Introduction

Metabolic dysfunction-associated steatotic liver disease (MASLD) has become the most prevalent chronic liver disorder in western populations, affecting approximately 30% of adults and driven by its strong association with obesity and metabolic syndrome¹. The disease spectrum ranges from matabolic dysfunction-associated fatty liver (MAFL) to metabolic dysfunction-associated steatohepatitis (MASH), with the latter characterized by steatosis, inflammation, hepatocyte ballooning, and progressive fibrosis². Central to MASLD pathogenesis are hepatic macrophages, particularly the embryo-derived Kupffer cells (KCs) that reside in liver sinusoids^3,4. These self-renewing resident macrophages⁵, play crucial roles in lipid homeostasis, as evidenced by studies showing that depletion of CD207⁺ KCs leads to impaired triglyceride storage⁶. When KCs die during MASH progression, they are gradually replaced by monocyte-derived macrophages (MoMFs) that exhibit more inflammatory properties and contribute to liver injury^6,7.This transition from protective embryo-derived KCs (EmKCs) to inflammatory monocyte-derived KCs (MoKCs) represents a critical juncture in disease progression, yet the mechanisms regulating this shift remain poorly understood.

A key determinant of macrophage function is cellular metabolism. Macrophages dynamically switch between glycolytic and oxidative phosphorylation pathways to adapt to environmental changes⁸. During MASLD, hepatic macrophages increase their glycolytic activity, which may exacerbate inflammation and tissue damage^9-11. While glucose metabolism is known to influence macrophage polarization, its specific role in determining hepatic macrophage fate - particularly the balance between KCs and MoMFs - remains unknown. Chitinase 3-like 1 (Chi3l1/YKL-40) has emerged as an important regulator of macrophage biology, promoting cell survival through ERK1/2 and PI3K/Akt pathways while modulating anti-inflammatory cytokines like IL-10^12-16. However, its potential role in macrophage metabolic reprogramming, particularly in the context of hepatic glucose metabolism, has not been explored.

In this study, we identify a novel mechanism by which Chi3l1 governs hepatic macrophage fate through metabolic regulation. We demonstrate that Chi3l1 directly interacts with glucose to suppress its uptake in macrophages. Strikingly, this interaction selectively protects glucose-high KCs from cell death in MASLD conditions, while having minimal effect on glucose-low MoMFs. These findings reveal a previously unrecognized Chi3l1-mediated metabolic checkpoint that maintains KC populations, providing new insights into the pathogenesis of MASLD and potential therapeutic strategies.

Materials and methods

Animal experiments and procedures

Animals Chil1^-/- (strain no. T014402), Chil1^flox//flox (strain no. T013652), Lyz2-cre (strain no. T003822) and Clec4f-cre (strain no. T036801) mice with a C57BL/6J background were purchased from GemPharmatech. Rosa tdtomato mice (strain no. C001181) were purchased from Cyagen. Accordingly, C57BL/6J mice (strain no. N000013) were used as wild-type (WT) mice. To generate Clec4f^ΔChil1 mice, Chil1^flox//flox mice were crossed with Clec4f-cre mice and knock out efficiency was examined in KCs by qRT-PCR (Figure S2B). To generate Clec4f^{Rosa tdtomato} mice, Rosa tdtomato mice were crossed with Clec4f-cre mice to examine the expression specificity of Clec4f-cre (Figure S2C). To generate Lyz2^ΔChil1 mice, Chil1^flox//flox mice were crossed with Lyz2-cre mice and knock out efficiency was examined in BMDM by western blot (Figure S3B). All mouse colonies were maintained at the Animal Core Facility of Yunnan University. The animal studies were approved by the Yunnan University Institutional Animal Care and Use Committee (IACUC, Approval No. YNU20220314). Male mice aged 6-8 weeks were used in this study.

Construction of MASLD/MASH mouse model

Mice were provided a high-fat and high-cholesterol diet (Research Diet, d12108c, 40 kcal% fat and 1.25% cholesterol). Throughout the feeding period, the body weight and food consumption of the mice were observed and recorded weekly. Once the dietary intervention was completed, the mice were euthanized. Liver and murine serum samples were collected for further analysis. Alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels in the serum, as well as cholesterol (TC) and triglyceride (TG) levels in both serum and liver tissues, were quantified using commercially available kits (Nanjing Jiancheng Bioengineering Institute).

