Fig 2: CD36+ CAFs predict efficacy of HCC immunotherapy and targeting MIF synergizes with immunotherapy in HCC murine model.a–c The HCC initiation and progression were evaluated in Cd36 and Mif conditional knockout (Acta2Cre) mice. d The proportion of MDSCs was downregulated in Cd36 and Mif conditional knockout (Acta2Cre) mice. e, f The prediction performance of CD36+ CAFs in HCC immunotherapy. g–k CD36 inhibitor SSO sythesizes PD-1 blockade in C57/BJ6 spontaneous HCC model. l, m The changes of Tregs, MDSC, IFN-γ+, GZMB+ CD8+ T cells in four different groups. Data shown as mean ± S.E.M., one-way ANOVA following multiple comparison test was used, ***P < 0.001, **P < 0.01, *P < 0.05, and ns not significant.

Fig 3: CD36+ CAF-derived MIF potentiates the capacity of MDSCs to promote an immunosuppressive TME and tumor stemness via IL-6/STAT3 activation.a The schematic diagram showed co-culture of MDSCs precursors with condition media (CM) of CD36+CAFs, CD36–CAFs, CD36kdCAFs, CD36+CAFs+ISO-1 or blocking CD74. b, c MDSC proportion was measured by flow cytometry when co-cultured with vehicle Ctrl, CD36+ CAFs, CD36– CAFs, CD36kd CAFs, a combination of CD36+ CAFs and MIF inhibitor ISO-1, or a combination of CD36+ CAFs and CD74 blocking agents. d The schematic diagram showed co-injection of tumor cells and CD36+CAFs at the ratio of 30:1 in orthotopic HCC model. e Representative images of HCC tumors from the orthotopic HCC model. f Liver weight, tumor numbers, tumor volume and ratio of liver weight and mice weight from the orthotopic HCC model. CAFs transduced with the empty lentiviral vector as a control. g The proportion of effector CD8+ T cells from different groups in the orthotopic HCC model. h The western blot and ELISA assays showed iNOS was evaluated in MDSCs-WT, CD36+CAFs+MDSCs, a combination of CD36+CAFs+MDSCs and ISO-1 or αGr-1. i, j The proportion of CD69+CD8+ T cells was evaluated in MDSCs-WT, CD36+CAFs+MDSCs, a combination of CD36+CAFs+MDSCs and ISO-1 or blocking CD74 by flow cytometry. k, l GSEA and KEGG analysis shows top pathways enriched in MDSC-CD36+CAF-CM vs MDSC-WT. m Western blotting experiment shows NF-kB pathway changes in MDSCs treating with CD36+ CAFs. n ELISA assay shows IL-6 secretion in MDSCs treating with CD36+ CAFs, CD36kd CAFs, MIF or a combination of MIF and ISO-1. Data shown as mean ± S.E.M., one-way ANOVA following multiple comparison test was used, ***P < 0.001, **P < 0.01, *P < 0.05, and ns not significant.

Fig 4: CD36 mediates OxLDL uptake to promote MIF expression via the lipid peroxidation/p38/CEBPs axis in CD36+ CAFs.a Isolation of CAF subtypes from HCC tumors via flow cytometry. b The average proportion of CAF subtypes in human or murine HCC tumors by flow cytometry. c Heatmap showed specific gene markers in CAF subtypes from murine and human HCC tumors. d, e The qPCR and ELISA assays showed MIF expression was higher in CD36+CAFs among all fibroblasts. f The ELISA and western blot assays showed MIF expression was downregulated when CD36 was knockdown in CAFs. g The activity of reactive oxygen species pathway gene signatures in different CAF subclusters. h GSEA shows top enriched pathways in CD36high vs CD36low group and CD36kd vs WT group. i–k Uptake of OxLDL and lipid peroxidation in CD36+ or CD36kd CAFs was measured using fluorescently conjugated OxLDL and flow cytometry. l Human CD36+CAFs treated with vehicle Ctrl, LDL (60 μg/mL), or OxLDL (30 or 60 μg/mL) for 24 h and then washed in PBS and incubated with BODIPY 581/591 C11 for the lipid peroxidation assay. m CD36+ or CD36kd CAFs were treated with vehicle Ctrl, OxLDL (60 μg/mL), Toco (200 mM), SSO (100 mM), a combination of OxLDL (60 μg/mL) and Toco (200 mM), or a combination of OxLDL (60 mg/mL) and SSO (100 mM) for another 24 h. p38 phosphorylation (p-p38) was measured by flow cytometry, and the MFI of p-p38 was normalized to Ctrl. n The expression of p-p38 among CD36+CAFs, CD36kdCAFs and CD36–CAFs from in vivo HCC murine models. o CD36+CAFs were treated with vehicle Ctrl, OxLDL (60 μg/mL), SSO (100 mM), p38 inhibitor SB203580, a combination of OxLDL and SSO, or a combination of OxLDL and SB203580 for another 24 h. MIF secretion was measured by ELISA experiments, and the expression of MIF was nomalized to Ctrl. p CEBPA and CEBPD in CD36+CAFs modulated the transcriptional expression of MIF by ChIP assays. Data are mean ± s.d. of n = 3 independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001 by Student’s t test. n = 3 biological replicates. Data shown as mean ± S.E.M., one-way ANOVA following multiple comparison test was used, ***P < 0.001, **P < 0.01, *P < 0.05, and ns not significant.

Fig 5: Increasing MIF restrained mitophagy through disturbing the interaction of PINK1 and Parkin.a The PPIs of PINK1 (MIF was marked by red box). b The PPIs of MIF (PINK1 was marked by red box) c The protein interaction of PINK1-MIF and PINK1-Parkin was measured by co-immunoprecipitation and immunoblotting. The gray value ratios were listed below each blot, reflecting the degree of interaction. d Representative confocal images showed the colocalization of MIF (green) and PINK1 (red) in different groups of HK-2 cells. Scale bar: 5 μm. The correlation of colocalization was presented as the Rr value of Pearson’s correlation coefficient (negative correlation: −1.0 to 0, positive correlation: 0–1.0, the correlation strength increased with the absolute value of Rr). e Representative images of triple-labeling MIF (green), PINK1 (red), and Parkin (white) immunofluorescence showed the colocalization of MIF & PINK1 (yellow arrows) and PINK1 & Parkin (pink arrows). Scale bar: 10 μm. n = 3. *P <0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.