Antigen-presenting cancer associated fibroblasts enhance antitumor immunity and predict immunotherapy response

Abstract

Cancer-associated fibroblasts (CAF) play a crucial role in tumor progression and immune regulation. However, the functional heterogeneity of CAFs remains unclear. Here, we identify antigen-presenting CAFs (apCAF), characterized by high MHC II expression, in gastric cancer (GC) tumors and find that apCAFs are preferentially located near tertiary lymphoid structures. Both in vivo and in vitro experiments demonstrate that apCAFs promote T cell activation and enhances its cytotoxic and proliferative capacities, thereby strengthening T cell-mediated anti-tumor immunity. Additionally, apCAFs facilitate the polarization of macrophages toward a pro-inflammatory phenotype. These polarized macrophages, in turn, promote the formation of apCAFs, creating a positive feedback loop that amplifies anti-tumor immune responses. Notably, baseline tumors in immunotherapy responders across various cancer types exhibit higher levels of apCAFs infiltration. This study advances the understanding of CAFs heterogeneity in GC and highlights apCAFs as a potential biomarker for predicting immunotherapy response in pan-cancer.

Fig. 1. Identification of antigen-presenting CAFs in patients with GC.

A Schematic overview of analytical workflow and experimental design to discover the presence of apCAFs in patients with GC. Created in BioRender. Song, J. (2024) https://BioRender.com/n91k593. B UMAP plot depicting and graph-based clustering of CAFs from 38 patients. C Dotplot displaying the expression of marker genes for CAFs subclusters. D UMAP showing all identified CAFs subtypes based on marker genes. E Heatmap showing pathways enriched in CAFs subclusters. F Representative images of multiplex immunofluorescence characterizing the apCAFs in FFPE sections (green: CD74; red: PDGFRA). Multiplexed immunofluorescence assays are performed twice on tumor samples following assay optimization. G Flow cytometry of apCAFs and their expression profile in fresh gastric tumors.

Fig. 2. apCAFs were enriched in tumor tissues and associated with favorable prognosis in GC.

A Representative multiplex immunofluorescence images and the ratio of apCAFs in tumor (n = 71) and normal tissues (n = 71) (green: CD74; red: PDGFRA). Error bars represent the mean ± SEM. Statistical significance was determined using Mann–Whitney two-sided test. Source data are provided as a Source Data file. B The ratio of apCAFs quantified in tumor and normal tissues by flow cytometry (n = 10). Statistical significance was determined using two-sided paired t-test. Source data are provided as a Source Data file. C Clinical characteristics associated with apCAFs immunofluorescence staining in FUSCC cohort (n = 140). Statistical significance was determined using two-sided Chi-square test. Source data are provided as a Source Data file. D Survival analysis of patients with different infiltration levels of apCAFs in FUSCC cohort (n = 140). Statistical significance was determined using log-rank test. Source data are provided as a Source Data file. E Survival analysis of patients with different infiltration levels of apCAFs assessed through apCAFs signature in ACRG cohort (n = 300). Statistical significance was determined using log-rank test. Source data are provided as a Source Data file.

Fig. 3. apCAFs exhibited a predominant localization around TLS.

A H&E staining of tissue section for spatial transcriptoms. B Abundance estimation of various cell populations in spatial transcriptomics. C Abundance estimation and relationship of tertiary lymphoid structures (TLS) signature and apCAFs signature in spatial transcriptomics. Correlation was evaluated using the two-sided Spearman rank correlation coefficient. D Representative images of multiplex immunofluorescence characterizing the distribution of TLS and apCAFs (green: CD74; red: PDGFRA; white: CD19; purple: CD3). Multiplexed immunofluorescence assays are performed twice on tumor samples following assay optimization. E H&E staining and the abundance estimation and relationship of TLS signature and apCAFs signature in another spatial transcriptomics section from different patient. Correlation was evaluated using the two-sided Spearman rank correlation coefficient. F Representative H&E staining images of tissue sections (left) and the proportion of TLS-positive tumors (right) in high apCAFs group (n = 70) and low apCAFs group (n = 70). Statistical significance was determined using two-sided Chi-square test. Source data are provided as a Source Data file.

