Epigenetic repression of hepatocyte FoxO1 disrupts local immune homeostasis and promotes liver inflammation

Abstract

Background and Aims:

Disrupting liver immune homeostasis drives inflammation. Recent evidence shifts immunoregulatory focus to hepatocytes, though the mechanisms remain poorly defined. Forkhead box O1 (FoxO1) is a critical homeostasis regulator, but its function in liver immune homeostasis is unknown. We aimed to clarify the role of hepatocyte FoxO1 in liver immune homeostasis and inflammation.

Approach and Results:

Human liver FoxO1 expression and its association with inflammation were analyzed in patients with various inflammation-related liver diseases. Hepatocyte-specific Foxo1 knockout (FoxO1^△hepa) mice were established. Hepatocyte-specific gene interference was employed in alcoholic hepatitis and hepatic schistosomiasis murine models. Transcriptomic, single-cell RNA sequencing, and CUT&Tag analyses were performed to elucidate the underlying mechanisms. Hepatocyte FoxO1 levels in human inflammatory livers declined prevalently and were inversely correlated with inflammation and fibrosis. Around 15–18 weeks after birth, FoxO1^△hepa mice exhibited mild spontaneous hepatic inflammation with natural killer T (NKT) cell and neutrophil accumulation. NKT cell depletion in FoxO1^△hepa mice with alcoholic hepatitis or hepatic schistosomiasis (HS) significantly reduced neutrophil accumulation and protected against liver inflammation and damage. Mechanistically, FoxO1 promoted retinoic acid synthesis to induce hepatocyte CD1d expression, which is necessary for regulating NKT cell apoptosis. Innovatively, decreased JMJD1C expression in hepatocytes caused histone H3 lysine 9 (H3K9) dimethylation at the Foxo1 promoter, repressing its transcription and disrupting local immune homeostasis.

Conclusions:

Our findings uncover a hitherto unrecognized mechanism for hepatocyte-based control of liver inflammation, in which hepatocyte FoxO1 maintained by JMJD1C restrains local NKT cells and neutrophils via CD1d induction, providing promising targets for inflammatory liver diseases.

INTRODUCTION

The liver performs many essential functions, including metabolism, nutrient storage, detoxification, and synthesis of serum proteins and bile. It also possesses a remarkable capacity to maintain physiological homeostasis upon environmental changes.^1,2 Of particular concern, the liver constantly encounters dietary and commensal bacterial products with potential inflammatory effects that originate from the stomach, gut, and spleen. The hepatic immune system must tolerate these products to maintain liver homeostasis. Disruption of the delicate balance between tolerance and immunity in the liver drives hepatic inflammation and tissue damage.^3,4

The liver is not only a vital metabolic organ but also an important immunological organ, enriched with various immune cell types, including Kupffer cells, natural killer T (NKT) cells, and lymphocytes. In the healthy liver, most resident immune cells promote tolerance via the expression of tolerogenic factors (PD-L1, IL-10, TGF-β, etc.) and relatively low expression of major histocompatibility complex and costimulatory molecules, induction of T cell anergy or apoptosis, and expansion of regulatory T cells.^5,6 Recent research has increasingly shifted the immunological focus to hepatocytes, the predominant parenchymal cells in the liver, which account for ~60% of the liver’s cell number and 80% of its mass.⁷ Studies have demonstrated that hepatocytes can activate both local innate and adaptive immune responses by expressing various immune receptors (eg, major histocompatibility complex, pattern recognition receptors, and costimulatory and adhesion molecules) and producing complement proteins, chemokines (eg, MCP-1, CXCL1), and cytokines (eg, IL-6, IL-1β, and TNF-α) in response to damage signals and infections.^8,9 Meanwhile, emerging evidence suggests that hepatocytes can also contribute to immune tolerance, for example, by expressing checkpoint molecules or releasing immunoregulatory factors in the context of liver diseases.^10–12 However, the precise mechanisms by which hepatocytes coordinate immune responses and maintain the liver’s default tolerogenic milieu remain poorly understood, necessitating further investigations to deepen our understanding of the pathobiological processes underlying liver homeostasis.

Forkhead box O1 (FoxO1), a multifunctional transcription factor in the forkhead family, plays a critical role in orchestrating a variety of physiological and stress-related transcriptional programs in hepatocytes, including glucose and lipid metabolism, antioxidant responses, cell survival, and apoptosis.^13,14 Interestingly, hepatocyte FoxO1 expression has been found to decline markedly in alcoholic hepatitic (AH).¹⁵ Deficiency of FoxO1 in the liver or administration of FoxO1 inhibitor leads to exacerbation of hepatic inflammation in non-alcoholic fatty liver disease and liver ischemia and reperfusion injury.^15,16 However, little is known about how and to what extent hepatocyte FoxO1 regulates liver immune responses.

Here, we investigated the role of FoxO1 in hepatocyte-based control of liver immune homeostasis, revealing that hepatocytes regulate NKT cells and neutrophils to restrain liver inflammation through FoxO1-controlled CD1d expression. Furthermore, we identified JMJD1C as an epigenetic regulator critical for maintaining FoxO1 expression in hepatocytes. During inflammatory liver diseases, loss of JMJD1C expression resulted in epigenetic repression of hepatocyte FoxO1 expression, disrupting local immune homeostasis and promoting liver inflammation. Overall, we uncovered a hitherto unrecognized role of hepatocyte JMJD1C-FoxO1 in regulating hepatic immunity.

METHODS

Human liver specimens

Formalin-fixed paraffin-embedded liver sections from patients with biopsy-proven non-alcoholic steatohepatitis [NASH; n=5; NAFLD activity score (NAS): 2 patients had NAS=7, and 3 patients had NAS=6], primary biliary cholangitis (PBC; n=5), Hepatic schistosomiasis (HS; n=3), drug-induced liver injury (DILI; n=5), liver hepatocellular carcinoma (LIHC; n=5), and AH (n=5) were obtained from the Third People’s Hospital of Changzhou, Department of Pathology. Liver tissues were collected through ultrasonography-guided percutaneous needle liver biopsies or hepatic resection. Control liver samples were pathologically normal liver samples obtained from individuals undergoing routine liver biopsy during bariatric surgery. All patients provided informed consent. The study protocol was approved by the Nanjing Medical University Ethics Committee (No. 2024-756) and conforms to the ethical guidelines of the 1975 Declaration of Helsinki.

Animal experiments

All mice were housed under specific pathogen-free conditions at the Laboratory Animal Center of Nanjing Medical University (Nanjing, China). They were kept in a facility with the temperature and humidity strictly controlled and housed with a 12-h light/dark cycle and fed with a standard chow diet (62.4% carbohydrate, 22.7% protein, 14.9% fat). All animal experiments were performed according to the regulations of the Institutional Animal Care and Use Committee (IACUC) of Nanjing Medical University (Permit Number: IACUC-1905064). Maximum efforts were made to minimize suffering for all procedures.

