Posted by Nathan Edwards
Kevin Brown Chandler, Petr Pompach, Radoslav Goldman, and Nathan Edwards
Georgetown University, Washington, D.C.
The identification of N-linked glycopeptides from high-throughput tandem mass-spectrometry datasets remains a difficult task. Despite a common monosaccharide core and specific attachment motif (NXS/T) which significantly constrains valid peptide substrates and glycan structures, few tools for identifying intact N-linked glycopeptides from large-scale MS/MS spectra datasets are available. Such tools are sorely needed for the study of N-linked glycan microheterogeneity of glycoproteins, inaccessible to PNGase-F detached glycan or glycopeptide workflows. We have developed a glycopeptide identification tool, GlycoPeptideSearch (GPS), to match peptide-glycan pairs to CID MS/MS spectra using in silico glycoprotein digests and glycans from publicly available glycan structure databases.
The GPS algorithm uses characteristic glycan oxonium ions to select glycopeptide CID spectra, and intact-peptide glycopeptide fragments to identify the likely peptide substrate, so that the remaining precursor mass can be presumed to represent the attached glycan, and matching glycan structures retrieved from a glycan structure database. This strategy avoids the exploration of many irrelevant peptide-glycan pairs, speeding up the search, and boosting the specificity of the resulting identifications. We evaluate the false-discovery-rate of the glycopeptide identifications by considering decoy peptides without the N-linked glycan motif from the protein of interest, and rescale the number of decoy hits based on the number of target and decoy peptides considered.
We use GPS to study haptoglobin microheterogeneity. Human haptoglobin was digested by trypsin and GluC, and 3,468 CID MS/MS spectra acquired on a QSTAR Elite mass spectrometer (Applied Biosystems) from eleven, glycopeptide enriched, HILIC fractions. In all, GPS considered 15 HPT_HUMAN peptides and 1,289 human N-linked glycans from GlycomeDB. This automated search matched glycan-peptide pairs to 250 spectra at false discovery rate (FDR) 3.9%. The matched spectra represent more than 100 distinct precursor ions in varying charge states and provide evidence for 23 distinct glycans at four N-linked motif sites.