HepG2 cell lines (obtained from Dr. Pan, GIBH, CAS, Guangzhou, China) were maintained in Dulbecco’s Modified Essential Medium (DMEM)-L (Gibico, Grand Island, NY, USA) supplemented with 10% fetal bovine serum (FBS) (Nobimpex, Herbolzheim, Germany) at 37°C with humidified atmosphere of 5% CO₂. ASGR1^-/- HepG2 cells were constructed by clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9 [12]. To induce IR, the cells were incubated with 18 mM glucosamine (GLN) (Beyotime, Nantong, China) for 18 hours in DMEM-L without FBS [27,28], or 0.25 mM sodium palmitate (PA) (Sigma, St. Louis, MO, USA) in DMEM-H for 24 hours followed by incubation in DMEM-L for 2 hours [29]. Then, the cells were stimulated with 100 nM insulin (Sigma) for 20 minutes and harvested for quantitative real-time polymerase chain reaction (qPCR) and Western blot analyses.

RNA was prepared from liver tissues and 4.0×10⁵ cells by Trizol reagent (MCE, Monmouth Junction, NJ, USA). For more details on reverse transcription and qPCR, please refer to our previous work [12]. The relative mRNA expression was normalized to a referenced gene (β-actin). The sequence of primers was provided in Supplementary Table 1.

Liver tissues and 8.0×10⁵ cells were lysed on ice by radioimmunoprecipitation assay (RIPA) buffer (Beyotime) supplemented with protease inhibitor. For more details on protein sample preparation and electrophoresis, please refer to previous work [30]. The used antibodies were listed in Supplementary Table 2. Protein bands were analyzed using Immobilon Western Chemiluminescent horseradish peroxidase (HRP) Substrate (Millipore, Burlington, MA, USA).

The cells were starved for 12 hours, and about 10 million cells were harvested. A total of two tubes within each group were collected for RNA isolation. High-fat diet (HFD)-fed wild type (WT) and Asgr1^-/-mice were fasted overnight and collected liver tissues. Each sample was used to extract total RNA, and the RNA quality was determined by NanoDrop, Qubit (Thermo Fisher Scientific, Waltham, MA, USA), and Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). After the libraries were constructed, the sequencing was processed by IGE Biotechnology Ltd. (Guangzhou, China). All the RNA sequencing raw and processed data were provided in the supplementary materials.

Glycogen content was measured by Glycogen Assay Kit (Solarbio, Beijing, China). The cells and weighted liver tissues were lysed with extract buffer. Then the samples were boiled for 20 minutes, and the test tube was shaken every 5 minutes. After the sample was completely dissolved and cooled, the volume was fixed to 5 mL with distilled water, mixed well, and centrifuged at 8,000 ×g for 15 minutes. Glycogen concentrations was normalized by liver weight or protein concentration of cells.

Malondialdehyde (MDA) levels in mice liver were measured by MDA assay kit (Beyotime). The liver tissues were homogenized with Western and IP cell lysate (Beyotime) on ice. The ratio of liver weight to lysate was 1:9, after homogenization, the mixture was centrifuged (1.2×10⁴ ×g) at 4°C for 15 minutes. The protein content of liver tissues was determined by bicinchoninic acid (BCA) protein concentration assay kit (Thermo Fisher Scientific).

Superoxide dismutase (SOD) activity in mice liver was measured by the total superoxide dismutase assay kit with WST-8 (Beyotime). The liver samples were homogenized in pre-cooled phosphate buffer saline then the homogenate was centrifuged (1.2×10⁴×g) at 4°C for 15 minutes for subsequent assays.

Paraffin-embedded liver tissues were sectioned into 4 μm, and stained with standard hematoxylin and aqueous eosin (Beyotime). The optimal cutting temperature compound-embedded liver tissues were cut into 20 μm and stained with Oil Red O solution (Sigma-Aldrich) counterstained with hematoxylin. The morphologic characteristics of the liver tissues were acquired using a light microscope (Leica, Wetzlar, Germany).

Serum was separated from mice and frozen at –80°C until required. Serum insulin concentration was measured by enzymelinked immunosorbent assay (ELISA) (Millipore). Aspartate aminotransferase (AST)/alanine aminotransferase (ALT) levels were measured using an AST/ALT Assay Kit (Nanjing Jiancheng Bioengineering Institute, Nanjing, China).

Serum lipid levels were measured by an automatic biochemical detector (Hitachi, Tokyo, Japan). Liver lipid were measured by using the tissue triglyceride (TG)/total cholesterol (TC) content assay kit (Applygen, Beijing, China).

Glucose tolerance test (GTT) or insulin tolerance test (ITT) was performed on 16-hour-fasted or 6-hour-fasted mice, and 2 g/kg body weight glucose (Sigma-Aldrich) or with 0.8 IU/kg body weight insulin (Sigma-Aldrich) was intraperitoneally injected to mice, respectively. Blood glucose was monitored from the tail veins of mice with a glucometer at different times after the glucose or insulin administration [31]. The calculation of the area under the curve (AUC) or area above the curve (AAC) were generated by the glucose value of each time point with the subtraction of those initial glucose value [32].