For establishing diet-induced AH, male C57BL/6 mice (7–8 weeks old) were subjected to a chronic-plus-binge alcohol feeding model using Lieber-DeCarli’82 diet (F1258SP, BioServ, NJ) containing 5% (vol/vol) ethanol (36% ethanol-derived calories) ad libitum for 8 weeks plus multiple binges of ethanol gavage (5 g/kg body weight), according to the method established previously.¹⁷

For establishing the HS model, male C57BL/6 mice (7–8 weeks old) were infected percutaneously with 12 ± 2 cercariae of the Chinese mainland strain of Schistosoma japonicum (S. japonicum) from infected snails (Oncomelania hupensis), which were obtained from Jiangsu Institute of Parasitic Diseases (Wuxi, China).

Statistical analysis

Data are expressed as the mean ± SEM for each group. All statistical calculations were performed using SPSS software (version 25). Statistical differences between the 2 groups were analyzed using the Student's t test. Statistical differences among multiple groups were evaluated by one-way analysis of variance (ANOVA). The expression correlations among genes were analyzed by using the Spearman's test. p<0.05 was considered significant.

Additional materials and methods are provided in Supplemental Information, http://links.lww.com/HEP/K309.

RESULTS

Universal decline of hepatocyte FoxO1 expression across human inflammatory diseases

The FoxO family comprises FoxO1, FoxO3, FoxO4, and FoxO6. FoxO1, FoxO3, and FoxO4 show similar biological behaviors.¹⁴ Various etiological factors (tumors, pathogens, drugs, ethanol, and lipid metabolites) can lead to liver inflammation. First, using online databases [Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA)], we studied the expression profile of FOXO genes in human livers under diverse inflammation-related conditions. Results showed that FOXO1 also exhibited a significant reduction in the livers of patients with liver hepatocellular carcinoma (LIHC), AH, simple steatosis (SS), and non-alcoholic steatohepatitis (NASH) compared with normal tissues. In contrast, a marginally significant or non-significant reduction was observed for FOXO3 and FOXO4 (Figures 1A–C). Moreover, hepatic FOXO1 mRNA levels maintained a continuous and significant decrease across the chronic liver disease stages, including fibrosis, cirrhosis, and acute-on-chronic liver failure (Figure 1D).

FIGURE 1.

Universal decline of hepatocyte FoxO1 expression across human liver diseases. (A–D) FOXO1, FOXO3, and FOXO4 expression levels in the liver of patients with liver cancer (A), alcoholic hepatitis (B), simple steatosis and non-alcoholic steatohepatitis (C), and fibrotic liver disease (D), based on publicly available datasets. (E) scRNA-seq analysis of FOXO1 expression in major cell populations of human liver, visualized using UMAP clustering. Images were generated via an online platform (https://cells.ucsc.edu). Left: UMAP showing liver cell clusters; right: FOXO1 expression mapped onto the clusters. (F, G) Immunohistochemical detection of FoxO1 expression in human liver sections from individuals with normal liver (healthy control) and patients with NASH, primary biliary cholangitis (PBC), schistosomiasis, drug-induced liver injury (DILI), hepatocellular carcinoma (HCC), or alcoholic hepatitis. Representative immunohistochemical images (F). Scale bar: 100 μm. Violin plot quantifying the percentage of FoxO1-positive hepatocytes (G). (H) Spearman correlation analysis of FOXO1 and TNF expression in human datasets of viral hepatitis (HBV, GSE84044; HCV, GSE15331), NASH (GSE63067), and alcoholic hepatitis (GSE143318), and in Schistosoma japonicum-infected mouse livers. (I) Spearman correlation analysis of FOXO1 and fibrosis-related genes (ACTA2, COL1A1, COL1A2 and COL3A1) in liver samples from cirrhosis patients (GSE63898). *p<0.05, **p<0.01, ***p<0.001, **** p<0.0001, and ns indicates not significant. Abbreviations: FoxO1, Forkhead box O1; LIHC, liver hepatocellular carcinoma; NASH, non-alcoholic steatohepatitis; NK, natural killer; PBC, primary biliary cholangitis; scRNA-seq, single-cell RNA sequencing; SS, simple steatosis; TCGA, The Cancer Genome Atlas; UMAP, Uniform Manifold Approximation and Projection.

We then re-analyzed published single-cell RNA sequencing (scRNA-seq) results of healthy human livers,¹⁸ showing that FoxO1 is predominantly expressed in hepatocytes and endothelial cells compared with other liver cell populations (Figure 1E). Immunohistochemistry indicated that hepatocyte FoxO1 expression was decreased in liver tissues from patients with biopsy-proven NASH, primary biliary cholangitis (PBC), HS, drug-induced liver injury (DILI), hepatocellular carcinoma (HCC), and AH, compared with those without chronic liver inflammation (Figures 1F, G). These data suggest a universal decline in hepatocyte FoxO1 expression in human inflammatory livers.

To further explore the relevance of hepatocyte FoxO1 and liver inflammation, we analyzed the gene expression levels of hepatic FOXO1 and TNF in inflammatory liver diseases using public databases. Analyses revealed consistent negative correlations between hepatic FOXO1/Foxo1 and TNF/Tnf expression levels in patients with viral hepatitis (HBV and HCV), NASH and AH, or mice with HS (Figure 1H). Moreover, hepatic FOXO1 was negatively correlated with fibrosis-related genes (ACTA2, COL1A1, COL1A2 and COL3A1) in liver samples of human cirrhosis (Figure 1I). These data strongly suggest a potential role of hepatic FoxO1 in protecting against inflammation and fibrosis in liver diseases.

Hepatocyte FoxO1 deficiency aggravates hepatic inflammation and damage, while its rescue alleviates these conditions in inflammatory liver diseases

To characterize the role of hepatocyte FoxO1 in liver inflammation, we crossed FoxO1^f/f mice with albumin-Cre transgenic mice to produce hepatocyte-specific FoxO1-knockout (FoxO1^△hepa) mice. Hepatocyte-specific deletion of FoxO1 in mice was confirmed by immunoblot analysis of the liver, heart, and spleen (Supplemental Figure S1A, http://links.lww.com/HEP/K309). Immunoblot analysis of hepatocytes and non-hepatocytes from FoxO1^△hepa mice further confirmed the specificity of FoxO1 deletion in hepatocytes (Supplemental Figure S1B, http://links.lww.com/HEP/K309). Beginning at ~15–18 weeks after birth, FoxO1^△hepa mice started to exhibit spontaneously increased inflammatory infiltrates in the liver, accompanied by damage features, such as hepatocellular edema and necrosis (Supplemental Figure S1C, http://links.lww.com/HEP/K309), as well as elevated serum ALT and AST levels (Supplemental Figure S1D, http://links.lww.com/HEP/K309). Moreover, FoxO1^△hepa mice showed increased hepatic IFN-γ and TNF-α levels (Supplemental Figure S1E, http://links.lww.com/HEP/K309). These results demonstrate that FoxO1^△hepa mice display spontaneous and mild liver inflammation and damage.

