Animal models

To establish a chemically induced HCC model, we injected diethylnitrosamine (DEN) (25 mg/kg) intraperitoneally once into two-week-old male C57BL/6 mice. Additionally, we injected 10% CCl4 (0.5 mL/kg) intraperitoneally starting at 4 weeks of age once a week for 20 weeks. The mice were euthanized, and tissues were collected at 24 weeks of age.

For the orthotopic HCC model, we implanted 3 × 106 Hepa 1–6 cells (1:1 mixed Matrigel, Corning) into the livers of 4-week-old male wild-type (WT) and Trem2−/− C57BL/6 mice (NM-KO-190402, Shanghai Model Organisms Center, Inc.) and euthanized them for tissue harvesting three weeks later. Trem2 gain-of-function experiments were conducted via intravenous tail vein injection of Trem2-expressing adeno-associated viruses (AAV-Trem2; AAV8-F4/80-Trem2, Hanheng Biotechnology) or AAV-vehicle (1.2 × 1011 vector genomes) seven days before the orthotopic injection of Hepa 1–6 cells.

To establish a subcutaneous HCC model, we injected 3 × 106 Hepa 1–6 cells (1:1 mixed Matrigel, Corning) into the ventral blood supply-rich area of four-week-old male mice. Tumor growth was continuously observed, and tumor dimensions were recorded for three weeks. Then, the mice were euthanized for tissue harvesting. CD8+ T cell depletion was conducted by intraperitoneal injection of InVivoMAb anti-mouse CD8α (200 µg/mouse; BE0061, BioXCell) every three days for a total of six times, starting three days after tumor cell inoculation. The tumor volume was estimated using the following formula: (length * width2)/2.

The grouping of all the animals was randomized. The experiments were approved by the Animal Experimental Ethical Inspection of the First Affiliated Hospital, Shihezi University.

Human samples

Formalin-fixed paraffin-embedded samples from 109 patients with HCC who did not receive radiotherapy or chemotherapy and underwent surgical resection between 2011 and 2019 were collected at the First Affiliated Hospital of Shihezi University. The study was approved by the Science and Technology Ethical Committee of the First Affiliated Hospital, Shihezi University.

Immunohistochemistry (IHC)

Sections of paraformaldehyde-fixed, paraffin-embedded mouse liver samples (4 μm thick) and human HCC tissue microarrays were subjected to antigen retrieval using a pressure cooker. Endogenous peroxidase activity was inhibited via the addition of 3% hydrogen peroxide. Nonspecific binding was blocked using 10% goat serum. The sections were subsequently incubated with antibodies against mouse Trem2 (1:300, ab86491, Abcam), CD8 (1:800, #98941, Cell Signaling Technology), CD4 (1:800, BS6982, Bioworld Technology), FOXP3 (1:800, ab215206, Abcam), cleaved caspase-3 (1:800, #9661, Cell Signaling Technology), TGF-β1 (1:800, ab215715, Abcam), PKM2 (1:1600, #4053, Cell Signaling Technology), and human TREM2 (1:1000, #91068, Cell Signaling Technology) at 4 °C overnight. The sections were then incubated with secondary antibody (ZSGB-BIO) and DAB substrate and counterstained with hematoxylin. Images were acquired via a KF-PRO-005 Digital Slide Scanner (KFBIO). The final staining scores of CXCL10 and PKM2 were calculated by multiplying the score for the percentage of positive cells (range, 0–4; 0, negative or ≤ 5%; 1, 6–25%; 2, 26–50%; 3, 51–75%; and 4, 76–100%) by the intensity score (range, 0–3; 0, negative; 1, weak; 2, moderate; and 3, strong) [17]. The final staining score of TREM2 was the product of the intensity score (range, 0–3; 0, negative; 1, weak; 2, moderate; and 3, strong) and the percentage of positive cells (range, 0–100) as previously described [18]. The staining score of TGF-β1 was determined as the percentage of positive cells per field.

Cell culture

Bone marrow cells were collected by washing the femurs and tibias of WT and Trem2−/− mice with PBS. After the red blood cells were lysed, the cell suspension was filtered through a 70-µm sieve. The cells were cultured in DMEM supplemented with murine recombinant M-CSF (40 ng/mL; 315-02, PeproTech) at 37 °C in a humidified 5% CO2 atmosphere for seven days, differentiating into mouse bone marrow-derived macrophages (BMDMs). Fresh complete DMEM was used for culture for 24 h, and the supernatant of the medium was collected as conditioned medium (CM).

