X, Du J, Chu ES, Go MY, Sung JJ, Yu J. CXC chemokine receptor 3 promotes steatohepatitis in mice through mediating inflammatory cytokines, macrophages and autophagy. J Hepatol. 2016 Jan;64(1):160-70. doi: 10.1016/j.jhep.2015.09.005.\u003c/li\u003e\n \u003cli\u003eWang F, Zhang X, Liu W, Zhou Y, Wei W, Liu D, Wong CC, Sung JJY, Yu J. Activated Natural Killer Cell Promotes Nonalcoholic Steatohepatitis Through Mediating JAK/STAT Pathway. Cell Mol Gastroenterol Hepatol. 2022;13(1):257-274. doi: 10.1016/j.jcmgh.2021.08.019.\u003c/li\u003e\n \u003cli\u003eSparano C, Solís-Sayago D, Zangger NS, Rindlisbacher L, Van Hove H, Vermeer M, Westermann F, Mussak C, \u003cem\u003eet al\u003c/em\u003e. Autocrine TGF-β1 drives tissue-specific differentiation and function of resident NK cells. J Exp Med. 2025 Mar 3;222(3):e20240930. doi: 10.1084/jem.20240930.\u003c/li\u003e\n \u003cli\u003eShao J, Yan Y, Ding D, Wang D, He Y, Pan Y, Yan W, Kharbanda A, Li HY, Huang H. Destruction of DNA-Binding Proteins by Programmable Oligonucleotide PROTAC (O'PROTAC): Effective Targeting of LEF1 and ERG. Adv Sci (Weinh). 2021 Oct;8(20):e2102555. doi: 10.1002/advs.202102555.\u003c/li\u003e\n \u003cli\u003eFranzén O, Gan LM, Björkegren JLM. PanglaoDB: a web server for exploration of mouse and human single-cell RNA sequencing data. Database (Oxford). 2019 Jan 1;2019:baz046. doi: 10.1093/database/baz046.\u003c/li\u003e\n \u003cli\u003eVeldhoen M. Interleukin 17 is a chief orchestrator of immunity. Nat Immunol. 2017 May 18;18(6):612-621. doi: 10.1038/ni.3742. PMID: 28518156.\u003c/li\u003e\n \u003cli\u003eKennedy PR, Arvindam US, Phung SK, Ettestad B, Feng X, Li Y, \u003cem\u003eet al\u003c/em\u003e. Metabolic programs drive function of therapeutic NK cells in hypoxic tumor environments. Sci Adv. 2024 Nov;10(44):eadn1849. doi: 10.1126/sciadv.adn1849.\u003c/li\u003e\n \u003cli\u003eTincopa MA, Anstee QM, Loomba R. New and emerging treatments for metabolic dysfunction-associated steatohepatitis. Cell Metab. 2024 May 7;36(5):912-926. doi: 10.1016/j.cmet.2024.03.011. Epub 2024 Apr 11. Erratum in: Cell Metab. 2024 Jun 4;36(6):1430. doi: 10.1016/j.cmet.2024.04.016.\u003c/li\u003e\n \u003cli\u003eSen U, Coleman C, Sen T. Stearoyl coenzyme A desaturase-1: multitasker in cancer, metabolism, and ferroptosis. Trends Cancer. 2023 Jun;9(6):480-489. doi: 10.1016/j.trecan.2023.03.003.\u003c/li\u003e\n \u003cli\u003eHuby T, Gautier EL. Immune cell-mediated features of non-alcoholic steatohepatitis. Nat Rev Immunol. 2022 Jul;22(7):429-443. doi: 10.1038/s41577-021-00639-3. Epub 2021 Nov 5.\u003c/li\u003e\n \u003cli\u003eHu L, Han M, Deng Y, Gong J, Hou Z, Zeng Y, Zhang Y, He J, Zhong C. Genetic distinction between functional tissue-resident and conventional natural killer cells. iScience. 2023 Jun 20;26(7):107187. doi: 10.1016/j.isci.2023.107187.\u003c/li\u003e\n \u003cli\u003eNie Q, Luo X, Wang K, Ding Y, Jia S, Zhao Q, Li M, Zhang J, Zhuo Y, Lin J, Guo C, Zhang Z, Liu H, Zeng G, You J, Sun L, Lu H, Ma M, Jia Y, Zheng MH, Pang Y, Qiao J, Jiang C. Gut symbionts alleviate MASH through a secondary bile acid biosynthetic pathway. Cell. 2024 May 23;187(11):2717-2734.e33. doi: 10.1016/j.cell.2024.03.034.\u003c/li\u003e\n \u003cli\u003eHorn P, Tacke F. Metabolic reprogramming in liver fibrosis. Cell Metab. 2024 Jul 2;36(7):1439-1455. doi: 10.1016/j.cmet.2024.05.003. Epub 2024 May 31.\u003c/li\u003e\n \u003cli\u003eGallage S, Avila JEB, Ramadori P, Focaccia E, Rahbari M, Ali A, Malek NP, Anstee QM, Heikenwalder M. A researcher's guide to preclinical mouse NASH models. Nat Metab. 