ASGR1 knockout mice were generated in Cyagen (Guangzhou, China) [12]. All mice were of C57BL/6 genic background. Mice were maintained under a standard environment (12-hour light-dark cycle) with free access to normal chow diet and water. Considering the results of our previous studies [12], the male mice were used for experiments to reduce the potential effects of hormones on metabolism. Mice (6 weeks old) were randomly assigned and administrated with HFD (fat, 39.9%; cholesterol, 1.25%; Trophic Animal Feed High-Tech Co. Ltd., Nantong, China) for 8 weeks. For analysis of the hepatic insulin signaling status, those mice (HFD 8 weeks) were fasted for 6 hours, then injected with either vehicle or insulin (0.8 IU/kg) intraperitoneally, and the mice were euthanized 15 minutes later.

All animal studies were reviewed and approved by the Animal Care and Use Committee of GIBH, CAS (IACUC approval: 2019012).

The data were analyzed with GraphPad Prism 9.0 (GraphPad Software Inc., San Diego, CA, USA). The Student’s t-test was used to assess the significance between two groups, and the one-way analysis of variance (ANOVA) was applied for three groups. All data were represented as mean±standard deviation, and P<0.05 was considered statistically significant.

We reported that ASGR1 deficiency significantly reduced plasma lipid content under normal physiological condition [12]. To further explore the physiological functions of ASGR1, we first tested whether the HFD influences the ASGR1 expression and found that there were no significant changes between chow and HFD-fed WT mice (Supplementary Fig. 1A and B). Then, ASGR1 deficiency and WT mice were fed with HFD for 8 weeks in total, and at the end of the sixth and seventh week, the GTT and ITT assays were performed, respectively. Except there were significant increased glucose levels in short phase of GTT (15 and 30 minutes) of those knockout mice, the overall AUC were clearly declined, particularly in Asgr1^-/- mice (Fig. 1A). However, the ITT results indicated that the ASGR1-deficient mice revealed significantly improved insulin sensitivity with comparison of those WT mice, evidenced by the significantly increased AAC (Asgr1^+/-, 65.5%; Asgr1^-/-, 77.7%) (Fig. 1B). It is well known that the decreased of serum insulin level is an indicator of improved insulin sensitivity under metabolic stress [33,34]; therefore, the serum insulin levels were measured, and the serum insulin levels in Asgr1^+/- and Asgr1^-/- mice were significantly lower compared with those of WT mice (Fig. 1C). Furthermore, an important systemic IR index, the homeostasis model assessment of insulin resistance index (HOMA-IR) was also decreased in ASGR1-deficient mice (Fig. 1D). While there were no significant differences among groups in the body weight and food intake (Supplementary Fig. 1C and D), and ASGR1 deficiency had no significant changes of serum lipid under HFD condition (Supplementary Fig. 1E-H). These results suggested that deficiency of ASGR1 alleviated HFD-induced systemic IR.

To determine whether the improved systemic insulin sensitivity under ASGR1-deficiency has an organ-specific manner, we analyzed insulin signaling status in major insulin response tissues. It was found that insulin-induced phosphorylation of insulin receptor β (pInsRβ) and AKT was only increased significantly in livers (the pS-473 AKT, Asgr1^+/-, 70%; Asgr1^-/-, 50%) (Fig. 2A), but neither in adipose nor muscle tissues of ASGR1-deficient mice (Fig. 2B). This is consistent with previous report that hepatic IR is a key factor in the pathogenesis of systemic IR syndrome [4]. Furthermore, we observed that ASGR1 deficiency had no effect on the levels of AST and ALT (Supplementary Fig. 2A and B). However, although there were no significant differences among groups in liver mass and TG content, Asgr1^-/- mice had higher TC content in liver (Supplementary Fig. 2C-F). Abnormal lipid metabolism induced the production of reactive oxygen species, which in turn contributed to IR [35]. Therefore, we examined oxidative stress in liver. The levels of MDA were decreased in ASGR1-deficient mice, especially in Asgr1^+/- mice, along with increased SOD activity in ASGR1-deficient mice (Supplementary Fig. 2G and H). From another point of view, it confirms that ASGR1 deficiency can alleviate hepatic IR. In conclusion, the ASGR1 deficiency relieves systemic IR mainly via improved hepatic insulin sensitivity. To further explore the regulatory effects of ASGR1 deficiency on hepatocyte insulin signal transduction under IR status, the HepG2 and ASGR1^-/- HepG2 cell lines were exposed to GLN for inducing IR condition. We found that the pS-473 AKT was significantly higher in ASGR1^-/- HepG2 cells (increased 50%) upon insulin stimulation compared with the HepG2 cells (Fig. 2C), it also can be duplicated in PA-induced IR condition (Supplementary Fig. 3). These results suggested that ASGR1 deficiency significantly improved insulin sensitivity of hepatocytes, thus alleviated hepatic IR under metabolic stress condition.