Statistical analysis

Data are presented as mean ± standard error of the mean (SEM) in all graph figures. Statistical analyses were conducted using the SPSS statistics software (Version 22). To compare the two groups, an unpaired two-tailed Student’s t-test was used. One-way analysis of variance (ANOVA) was performed for comparisons involving three or more groups. For patients with MASLD liver, the samples were tested using the Mann-Whitney test. Statistical significance was set at p < 0.05 and p value is indicated. All cell culture results represent at least three independent experiments.

Additional Methods

Additional detailed methods can be found in the Supporting Information.

Results

Hepatic macrophages express Chi3l1

To investigate the dynamic changes in hepatic macrophages, we performed BD Rhapsody scRNA-seq on non-parenchymal cells (NPCs) isolated from healthy livers of mice fed a normal chow diet (NCD) and from livers of mice with MASLD induced by a high-fat, high-cholesterol (HFHC) diet. After quality control and filtration, we retained 23,312 cells from NCD livers and 6,567 cells from HFHC livers for downstream analysis. Using a graph-based clustering approach, we identified 32 distinct cell populations, visualized via uniform manifold approximation and projection (UMAP) (Figure 1A). Monocyte/macrophage subsets were further defined based on lineage-specific markers: Monocytes expressed Ly6c2, Chil3, S100a6, Ccr2, Itgam, and Cx3cr1 but lacked macrophage markers. KCs were marked by Cd68, Vsig4, Clec4f, Timd4, Adgre1, and Clec1b. MoMFs were negative for KCs markers but positive for macrophage markers such as Ccr2, Cx3cr1, Cd9, Itgax, Gpnmb, Cd68, and Adgre1 (Figure 1B, C; UMAP in Fig S1A)^17,18. ScRNA-seq analysis revealed high expression of Chil1 (Chil1 being the gene name for Chi3l1) in hepatic macrophages (Figure 1D). To validate this, we performed immunofluorescence staining for Chi3l1 in hepatic macrophages using antibodies against Timd4, F4/80, and Chi3l1. Our results confirmed Chi3l1 expression in hepatic macrophages under both NCD and HFHC conditions (Figure 1E). Consistently, patients with MAFL or MASH exhibited elevated hepatic Chil1 mRNA levels, which correlated with MASLD severity and fibrosis stage (Figure 1F, G). These findings suggest that Chi3l1 is expressed in hepatic macrophages and may play a role in MASLD progression.

Figure 1.

Deficiency of Chi3l1 in Kupffer cells promotes insulin resistance and hepatic lipid accumulation

Given that hepatic macrophages are the primary source of Chi3l1 in the liver, we investigated its functional role by generating mice with conditional knockout (cKO) of Chil1 in either KCs or MoMFs. First, we generated Clec4f ^ΔChil1 mice by crossing Chil1^fl/fl mice with Clec4f-cre mice¹⁹, achieving selective Chil1 ablation in KCs (Figure S2A). When fed an HFHC diet, Clec4f^ΔChil1 mice exhibited: Increased hepatic lipid deposition, evident by H&E and Oil Red O staining as early as 8 weeks (Figure 2A, B). Elevated metabolic disturbances at 16 weeks, including higher liver index (liver/body weight), serum ALT, and increased cholesterol/triglycerides in both liver and serum (Figure 2C). Impaired glucose metabolism, with worsened glucose tolerance (IGTT) and insulin resistance (ITT) (Figure 2D, B). These results demonstrate that Chi3l1 deficiency in KCs drives lipid accumulation and insulin resistance.

Figure 2.

To assess the role of Chi3l1 in MoMFs, we generated Lyz2^ΔChil1 mice (Chil1^fl/fl × Lyz2-Cre¹⁹) (Figure S3A). After 20 weeks on an HFHC diet: Hepatic lipid accumulation (H&E, Oil Red O) was comparable to controls (Figure S4A, B). Metabolic parameters (liver index, ALT, serum/liver lipids) showed no significant differences (Figure S4C). Lyz2^ΔChil1 mice showed glucose intolerance (IGTT) but normal insulin sensitivity (ITT) (Figure S4D, E). Thus, Chi3l1 loss in MoMFs does not substantially contribute to metabolic dysregulation.

