Multimodal analysis of glycan dynamics mediating ovarian cancer metastasis

During metastatic outgrowth, cancer cells undergo molecular changes including alteration of glycosylation. Here, glycolipids and glycoproteins can act as key regulators for different physiopathological processes. The functional role of these glycosylations in the context of ovarian cancer (OC) progression is not well understood. OC is a highly metastatic disease and the leading cause of death among all gynaecological malignancies. In regards to metastasis, OC cells disseminate from the primary site within the peritoneal cavity through passive, lymphatic and hematogenous spread. During this dissemination, cancer cells undergo epithelial-to-mesenchymal transition (EMT), a process where cancer cells lose their epithelial features to gain mesenchymal characteristics with the aim to increase their capacities to migrate and invade distant organs. The resulting mesenchymal cells may revert back to an epithelial cell state, a process named MET suggested to be necessary for maintaining growth at distant sites. Beside differential gene and protein expression accompanied with reversible EMT, altered cellular glycosylation has been described in this context, however, their contribution to cancer cell plasticity and metastasis in ovarian cancer remains uncertain.
In our first study, we summarised the current literature suggesting that different GSL classes define epithelial and mesenchymal ovarian cancer cells. Following on the hypothesis of EMT-dependent GSL dynamics, we first investigated how glycosphingolipids (GSLs) modulate EMT in ovarian cancer. Using a novel MALDI-MSI protocol to spatially visualise GSLs, we report that sialylated GSLs (gangliosides) are mostly associated with mesenchymal-like stroma cells in contrast to tumour regions being rich in globosides and characterised by maintained epithelial marker expression during disease progression. Investigation of transcriptomic datasets revealed ST8SIA1, a key enzyme for ganglioside synthesis to be upregulated in mesenchymal cancer samples. Following the genetic loss of ST8SIA1 enriched ovarian cancer cells for epithelial features in vitro and in vivo, we observed increased globosides accompanied with a reduction of ErkY202/T204 and AKTS124 phosphorylation. Downstream phosphoproteomic analysis revealed different signalling pathways that may promote EMT and MET independently. Finally, additional gene set enrichment analysis and extracellular calcium treatment revealed calcium-dependent signatures in mesenchymal cells.
Altered N-glycans on the tumour and microenvironment have been identified to mediate critical events in cancer progression. Thus, we asked if we could observe the same pattern with N-glycans. However, the analysis of sialylated N-glycans using MALDI-MSI in the positive mode remains technically challenging due to the instability of the sialic acid residue. We therefore established a method to stabilise sialic acid on N-glycans in formalin-fixed and paraffin-embedded (FFPE) cancer tissues. Through this straightforward, sensitive and robust workflow, we spatially analysed N-glycans from complex and matched heterogeneous tumour tissue sections (n = 4 patients) and found enrichment of high-mannose glycans in the tumour while sialylated glycans were preferentially expressed in the stroma region, confirming a specific signature of sialylated glycans in mesenchymal-like stroma cells. Moreover, we observed an increase of N-glycan branching comparing primary and metastatic sites demonstrating a higher glycan complexity throughout the disease. To determine the underlying glycoprotein associated with this specific glycan signature, FFPE tissues were further subjected to glycoproteomic analysis. We identified 1000 glycopeptides, the most abundant from the extracellular matrix origin, which were associated with high-mannose.
Altogether my PhD thesis highlights the importance of glycosylation on lipids and proteins in ovarian cancer metastasis. We showed through genetic modulation that GSLs emerge as a potential target that may govern cell differentiation and could be used as a therapy. Moreover, we demonstrated that sialic acids on both GSLs and N-glycans can be used to discriminate mesenchymal-like stroma cells. Finally, we pointed out that N-glycans on specific proteins were altered at metastatic sites compared to the primary tumour.