Macrophages restrict tumor immune infiltration 3 by controlling collagen topography through a Tcf4-Col3 axis

During tumorigenesis, the extracellular matrix is profoundly remodeled. While the impact of such remodeling on tumor growth and invasion is well-described, less is known on its consequences on immune infiltration. Combining tissue imaging and machine-learning, we show that immune cell localization in tumors can be predicted by the local topography of fibrillar collagens. Such topographies are dictated by a fibrotic pathway driven by the transcription factor Tcf4 in both cancer and stromal cells, which promotes collagen 3 deposition, resulting in intermingled collagen networks that favor intra-tumoral infiltration of T cells and neutrophils. Macrophages inhibit this pathway, highlighting their key structural role in shaping the tumor extracellular matrix. Human database analyses revealed a strong correlation between Tcf4, collagen 3, and tumor infiltration by T cells and neutrophils in different solid tumors, attesting the clinical relevance of our findings. Collagen network topographies thus emerge as a major regulator of antitumor immunity . Overall design: D14 MC38 tumors from CD64DTR and CD64WT were harvested at day 14 d.p.i., digested in Liberase TL 0,25mg/ml (Roche, #5401020001) DNAse 0,5mg/ml (Sigma, DN25-100MG) in CO2 independent medium (Gibco, 18045088) using GentleMacs Octo dissociator pre-set program (37C_m_TDK1, 41min at 37° under agitation). Cells were filtered on a 100 µm cell strainer and red blood cells were lysed using red blood cells lysis bufer 1X (Biolegend, 420301) diluted in water. After centrifugation (5min, 1500rpm), cells were incubated 10 min with Fcblock solution diluted in FACS bufer (BD, 553142, 1:100) and stained either with (1) DAPI (Invitrogen, 11534886, 1:10000) for the first experiment (batch #1), or (2) with DAPI, anti-CD45 APC-Cy7 (BD, 557659, 1:200) and anti-CD64 APC (Biolegend,139306, 1:200) (for the second experiment (batch #2) diluted in FACS bufer (PBS 1X, 0,5%BSA, 2mM EDTA). For batch#1, DAPI- cells were sorted while for batch#2, DAPI-CD45-, DAPI-CD45+CD64+, DAPI- CD45+CD64- populations were sorted separately. For batch #1, 15 000 cells form CD64WT and CD64DTR tumors were loaded. For Batch #2, 10000 cells form CD64WT and CD64DTR tumors were loaded. Sequencing was performed on a Novaseq 6000, with the Single Cell 3' V3.1 Next-GEM kits. Batch #1 and DAPI-CD45- from batch #2 were sequenced at 50,000 reads / cell, DAPI-CD45+CD64+, DAPI- CD45+CD64- from batch #2 were sequenced at 100,000 reads / cell. ScRNAseq data were analyzed with R using the Seurat suite (v4). Cells were then filtered out when expressing less than 1000 genes, or when expressing more than 10% of mitochondrial genes. Doublets were identified using the DoubletFinder package. Clusters containing less than 100 cells were excluded from the analysis.