ScRNA-seq reveals upregulated glucose metabolism-related transcripts in KCs, correlating with cell death signatures

To dissect the distinct metabolic and functional profiles between KCs and MoMFs during MASLD progression, we analyzed our scRNA-seq data. Consistent with prior studies^6,7, we observed decreased KCs numbers but increased MoMFs and monocytes in HFHC-fed mice compared to NCD controls (Figure 3A). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis showed that while both cell types exhibited activation of phagocytosis-related pathways (lysosome, phagosome, endocytosis, and efferocytosis), they displayed divergent cell fate patterns (Figure 3B). KCs showed strong cell death signatures, whereas MoMFs maintained proliferative activity without evidence of cell death (Figure 3B). Monocytes showed strong cell death and proliferative activity (Figure 3B). Given the significant role of metabolic regulation in cell fate,²⁰ we compared pathways involved in glucose metabolism, cell death, and cell proliferation (Figure 3C). Notably, glucose metabolism pathways were significantly more active in KCs and monocytes compared to MoMFs (Figure 3C). Moreover, the cell proliferation pathway was highly activated in monocytes and consistently activated in MoMFs but not in KCs (Figure 3C). Gene Set Variation Analysis (GSVA)-based correlation analysis revealed a striking association between glucose metabolism and cell death pathways (Figure 3D). These findings demonstrate distinct glucose metabolic activation patterns between KCs and MoMFs, which may underlie their divergent cell fates in MASLD progression.

Figure 3.

Chi3l1 deficiency promote KCs death during MASLD

To investigate the role of Chi3l1 in KCs survival during MASLD, we generated Chil1^-/- mice, as Chi3l1 is a secreted protein^21,22. The successful knockout was confirmed by qRT-PCR analysis of liver tissue (Figure S5A, B). We then performed scRNA-seq on NPCs isolated from Chil1^-/- mice fed an HFHC diet for 16 weeks. After quality control, 6,813 high-quality cells were retained for analysis. Using established KC markers (Figure 1C), we conducted GSVA to examine metabolic pathways. This revealed enhanced cell death pathways in KCs from HFHC-fed mice, with significantly greater apoptosis signatures in Chil1^-/- KCs compared to wild-type (WT) controls (Figure 4A). The increased apoptosis was further supported by upregulation of pro-apoptotic genes in Chil1^-/- KCs (Figure 4B).

Figure 4.

Flow cytometry analysis validated these findings, using the gating strategy shown in Figure S6A. While WT and Chil1^-/- mice showed similar KC numbers at baseline, dramatic differences emerged during HFHC feeding. WT KC numbers remained stable at 8 weeks but decreased by 50% at 16 weeks. In contrast, Chil1^-/- mice exhibited accelerated KCs loss, with a 30% reduction by 8 weeks progressing to 60% by 16 weeks (Figure 4C, B). Notably, MoMFs populations remained comparable between groups at early timepoints but showed greater reduction in Chil1^-/- mice at 16 weeks (Figure 4C, B).

Histological analysis through Clec4f/TUNEL co-staining provided further confirmation. While WT livers showed no TUNEL⁺ KCs at baseline, 40% were TUNEL⁺ by 8 weeks, increasing to 55% at 16 weeks. Chil1^-/- mice displayed elevated KC apoptosis even at baseline, with significantly more TUNEL+ KCs (approximately 2-fold higher than WT) by 8 weeks (Figure 4E). Together, these results demonstrate that Chi3l1 deficiency accelerates HFHC-induced KCs death, leading to premature KCs depletion during MASLD progression.

