{"paper_id":"3b21efd1-c40f-41cb-9475-7792748fef73","body_text":"Electron-Activated Dissociation and Collision-Induced Dissociation \nGlycopeptide Fragmentation for Improved Glycoproteomics  \n \nKyle L. Macauslane1, Cassandra L. Pegg1, Amanda S. Nouwens1, Edward D. Kerr1, Joy \nSeitanidou1, Benjamin L. Schulz1,* \n1. School of Chemistry and Molecular Bioscience, The University of Queensland, Brisbane, \nAustralia \n*, corresponding author: b.schulz@uq.edu.au \n \n  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\nAbstract \nTandem mass spectrometry  coupled with liquid chromatography (LC -MS/MS) has proven a \nversatile tool for the identification and quantification of proteins and their post-translational \nmodifications (PTMs) . Protein glycosylation is a critical PTM for the stability and  biological \nfunction of many proteins, but full characterisation of site-specific glycosylation of proteins \nremains analytical ly challenging. Collision induced dissociation (CID) is the most common \nfragmentation method used in LC-MS/MS workflows, but loss of labile modifications render CID \ninappropriate for  detailed characterisation of site-specific glycosylation. Electron -based \ndissociation ( ExD) methods provide alternative s that retain intact glycopeptide fragments for \nunambiguous site localisation, but these methods often underperform CID due to  increased \nreaction times and reduced efficiency. Electron activated dissociation (EAD) is another strategy \nfor glycopeptide fragmentation. Here, we use a ZenoTOF 7600 SCIEX instrument to compare the \nperformance of various fragmentation techniques for the analysis of a complex mixture of \nmammalian O- and N-glycopeptides. We found CID fragmentation identified the most  \nglycopeptides and generally produced higher quality spectra, but EAD provided improved \nconfidence in glycosylation site localisation. Supplementing EAD with CID fragmentation  \n(EAciD) further increased the number and quality of glycopeptide identifications, while retaining \nlocalisation confidence. These methods will be useful for glycoproteomics workflows for either \noptimal glycopeptide identification or characterisation.  \n  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\nIntroduction \n \nGlycosylation is a post -translational modification (PTM) of most eukaryotic secreted and \nmembrane proteins that is critical for protein folding, stability, and mediating diverse functions [1-\n3], and is therefore of both physiological and pathological importance [4-7]. Protein glycosylation \nis chemically and biosynthetically diverse, with the most well -studied and common forms in \nmammalian glycoproteins being N- and O-glycosylation. As secretory and membrane polypeptides \nare translocated into the endoplasmic reticulum (ER) they can be co- and post-translocationally N-\nglycosylated with a Glc 3Man9GlcNAc2 glycan at select asparagine residues  [8]. The likelihood \nthat a specific asparagine residue will be N-glycosylated is much higher if it is located in a \nsequence motif known as a glycosylation sequon (N-X-S/T; X≠P), which has high affinity for the \npeptide acceptor binding site of the oligosaccharyltransferase (OST) enzyme that catalyses N-\nglycosylation [9-11]. In the  Golgi, proteins can be O-glycosylated with GalNAc at serine and \nthreonine residues, typically in serine/threonine-rich mucin domains [12]. Both N- and O-glycans \ncan be further processed by a biosynthetic network of glycosylhydrolases and glycosyltransferases \nas glycoproteins traffic through the Golgi. Not all potential sites of glycosylation are necessarily \noccupied, and both the glycan occupancy and structures at a site can vary due to diverse regulatory \nprocesses and are subject to the overall health of a cell or organism [13-15]. \n \nIn mammals, p rocessing of N-glycans gives rise to three major classes of glycan structures : \noligomannose, hybrid, and complex. Each contains a common core (Man3GlcNAc2) that can also \nbe modified with core fucose (Fuc) [16]. Oligomannose glycans feature terminal branches \ncomprised entirely of mannose (Man), complex glycans have trimming and substitution and \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\ninstead terminate in sugars such as sialic acid ( N-acetylneuraminic acid, NeuAc), while hybrid \nglycans feature both substituted and unsubstituted antennae [16]. O-GalNAc glycans can have \neight core structures built from the initial GalNAc, with terminal structures similar to those found \non N-glycans [17].  \n \nGlycoproteomics aims to identify, characterise, and quantify all features of a glycoproteome. That \nis, the ultimate goal involves not only the identification of glycoproteins  from complex samples, \nbut also characterisation of both the occupancy of individual sites of glycosylation \n(macroheterogeneity), and of the structure and composition of any attached glycans at specific \nsites (microheterogeneity) [18]. Hence, the ideal glycoproteomic workflow involves the \ncharacterisation of the underlying peptide  backbone, the attached glycan, and of the site(s) of \nmodification [16]. It is for this reason intact glycopeptide analysis with LC-MS/MS is the method \nof choice in glycoproteomics, as only analysis of intact glycopeptides can  permit a complete \ncharacterisation [19]. \n \nLC-MS/MS based glyco/proteomics uses fragmentation  of precursor glyco/peptide ions to \ncharacterise the resultant product ions. The dissociation techniques used in peptide fragmentation \ncan broadly be categorised in two groups: thermal- and radical-driven dissociation techniques. The \nmost commonly used fragmentation method in glycoproteomics is thermally driven , beam-type \ncollision induced dissociation (CID  or alternatively HCD ) [20]. In CID, collision of accelerated \nprecursor ions with neutral gas molecules imparts the internal vibrational energy that induces bond \nfragmentation [21]. CID results in peptide (a-, b- and y-type ions) and glycosidic bond (B- and Y-\ntype ions) fragmentation events  [16, 22] . I n particular, low mass oxonium ions arising from \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\nfragmentation of mono/disaccharides are a fundamental feature of glycopeptide fragmentation \nwith CID that enable identification of the monosaccharides present in the attached glycan  [23]. \nCID fragmentation results in preferential cleavage of the weakest covalent bonds within a molecule, \nand as such occurs preferentially at glycosidic bonds over peptide C-N (amide) bonds. \nGlycopeptide CID fragmentation spectra are therefore rich in information about the glycan, but \nrelatively poor in information about the peptide  [24]. Nonetheless, rapid acquisition rates allow \nfor in-depth identification of glycopeptides in complex samples [20]. However, the lack of peptide \nfragment ions makes localisation of the site of glycosylation difficult [25].  \n \nRadical-driven technique s, often called “soft” dissociation methods , such as electron capture \ndissociation (ECD) and electron transfer dissociation (ETD) , generally fall under the collective \nterm electron-driven dissociation (ExD). These techniques use electrons to impart a radical state \nthat induces fragmentation between the  stronger N–C⍺ (amine) b ond, resulting in preferential \ncleavage of the peptide backbone and allowing localisation of  modifications [26-28]. ExD \nfragmentation results in peptide bond fragmentation to generate c - and z-type fragment ions, but \nlittle to no glycosidic bond fragmentation results  [29]. Although this can complicate \ncharacterisation of the glycan due to the absence of B-, Y- and oxonium ions, intact glycopeptide \nfragments are retained, which can be used to unambiguously determine the site of glycosylation. \nExD can also provide more complete  peptide backbone characterisation , as  the N-C⍺\t\nfragmentation in ExD is less dependent on peptide features such as charge state  and amino acid \nsequence than in CID [30]. However, ExD techniques are  potentially limited by several \ndisadvantages. Peptides with low precursor charge state s can yield low  fragment ion intensities  \n[31]. This is especially problematic in the case of d issociation of doubly charged precursors , as \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\nelectron capture necessitates the generation of at least one neutral fragment