2022 Dec;4(12):1632-1649. doi: 10.1038/s42255-022-00700-y.\u003c/li\u003e\n \u003cli\u003eZhang X, Fan L, Wu J, Xu H, Leung WY, Fu K, Wu J, Liu K, Man K, Yang X, Han J, Ren J, Yu J. Macrophage p38α promotes nutritional steatohepatitis through M1 polarization. J Hepatol. 2019 Jul;71(1):163-174. doi: 10.1016/j.jhep.2019.03.014.\u003c/li\u003e\n \u003cli\u003eZhang Z, Yuan Y, Hu L, Tang J, Meng Z, Dai L,\u003cem\u003e\u0026nbsp;et al\u003c/em\u003e. ANGPTL8 accelerates liver fibrosis mediated by HFD-induced inflammatory activity via LILRB2/ERK signaling pathways. J Adv Res. 2023 May;47:41-56. doi: 10.1016/j.jare.2022.08.006.\u003c/li\u003e\n \u003cli\u003eGallage S, Ali A, Barragan Avila JE, Seymen N, Ramadori P, Joerke V, Zizmare L, \u003cem\u003eet al\u003c/em\u003e. A 5:2 intermittent fasting regimen ameliorates NASH and fibrosis and blunts HCC development via hepatic PPARα and PCK1. Cell Metab. 2024 Jun 4;36(6):1371-1393.e7. doi: 10.1016/j.cmet.2024.04.015.\u003c/li\u003e\n \u003cli\u003eKamizono S, Duncan GS, Seidel MG, Morimoto A, Hamada K, Grosveld G, Akashi K, Lind EF, Haight JP, Ohashi PS, Look AT, Mak TW. Nfil3/E4bp4 is required for the development and maturation of NK cells in vivo. J Exp Med. 2009 Dec 21;206(13):2977-86. doi: 10.1084/jem.20092176.\u003c/li\u003e\n \u003cli\u003eWang F, Zhang X, Liu W, Zhou Y, Wei W, Liu D, Wong CC, Sung JJY, Yu J. Activated Natural Killer Cell Promotes Nonalcoholic Steatohepatitis Through Mediating JAK/STAT Pathway. Cell Mol Gastroenterol Hepatol. 2022;13(1):257-274. doi: 10.1016/j.jcmgh.2021.08.019.\u003c/li\u003e\n \u003cli\u003eWang S, Gao J, Yang M, Zhang G, Yin L, Tong X. OPN-Mediated Crosstalk Between Hepatocyte E4BP4 and Hepatic Stellate Cells Promotes MASH-Associated Liver Fibrosis. Adv Sci (Weinh). 2024 Dec;11(47):e2405678. doi: 10.1002/advs.202405678.\u003c/li\u003e\n \u003cli\u003eSchnell S, Démollière C, van den Berk P, Jacobs H. Gimap4 accelerates T-cell death. Blood. 2006 Jul 15;108(2):591-9. doi: 10.1182/blood-2005-11-4616.\u003c/li\u003e\n \u003cli\u003eHeinonen MT, Laine AP, Söderhäll C, Gruzieva O, Rautio S, Melén E, Pershagen G, Lähdesmäki HJ, Knip M, Ilonen J, Henttinen TA, Kere J, Lahesmaa R; Finnish Pediatric Diabetes Registry. GIMAP GTPase family genes: potential modifiers in autoimmune diabetes, asthma, and allergy. J Immunol. 2015 Jun 15;194(12):5885-94. doi: 10.4049/jimmunol.1500016.\u003c/li\u003e\n \u003cli\u003eHuang DQ, El-Serag HB, Loomba R. Global epidemiology of NAFLD-related HCC: trends, predictions, risk factors and prevention. Nat Rev Gastroenterol Hepatol. 2021 Apr;18(4):223-238. doi: 10.1038/s41575-020-00381-6.\u003c/li\u003e\n \u003cli\u003ePovero D, Chen Y, Johnson SM, McMahon CE, Pan M, Bao H, \u003cem\u003eet al\u003c/em\u003e. HILPDA promotes NASH-driven HCC development by restraining intracellular fatty acid flux in hypoxia. J Hepatol. 2023 Aug;79(2):378-393. doi: 10.1016/j.jhep.2023.03.041. Epub 2023 Apr 13.\u003c/li\u003e\n \u003cli\u003eHuang Y, Xie Y, Zhang Y, Liu Z, Jiang W, Ye Y, \u003cem\u003eet al\u003c/em\u003e. Single-cell transcriptome reveals the reprogramming of immune microenvironment during the transition from MASH to HCC. Mol Cancer. 2025 Jun 11;24(1):177. doi: 10.1186/s12943-025-02370-2.\u003c/li\u003e\n \u003cli\u003eTrivedi P, Wang S, Friedman SL. The Power of Plasticity-Metabolic Regulation of Hepatic Stellate Cells. Cell Metab. 