The role of sialylation in stromal cell-mediated immunosuppression in the colorectal tumour microenvironment

Abstract

Immunosuppressive tumour microenvironments (TME) inhibit the effectiveness of many immunotherapies. The hypersialylation of cancer cells contributes significantly to this suppressive environment, aiding in tumour immune evasion. Sialic acids are bound by siglec (sialic acid–binding, immunoglobulin (Ig)-like lectin) receptors expressed by immune cells such as macrophages and T cells. This binding initiates a downstream inhibitory signalling response similar to what is seen with PD-1/PD-L1 binding. Cancer associated fibroblasts, or CAFs, are a highly immunosuppressive cell type found in the colorectal cancer TME and are associated with tumour progression and a worse prognosis, however, the mechanisms by which they mediate their immunosuppressive properties have not yet been fully elucidated. This PhD aimed to investigated if the sialylation profile of CAFs and tumour-conditioned stromal cells contribute to their immunosuppressive properties; and if this could provide a novel target, overcoming immunotherapy resistance commonly associated with approved immunotherapies. Colorectal cancer cell line secretome, +/- an inflammatory TNF-α stimulus, was generated and used to condition primary BALB/c bone marrow mesenchymal stromal cells (BM-MSCs). BM-MSCs recruited into the colorectal cancer TME are known to be precursors to CAFs, and our model reflects this recruitment process. Intestinal stromal cells were isolated from primary colorectal cancer patient resections and were termed cancer associated fibroblasts (CAFs) or normal associated fibroblasts (NAFs) based on their location relative to the primary tumour. The sialylation profile of these cells were assessed through the use of lectins and Fc chimeras staining on flow cytometry and an elevated level of α2,6 and α2,3 linked sialic acid was confirmed in both tumour conditioned mouse stromal cells and CAFs compared to controls. BALB/c BM-MSCs conditioned with tumour secretome had elevated expression of both α 2,6 linked sialic acids and siglec-E ligands. When these same cells were cultured with inflammatory tumour secretome, there was a significant increase in α2,3 linked sialic acid. In primary intestinal stromal cells, CAFs had significantly higher levels of α2,6 linked sialic acids and siglec-9 ligands compared to patient matched NAFs. Following this, the use of a sialyltransferase inhibitor, P-3FAX-Neu5Ac, was shown to significantly reduce cell surface sialic acid expression on stromal cells +/- tumour conditioning, without affecting cell size or viability. Interestingly, the use of the sialyltransferase inhibitor could significantly increase the detection of PD-L1 by both flow cytometry and western blotting, suggesting the PD-L1 may be sialylated. Stromal cells +/-tumour secretome conditioning were cultured with healthy splenocytes (mouse) or PBMCs (human) and their immunosuppressive properties were assessed by flow cytometry. CAFs significantly induced a more exhausted and immunomodulatory T cell phenotype, highlighted by increased expression of exhaustion markers, PD-1, TIM-3 and LAG-3, and immunomodulatory receptors Siglec-7 and -9 when compared with NAFs. To elucidate the role of sialylation on CAF-mediated immunosuppression, NAFs and CAFs were treated with the sialyltransferase inhibitor (SI) P-3FAX-Neu5Ac prior to co-culture. Reduction of sialic acid expression on NAFs/CAFs was confirmed by flow cytometry and the SI-treated NAFs/CAFs were then co-cultured with allogeneic T cells to assess the functional consequences of reduced NAF/CAF sialylation. SI-treated CAFs induced significantly less CD4+TIM-3 + and both CD4+ LAG-3 + and CD8+ LAG-3 + T cells compared to their untreated counterparts. Interestingly, SI-treated CAFs also induced significantly less Siglec-7 and -9 receptor-expressing CD8+ T cells. An in vivo mouse model of CT26 colorectal cancer was designed to assess the role of highly sialylated stromal cells. Mice were subcutaneously injected with CT26 cells and conditioned stromal cells, where the stromal cells had undergone prior desialylation in the appropriate groups. Mice with highly stromal dense tumours had significantly supressed T cells, with supressed CD25 and granzyme B expression, both intratumourally and distally in the spleen and draining lymph node. This suppression was sialylation dependent, with complete loss of suppression in tumours with desialylated stromal cells. Macrophages and NK cells in the tumour, draining lymph nodes and spleens expressed significantly less siglec-G in mice which received co- injection of CT26 cells and tumour-conditioned MSCs, which was dependent on the stromal cells’ sialylation profile. These results demonstrate that stromal cells in the tumour-microenvironment can regulate immune cells in a sialylation dependent manner and that their immunosuppressive effects may be reversed through the targeting of sialylation on the stromal cell surface.