Next, we were especially interested in determining the effects of hepatocyte FoxO1 deficiency on inflammatory liver diseases. We established AH and HS in mice, both of which are well-defined animal models of typical inflammation-driven liver diseases.^19,20 Interestingly, we observed substantial declines in hepatocyte FoxO1 mRNA and protein levels in both models (Supplemental Figure S2, http://links.lww.com/HEP/K309). Hepatocyte FoxO1 deficiency resulted in enhanced inflammatory cell infiltration in the liver of mice with AH or HS (Figure 2A, upper panel), leading to the exacerbation of fibrosis as assessed by Sirius red staining (Figures 2A lower panel, B) and α-SMA immunoblot (Figurea 2C, D), and the aggravation of liver damage (ALT and AST; Figures 2E, F). In addition, the levels of inflammatory mediators, including TNF-α, IFN-γ, and IL-1β, were significantly higher in both the livers and sera of FoxO1^△hepa mice with AH or HS than in those of control mice (Figures 2G–J). Of note, the fold-change analysis revealed that FoxO1^△hepa mice exhibited higher or slightly higher deterioration of liver fibrosis (α-SMA) and damage (ALT and AST) than FoxO1^f/f mice following the establishment of AH or HS (Supplemental Tables S1–S3, http://links.lww.com/HEP/K310, http://links.lww.com/HEP/K311, http://links.lww.com/HEP/K312). These findings suggest that hepatocyte FoxO1 deficiency exacerbates liver inflammation and damage in inflammatory liver diseases (AH and HS).

FIGURE 2.

Hepatocyte FoxO1 deficiency aggravates hepatic inflammation and damage in inflammatory liver diseases. (A, B) Representative H&E and Sirius red staining of liver sections from FoxO1^f/f and FoxO1^△hepa mice with alcoholic hepatitis (AH; ethanol/EtOH) or hepatic schistosomiasis (HS; Schistosoma japonicum). Scale bar: 200 µm. The quantification of the Sirius red–positive area (% area; B). (C, D) Immunoblot analysis of α-SMA expression in FoxO1^f/f and FoxO1^△hepa mice with AH or HS. (E, F) Serum ALT and AST levels in FoxO1^f/f and FoxO1^△hepa mice with AH or HS. (G–J) RT-PCR analysis of hepatic Ifng and Tnf mRNA expression (G, H), and ELISA analysis of serum IL-1β and TNF-α levels (I, J), in FoxO1^f/f and FoxO1^△hepa mice with AH or HS. Mice were sacrificed at the age of 15 weeks. The data are representative of 2 independent experiments. n=5 mice/group. *p<0.05, **p<0.01, ***p<0.001, and ns indicates not significant. Abbreviations: FoxO1, Forkhead box O1; H&E, hematoxylin and eosin; RT-PCR, reverse transcription–polymerase chain reaction; α-SMA, alpha–smooth muscle actin.

Given the significant loss of hepatocyte FoxO1 in inflammatory liver diseases, we explored whether rescuing its expression could alleviate liver inflammation. We constructed an adeno-associated virus serotype (AAV) vector expressing Foxo1 under the control of the thyroxine-binding globulin (TBG) promoter to induce hepatocyte-specific overexpression of FoxO1 in mice with AH or HS. The experimental design is shown in Supplemental Figures S3A, B, http://links.lww.com/HEP/K309. Successful overexpression of FoxO1 in hepatocytes was verified using reverse transcription–polymerase chain reaction (RT-PCR) and immunoblot (Supplemental Figures S3C–F, http://links.lww.com/HEP/K309). Histopathological analysis [hematoxylin and eosin (H&E) and Sirius red staining] revealed that hepatocyte FoxO1-overexpression in mice with AH or HS attenuated inflammatory infiltration and hepatic fibrosis (Supplemental Figures S3G, H, http://links.lww.com/HEP/K309). Immunoblot analysis also showed reduced expression of α-SMA in the livers of hepatocyte FoxO1-overexpressing AH or HS mice compared with those of control mice (Supplemental Figures S3I, J, http://links.lww.com/HEP/K309). Furthermore, AH or HS mice with hepatocyte FoxO1-overexpression exhibited a marked decrease in liver damage, as indicated by ALT and AST serum levels (Supplemental Figures S3K, L, http://links.lww.com/HEP/K309), concomitant with reduced expression of inflammatory mediators (TNF-α, IFN-γ, and IL-1β) in the liver and serum (Supplemental Figures S3M–P, http://links.lww.com/HEP/K309). These results demonstrate that rescue of decreased hepatocyte FoxO1 alleviates liver inflammation and damage in inflammatory liver diseases (AH and HS).

Hepatocyte FoxO1 restrains hepatic inflammation and damage in a manner dependent on NKT cells

To gain insights into the roles of hepatocyte FoxO1 in regulating inflammation, we characterized immune cell populations in the liver of FoxO1^△hepa mice. Notably, among immune cells, including CD3⁻NK1.1⁺ NK cells, only CD3⁺NK1.1⁺ NKT cells and CD11b⁺Ly6G⁺ neutrophils were significantly accumulated in the liver of FoxO1^△hepa mice (Figures 3A–C). However, there were no significant differences in splenic immune cell populations, including NKT cells and neutrophils, between FoxO1^△hepa and FoxO1^f/f mice (Supplemental Figure S4, http://links.lww.com/HEP/K309). In addition, hepatic CD3⁺NK1.1⁺ NKT cells and CD11b⁺Ly6G⁺ neutrophils were decreased in mice with hepatocyte FoxO1-overexpression (Figures 3D, E).

FIGURE 3.

Hepatocyte FoxO1 restrains hepatic inflammation and damage in a manner dependent on NKT cells. (A–C) Representative flow cytometry (FCM) dot plots and quantification of hepatic immune cell populations, including CD19⁺ B cells, CD3⁺ T cells, CD4⁺ T cells, CD8⁺ T cells, F4/80⁺CD11b⁺ macrophages (A); CD3⁺NK1.1⁺ NKT cells, CD3⁻NK1.1⁺ NK cells (B), and CD11b⁺Ly6G⁺ neutrophils (C) in FoxO1^f/f and FoxO1^△hepa mice. (D, E) FCM analysis of CD3⁺NK1.1⁺ NKT cells (D) and CD11b⁺Ly6G⁺ neutrophils (E) in the liver of C57BL/6 mice injected with AAV-TBG-FoxO1 or AAV-Ctrl via the tail vein. (F–I) Representative FCM dot plots and quantification (proportion and mean fluorescence intensity, MFI) of IFN-γ⁺ (F), TNF-α⁺ (G), IL-6⁺ (H), and IL-4⁺ (I) NKT cells (CD3⁺NK1.1⁺) in the livers of FoxO1^f/f and FoxO1^△hepa mice. (J–P) Effects of NKT cell depletion in FoxO1^△hepa mice with Schistosoma japonicum infection using an anti-NK1.1 antibody. Representative FCM dot plots and quantification of CD3⁺NK1.1⁺ NKT cells (J) and CD11b⁺Ly6G⁺ neutrophils (K). RT-PCR analysis of hepatic Ifng and Tnf mRNA levels (L). ELISA analysis of serum levels of IFN-γ and IL-1β (M). Representative Sirius red staining of liver sections and the quantification of the Sirius red–positive area (% area; N). Immunoblot analysis and quantification of α-SMA in the liver (O). Serum ALT and AST levels (P). Mice were sacrificed at the age of 18 weeks in panels A–I, or at the age of 15 weeks in panels J–P. n=5–6 mice/group. The data are representative of 2 independent experiments. *p<0.05, **p<0.01, ***p<0.001, and ns indicates not significant. Abbreviations: AAV, adeno-associated virus; Ctrl, control; FoxO1, Forkhead box O1; NKT, natural killer T; RT-PCR, reverse transcription–polymerase chain reaction; TBG, thyroxine-binding globulin; α-SMA, alpha–smooth muscle actin.