Fig. 4. apCAFs modulated immune profile and promoted antitumor immune response.

A Schematic representation of apCAFs adoptive transfer in vivo experiments. Created in BioRender. Song, J. (2024) https://BioRender.com/j06i038. B Tumor growth curves showing tumor volume in mice from control group, apCAFs group and MHC II^- CAFs group (n = 5 mice/per group). Error bars represent the mean ± SD. Statistical significance was determined using one-way ANOVA, followed by a two-sided Dunnett’s test to compare the specific differences between the groups. Source data are provided as a Source Data file. C Distribution of immune cell clusters in tumors from control, apCAFs and MHC II^- CAFs groups (n = 2 mice/per group). D The proportion immune cell clusters in tumors from control, apCAFs and MHC II^- CAFs groups (n = 2 mice/per group). E Relative proportion of total CD4⁺ T, CD69⁺CD4⁺ T cells, IFN-γ⁺CD4⁺ T cells and GZMB⁺CD4⁺ T cells in tumors from control, apCAFs and MHC II^- CAFs groups (n = 5 mice/per group). Error bars represent the mean ± SEM. Statistical significance was determined using one-way ANOVA, followed by a two-sided Dunnett’s test to compare the specific differences between the groups. Source data are provided as a Source Data file. F Relative proportion of total CD8⁺ T, GZMB⁺CD8⁺ T cells, IFN-γ⁺CD8⁺ T cells and TNF⁺CD8⁺ T cells in tumors from control, apCAFs and MHC II^- CAFs groups (n = 5 mice/per group). Error bars represent the mean ± SEM. Statistical significance was determined using one-way ANOVA, followed by a two-sided Dunnett’s test to compare the specific differences between the groups. Source data are provided as a Source Data file. G The concentration of IFN-γ within tumors from control, apCAFs and MHC II^- CAFs groups (n = 5 mice/per group). The box is bounded by the first and third quartile with a horizontal line at the median and whiskers extend to the maximum and minimum value. Statistical significance was determined using one-way ANOVA, followed by a two-sided Dunnett’s test to compare the specific differences between the groups. Source data are provided as a Source Data file. H The concentration of TNF within tumors from control, apCAFs and MHC II^- CAFs groups (n = 5 mice/per group). The box is bounded by the first and third quartile with a horizontal line at the median and whiskers extend to the maximum and minimum value. Statistical significance was determined using one-way ANOVA, followed by a two-sided Dunnett’s test to compare the specific differences between the groups. Source data are provided as a Source Data file. I Representative images of multiplex immunofluorescence showing T cell marker (green: CD3) and cytotoxic molecule (red: GZMB) expression in the high apCAFs group (n = 80) and low apCAFs group (n = 60) from the FUSCC cohort. J Relative proportion of overall T cells (CD3⁺ cells) and cytotoxic T cells (GZMB⁺CD3⁺ cells) in high apCAFs group (n = 70) and low apCAFs group (n = 70) in FUSCC cohort. Error bars represent the mean ± SEM. Statistical significance was determined using Mann–Whitney two-sided test. Source data are provided as a Source Data file.

Fig. 5. apCAFs augmented T cell-mediated anti-tumor immune response.