To confirm the effect of ASGR1 on the insulin signal pathway, RNA sequencing was performed on those HFD-fed Asgr1^-/- and WT livers (the detailed data of RNA-seq is listed in Supplementary Table 3). Gene set enrichment analyses showed that the genes related to PI3K-AKT signaling were significantly enriched at the top of the differentially expressed genes (DEGs), with a high normalized enrichment score (=–1.53) (Supplementary Fig. 4A), indicating that the PI3K-AKT signaling pathway was significantly improved in Asgr1^-/- mice. Among those DEGs, the expressions with significant changes were 268 genes in total (fold change >2, q value <0.05) (Supplementary Fig. 4B). Importantly, these DEGs were indeed enriched in the insulin signal pathway (Supplementary Fig. 4C), that were consistent with other results from cell and mice models under metabolic stress.

To further demonstrate the effects of ASGR1 on hepatocyte metabolism, transcriptome analyses were also performed on ASGR1^-/- HepG2 and HepG2 cells (detailed in Supplementary Table 4). The similarity of transcriptome between ASGR1^-/- HepG2 and HepG2 cells was over 98% (Fig. 3A), except there were 249 DEGs (80 up, 169 down) with significant changes (fold change >2, P<0.05) in the ASGR1^-/- HepG2 compared to the HepG2 cells. Hierarchical clustering by significant DEGs was shown by heatmap (Fig. 3B). Gene Ontology (GO) analysis and the Kyoto Encyclopedia of Genes and Genomes (KEGG) were used to characterize the functions of the DEGs and reveal pathways with significant enrichment, respectively [36]. According to the significant DEGs, the distributions between the DEGs and certain specific GO classification terms and pathways were calculated. As expected, those dysregulated genes were enriched in insulin responses (Fig. 3C), particularly in PI3K-AKT signaling (Fig. 3D and Supplementary Table 5). Once again, it indicates that ASGR1 participated in regulating hepatic insulin signaling and glucose metabolism.

Based on suggestions from the transcriptome analyses, we shed light on the status of hepatic glucose metabolism, in particular on the two main pathways of HGP, glycogenolysis and gluconeogenesis. After 8 weeks of HFD feeding, the glycogen analyses found that deficiency of ASGR1 significantly increased hepatic glycogen concentrations (Asgr1^+/-, increased 67%; Asgr1^-/-, increased 25%) (Fig. 4A), and the mRNA expression level of glycogen phosphorylase (Pygl, response for glycogenolysis) was significantly reduced in ASGR1-deficient mice (Fig. 4B). While the mRNA expression of glycogen synthase 2 (Gys2) and glucokinase (Gck) (responsible for glycogen synthase) had no significant differences among groups (Supplementary Fig. 5). However, the expressions of two key gluconeogenic genes, Pck1 and G6pc were significantly reduced in ASGR1 deficiency mice liver (Fig. 4C). It suggested that ASGR1 deficiency contributed to reductions of glycogenolysis and gluconeogenesis under IR condition. Further analysis of glucose metabolism in hepatocytes with GLN induction was consistent with the results of animal model, concretely, the glycogen content increased about 59% in ASGR1^-/- HepG2 compared to the HepG2 cells (Fig. 4D), and the mRNA of PYGL showed a decreased trend (Fig. 4E); moreover, the mRNA of PCK1 and G6PC were reduced, along with their declined proteins (PEPCK1 and G6Pase, downed 90% and 43%, respectively) (Fig. 4F and G). Mechanistically, these results indicated that deficiency of ASGR1 led to improved hepatocellular insulin signal transduction linked with decreased glycogenolysis and gluconeogenesis under IR status.

To investigate whether the effects of ASGR1 on hepatic glucose metabolism and insulin signaling are metabolic stress-related, we further analyzed the insulin signal pathway status in the liver of mice fed with a standard diet, and found that there were no significant differences in the phosphorylation of the InsRβ and AKT among groups (Fig. 5A). However, the liver glycogen content was increased in ASGR1-deficient mice (Asgr1^+/-, 40%; Asgr1^-/-, 100%) (Fig. 5B), and the mRNA expression of Gys2 and Gck were significantly down-regulated, while the mRNA expression level of Pygl showed a slight downregulation in ASGR1-deficient mice (Fig. 5C). Moreover, the mRNA of Pck1 and G6pc, as well as their proteins were also decreased in ASGR1-deficient mice (G6Pase and PEPCK1, 45% and 34% in Asgr1^+/-; 47% and 56% in Asgr1^-/-, respectively) (Fig. 5D and E).

Those phenotypic changes in mice were further investigated in cell lines. Under routine culture condition, the insulin signaling had no significant changes between cells with or without ASGR1 (Fig. 6A and Supplementary Fig. 6); however, the glycogen content was 60% higher in ASGR1^-/- HepG2 cells compared with HepG2 cells (Fig. 6B), and mRNA of PYGL was significantly decreased (Fig. 6C), accompanied with decreased expression of G6PC and PCK1 at mRNA and protein levels (G6Pase and PEPCK1, 28% and 90%, respectively) (Fig. 6D and E). These observations suggest that ASGR1 deficiency modulates hepatic glucose metabolism mainly by decreased glycogenolysis and gluconeogenesis in hepatocytes that were independent of metabolic stress.