Notably, immunofluorescence histochemical analysis revealed increased numbers of hepatic Vα24-Jα18⁺ NKT cells and MPO⁺ neutrophils in AH and HS patients (Supplemental Figures S5A, B, http://links.lww.com/HEP/K309). Flow cytometry (FCM) analysis also showed increased hepatic NKT cells and neutrophils in AH and HS mice compared with normal mice (Supplemental Figures S5C, D, http://links.lww.com/HEP/K309). FoxO1 deficiency in hepatocytes of HS mice resulted in increased total liver mononuclear cells (LMNCs; Supplemental Figure S6A, http://links.lww.com/HEP/K309), as well as increased NKT cells (Supplemental Figures S6B, C, http://links.lww.com/HEP/K309) and neutrophils (Supplemental Figures S6D, E, http://links.lww.com/HEP/K309) in the liver. NKT cells, a common liver-resident cell type, can rapidly secrete proinflammatory cytokines to initiate inflammation.²¹ We observed that liver NKT cells from FoxO1^△hepa mice expressed significantly higher levels of IFN-γ and TNF-α than those from FoxO1^f/f mice (Figures 3F, G), while there were no differences in IL-6⁺ NKT cells between the groups and fewer IL-4⁺ NKT cells in FoxO1^△hepa mice (Figures 3H, I).

Neutrophil accumulation in the liver is mainly due to recruitment from the peripheral blood and is triggered by local inflammatory mediators (TNF-α, IFN-γ, and IL-1β).²² To investigate the potential role of proinflammatory NKT cells in neutrophil recruitment, NKT cells were depleted in FoxO1^f/f and FoxO1^△hepa mice with AH or HS using an anti-NK1.1 antibody. Flow cytometric analysis confirmed the efficiency of NKT cell depletion (Figure 3J and Supplemental Figure S7A, http://links.lww.com/HEP/K309). As expected, we detected a markedly reduced frequency of hepatic neutrophils in both FoxO1^f/f and FoxO1^△hepa mice following NKT cell depletion (Figure 3K and Supplemental Figure S7B, http://links.lww.com/HEP/K309), concomitant with reduced levels of inflammatory mediators (IFN-γ, TNF-α, and IL-1β; Figures 3L, M, and Supplemental Figures S7C, D, http://links.lww.com/HEP/K309) as well as diminished liver fibrosis and damage (Figures 3N–P and Supplemental Figures S7E–G, http://links.lww.com/HEP/K309). Notably, following NKT cell depletion, no significant differences were observed between FoxO1^f/f and FoxO1^△hepa mice with AH or HS (Figures 3K–P and Supplemental Figures S7B–G, http://links.lww.com/HEP/K309). Collectively, these results demonstrate that hepatocyte FoxO1 regulates hepatic inflammation and damage in an NKT cell-dependent manner.

Hepatocyte FoxO1 maintains liver NKT cell homeostasis through modulation of their apoptosis

We further explored the mechanism underlying NKT cell accumulation in the livers of FoxO1^△hepa mice. Liver NKT cells are depleted in mice lacking CXCL16, mainly responsible for NKT cell recruitment and retention in the liver.^23,24 Therefore, we investigated the hepatic expression of CXC chemokine ligand 16 (CXCL16). The results showed no difference in CXCL16 mRNA expression in the liver between FoxO1^f/f and FoxO1^△hepa mice (Figure 4A). Next, we investigated whether NKT cell death or proliferation contributes to the accumulation of NKT cells in the liver of FoxO1^△hepa mice. The FCM gating strategy used to analyze NKT or non-NKT cells in mouse LMNCs is shown (Figure 4B). No difference in apoptotic rate was observed in hepatic non-NKT cells between FoxO1^f/f and FoxO1^△hepa mice (Figures 4C, D). In contrast, FoxO1^△hepa mice, compared with FoxO1^f/f mice, exhibited a significant reduction in the apoptosis of hepatic NKT cells, revealed by a decreased percentage of Annexin V⁺ NKT cells in the liver (Figures 4C, E). In addition, neither hepatic NKT nor non-NKT cells displayed a difference in the nuclear expression of Ki67 between FoxO1^f/f and FoxO1^△hepa mice (Figures 4F–H), suggesting comparable proliferative activity. Notably, in the murine model of HS, we also observed significantly decreased and increased apoptosis of liver NKT cells in FoxO1^△hepa mice and wild-type mice with hepatocyte FoxO1-overexpression, respectively (Figures 4I–L). These findings suggest that hepatocyte FoxO1 maintains liver NKT cell homeostasis through modulation of apoptosis.

FIGURE 4.

Hepatocyte FoxO1 maintains liver NKT cell homeostasis through modulation of their apoptosis. (A) Relative mRNA levels of Cxcl16 in the livers of FoxO1^f/f and FoxO1^△hepa mice. (B) FCM gating strategy for analyzing NKT cells (CD3⁺NK1.1⁺) and non-NKT cells (CD3⁺NK1.1⁻, CD3⁻NK1.1⁺, and CD3⁻NK1.1⁻) in mouse liver mononuclear cells (LMNCs). (C–E) Representative FCM dot plots (C) and quantification of Annexin V-positive cells in non-NKT cells (D) and NKT cells (E) in the liver of FoxO1^f/f and FoxO1^△hepa mice. (F–H) Representative FCM dot plots (F) and quantification of Ki67-positive cells in non-NKT cells (G) and NKT cells (H) in the liver of FoxO1^f/f and FoxO1^△hepa mice. (I–L) Apoptosis of liver NKT cells was evaluated by FCM. Representative FCM histograms and quantification of Annexin V⁺ NKT cells in the livers of Schistosoma japonicum-infected FoxO1^f/f mice and FoxO1^△hepa mice (I, J), and wild-type mice injected with AAV-Ctrl or AAV-TBG-FoxO1 (K, L). n=5 mice/group. Mice were sacrificed at the age of 18 weeks. The data are representative of 2 independent experiments. ***p<0.001 and ns indicates not significant. Abbreviations: AAV, adeno-associated virus; Ctrl, control; FCM, flow cytometry; FoxO1, Forkhead box O1; NKT, natural killer T; TBG, thyroxine-binding globulin.

