Fig 2: Survival rate of anti-rheumatic drugs between RA patients with and without anti-cit-OPN antibodies. A Positivity of anti-cit-OPN antibody from RA sera collected over 3 consecutive years, measured by ELISA. B Relationship between serum anti-CCP antibody and anti-cit-OPN antibody using stored serum in the 2016 cohort. C Survival rate of TNF inhibitors due to lack of efficacy between RA patients with and without anti-cit-OPN antibodies in the 2016 cohort. Patients in whom TNF inhibitors were discontinued within 90 days were excluded. D Survival rate of CTLA4-Ig due to lack of efficacy between RA patients with and without anti-cit-OPN antibodies. Patients in whom the drug was discontinued within 90 days were excluded

Fig 3: Single-cell RNA-sequencing analysis associated the immune co-expression network with liver macrophages. (A) Two-dimensional UMAP visualization of the liver cancer cells in the scRNA-seq GSE166635 cohort. (B) Expression of markers for cell types presented in (A). (C) Enrichment of the intersected genes between LIHC and LIRI consensus co-expression clusters (see Figures 1D, E ) in the scRNA-seq discovery cohort evaluated by ssGSEA. (D) Heatmap shows the expression of the top-ranked markers for C1QA +, SPP1 +, and VCAN + liver macrophages. (E) Expression of the markers for C1QA +, SPP1 +, and VCAN + macrophages in the macrophage subsets Mφ-C1-THBS1 and Mφ-C2-C1QA in GSE140228 cohort (8). (F) Overrepresentation of the markers for C1QA +, SPP1 +, and VCAN + macrophages in the macrophage subsets FOLR2 + TAMs (TRMs, tissue-resident macrophages), FOLR2 + TAMs (MDMs, monocyte-derived macrophages), SPP1 + TAMs, and MT1G + TAMs in GSE156337 cohort (39). Significance was evaluated by hypergeometric testing followed by Benjamini-Hochberg correction (adjusted p-value).

Fig 4: SPP1 + macrophages interact with T cells through the SPP1 – CD44 ligand-receptor pair. (A) Cell-cell interactions between 18 cell types in HCC. Edges represent the total interaction strength between two cell types; node size indicates the number of cells in one cell type. Interaction maps from C1QA +, SPP1 +, and VCAN + macrophages to the other cell types are shown independently. (B) Dot plot shows the communication probability based on the ligand-receptor pair between C1QA +, SPP1 +, VCAN + macrophages, and T cells. Only significant ligand-receptor pairs are shown as dots. (C) Violin plots show the expression of SPP1 in the 18 cell types. (D) Western blot shows the level of CD44 in CD8+ T cells treated with 400ng/ml of purified SPP1. Band intensity was normalized with GAPDH.

Fig 5: A high fraction of SPP1 + macrophages is harmful to HCC patients’ survival. (A) Relative fractions of C1QA +, SPP1 +, VCAN +, and cycling macrophages and DCs in the LIHC and LIRI cohorts estimated by CIBERSORTx. (B) Comparison of the relative fractions of the five subtypes in (A) between normal tissues and tumors in LIHC and LIRI cohorts. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by Wilcoxon rank-sum test. (C) Forest plot shows the hazard ratio (HR) and 95% confidence interval (CI) of the association of relative macrophage/DC content to patients’ survival, evaluated by univariate Cox survival model. (D) Cell differentiation trajectories between macrophage subtypes and DCs. The zero-pseudo time was determined as the cell expressed the highest monocyte marker CD14. (E) The top 10 enriched TFs for C1QA +, SPP1 +, and VCAN + macrophages identified by SCENIC. (F) Average (relative) expression of predefined functional markers in the macrophages/DCs from the GSE166635 cohort.