This journal is © The Royal Society of Chemistry 2020 Mol. Omics, 2020, 16, 147--155 | 147 Cite this: Mol. Omics, 2020, 16, 147 The effectiveness of filtering glycopeptide peak list files for Y ions† Robert J. Chalkley, * a Katalin F. Medzihradszky, ab Zsuzsanna Darula, b Adam Pap bc and Peter R. Baker a Intact glycopeptide analysis is becoming more common with developments in mass spectrometry instrumentation and fragmentation approaches. In particular, collision-based fragmentation approaches such as higher energy collisional dissociation (HCD) and radical-driven fragmentation approaches such as electron transfer dissociation (ETD) provide complementary information, but bioinformatic strategies to utilize this combined information are currently lacking. In this work we adapted a software tool, MS-Filter, to search HCD peak list files for predicted Y ions based on matched EThcD results to propose additional glycopeptide assignments. The strategy proved to be extremely powerful for O-glycopeptide data, and also of benefit for N-linked data, where it allowed rescue of low confidence results from database searching. Introduction Intact glycopeptide analysis using mass spectrometry presents significant challenges due to the differing behavior of the peptide and glycan components in the hybrid species. Nevertheless, when analyzing a single protein, collision induced dissociation (CID) strategies have been very effective. In the characteristic fragmenta- tion of glycopeptides using these approaches one of the most prominent fragments in O-linked glycopeptide spectra is typically the gas-phase-deglycosylated peptide (Y 0 ), whereas for N-linked glycopeptides a similarly abundant ion is commonly the peptide retaining the core GlcNAc (Y 1 ) [nomenclature: 1 ]. When analyzing an isolated protein, the masses of the unmodified versions of potentially glycosylated peptides can be calculated, so it is straight- forward to use the smallest Y ion mass to identify the peptide, one can infer the mass of the glycan based on the mass difference between the observed precursor and the Y ion, then the CID fragmentation provides information on the glycan(s) composition. Resonance activation CID in ion trap yields B and Y fragments via single bond cleavages, identifying terminal groups, thus providing knowledge about branching in the oligosaccharide(s). 2,3 Beam- type CID (HCD) can produce multiple bond cleavages, and thus provides some information on the direct connection of certain sugar units in form of internal oxonium ions. 3,4 Analysis of complex mixtures of glycoproteins is currently mostly performed by enzymatic release of glycan species then analysis of peptides and glycans separately. However, intact glycopeptide analysis for these types of samples is starting to become more common due to the development of improved fragmentation methods in mass spectrometers that allow formation of fragments from both peptide and glycan compo- nents in the same spectrum, namely EThcD and stepped HCD fragmentation. 5–7 A few datasets identifying hundreds of unique glycopeptides have now been published, although there are still challenges with controlling the reliability of results from software doing these analyses. 5–9 In this work we sought to investigate software strategies that make use of small Y ions to (a) increase the number of glycopeptide identifications; and (b) improve the reliability of reported results in analyses of complex glycopeptide datasets. Using Y ions to try to identify glycopeptides from CID-type fragmentation spectra is not novel. However, using EThcD data to derive the list of glycosylated peptides for querying HCD data is a new approach. Prior strategies have all employed PNGase F to deglycosylate a fraction of the sample, then analyzed this to derive a list of potential glycopeptides to consider in the intact glycopeptide analysis. Some researchers additionally performed glycomic analysis to derive a list of glycosylations to consider, 10,11 whereas others have used a database of known glycans as a reference. 12 These approaches require significantly more sample handling and acquisition and can only be used for N-glycosylation analysis. We developed new features in the MS-Filter tool within Protein Prospector, then tested these on intact O-linked and N-linked glycopeptide datasets to see the effect of different filtering para- meters on the resulting glycopeptide results. a Department of Pharmaceutical Chemistry, School of Pharmacy, University of California San Francisco, USA. E-mail: chalkley@cgl.ucsf.edu b Laboratory of Proteomics Research, Biological Research Centre, Temesvari krt. 62, H-6726 Szeged, Hungary c Doctoral School in Biology, Faculty of Science and Informatics, University of Szeged, Kozep fasor 52, H-6726 Szeged, Hungary † Electronic supplementary information (ESI) available. See DOI: 10.1039/ c9mo00178f Received 9th December 2019, Accepted 6th February 2020 DOI: 10.1039/c9mo00178f rsc.li/molomics Molecular Omics RESEARCH ARTICLE Published on 06 February 2020. Downloaded on 9/29/2021 8:28:46 AM. View Article Online View Journal | View Issue