FoxO1 induces CD1d expression via all-trans retinoic acid signaling in hepatocytes

Hepatocytes have previously been reported to promote apoptosis of liver-resident NKT cells in a manner dependent on hepatocellular CD1d under physiological conditions.²⁵ Our previous study also revealed hepatocyte CD1d-dependent regulation of NKT cell homeostasis, which modulates liver immunopathology in murine schistosomiasis.²⁶ Therefore, we further investigated whether hepatocyte FoxO1 regulates NKT cell homeostasis by controlling hepatocyte CD1d expression. Primary hepatocytes from FoxO1^△hepa mice exhibited a substantial downregulation in CD1d mRNA and protein expression (Figures 5A, B). Corresponding to the above data showing a considerable decrease in hepatocyte FoxO1 expression in mice with AH or HS (Supplemental Figure S2, http://links.lww.com/HEP/K309), CD1d mRNA and protein levels in hepatocytes were also dramatically and synchronously decreased in both models (Supplemental Figures S8A–D, http://links.lww.com/HEP/K309). Interestingly, significant positive correlations were observed between hepatic Foxo1 and Cd1d1 gene expression levels in HS models based on online databases (Supplemental Figure S8E, http://links.lww.com/HEP/K309). We also observed marked increases in CD1d expression in FoxO1-overexpressing hepatocytes from mice with AH or HS (Figures 5C–F).

FIGURE 5.

FoxO1 induces CD1d expression via atRA signaling in hepatocytes. (A, B) Relative mRNA and protein levels of FoxO1 and CD1d in hepatocytes from FoxO1^f/f or FoxO1^△hepa mice were determined by RT-PCR (A) and immunoblot (B). (C–F) Relative mRNA and protein levels of FoxO1 and CD1d in hepatocytes from mice with hepatic schistosomiasis (C, D) or alcoholic hepatitis (E, F) injected with AAV-TBG-FoxO1 or AAV-Ctrl were determined by RT-PCR (C, E) and immunoblot (D, F). (G, H) AML12 hepatocytes were transfected with FoxO1-specific siRNA or scrambled siRNA. FoxO1 and CD1d expression levels in transfected cells were determined by RT-PCR (G) and immunoblot (H). (I, J) AML12 hepatocytes were transfected with pcDNA-3.1 or pcDNA-FoxO1-AAA. FoxO1 and CD1d expression levels in transfected cells were determined by RT-PCR (I) and immunoblot (J). (K) KEGG pathway enrichment analysis of differentially expressed genes in hepatocytes from FoxO1^f/f and FoxO1^△hepa mice. (L) Heatmap of atRA biosynthesis-related genes (Rdh family) related to the KEGG term “retinol metabolism.” (M) Relative mRNA levels of Rdh1, Rdh10, and Rdh11 in hepatocytes from FoxO1^f/f and FoxO1^△hepa mice were validated by RT-PCR. (N–R) FoxO1^△hepa mice were intraperitoneally injected with atRA or vehicle (corn oil). Immunoblot analysis of FoxO1 and CD1d protein levels in hepatocytes was performed. The protein levels were normalized to GAPDH (N). Representative FCM dot plots and quantification of liver CD3⁺NK1.1⁺ NKT cells (O) and CD11b⁺Ly6G⁺ neutrophils (P). The relative mRNA levels of Ifng and Tnf in the liver were determined by RT-PCR (Q). Immunoblot analysis and quantification of IFN-γ and TNF-α expression in the liver (R). n=3–6 /group. Mice were sacrificed at the age of 18 weeks (panels A, B, and K–R) and 15 weeks (panels C–F), respectively. The data are representative of 2 independent experiments. *p<0.05, **p<0.01, ***p<0.001, and ns indicates not significant. Abbreviations: atRA, all-trans retinoic acid; AAV, adeno-associated virus; Ctrl, control; FCM, flow cytometry; FoxO1, Forkhead box O1; KEGG, Kyoto Encyclopedia of Genes and Genomes; RT-PCR, reverse transcription–polymerase chain reaction; TBG, thyroxine-binding globulin.

To further explore whether FoxO1 controls CD1d expression, we knocked down or overexpressed FoxO1 in AML12 hepatocytes. The results showed that FoxO1-knockdown using specific siRNA significantly reduced CD1d mRNA and protein expression in AML12 hepatocytes (Figures 5G, H). In contrast, CD1d mRNA and protein levels were significantly increased after FoxO1-AAA overexpression (Figures 5I, J). These data demonstrate that FoxO1 induces CD1d expression in hepatocytes.

We further investigated in vitro whether hepatocyte FoxO1 promotes NKT apoptosis through surface CD1d, using an anti-CD1d blocking antibody (Supplemental Figure S9A, http://links.lww.com/HEP/K309). Consistent with the in vivo findings, NKT cells co-cultured with FoxO1^−/− hepatocytes demonstrated significantly reduced apoptosis (Supplemental Figures S9B, C, http://links.lww.com/HEP/K309), accompanied by enhanced activity, as evidenced by elevated IFN-γ and TNF-α levels in NKT cells (Supplemental Figures S9D–G, http://links.lww.com/HEP/K309). When co-cultured with FoxO1^f/f hepatocytes pretreated with anti-CD1d, NKT cells showed reduced apoptosis, whereas their activity remained unchanged (Supplemental Figure S9B–G, http://links.lww.com/HEP/K309). These findings demonstrate that hepatocyte FoxO1 modulates NKT cell apoptosis in a surface CD1d-dependent manner, whereas it regulates NKT cell activity independently of surface CD1d.

We further explored the underlying mechanism through which hepatocyte FoxO1 induces CD1d expression. FoxO1 has been reported to bind to the promoters of multiple genes and regulate their transcriptional programs.²⁷ However, chromatin immunoprecipitation sequencing (ChIP-Seq) studies of FoxO1 in cells, including hepatocytes, have suggested that FoxO1 may not directly bind to the Cd1d1 promoter region.^28–31 In addition, among the transcription factors, SP1, ELF-1, LEF-1, and PU.1, which have been reported to regulate CD1d transcription,^32–35 displayed no alterations in mRNA expression between FoxO1^△hepa mice and FoxO1^f/f mice (Supplemental Figure S10, http://links.lww.com/HEP/K309).

To further investigate the mechanism by which FoxO1 controls CD1d expression in hepatocytes, whole-transcriptome RNA-seq analysis was performed on hepatocytes from FoxO1^△hepa and FoxO1^f/f mice. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis revealed that the retinol metabolism pathway was significantly affected in FoxO1^△hepa mice (Figure 5K). All-trans retinoic acid (atRA) signaling has been reported to induce CD1d expression in immune cells.^36–38 Further investigations based on RNA-seq data were performed to screen for genes involved in hepatocyte atRA biosynthesis. Among atRA biosynthesis-related genes,³⁹ retinol dehydrogenase 10 (RDH10), RDH11, and RDH14 displayed decreased expression in hepatocytes from FoxO1^△hepa mice compared with those from FoxO1^f/f mice (Figure 5L). The Rdh genes (Rdh1, Rdh10, and Rdh11) were further validated using RT-PCR (Figure 5M).

