Membrane protein glycan markers of epithelial ovarian cancer: discrimination of serious tumours of ovary, peritoneum and tube

Epithelial ovarian cancer is characterised by a low 5-year survival rate of 43% of those diagnosed. The overall prognosis is poor due to many factors such as the lack of reliable and sensitive markers for early stage detection, the heterogeneity of the tumours, rapid metastasis of the disease which extends beyond the ovaries and the different cancer tissue origins (ovary, peritoneum and tube). As these cancers are derived from the epithelial surface, the comprehensive glycosylation of the cell membrane proteins and lipids from cell lines and serous cancer tissues was investigated to unravel cancer-specific biomarkers for improved diagnosis and personalised tumour-specific treatments. In the first phase of this study, N- and O-glycans were enzymatically released by PNGase F and reductive β-elimination, respectively, from extracted membrane glycoproteins derived from 'in vitro'-based cell line models of cancerous and non-cancerous ovarian cells. Released glycan alditols were separated using porous graphitised carbon (PGC) chromatography, analysed using electrospray ionisation mass spectrometry (LC-ESI-MS/MS) and characterised based on negative ion tandem MS fragmentation patterns and LC retention times. Glycan structural features such as bisecting N-acetylglucosamine (GlcNAc) and sialylated N-N-diacetyllactosamine (LacdiNAc)-type structures, together with increased levels of α2-6 sialylation were detected on membrane glycoproteins derived from ovarian cancer cells. The corresponding gene expression and epigenetic regulation of specific glycosyltransferases responsible for the biosynthesis of relevant N-glycan structures were also investigated using qRT-PCR and de-methylation by 5-Aza treatment. Their presence correlated with the corresponding glycosyltransferase gene expression of ST6GAL1, B4GALNT3, and MGAT3, in which MGAT3 was found to be also epigenetically regulated by DNA hypomethylation. The presence of specific glycan structures implicated in the preliminary cell-line models were further verified in membrane proteins of serous cancer tissues from different origins (14 ovarian, 14 peritoneal and 4 tubal). Several statistical analyses such as analysis of variance (ANOVA), principal component analysis (PCA), partial-least-square discrimination analysis (PLSDA) and receiver-operating curve (ROC) were employed to evaluate differences in the expression of glycan structures and to discriminate between all three serous cancers. Bi-antennary sialylated N-glycans with prominent α2-6 sialylation appeared to be the most common feature in all serous cancer tissues, whilst the unique expression of sialylated LacdiNAc N-glycans was detected specifically in serous ovarian-derived cancer tissues and these structures were statistically classified as 'highly accurate' biomarkers of serous ovarian cancer by ROC. The expression of 11 N-glycans, including the LacdiNAc-type glycans contributed significantly (p=0.00186) to the classification accuracy (78.6 %) of serous ovarian and peritoneal cancers. In addition to the structural identification of serous cancer membrane N- and O-glycans, a second analytical platform employing PGC-LC-ESI-MS/MS was also developed to characterise glycans released enzymatically (Endoglycoceramidase II) from PVDF-immobilised glycosphingolipids (GSLs) derived from cancer tissue samples and cell line. The analysis of the glycosylation of ovarian cancer membrane lipids revealed the presence of several isomeric and isobaric structures which were differentiated using specific diagnostic and structural feature ions produced by negative ion mode MS/MS fragmentation. This approach led to the identification of P blood group-related as well as fucosylated/non-fucosylated Type 1 and Type 2 antigens which were not expressed on N- and O-glycans of membrane proteins of serous cancers analysed in this study. These antigens were also implicated in the immune recognition by auto-antibodies found in the plasma and ascites fluid of ovarian cancer patients. The identification and understanding of the regulation of membrane protein and lipid glycan epitopes unique to ovarian cancer could be utilised to distinguish serous ovarian from peritoneum and tubal cancers. More importantly, such structures may facilitate the diagnosis and development of different drug targets to improve survival rates of this malignancy.