Single-cell resolution spatial analysis of antigen-presenting cancer-associated fibroblast niches
Xiongfeng CheniD, Zhuan ZhouiD, Luyu XieiD, Kailiang QiaoiD, Yiyue JiaiD, Si LiuiD, Zeynep YazganiD, Francesca RossiiD, Yang LiuiD, Bo ZhangiD, Patricio Marcelo PolancoiD, Herbert J. ZehiD, Alex C. KimiD, Huocong HuangiD
Abstract
Recent studies have identified a unique subtype of cancer-associated fibroblasts (CAFs) termed antigen-presenting CAFs (apCAFs), which remain the least understood CAF subtype. To gain a comprehensive understanding of the origin and function apCAFs, we construct a fibroblast molecular atlas across 14 types of solid tumors. Our integration study unexpectedly reveals two distinct apCAF lineages present in most cancer types: one associated with mesothelial-like cells and the other with fibrocytes. Using a high-resolution single-cell spatial imaging platform, we characterize the spatial niches of these apCAF lineages. We find that mesothelial-like apCAFs are located near cancer cells, while fibrocyte-like apCAFs are associated with tertiary lymphoid structures. Additionally, we discover that both apCAF lineages can up-regulate the secreted protein SPP1, which facilitates primary tumor formation and peritoneal metastasis. Taken together, this study offers an unprecedented resolution in analyzing apCAF lineages and their spatial niches.
02 /Specification & Classification
Click any tag to browse matching papers- Citations
- 63
- FWCI
- 17.12
- Field percentile
- top 1%
- Organism
- Assay
- Research Focus
- Approach
In-depth summary
Claude Opus 5.5 · xhigh effort · 6 Nov 2026 · from the full text of version 1 (17 Nov 2024)
Pan-cancer single-cell and spatial profiling resolves two antigen-presenting CAF lineages: tumor-adjacent mesothelial-like apCAFs and tertiary lymphoid structure-associated fibrocyte-like apCAFs, both secreting pro-tumorigenic SPP1.
Antigen-presenting cancer-associated fibroblasts (apCAFs) express MHC II molecules and regulate T cells, yet their cellular origins, conservation across human solid tumors, and spatial microenvironments remain poorly characterized. Integrating 560 human scRNA-seq samples with single-cell Xenium spatial transcriptomics distinguishes mesothelial-associated and fibrocyte-associated apCAF lineages occupying distinct niches and identifies stromal SPP1 as a therapeutic target.
Key results
ShownCross-tissue scRNA-seq integration across 14 solid tumor types identifies two conserved apCAF lineages with distinct origins and transcriptional regulators.
Across 560 human samples and over 2.5 million cells, fibroblast-like cells formed 13 clusters including fibrocyte-like F-apCAFs (c05; PTPRC, CD37, CD52, C1q; regulated by SPI1, RUNX3, GATA3) and mesothelial-like M-apCAFs (c06; MSLN, UPK3B, KRT19, CD24; regulated by BARX2, BHLHE41, HSF4, POU5F1), both enriched in tumors versus normal tissue. · CAF phenotypes are conserved across cancer types; Identification of two distinct lineages of…; Figure legends
ShownSingle-cell spatial transcriptomics shows F-apCAFs localize to tertiary lymphoid structures (TLS) whereas M-apCAFs occupy immunosuppressive tumor-adjacent mesothelial niches.
Across 799,455 cells in 8 human colon peritoneal metastasis (PM) samples profiled by 480-gene Xenium assays, F-apCAFs were the most abundant cell type in lymphocyte-enriched niches (29.2%) and localized closer to TLS than myCAFs (average distance 37.86 µm vs 66.06 µm), while M-apCAFs comprised 23.89% of mesothelial-enriched niches near tumor epithelium. · Spatial analysis of F-apCAF niches; Spatial analysis of M-apCAF niches; Figure legends
ShownGoblet cell-like iCMS3 colon cancer is strongly associated with peritoneal metastasis, apCAF enrichment, and inactive T cell infiltration.
GeoMx spatial profiling (96 ROIs from 8 PM patients) and imaging mass cytometry (3 iCMS2 and 3 iCMS3 samples) showed iCMS3 PM enriched for antigen presentation pathways and T cells lacking granzyme B, perforin, IFNγ, and Ki67; in stage T4 primary colon tumors, SPINK4+ invasion fronts predicted PM (relative risk=16.5, P=0.005). · Stromal heterogeneity of peritoneal metastasis; GeoMx spatial profiling; Figure legends
ShownBoth apCAF lineages up-regulate SPP1 during differentiation, and stromal Spp1 depletion or antibody blockade suppresses peritoneal metastasis and PDAC growth in mice.
F-apCAFs and M-apCAFs were the primary CAF sources of SPP1, part of a 6-gene cancer signature associated with poor survival; stromal Spp1 knockout (WT n=8 vs KO n=10 for MC38 PM; n=6/group for 6620c1 PDAC) or anti-SPP1 mAb (500 μg/dose) reduced tumor burden, decreased myCAFs, up-regulated Pi16 and Dpt, and increased T cell infiltration. · Identification of two distinct lineages of…; Stromal knockout of Spp1 results in a reduction…; Stromal SPP1 facilitates fibrotic barrier…; Figure legends
ShownIn human PDAC, M-apCAFs colocalize with chemoradiotherapy-resistant SPP1+ cancer cells and SPP1+ myeloid cells.