Furthermore, in vivo injection of atRA into FoxO1^△hepa mice significantly rescued CD1d expression in hepatocytes compared with vehicle injection into FoxO1^△hepa mice (Figure 5N). Moreover, atRA injection decreased NKT cell and neutrophil levels (Figures 5O, P), as well as the hepatic levels of proinflammatory cytokines (IFN-γ and TNF-α), in the liver of FoxO1^△hepa mice (Figures 5Q, R). In addition, atRA injection into FoxO1^f/f and FoxO1^△hepa mice with HS significantly increased CD1d expression in hepatocytes compared with vehicle injection (Supplemental Figure S11A, http://links.lww.com/HEP/K309). AtRA injection reduced hepatic NKT cells, as well as IFN-γ⁺ and TNF-α⁺ NKT cells, in both FoxO1^f/f and FoxO1^△hepa mice with HS (Supplemental Figures S11B–D, http://links.lww.com/HEP/K309). Notably, no significant differences were observed between FoxO1^f/f and FoxO1^△hepa mice with HS following atRA treatment (Supplemental Figures S11B–D, http://links.lww.com/HEP/K309). Similar changes in hepatic neutrophil accumulation (Supplemental Figure S11E, http://links.lww.com/HEP/K309), inflammation (Supplemental Figures S11F, G, http://links.lww.com/HEP/K309), fibrosis (Supplemental Figures S11H, I, http://links.lww.com/HEP/K309), and damage (Supplemental Figure S11J, http://links.lww.com/HEP/K309) were observed across the groups.

Together, these results demonstrate that FoxO1 induces CD1d expression in hepatocytes via atRA signaling, thereby regulating hepatic NKT cells, liver inflammation, and damage.

Hepatocyte FoxO1-mediated control of liver NKT cells and neutrophils is CD1d-dependent in inflammatory liver diseases

To further confirm that hepatocyte FoxO1 limits liver inflammation by controlling liver NKT cells and neutrophils through CD1d, we overexpressed FoxO1 and downregulated CD1d in the hepatocytes of mice with AH and HS. We constructed AAV vectors for TBG promoter-driven FoxO1 overexpression (AAV-FoxO1) or CD1d-knockdown (miR30-based shRNA targeting Cd1d1, AAV-CD1d-shRNA). The experimental design is shown in Figures 6A, B. The efficiency of FoxO1-overexpression and CD1d-knockdown in the hepatocytes of AH and HS mice was confirmed by immunoblot (Figures 6C, D) and immunofluorescence histochemistry (Supplemental Figure S12A–D, http://links.lww.com/HEP/K309). FCM analysis further confirmed that FoxO1-overexpression induces and CD1d-knockdown reduces hepatocyte surface CD1d expression after in vitro AAV transfection (Supplemental Figures 12E, F, http://links.lww.com/HEP/K309). Hepatocyte-FoxO1 overexpression alone significantly increased CD1d expression (Figures 6C, D and Supplemental Figures S12A–D, http://links.lww.com/HEP/K309) and simultaneously decreased NKT cells and neutrophils in the livers of mice with AH or HS (Figures 6E–H). However, these effects were reversed by CD1d-knockdown in hepatocytes (Figures 6C–H and Supplemental Figures 12A–D, http://links.lww.com/HEP/K309). Hepatocyte-FoxO1 overexpression did not have a further constraint on NKT cells and neutrophils in AH or HS mice with hepatocyte CD1d-knockdown, even exhibiting significantly more NKT cells and neutrophils (Figures 6E–H). Consistent with these observations, although FoxO1 overexpression alone reduced attenuated liver inflammation and fibrosis in AH and HS mice, CD1d-knockdown reversed these pathology-alleviating effects (Supplemental Figures S13, S14, http://links.lww.com/HEP/K309). Together, these results demonstrate that hepatocyte FoxO1 controls liver inflammation and damage through the CD1d-dependent control of local NKT cells and neutrophils in inflammatory liver diseases (AH and HS).

FIGURE 6.

Hepatocyte FoxO1-mediated control of liver NKT cells and neutrophils is CD1d-dependent in inflammatory liver diseases. (A, B) Schematic of the experimental methods, showing the injection of AAV8 vectors (AAV-TBG-FoxO1, AAV-CD1d-shRNA, or AAV-Ctrl) in murine models of AH (A, EtOH) and HS (B, Schistosoma japonicum). (C, D) Immunoblot analysis of FoxO1 and CD1d expression in hepatocytes from mice with AH (C) or HS (D) following AAV8 vector injection. (E–H) FCM analysis and quantification of the proportions of NKT cells (E, G) and neutrophils (F, H) in the livers of AH mice (E, F) and HS mice (G, H) following AAV8 vector injection. n=5 mice/group. Mice were sacrificed at the age of 15 weeks. The data are representative of 2 independent experiments. *p<0.05, **p<0.01, ***p<0.001, and ns indicates not significant. Abbreviations: atRA, all-trans retinoic acid; AAV, adeno-associated virus; AH, alcoholic hepatitis; Ctrl, control; EtOH, ethanol; FCM, flow cytometry; FoxO1, Forkhead box O1; HS, hepatic schistosomiasis; shRNA, short hairpin RNA; TBG, thyroxine-binding globulin.

Decreased JMJD1C expression leads to the transcriptional repression of Foxo1 in inflammatory liver diseases

Finally, it is of immense value to explore the deep mechanism underlying the common decline of hepatocyte FoxO1 expression in human inflammatory livers. Dysregulation of the epigenome drives aberrant gene expression that may exert pathologic effects in liver diseases.⁴⁰ Using human databases (GEO and TCGA), we analyzed common transcriptional alterations of epigenetic enzymes in multiple inflammatory livers, including AH, NASH, LIHC, and liver fibrosis, with Venn diagrams depicting 12 upregulated and 6 downregulated genes (Figure 7A). Among these 18 altered epigenetic enzyme genes, we further identified that Jmjd1c, the gene encoding H3K9me2 demethylase, was the most significantly downregulated gene in murine AH (Figure 7B). Furthermore, by utilizing published scRNA-seq data from the liver of normal and S. japonicum-infected mice, we performed cell clustering analysis and identified 10 cell clusters (Figures 7C, D), which revealed the downregulation of Jmjd1c in hepatocytes following infection (Figure 7E). Immunohistochemical analysis of the liver sections revealed decreased JMJD1C⁺ hepatocytes in patients infected with S. japonicum (Figure 7F). A substantial reduction in Jmjd1c expression in the hepatocytes of S. japonicum-infected mice was confirmed by RT-PCR, which decreased to a much lower level at 8 weeks after infection (Figure 7G). Whole-transcriptome RNA-seq analysis revealed a significant positive correlation between Jmjd1c and Foxo1 expression in the livers of S. japonicum-infected mice (Figure 7H). RT-PCR analysis further demonstrated a stronger correlation between these genes in the livers of S. japonicum-infected mice at various infection phases (Figure 7I).

FIGURE 7.