Molecular interaction between Chi3l1 and glucose

Our investigation into Chi3l1-mediated KCs survival revealed an unexpected structural relationship: Chi3l1 binds to glucose, which is structurally analogous to chitin, a polysaccharide well known to bind Chi3l1(Figure 5A). Bioinformatics analysis using the STITCH database further supported this observation, predicting a high probability of direct Chi3l1-glucose interaction (Figure 5B). To experimentally validate this interaction, we performed pull-down assays using biotin-labeled glucose incubated with plasma from HFHC-fed mice. Streptavidin bead isolation followed by anti-Chi3l1 Western blotting demonstrated specific binding between Chi3l1 and biotin-glucose, but not biotin alone (Figure 5C, B). This interaction was competitively inhibited by unlabeled glucose, confirming specificity (Figure 5D). Quantitative analysis using microscale thermophoresis with recombinant mouse Chi3l1 (rChi3l1) yielded a dissociation constant (Kd) of 4.95 mM for the Chi3l1-glucose interaction (Figure 5E). Notably, circulating Chi3l1 levels were significantly elevated in serum from HFHC-fed mice compared to baseline (Figure 5F), suggesting a potential physiological role for this interaction in metabolic regulation. These findings establish Chi3l1 as a novel glucose-binding protein that may participate in glucose homeostasis during MASLD progression.

Figure 5.

Chi3l1 limits glucose uptake and protects hepatic macrophages from cell death

To elucidate the functional consequences of Chi3l1-glucose binding, we examined glucose metabolism in hepatic macrophages. Using the fluorescent glucose analog 2-NBDG²³, we performed uptake assays in KCs following 12-hour glucose starvation. While glycogen droplet size remained unchanged in untreated KCs regardless of rChi3l1 supplementation (Figure 6A), 2-NBDG exposure significantly increased glycogen accumulation. This effect was markedly suppressed by rChi3l1 co-treatment (Figure 6A), a phenotype replicated in BMDM (Figure 6B). These results demonstrate that Chi3l1 restricts glucose uptake and subsequent glycogen storage.

Figure 6.

Further validation using Stbd1 (a glycogen-binding protein²³) immunofluorescence revealed minimal glycogen foci in glucose-deprived BMDM, with no rChi3l1-dependent differences. High-glucose conditions, however, triggered robust glycogen aggregation, which was significantly attenuated by rChi3l1 (Figure 6C). Concordantly, extracellular acidification rate (ECAR) measurements showed reduced basal and total glycolytic capacity in rChi3l1-treated BMDMs (Figure 6D,E), confirming Chi3l1’s role in limiting glucose metabolism.

To test whether Chi3l1-glucose binding influence cell survival, we employed a palmitic acid (PA)-induced lipotoxicity cell-based model to better mimic the in vivo environment. rChi3l1 supplementation reduced PA-induced cleavage of caspase-3 (Figure 6F) and decreased KCs death (calcein/PI staining, Figure 6G). These findings collectively establish that Chi3l1 safeguards KCs by modulating glucose uptake, thereby suppressing metabolic stress-induced death.

Discussion

Our findings establish Chi3l1 as a critical metabolic regulator that controls hepatic macrophage fate through a novel glucose-dependent mechanism in MASLD. Using cell-specific knockout models, we uncovered a fundamental dichotomy in Chi3l1 function: selective ablation in KCs dramatically accelerated MASLD progression and metabolic dysfunction, whereas deletion in MoMFs produced minimal metabolic effects. Single-cell transcriptomics revealed the molecular basis for this cell-type specificity - KCs exhibit a glucose-hungry metabolic phenotype that renders them uniquely dependent on Chi3l1-mediated regulation, while MoMFs maintain a relatively glucose-independent metabolic program. At the mechanistic level, we demonstrate that Chi3l1 functions as a physiological glucose sensor, directly binding extracellular glucose to limit its cellular uptake. This interaction establishes a crucial metabolic safeguard that specifically protects glucose-dependent KCs from lethal metabolic stress while sparing glucose-independent MoMFs. Through this precise modulation of glucose availability, Chi3l1 maintains metabolic homeostasis and preserves KCs populations during chronic dietary challenge (Figure 7).

Figure 7.