that cannot be detected \n[32]. ExD methods are therefore most efficient for highly charged peptide precursor ions. However, \nthe addition of neutral glycans in glycopeptides contribute to the mass of the precursor without \ntypically increasing the positive charge  [33]. Further, in the case of sialylated glycopeptides, the \naddition of a negatively charged glycan  can decrease the overall charge  [33]. Reaction times for \nExD techniques are also longer than CID, with up to hundreds of milliseconds required for each \nspectrum, which reduces the number of spectra that can be generated in  an LC-MS/MS analysis. \nIt is for these reasons CID has typically been favoured over ExD methods for glycopeptide analysis. \n \nAnother group of fragmentation strategies worth considering are the combinations of ExD and \ncollisional dissociation [34]. ExD and CID are complementary techniques  which enable analysis \nof the peptide backbone with labile structures intact and the glycan structure , respectively [29]. \nCombining the two (i.e. electron transfer/higher -energy collision dissociation ; EThcD) provides \nspectra with fragment ions typical of both dissociation mode s, although the duty cycle costs \nassociated with large reaction times in ExD are still retained.  \n \nThe ZenoTOF 7600 (S CIEX) allows a variant of ExD known as electron activated dissociation \n(EAD) [35]. One key feature of EAD is the ability to leverage several tuneable parameters for \nspecific experimental needs. Here, we optimize these parameters to allow in -depth analysis of a \ncomplex mixture of mammalian O- and N-glycopeptides. \n \n  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\nMethods \nCell Culture  \nHuman A549 cells  (purchased from the ATCC)  were cultured in Dulbecco’s Modified Eagle’s \nMedium supplemented with 50 U/mL penicillin -streptomycin (1% v/v) and foetal -bovine serum \n(10% v/v). Cells were cultured at 37 °C with injection of 5% CO2. Cells with a passage number < \n25 were used to produce secreted proteins. In brief, A549 cells were grown until exceeding 90% \nconfluence of a T175 flask (~2x107 cells). At this point, cells were washed with chilled PBS four \ntimes to remove growth serum proteins, and then supplemented with serum-free media. Spe nt \nmedia was collected from the cells after 24 hrs , concentrated with 10 kDa MWCO Amicon Ultra \nfilter units (Millipore) via centrifugation at 4 ,000 rcf for 30 min , and exchanged into 50 mM \nHEPES buffer pH 7.4. Protein concentrations were determined using the Qubit Protein \nQuantification Assay Kit (Thermo Fisher Scientific). \n \nProtein Sample Preparation \nProteins (90 µg aliquots) were denatured and reduced by addition of 2x lysis buffer to give final \nconcentrations of 1% SDS, 50 mM Tris-HCl buffer pH 8.0, and 10 mM DTT, and incubated at \n95 °C for  10 min. After cool ing to room temperature, proteins were alkylated by addition of \nacrylamide to a final concentration of 25 mM and incubation at 30 °C with shaking at 1500 rpm \nfor 1 h . Excess acrylamide was quenched by the addition of DTT to an additional final \nconcentration of 5 mM. Proteins were precipitated by addition of four volumes of \nmethanol/acetone (1:1 v/v) and incubation at -20 °C for 16 h. Precipitated samples were centrifuged \nat 18,000 rcf for 10 min to pellet proteins and the supernatant was discarded. Proteins were digested \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\nby resuspension in 50 mM ammonium bicarbonate with porcine trypsin at a 1:20 enzyme to protein \nratio and incubation at 37 °C with shaking at 1500 rpm for 16 h.  \n \nGlycopeptide Enrichment \nPeptides were dried using vacuum centrifugation and resuspended in 500 µL 0.1% trifluoroacetic \nacid (TFA) for sample clean-up with 50 mg Sep-Pak columns (Waters). Peptides underwent HILIC \nenrichment for glycopeptides as previously described [36], with a 1 min incubation period \nfollowing sample addition. HILIC enrichment was performed in 80% ACN using \nPolyHYDROXYETHYL A™, 100 Å pore diameter, 12 µm particle diameter (PolyLC) HILIC \nbeads. Eluted sample s enriched for glycopeptides were dried with vacuum centrifugation and \nresuspended in 0.1% formic acid for LC-MS/MS analysis. \n \nLC-MS/MS Analysis  \nSamples were analysed with a ZenoTOF 7600 (S CIEX) mass spectrometer coupled to a Acquity \nUPLC M-Class system (Waters). Approximately 1-2 µg of enriched glycopeptide sample s were \ninjected onto a Waters nanoEase M/Z HSS T3 C18 column (300 µm x 150 mm, 1.8 µm, 100Å). \nThe mobile phases were A: 0.1% formic acid in water, and B: 0.1% formic acid in acetonitrile. \nPeptides were loaded in 5% B. The LC was held at 5% B for 30 s and peptides were eluted with a \nlinear gradient from 5-35% B over 22 min at a flow rate of 5 µL/min. ZenoTOF source conditions \nincluded: spray voltage , 5000V; Gas 1 and G as 2, 20 psi ; curtain gas , 35 psi ; CAD gas , 7; \ntemperature, 150 ˚C; and column temperature, 35 ˚C.  MS1 spectra were acquired with: mass range, \n300-2250 m/z; accumulation time, 0.2 s; Declustering Potential, 80 V;  Collision Energy, 10 V; \nand time bins to sum , 8. The 20 most intense monoisotopic precursors with intensity > 100 cps, \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\nfrom a candidate mass range of 800 -2250 m/z, and with charge states 2-6 were selected for MS2 \nfragmentation with one of four fragmentation regimes: CID, EAD and two EAD methods with \nsupplemental CID fragmentation termed default and low energ y EAciD. For all fragmentation \nmethods parameters wer: Zeno trapping, on; Zeno threshold, 100,000; TOF mass range, 50-4500 \nm/z; Dynamic background Subtraction , on; and former candidate ion exclu sion, for 6 s after 2 \noccurrences. EAD simultaneous trapping mode was used for EAD and EAciD methods, the latter \nhaving dynamic collision energy enabled. The default equation for dynamic collision energy was \nemployed for CID and default EAciD methods, and a modified equation with half the gradient was \nused for low energ y EAciD. Tuneable parameters  for EAD -based methods were as follows: \nelectron beam current, 5000 nA; electron kinetic energym 8 eV; radio frequency, 200 Da; reaction \ntime, 20 ms; and accumulation timem 50 ms. For CID fragmentation, an accumulation time of 50 \nms was also used . Injections were made in triplicate, with randomised injection order within \nreplicates for each fragmentation method. \n \nGlycopeptide Identification \nGlycopeptides were identified using Byonic (Protein Metrics, v4.3.4) searching against the human \ndatabase of high confidence proteins  (20,350 proteins, downloaded from UniProt 25/11/2021 ) \nappended with porcine trypsin and 199 bovine serum proteins [37], and a glycan database provided \nby Byonic  containing 182 human N-glycans and 6 common O-glycans ( set as “rare 1 ” \nmodifications). Cleavage was set as fully specific, C-terminal to arginine and lysine, and \npermitting two missed cleavage events. Propionamide at cysteine was set as a fixed modification, \nand variable modifications were oxidation at methionine and deamidation of asparagine ( set as \n“common 1” modifications). A total of two common modifications and 1 rare modification were \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\npermitted per glycopeptide. A precursor mass tolerance of 20 ppm, and a fragment mass tolerance \nof 0.1 Da were used. Fragmentation mode was set to HCD for CID, EThcD for EAciD and ECD \nfor EAD methods respectively.  \n \nStatistical Analysis and Data Visualisation \nPrincipal component analysis and heatmaps of glycopeptide peptide spectral matches ( GPSMs) \nand their assigned Byonic scores were generated with ClustVis [38]. Non-linear regression \nanalysis was performed to model the score distributions of GPSMs. Two-tailed unpaired Student’s \nt-tests were performed for comparing the means of two groups  and Fisher exact tests were \nperformed to test the significance of association of categorical variables.  