2021 Feb 2;33(2):242-257. doi: 10.1016/j.cmet.2020.10.026. Epub 2020 Nov 23.\u003c/li\u003e\n \u003cli\u003eMassagué J, Sheppard D. TGF-β signaling in health and disease. Cell. 2023 Sep 14;186(19):4007-4037. doi: 10.1016/j.cell.2023.07.036.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8654560/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8654560/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Metabolic dysfunction-associated steatohepatitis (MASH) is a systemic metabolic disorder associated with obesity that leads to liver disease (such as hepatocyte damage and fibrosis) and hepatocellular carcinoma. During the progression of MASH, the accumulation of metabolites triggers hepatocyte damage and inflammation. however, the mechanisms underlying MASH-related hepatic damage remain incompletely understood. Using several mouse models that simulate critical features of human MASH (hereinafter referred to as MASH mice) and single-cell transcriptomics sequencing, we identify a distinct NK cell subset critically implicated in hepatic immunopathology. We found NK cells with tissue residency characteristics (IL21R, RGS1) and effector functions (GZMK) aggregated in the livers of MASH mice. These CXCR3+ NK cells exhibited reduced activity of the LEF1 transcription factor and were found with elevated abundance in the context of MASH, as observed in both experimental mice and human subjects. The mechanism involves IL-21 sensitizes the responsiveness of hepatic CXCR3⁺ NK cells to metabolic stimuli, such as Acetate and CXCL9, by downregulating LEF1 and upregulating CXCR3, collectively triggering immunopathological hepatic damage. Hepatic CXCR3+LEF1low NK cells from MASH patients and mice exhibit similar transcriptional profiles. During MASH progression, CXCR3+ LEF1lowNK cells were recruited to and retained in the liver, were metabolically reprogrammed in the hepatic microenvironment, and mediated MHC class I-independent auto-destructive killing of hepatocytes. CXCR3+LEF1lowNK cell cytotoxicity fundamentally diverges from canonical anti-tumor NK cell immunity, mechanistically distinguishing auto-destructive versus protective NK cell functional modalities.","manuscriptTitle":"CXCR3+ LEF1low NK cells cause immunopathological hepatic damage in MASH","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-11 05:48:41","doi":"10.21203/rs.3.rs-8654560/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-communications","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"NCOMMS","sideBox":"Learn more about [Nature Communications](http://www.nature.com/ncomms/)","snPcode":"","submissionUrl":"https://mts-ncomms.nature.com/","title":"Nature Communications","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Communications","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"6cd00466-db1d-46fa-bc5b-487eab2f1aab","owner":[],"postedDate":"February 11th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":62188802,"name":"Biological sciences/Cell biology/Cell signalling/Nutrient signalling"},{"id":62188803,"name":"Biological sciences/Cell biology/Mechanisms of disease"},{"id":62188804,"name":"Biological sciences/Cell biology/Cell signalling/Lipid signalling"}],"tags":[],"updatedAt":"2026-02-11T05:48:42+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-11 05:48:41","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8654560","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8654560","identity":"rs-8654560","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}