A Distribution (up) and proportion (down) of CD4⁺ T cells subclusters in tumors from control, apCAFs and MHC II^- CAFs groups (n = 2 mice/per group). B Genes expression heatmap of CD4⁺ T cells in tumors from control, apCAFs and MHC II^- CAFs groups (n = 2 mice/per group). C Gene expression dynamics along the CD4⁺ T cells trajectory. D Schematic representation of apCAFs and CD4⁺ T cells co-cultured experiments. OVA: ovalbumin. Created in BioRender. Song, J. (2024) https://BioRender.com/l34h504. E The expression of CD69, GZMB and IFN-γ in CD4⁺ T cells co-cultured with apCAFs or MHC II^- CAFs, which was measured by flow cytometry (n = 3 biological replicates for each experiment. Data are representative of 3 independent experiments.). Error bars represent the mean ± SEM. Statistical significance was determined using one-way ANOVA, followed by a two-sided Dunnett’s test to compare the specific differences between the groups. Source data are provided as a Source Data file. F The cytotoxicity scores of CD8⁺ T cells (n = 889) in tumors from control, apCAFs and MHC II^- CAFs groups. Statistical significance was determined using one-way ANOVA, followed by a two-sided Dunnett’s test to compare the specific differences between the groups. G Representative images of CFSE-labeled CD8⁺ T cells and the proportion of proliferated CD8⁺ T cells after co-culture with control CD4⁺ T cells or apCAFs-treated CD4⁺ T cells (n = 3 biological replicates for each experiment. Data are representative of 3 independent experiments.). Error bars represent the mean ± SEM. Statistical significance was determined using unpaired two-sided t-test. Source data are provided as a Source Data file. H Representative flow cytometry strategy and relative proportion of GZMB⁺CD8⁺ T cells, IFN-γ⁺CD8⁺ T cells and TNF⁺CD8⁺ T cells in total CD8⁺ T cells after co-culture with control CD4⁺ T cells or apCAFs-treated CD4⁺ T cells (n = 3 biological replicates for each experiment. Data are representative of 3 independent experiments.). Error bars represent the mean ± SEM. Statistical significance was determined using Mann–Whitney two-sided test. Source data are provided as a Source Data file. I The proportion of tumor cells killed by CD8⁺ T cells after co-culture with control CD4⁺ T cells or apCAFs-treated CD4⁺ T cells (n = 3 biological replicates for each experiment. Data are representative of 3 independent experiments.). Error bars represent the mean ± SEM. Statistical significance was determined using Mann–Whitney two-sided test. Source data are provided as a Source Data file.

Fig. 6. apCAFs and pro-inflammatory macrophages exhibit positive feedback regulation in the tumor microenvironment.

A Distribution (left) and proportion (right) of macrophages subclusters in tumors from control, apCAFs and MHC II^- CAFs groups (n = 2 mice/per group). B Expression levels of pro-inflammatory phenotype and pro-inflammatory phenotype related genes among various subclusters of macrophages. C Representative images of multiplex immunofluorescence (green: CD74; red: PDGFRA; white: CD68; purple: CD86) and distance from apCAFs or MHC II^- CAFs to pro-inflammatory macrophages. Statistical significance was determined using Mann–Whitney two-sided test. Multiplexed immunofluorescence assays are performed twice on tumor samples following assay optimization. D Spatial feature plot of pro-inflammatory macrophages signature in sample 1. ST: Spatial transcriptomics. E Correlation of signature score of apCAFs and pro-inflammatory macrophages in spatial transcriptomics data of sample 1. Correlation was evaluated using the two-sided Spearman rank correlation coefficient. F Correlation of signature score of apCAFs and pro-inflammatory macrophages in ACRG cohort (n = 300). Correlation was evaluated using the two-sided Spearman rank correlation coefficient. G Heatmap showing ligands activity and regulatory potential of the prioritized ligands in apCAFs to macrophages. H Schematic representation of apCAFs or MHC II^- CAFs and macrophages co-cultured experiments. Created in BioRender. Song, J. (2024) https://BioRender.com/r25u487. I The expression of pro-inflammatory genes and anti-inflammatory genes in macrophages after co-culture with apCAFs or MHC II^- CAFs (n = 3 in each group). J The proportion of pro-inflammatory macrophages (CD86⁺ cells and MHC II⁺ cells) in macrophages after co-culture with apCAFs or MHC II^- CAFs from five different patients. Error bars represent the mean ± SEM. Statistical significance was determined using one-way ANOVA, followed by a two-sided Dunnett’s test to compare the specific differences between the groups. Source data are provided as a Source Data file. K The prediction of ligand-receptor interaction activity from macrophage subclusters to apCAFs through “CellChat” algorithm. Pro-inflam Macro: pro-inflammation macrophages; Anti-inflam Macro: Anti-inflammation macrophages. L The prediction of ligand-receptor interactions activity between pro-inflammatory macrophages and apCAFs through “CellCall” algorithm. Pro-inflam Macro: pro-inflammation macrophages; Anti-inflam Macro: Anti-inflammation macrophages. M Schematic representation of control macrophages or pro-inflammatory macrophages and pan-CAFs co-cultured experiments. Pro-inflam Macro: pro-inflammation macrophages. Created in BioRender. Song, J. (2024) https://BioRender.com/h05n316. N The proportion of apCAFs in pan-CAFs from five different patients after co-culture with control macrophages or pro-inflammatory macrophages. Error bars represent the mean ± SEM. Statistical significance was determined using unpaired two-sided t-test. Source data are provided as a Source Data file. Pro-inflam Macro: pro-inflammation macrophages.