Decreased JMJD1C expression leads to the transcriptional repression of Foxo1 in inflammatory liver diseases. (A) Venn diagram illustrating the common upregulated (left: 12 genes) and downregulated (right: 6 genes) epigenetic enzyme genes (from the dbEM database) across human liver disease datasets (alcoholic hepatitis, liver cancer, non-alcoholic steatohepatitis, and liver fibrosis). (B) Volcano plot showing the 18 commonly altered epigenetic enzyme genes and Foxo1 from the sequencing dataset of alcoholic hepatitis mice (GSE230175). (C–E) Reanalysis of liver scRNA-Seq data (GSE220286) from Schistosoma japonicum-infected mice. UMAP plots show sample compositions (top) and cell clusters (bottom) (C). The heatmap shows marker genes across 10 cell clusters (D). Box plot (left) and heatmap (right) of Jmjd1c expression in hepatocytes from normal and S. japonicum-infected mice (E). (F) Immunohistochemical staining of JMJD1C in the liver sections of S. japonicum-infected patients. Red arrows indicate JMJD1C-positive hepatocytes and blue arrows indicate JMJD1C-negative hepatocytes. Scale bars: 100 and 50 µm. Quantification of JMJD1C-positive hepatocytes is shown in the bar chart. n=5 and 3 patients/group, respectively. (G) A violin plot shows Jmjd1c mRNA expression in hepatocytes from S. japonicum-infected mice at 5 and 8 weeks post-infection compared with normal mice. n=6 mice/group. (H, I) Spearman correlation analysis of Jmjd1c and Foxo1 expression in the liver of S. japonicum-infected mice based on the transcriptome sequencing dataset (GSE59276; H) and liver samples collected at different time points post-infection (0 to 18 weeks, n=24; I). (J, K) Immunoblot analysis of JMJD1C, H3K9me1, and H3K9me2 expression in hepatocytes from S. japonicum-infected mice (J) and alcoholic hepatitis mice (K). The bar charts on the right show statistical data. n=5 mice/group. (L, M) Primary mouse hepatocytes were treated with 2 µM JMJD1C inhibitor (193D7) or vehicle (DMSO) for 48 hours. Foxo1 expression was analyzed by RT-PCR (L). FoxO1 and H3K9me2 expression was analyzed by immunoblot, with statistical data displayed in bar charts (M). n=3/group. (N, O) Primary mouse hepatocytes were transfected with JMJD1C-specific siRNA (JMJD1C siRNA1/2) or scrambled siRNA for 48 hours. Foxo1 and Jmjd1c mRNA expression was measured using RT-PCR (N), and JMJD1C, FoxO1, and H3K9me2 expression was analyzed by immunoblot, with statistical analyses shown in bar charts (O). (P) Heatmap shows H3K27me2 CUT&Tag reads across gene-coding regions (−3 kb TSS to +3 kb TES) in primary murine hepatocytes treated with JMJD1C inhibitor (193D7) or vehicle (DMSO) for 48 hours. (Q) Snapshot of H3K9me2 deposition on the Foxo1 gene locus from CUT&Tag-seq data. (R) Primary mouse hepatocytes were transfected with JMJD1C-specific or scrambled siRNA for 48 hours. CUT&Tag-qPCR was performed to assess H3K9me2 enrichment in the Foxo1 gene promoter region. The data are representative of 2 independent experiments. n=3/group. *p<0.05, **p<0.01, ***p<0.001, and ns indicates not significant. Abbreviations: AH, alcoholic hepatitis; EtOH, ethanol; FoxO1, Forkhead box O1; LIHC, liver hepatocellular carcinoma; NASH, non-alcoholic steatohepatitis; RT-PCR, reverse transcription–polymerase chain reaction; scRNA-Seq, single-cell RNA sequencing; siRNA, small interfering RNA; TCGA, The Cancer Genome Atlas; TES, transcription end site; TSS, transcription start site; UMAP, Uniform Manifold Approximation and Projection.

JMJD1C demethylates H3K9me2 and H3K9me1, both associated with transcriptional repression.^41,42 As expected, concomitant with the substantial reduction of JMJD1C, a marked increase in H3K9me2 but not H3K9me1 was observed in the hepatocytes of both HS and AH mice (Figures 7J, K). Treatment with 193D7, a specific JMJD1C inhibitor, effectively decreased FoxO1 expression in primary mouse hepatocytes, coinciding with increased H3K9me2 levels (Figures 7L, M). Consistently, similar results were obtained using siRNA-targeted Jmjd1c knockdown (Figures 7N, O).

We further overexpressed JMJD1C while downregulating FoxO1 in the hepatocytes of mice with AH and HS via tail vein injection of AAV8 vectors. The efficiency of JMJD1C-overexpression and FoxO1-knockdown in hepatocytes of AH and HS mice was confirmed by RT-PCR and immunoblot (Supplemental Figures S15, S16A, B, http://links.lww.com/HEP/K309). Hepatocyte JMJD1C-overexpression alone substantially reduced H3K9me2 levels while simultaneously enhancing FoxO1 and CD1d expression (Supplemental Figures S15B, S16B, http://links.lww.com/HEP/K309). Correspondingly, hepatocyte JMJD1C-overexpressing AH and HS mice exhibited diminished NKT cell numbers and activity (Supplemental Figures S15, S16C–E, http://links.lww.com/HEP/K309), decreased neutrophils (Supplemental Figures S15F, S16F, http://links.lww.com/HEP/K309), and attenuated liver inflammation (Supplemental Figures S15, S16G, H, http://links.lww.com/HEP/K309), fibrosis (Supplemental Figures S15, S16I–L, http://links.lww.com/HEP/K309), and damage (Supplemental Figures S15M, S16M, http://links.lww.com/HEP/K309). However, all these effects of JMJD1C-overexpression were significantly reversed by FoxO1-knockdown in hepatocytes (Supplemental Figures S15, S16B–M, http://links.lww.com/HEP/K309). These data demonstrate that JMJD1C is critical for FOXO1-dependent CD1d expression, subsequently regulating NKT cells and hepatic inflammation.

Finally, to determine whether H3K9me2 directly regulated Foxo1, anti-H3K9me2 CUT&Tag analysis was performed using hepatocytes from S. japonicum-infected mice. A promoter heatmap revealed that the enrichment of H3K9me2 predominantly binds to the transcription start sites (TSS) at the genomic level (Supplemental Figure S17A, http://links.lww.com/HEP/K309). H3K9me2 enrichment peaks were identified in the promoter region of Foxo1 in hepatocytes from S. japonicum-infected mice (supplemental Figure S17B, http://links.lww.com/HEP/K309), indicating the methylation of the Foxo1 promoter. CUT&Tag analysis further revealed that treatment with the specific JMJD1C inhibitor 193D7 increased H3K9me2 enrichment at the TSS and led to higher H3K9me2 enrichment peaks in the promoter region of Foxo1 in primary mouse hepatocytes (Figures 7P, Q). Consistently, CUT&Tag-qPCR revealed that Jmjd1c knockdown also increased H3K9me2 enrichment in the Foxo1 promoter region (Figure 7R). These results demonstrate that decreased JMJD1C expression leads to increased H3K9me2 levels and thereby the transcriptional repression of Foxo1 in inflammatory liver diseases.

DISCUSSION

The liver, which contains diverse immune cell populations, maintains a tolerogenic environment under homeostasis.⁴³ Recent evidence shifts the focus of immunoregulation from resident immune cells to hepatocytes. In addition to metabolic roles, hepatocytes play a crucial role in sustaining hepatic immune tolerance.^12,44 Hepatic inflammation is a key driver of most liver diseases.⁶ However, the potential significance of hepatocytes in resolving hepatic inflammation remains poorly understood. FoxO1 is well known for regulating glucose and lipid homeostasis.¹³ Our study identified a hitherto unrecognized role of hepatocyte FoxO1 in maintaining hepatic immune homeostasis and resolving inflammation by constraining NKT cells and neutrophils in a CD1d-dependent manner, representing an innovative perspective on the interaction between hepatocytes and local immune cells in the liver.

