engleski

Network inference from glycoproteomics data reveals new reactions in the IgG glycosylation pathway

Immunoglobulin G (IgG) is major player in the human immune response and is one of the most studied glycoproteins. However, the molecular mechanisms that regulate its glycosylation are still not fully understood: the known IgG glycosylation pathway might still be incomplete and, to the best of our knowledge, an experimental validation on IgG is not feasible. Therefore, in this study, we analyze LC-ESI-MS measurements of plasma IgG Fc glycans to statistically infer possible unknown enzymatic reactions in the IgG glycosylation pathway. First, we calculate a glycan correlation network using Gaussian Graphical Models (GGMs), which are based on partial correlations and allow to remove spurious effects due to the presence of confounders. We show that glycan pairs that have significant partial correlations correspond to the substrate and the product of single enzymatic reactions in the known IgG glycosylation pathway. We exploit this finding to build a rule-based approach for pathway inference. Since we observe a tight association between the calculated GGM and the known glycosylation pathway, we investigate whether unexpected significant correlations could represent true but still undiscovered reactions. We cluster all possible missing reactions into six enzymatic rules, according to the features of the involved glycoforms, for example the presence of fucose or bisecting N- acetylglucosamine. We then evaluate whether the addition of these new reaction rules to the known glycosylation pathway associates stronger with the calculated GGM than the known pathway alone. Remarkably, we find evidence that two among the six hypothetical rules considered are likely to occur, namely fucosylation of galactosylated, bisected glycans and galactosylation of monosialylated glycans. Statistical significance of the results is tested via bootstrapping and all findings are replicated in four large-scale cohorts. For validation, we perform GWAS analysis in a fifth cohort using substrate-product glycan ratios as quantitative traits. We find significant associations between ratios from our predicted reactions and genetic variants in the region of the corresponding glycosyltransferase genes, providing strong evidence that those reactions take part in the IgG glycan synthesis in vivo.

glycosylation pathway ; Gaussian Graphical Models ; GWAS validation ; Immunoglobulin G ; pathway inference

nije evidentirano