Xenium spatial profiling of 3 treatment-naive and 3 chemoradiotherapy-treated human PDAC samples revealed that SPP1+ cancer cell 2 persisted after chemoradiotherapy, with M-apCAFs enriched in SPP1+ tumor areas alongside SPP1+ myeloid cells and F-apCAFs restricted to TLS regions. · M-apCAFs are associated with SPP1-positive…; Figure legends
Methods, data and reproducibility
The authors integrated public pan-cancer scRNA-seq datasets with Seurat and Harmony to build a fibroblast atlas, inferring trajectories (Monocle3), regulons (pySCENIC), and ligand-receptor pairs (CellPhoneDB). They mapped apCAF niches in human colon peritoneal metastasis and PDAC via GeoMx, imaging mass cytometry, and 480-gene Xenium spatial profiling (RCTD), plus Spp1 KO and anti-SPP1 mouse models.
- Pan-cancer integrated scRNA-seq atlas of primary human tumors, adjacent normal tissues across 13 organs, normal peritoneum, and peritoneal metastasis · human · n = 560 samples, over 2.5 million cells
- Nanostring GeoMx Human Whole Transcriptome Atlas spatial RNA profiling of colon cancer-derived peritoneal metastasis · human · n = 8 patients, 96 ROIs (12 per sample) · GSE274609, GSE274623, GSE274673
- 10x Genomics Xenium In Situ 480-gene single-cell spatial profiling of colon cancer peritoneal metastasis and adjacent tissue · human · n = 8 samples, 799,455 cells · GSE274609, GSE274623, GSE274673
- 10x Genomics Xenium In Situ 480-gene single-cell spatial profiling of treatment-naive and chemoradiotherapy-treated PDAC · human · n = 6 samples (3 treatment-naive, 3 chemoradiotherapy-treated) · GSE274609, GSE274623, GSE274673
- Fluidigm Hyperion imaging mass cytometry of 20 immune markers in iCMS2 and iCMS3 peritoneal metastasis FFPE sections · human · n = 6 samples (3 iCMS2, 3 iCMS3; 2 images per sample)
- 10x Genomics scRNA-seq of orthotopic 6620c1 PDAC tumors from wildtype and Spp1 knockout C57BL/6 mice · mouse · n = 6 tumors per group (3 pooled libraries per group, ~10,000 cells per library) · GSE274609, GSE274623, GSE274673
- RNA-seq of flow-sorted GFP+ PanMeso pancreatic mesothelial cells co-injected with murine PDAC lines (BMFA3 or CT1BA5) · mouse · GSE196740
- Methods & tools
- Cell Ranger, Seurat, DoubletFinder, Harmony, AUCell, Monocle3, pySCENIC, CellPhoneDB, CytoTRACE, SpatialDecon, spacexr (RCTD), CellProfiler / histoCAT
- Reagents & environment
- R v4.3.2, Seurat v5.0.1 and v4.3, Cell Ranger v7.2.0, DoubletFinder v2.0.3, Harmony v1.2.0, AUCell v1.24.0, psych v2.4.3, pheatmap v1.0.12, Monocle3 v2.30.1, pySCENIC v0.12.1, CellPhoneDB v3.0, GeoMx NGS Pipeline v2.3.3.10, CellProfiler v4.2.5, spacexr v2.2.1, SAS v9.4, GraphPad Prism v10.0.0
- Reproducibility gaps
- No custom analysis code repository is provided; spatial and scRNA-seq datasets are deposited under GEO accessions GSE274609, GSE274623, GSE274673, and GSE196740, and reference iCMS epithelial data under Synapse ID syn26844071.
Limitations (4)
- Patient age and gender were undisclosed for the human FFPE tissue cohorts (T4 colon cancer, peritoneal metastasis, and PDAC). · stated by the authors
- TLS formation in human PDAC was rare across the profiled cohort, with only 1 of 6 Xenium-assayed PDAC samples containing tertiary lymphoid structures. · stated by the authors
- Human spatial transcriptomics cohorts had small sample sizes, profiling 8 colon cancer peritoneal metastasis samples and 6 PDAC samples on Xenium. · noted in this summary
- In vivo functional perturbation of Spp1 relied on whole-body Spp1 knockout mice and systemic anti-SPP1 monoclonal antibody treatment rather than apCAF-specific conditional deletion. · noted in this summary
Who should read:Read this if you study cancer-associated fibroblast heterogeneity, tumor-immune spatial niches, or stromal therapeutic targets in peritoneal metastasis and pancreatic cancer.
Reproducibility ledger
Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.
Sign in to file a reproducibility report.
Related papers
- Spatially organized cancer-associated fibroblast subtypes partition cutaneous carcinomas into immune-active and contracted, immune-repressed niches
Bertram Aschenbrenner, Agnes Forsthuber, … Beate M. Lichtenberger·bioRxiv·Jun 4, 2026
- Cancer associated fibroblast subtypes modulate the tumor-immune microenvironment and are associated with skin cancer malignancy
A. Forsthuber, B. Aschenbrenner, … B. M. Lichtenberger·bioRxiv·May 3, 2023
- Spatial Transcriptomics Reveals Spatially Diverse Cancer-Associated Fibroblast in Lung Squamous Cell Carcinoma Linked to Tumor Progression
Hongyoon Choi, Kwon Joong Na, … Young Tae Kim·bioRxiv·May 16, 2024
- Development of a multiplex immunofluorescence panel to study heterogenous cancer-associated fibroblast subtypes with spatial resolution
Amy Burley, Tatiany Silveira, … Anna Clare Wilkins·bioRxiv·Jul 1, 2026
- Spatial positioning and matrix programs of cancer-associated fibroblasts promote T cell exclusion in human lung tumors
J. A. Grout, P. Sirven, … H. Salmon·bioRxiv·Jan 20, 2022
- Deciphering the temporal heterogeneity of cancer-associated fibroblast subpopulations in breast cancer
F. A. Venning, K. W. Zornhagen, … J. T. Erler·bioRxiv·May 5, 2020