Interestingly, our study found that hepatocyte FoxO1 expression declines prevalently across various human liver diseases (viral hepatitis, AH, NASH, etc.) and inversely correlates with hepatic inflammation. Some previous studies reported elevated hepatic FoxO1 expression in high-fat diet-fed mice and patients with non-alcoholic fatty liver disease (NAFLD),^45–47 likely due to the susceptibility of FoxO1 expression to various metabolic syndrome-related comorbidities,¹³ particularly in early stages of NAFLD without serious liver injury. Notably, single-cell sequencing of liver samples from patients at different stages of NAFLD revealed a progressive decline in hepatocyte FoxO1 levels with disease progression,⁴⁸ consistent with our findings in NASH patients and some other studies.^16,49,50

Using conditional knockout mice, we demonstrated that FoxO1 deficiency in hepatocytes specifically led to an increased accumulation of local NKT cells and neutrophils, resulting in spontaneous liver inflammation beginning at ~15–18 weeks of age. Hence, the loss of hepatocyte FoxO1 in human liver diseases may disturb liver immune homeostasis, thereby aggravating inflammation. While a previous study showed that hepatocyte FoxO1 inhibits pyroptosis to alleviate liver injury in AH mice,¹⁵ our study uncovered a previously unrecognized mechanism by which hepatocytes regulate local NKT cells and neutrophils, a process essential for the control of liver inflammation. Moreover, diet restriction has been shown to exert remarkable protective effects against liver ischemia-reperfusion injury through upregulation of FoxO1 in hepatocytes.⁵¹ Here, we provide in-depth insight into the underlying mechanisms of hepatocyte FoxO1-mediated liver protection against inflammatory damage. Paradoxically, a recent study demonstrated that in a carbon tetrachloride-induced liver fibrosis model, hepatocyte FoxO1 promotes TGF-β expression to activate hepatic stellate cells and exacerbate fibrosis.⁵² This discrepancy may arise from carbon tetrachloride, which primarily induces liver damage through radical toxicity, rather than proinflammatory damage by local NKT cells and neutrophils. In this process, the regulation of local inflammatory cells by FoxO1 cannot become the dominant effect. In addition, a previous study demonstrated that chemically inhibiting FoxO1 protected mice from NAFLD.⁵³ This protective effect could be attributed to FoxO1 suppression in macrophages, where FoxO1 has been shown to promote liver inflammation^54,55 and may play a dominant role in NAFLD progression, in contrast to the role of hepatocyte FoxO1 in our study. Therefore, FoxO1 has divergent functions across cell types, and more importantly, cell-specific interventions on FoxO1 may be crucial for future NAFLD therapy.

Tight control over hepatic NKT cells is essential for preventing inappropriate liver inflammation.²¹ Herein, we found that hepatocyte-specific FoxO1 deficiency significantly reduced NKT cell apoptosis without affecting their proliferation or chemotaxis. Ours and previous studies have revealed that hepatocyte CD1d-dependent dual regulation of both hepatic NKT cell and hepatocyte apoptosis limits liver inflammation.^25,26,56 A previous study indicated that the regulation of hepatocyte CD1d on NKT cells varies depending on the lipid antigens presented, implying that CD1d-mediated control of NKT cells is dictated by the microenvironmental balance between NKT cell-activating and non-activating lipids.²⁵ However, which lipid antigen(s) presented by hepatocyte CD1d promote NKT cell apoptosis remains an open challenge within the field and warrants future investigation.

Notably, CD1d expression in hepatocytes was markedly downregulated in FoxO1^△hepa mice. However, the mechanisms regulating CD1d expression of hepatocytes in physiological and pathological settings remain unclear. Herein, we uncovered a hitherto unknown role of FoxO1 in inducing CD1d expression in hepatocytes, revealing a previously unrecognized role of FoxO1 in endowing hepatocytes with the unique ability to regulate liver inflammation. Unexpectedly, neither previous ChIP-Seq data²⁸ nor our ChIP assay (data not shown) for FoxO1 in hepatocytes supports the hypothesis that FoxO1 directly binds to the CD1d gene promoter region and initiates its transcription. Instead, our further investigation revealed that FoxO1 indirectly induces CD1d expression by upregulating RDHs, which promote atRA biosynthesis.^37,38 These findings provide new insights into hepatic inflammation and identify potential novel therapeutic targets.

Moreover, we used murine models of HS and AH, both well-established models for studying hepatic inflammation,^19,20 to investigate the role of hepatocyte FoxO1 under inflammatory conditions. Consistent with human results, FoxO1 expression in hepatocytes was dramatically decreased in both models, and hepatocyte-specific FoxO1 overexpression reduced the accumulation of NKT cells and neutrophils and attenuated inflammatory pathology in the liver. Thus, we speculated that hepatocyte FoxO1 reduction may contribute to the pathogenesis of these inflammatory liver diseases.

Further investigation into in-depth mechanisms underlying the downregulation of FoxO1 expression in hepatocytes during liver inflammation is warranted. Epigenomes, such as histone modifiers, play an essential role in liver function, and epigenetic changes in response to environmental cues have been implicated in developing liver disease.^40,57 However, epigenetic regulation of liver inflammation and fibrosis is poorly understood. In this study, we innovatively identified JMJD1C as a key H3K9 demethylase that is frequently dysregulated in multiple liver diseases. Appropriate histone methylation of the Foxo1 promoter in hepatocytes is essential for its transcription and liver homeostasis. We demonstrated that JMJD1C demethylates H3K9me2 and activates Foxo1 transcription in hepatocytes, which helps limit liver inflammation and maintain immune homeostasis. Thus, downregulation of JMJD1C in inflammatory livers leads to increased H3K9me2 levels at the Foxo1 promoter, thereby repressing its transcription. However, little is known regarding the regulation of JMJD1C expression to date. A recent study demonstrated that inflammatory cytokines in the tumor microenvironment promote JMJD1C expression in regulatory T cells.⁵⁸ In contrast, we revealed substantially reduced JMJD1C expression in hepatocytes within inflammatory livers. The underlying mechanisms driving this discrepancy warrant further investigation.

In summary, as illustrated in the Graphical Abstract, this study uncovers a previously unidentified role in hepatocyte-based regulation of hepatic inflammation. Specifically, hepatocyte FoxO1 induces CD1d expression, which in turn restrains local NKT cells and subsequent neutrophils under both physiological and pathological conditions. Mechanistically, our in-depth investigation reveals that JMJD1C plays a critical role in maintaining FoxO1 expression in hepatocytes, thereby contributing to liver immune homeostasis. These findings provide important insights into the molecular regulation of liver immunity and offer potential therapeutic targets for treating inflammatory liver diseases.

Supplementary Material

hep-84-469-s001.pdf ^{(9.8MB, pdf)}

hep-84-469-s002.docx ^{(17.6KB, docx)}

hep-84-469-s003.docx ^{(17.5KB, docx)}

hep-84-469-s004.docx ^{(18.8KB, docx)}