Authenticated human monocytic THP-1 and mouse Hepa 1–6 cells without mycoplasma contamination were obtained from the Cell Bank, Type Culture Collection, Chinese Academy of Sciences. THP-1 cells were maintained in RPMI-1640 medium (Gibco) supplemented with 10% fetal bovine serum (FBSST-01033, Cyagen), 5 µM 2-mercaptoethanol (21985023, Gibco), and 1% penicillin-streptomycin. THP-1 cells were exposed to phorbol 12-myristate-13-acetate (PMA; CS0001, Multi Sciences) at 50 ng/mL for 48 h to differentiate into macrophage-like cells. Hepa 1–6 cells were maintained in complete DMEM (D6429, Sigma-Aldrich) supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin.

Cytokine assay

The cell culture supernatant of the BMDMs was collected and centrifuged to remove particulates. Next, cytokines and chemokines in the cell culture supernatants were profiled via the Mouse Cytokine Array, Panel A (ARY006, R&D Systems), according to the manufacturer’s instructions. Images were captured using a chemiluminescence imaging system (Clinx Science Instruments).

Real-time PCR

Total RNA was extracted using the E.Z.N.A. Total RNA Kit (R6834-01, Omega Bio-Tek) and reverse transcribed into cDNA via the HiFiScript cDNA Synthesis Kit (CW2569M, CWBIO) following the manufacturer’s instructions. Gene expression analyses were conducted via TB Green Premix Ex Taq II (Takara) on a CFX96 Touch Real-Time PCR Detection System (Bio-Rad Laboratories). ACTB served as the internal control. The primers used are listed in Supplementary Table 1.

Western blotting

Protein lysates were prepared from cells using RIPA lysis buffer (R0010, Solarbio) containing protease and phosphatase inhibitors (P1260, Solarbio). The protein concentration was determined via a BCA protein assay (CW0014S, CWBIO). Proteins were separated by 10% or 12% SDS‒polyacrylamide gel electrophoresis and transferred onto Immobilon-FL PVDF membranes (IPLH00010, Millipore). The membranes were incubated overnight at 4 °C with primary antibodies against human TREM2 (1:1000, #91068, Cell Signaling Technology), mouse Trem2 (1:1000, AF1729, R&D Systems), PKM2 (1:1000, #4053, Cell Signaling Technology), and β-actin (1:1000, #TA-09, ZSGB-BIO). The membranes were subsequently incubated with secondary antibodies for 2 h at room temperature. The bands were detected using ECL Western Blotting Substrate (BL520A, Biosharp).

Cell proliferation assay

Hepa 1–6 cells were seeded in 96-well plates at a density of 3000 cells per well, with BMDM CM mixed with complete medium (1:2). Cell proliferation was assessed at 0, 24, 48, and 72 h using the Enhanced Cell Counting Kit-8 (BL1055B, Biosharp) according to the manufacturer’s instructions.

Data collection and processing

The Cancer Genome Atlas Liver Hepatocellular Carcinoma (TCGA-LIHC) dataset was utilized for gene expression correlation analysis, survival analysis, and enrichment analysis, and the results were then validated with the GSE14520 dataset obtained from the Gene Expression Omnibus (GEO). The list of genes encoding secreted proteins was compiled from the Human Protein Atlas (https://www.proteinatlas.org/). Molecules capable of regulating the glycolytic pathway were identified via a literature search (Supplementary Table 2).

Gene set enrichment analysis

Gene set enrichment analysis for all the hallmark gene sets associated with TREM2 expression in TCGA-LIHC and GSE14520 datasets was conducted, and the data were visualized; both analysis and data visualization were performed using the “GSVA” package in R [19].

Identification of differentially expressed genes

We used the “DEseq2” package in R to identify differentially expressed genes [20]. Differential expression was analyzed between WT BMDMs and Trem2−/− BMDMs based on log2-transformed gene expression levels. Genes with |log FC| ≥ 0.585 and P < 0.05 were considered significantly differentially expressed.

ELISA

The cell culture supernatant was diluted 30 times, and the IL-1β concentration was assessed using the Mouse IL-1β ELISA Kit (EK201B, MULTISCIENCES) according to the manufacturer’s instructions. Finally, the OD values were measured via an enzyme-linked immunosorbent assay. The final OD value = OD450 value-OD630 value.

Glucose consumption and lactic acid production

A total of 5 × 105 cells were cultured in 6-well plates for 24 h and then cultured for another 24 h with BMDM CM mixed with complete medium (1:2) with or without raleukin (50 ng/mL; HY-108841, MedChemExpress). The medium was collected, and the changes in the glucose consumption and lactate production capacity of the cells were detected via the Glucose Assays Kit-WST (G264, DojinDo) and Lactic Acid (LA) Content Assay Kit (BC2235, Solarbio) [21, 22].

Statistical analysis

Numerical data are presented as the means ± SDs and were analyzed via GraphPad Prism and R software. The Kaplan‒Meier method was used to analyze patient survival, and the log-rank test was performed for statistical analysis. A t test or ANOVA was used to analyze the significance of differences between groups. Correlations between genes were assessed according to Spearman correlation coefficients. Differences were considered statistically significant at a P value < 0.05.