Defining fibroblast niches in colorectal cancer
Tyrrell H.
Colorectal cancer (CRC) is a heterogenous disease and high stromal content and specifically fibroblasts have been associated with advanced stage and poor prognosis. Fibroblasts are versatile cells that can alter from quiescent cells to myofibroblasts, that secrete extracellular matrix components and become contractile in situations such as wound healing, or alter to inflammatory fibroblasts, that secrete cytokines and chemokines to recruit immune cells. These phenotypes have been identified across multiple types of cancer through scRNAseq, and the key cytokines driving these clusters are hypothesised as TGFβ for myofibroblastic cancer associated fibroblasts (myCAFs) and IL-1 for inflammatory CAFs (iCAFs). However, few studies have identified them in situ in CRC and no fibroblast subsets have been conclusively associated with prognosis in this disease. It is also unknown whether the fibroblast clusters seen on scRNAseq represent permanent differentiation states or temporary responses to stimulation. Here we use multiplex imaging of 60 rectal cancer sections to identify myCAF-like and iCAF-like stromal cells in situ. We have shown differences in their abundance with the Consensus Molecular subtypes (CMS), with iCAFs being found in CMS3 and myCAF-like stromal cells in CMS4. We have also stained adjacent slides with a panel to look for immune subsets and found greater numbers of both T-cells and neutrophils in CMS1 disease. We have found that the iCAF-like stromal cells are found in close association with immune cells, particularly neutrophils, consistent with the hypothesis of IL-1 driving this phenotype. We have also confirmed that treating primary stromal cells in vitro with IL-1β gives them an inflammatory phenotype that allows them to map more closely to iCAF clusters seen in scRNAseq. The myCAF-like stromal cells were identified deep within stromal tracts. We have also identified a CD56+ stromal subset, found surrounding the tumour nests and associated with CMS2 CRC. We also hypothesised that fibroblasts from CRC tumours may be hyper-responsive to these key cytokines, TGFβ and IL-1β, compared to the matched normal adjacent fibroblasts, consolidating pathological phenotypes. However, we found no evidence that once in vitro there were any differences in the stromal cells from tumour and normal adjacent, either at baseline or in terms of response to cytokine stimulation. This suggests that the CAF clusters seen are temporary activation states depending on the micro-environment the cells were exposed to immediately prior to sequencing. Additionally, we hypothesised that TGFβ1, TGFβ2 and TGFβ3, may be having differential effects on fibroblasts. However, on bulk RNAseq we found no evidence that the transcriptional response to TGFβ stimulation is of a different character or magnitude between the three isoforms.