The transcriptional landscape of glycosylation-related genes in cancer

Changes in glycosylation patterns have been associated with malignant transformation and clinical outcome in several types of cancer, although no comprehensive analysis has been performed in a pan-cancer setting. Here, we performed an extensive transcriptomic analysis of glycosylation related genes (such as enzymes involved in synthesis and degradation of glycoconjugates, transporters, mucins and galectins), using publicly available bulk and single cell transcriptomic data sets from tumor samples and cancer cell lines. We identified genes and pathways associated with different tumor types, which may represent novel diagnostic biomarkers as 2-3 sialylation for Melanoma, MUC21 for Lung adenocarcinoma and Galectin-7 for Squamous cell carcinomas (SCC). Accordingly, serum levels of Galectin-7 in patients with lung cancer were elevated in SCC respect to adenocarcinomas, supporting its biomarker potential. Moreover, we characterized the contribution of different cell types to the overall glycosylation profiles observed by performing the integration and analysis of 14 single cell RNA-seq datasets. This led us to identify that cancer cells are responsible for the specific tumor glyco-codes identified in bulk transcriptomics, while stromal and immune cells contribute in a conserved manner across various malignancies. Furthermore, our results suggest that the glycosylation-related genes and pathways expressed by cancer cells are influenced by the cell of origin and the oncogenic pathways that led to malignant transformation. Lastly, we described the association of different glycosylation-related genes and pathways with the clinical outcome of patients. Interestingly, while the expression of genes associated to some pathways (as proteoglycan biosynthesis) are consistently associated with a more aggressive disease, the correlation of others pathways with the survival of patients depends on the particular tumor type. Remarkably, the expression of genes associated with the synthesis of CMP-sialic acid was correlated with lower survival of patients in Uveal Melanoma and PDAC, while the opposite was observed for colorectal cancer. The extensive transcriptomic analysis of glycosylation pathways in cancer that we report here can serve as a resource for future research aimed to unravel the glyco-code in cancer related to clinical outcome or biomarker development.