Our study reveals fundamental differences in metabolic requirements between hepatic macrophage subsets that provide new insights into MASLD pathogenesis. We demonstrate that KCs and MoMFs play stage-specific roles in disease progression, with KCs serving as critical regulators of early metabolic homeostasis while MoMFs appear more involved in later inflammatory phases. This temporal specialization explains the striking dichotomy observed in our genetic models - KCs-specific Chi3l1 deletion dramatically exacerbated metabolic dysfunction, whereas MoMFs deletion showed minimal effects. The heightened glucose metabolism of KCs during MASLD renders them uniquely vulnerable to dietary stress. Chi3l1 serves as a crucial metabolic buffer in this context, directly protecting KCs through glucose modulation as evidenced by reduced glycogen accumulation and attenuated glycolytic flux. Our findings using the HFHC model complement previous findings in fibrogenic CDAA-HFAT models²⁴ or MCD/CCL4 models²⁵ or human livers²⁶, collectively suggesting Chi3l1 may have dual roles in MASLD - maintaining metabolic balance through KCs in early disease while potentially influencing fibrogenesis via MoMFs in advanced stages. The accelerated KCs death in knockout models provides direct experimental evidence linking macrophage survival to metabolic outcomes, resolving key questions about MASLD progression mechanisms.

The structural characteristics of Chi3l1 have been extensively studied. Chi3l1 is a homodimeric protein composed of two identical subunits. Each subunit consists of a catalytic domain and a carbohydrate-binding domain. The catalytic domain contains an active site capable of hydrolyzing chitin, a polymer of N-acetylglucosamine^27,28. The carbohydrate-binding domain is responsible for the interaction between Chi3l1 and various carbohydrate ligands²⁷. It is likely that Chi3l1 also interacts with glucose through its carbohydrate-binding domain. However, the precise molecular interactions between Chi3l1 and glucose are still not fully elucidated, and further research is needed to explore these interactions in greater detail. Furthermore, our data demonstrated that Chi3l1 is upregulated in the serum of mice fed a high-fat, high-cholesterol (HFHC) diet for 16 weeks (Fig. 3H). However, continuing to feed the mice for 24 weeks did not further increase the level of Chi3l1 (data not shown). This suggests that the level of Chi3l1 initially increases but reaches a plateau stage over time. These findings indicate that while Chi3l1 has the capability to bind with glucose, this interaction may be limited in living organisms.

Our findings carry important translational potential for MASLD treatment. The discovery of Chi3l1’s glucose-sensing function in KCs suggests two complementary therapeutic strategies: first, developing Chi3l1-based interventions to preserve KC populations during early metabolic dysfunction; second, creating cell-type-specific approaches that selectively modulate glucose metabolism in KCs while sparing MoMFs. The structural mapping of Chi3l1’s glucose-binding domain now enables rational design of small molecule mimetics to therapeutically enhance this protective pathway. Besides, several key questions emerge for future research to advance these therapeutic possibilities: (1) How glucose levels are coordinated with other death inducers such as lipid toxicity; (2) Whether competing carbohydrate ligands modulate Chi3l1’s glucose-sensing capacity in different metabolic states; (3) The clinical relevance of human Chi3l1 variants in MASLD susceptibility and progression. Addressing these questions will be crucial for translating our mechanistic insights into targeted therapies that account for the complex metabolic specialization of hepatic macrophage subsets.

Our findings reveal a novel metabolic checkpoint in which Chi3l1 selectively sustains KCs populations by modulating glucose metabolism, offering key insights into MASLD pathogenesis. The study highlights the therapeutic potential of targeting Chi3l1-glucose interactions to preserve protective KCs and curb MASLD progression. Future research should explore whether Chi3l1 supplementation or pharmacological modulation can rescue KCs viability, as well as investigate whether this mechanism extends to other macrophage-driven metabolic disorders, such as MASH or diabetes. By identifying cell type-specific metabolic vulnerabilities, this work paves the way for precision therapies that selectively manipulate macrophage subsets to treat liver disease.

Acknowledgements

We thank Dr. Bin Qi (Yunnan University) for suggestions and discussion. We thank Guangxun Meng (The Shanghai Institute of Immunity and Infection of the Chinese Academy of Sciences) for providing us with L929 cells. We thank Cynthia Ju (UTHealth) for advice in manuscript submission.