Fig. 7. Baseline intratumoral apCAFs predicts clinical benefits from immunotherapy.

A Schematic of GC patients treated with anti-PD1 therapy used as discovery cohort. Created in BioRender. Song, J. (2024) https://BioRender.com/m64n940. B Bar chart representing apCAFs signature in patients with high immune signature (n = 22) or low immune signature (n = 23) in discovery cohort. Error bars represent the mean ± SEM. Statistical significance was determined using unpaied t-test. Source data are provided as a Source Data file. C Bar chart representing apCAFs signature in responders (n = 12) and non-responders (n = 33) in discovery cohort. Error bars represent the mean ± SEM. Statistical significance was determined using unpaired two-sided t-test. Source data are provided as a Source Data file. R: responders; NR: non-responders. D ROC curves showing the value of apCAFs in predicting therapeutic response in discovery cohort. Source data are provided as a Source Data file. E Schematic of GC patients treated with neoadjuvant anti-PD1 plus chemotherapy used as validation cohort. Created in BioRender. Song, J. (2024) https://BioRender.com/a10h486. F Representative CT images of patients with different responses to neoadjuvant anti-PD1 plus chemotherapy in validation cohort. R: responders; NR: non-responders. G Representative images of multiplex immunofluorescence characterizing the apCAFs in patients with different responses in validation cohort (green: CD74; red: PDGFRA). H Bar chart representing the proportionof apCAFs in all cells of tumors with different responses to neoadjuvant anti-PD1 plus chemotherapy in validation cohort (n = 21). Error bars represent the mean ± SEM. Statistical significance was determined using unpaired two-sided t-test. Source data are provided as a Source Data file. R: responders; NR: non-responders. I ROC curves showing the value of apCAFs in predicting therapeutic response in validation cohort. Source data are provided as a Source Data file. J UMAP plots and proportion of CAFs clusters in pre-treatment and post-treatment tissues from patients with different responses in triple negative breast cancer cohort (n = 22). TNBC: triple negative breast cancer; R: responders; NR: non-responders. K Violin plots showing apCAFs signature in CAFs (n = 10825) from pre- and post-treatment tissues of patients with different responses. The box is bounded by the first and third quartile with a horizontal line at the median and whiskers extend to the maximum and minimum value. Statistical significance was determined using Mann–Whitney two-sided test. R: responders; NR: non-responders. L Predictive value of apCAFs to clinical benefits in melanoma patients receiving anti-PD1 therapy (n = 41). Error bars represent the mean ± SEM. Statistical significance was determined using unpaired two-sided t-test. Source data are provided as a Source Data file. R: responders; NR: non-responders. M Predictive value of apCAFs to clinical benefits in melanoma patients receiving anti-PD1 and anti-CTLA4 therapy (n = 32). Error bars represent the mean ± SEM. Statistical significance was determined using unpaired two-sided t-test. Source data are provided as a Source Data file. R: responders; NR: non-responders. N Predictive value of apCAFs to clinical benefits in non-small cell lung cancer patients receiving anti-PD1 therapy (n = 16). Error bars represent the mean ± SEM. Statistical significance was determined using Mann–Whitney two-sided test. Source data are provided as a Source Data file. NSCLC: non-small cell lung cancer; R: responders; NR: non-responders.