Two-way ANOVA and \nDunnett’s multiple comparison test were performed when comparing quantitative variable changes \nbased on two independent variables. Statistical testing was performed in GraphPad Prism (v 9.4.0). \n \n  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\nResults \nTuneable EAD Parameters \nThe major feature distinguishing EAD from other ExD methods is the ability to fine -tune the \nunderlying instrument parameters to suit specific experimental needs. We surveyed several \ntuneable EAD parameters to optimise conditions for glycopeptide identification . To obtain a \nsample with a complex mixture of mammalian O- and N-glycopeptides, we enriched glycopeptides \nfrom a tryptic digest of proteins secreted from human A549 cells. We repeatedly measured this \nsample with LC -MS/MS with EAD fragmentation and  assessed the effect on glycopeptide \nidentification by Byonic of individually changing each tuneable EAD parameter: radio frequency \n(RF), electron beam current, electron kinetic energy (KE), and reaction time. \n \nThe ZenoTOF 7600 instrument applies an RF voltage to trap ions in the EAD cell to increase the \nefficiency of EAD  fragmentation. A higher RF can be used to trap product ions across a wider \nmass range [39]. We tested RF from 100-300 Da and found that using a lower mass cut-off (LMCO) \nvalue of 200 Da was optimal for glycopeptide identification and spectral quality  (Figure 1A). \nIncreasing the electron KE increases the thermal energy input to the precursor ions, improving the \ndissociation of molecules when electron capture alone is insufficient. However, increasing the \nelectron KE too far reduces the cross-sectional area for electron capture, and results in the electrons \nmore frequently inducing vibrational dissociation and reducing c and z -type ion signals [28]. We \ntested KE from 0 -12 eV, and found that higher values 8 or 12 keV performed better than lower \nvalues (Figure 1A). The electron beam current parameter controls the flow of the electrons used \nto induce fragmentation in EAD. Precursor fragmentation is proportional to the electron beam \ncurrent, but secondary electron capture events that neutralise fragment ions and render them \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\nundetectable by the mass analyser are more frequent with higher electron beam current  [40]. We \ntested electron beam currents from 2500 -7500 nA and found 5000 nA was optimal for \nglycopeptide identification (Figure 1A). Increasing reaction time also improves the fragmentation \nefficiency in EAD, although acquisition times must be compatible with LC peak widths to achieve \ncomprehensive analyte coverage. We tested reaction times from  10-45 ms, and found that the \nlonger 20 or 45 ms reaction times were preferrable (Figure 1A). We subsequently refined the EAD \nparameters for KE and reaction time , in combination . We once again analysed our complex \nglycopeptide sample and tested four EAD methods varying combinations of KE (8 and 12 eV) and \nreaction times (20 and 45 ms) . We observed a  greater number of unique glycopeptide \nidentifications in the methods with the shorter reaction time of 20 ms (Figure 1B). With a 20 ms \nreaction time, we found that adjusting KE did not significantly change the overall number of \nglycopeptide identifications. However, glycopeptide identifications with a KE of 8 eV tended to \nhave higher  scoring GPSMs than the 12 eV method (Figure 1C). Therefore, we selected an \noptimized EAD fragmentation method with a KE of 8 eV, reaction time of 20 ms, RF of 200 Da , \nand electron beam current of 5000 nA for maximum identification of glycopeptides with high \nquality MS/MS spectra.  \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\nFigure 1. Optimisation of EAD tuneable parameters for glycopeptide characterisation. (A)  Distributions of \nByonic scores of unique glycopeptide PSMs from LC -MS/MS analysis of a complex glycoproteome of enriched \nglycopeptides from proteins secreted by human A549 cells, varying radio frequency (RF), electron kinetic energy \n(KE), electron beam cu rrent, and reaction time, in the absence of any other changes. (B) Mean number of unique \nglycopeptide PSMs ± SEM (n=3) with the joint modifications of KE and reaction times. (C) Distributions of Byonic \nscores of unique glycopeptide PSMs with the modificati on of both KE and reaction times. Distributions show PSM \ncounts in bin widths of 50. Non -linear curves were fit to the data using a sum of two gaussian curves model (Prism \nv9.4.0). \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\nComparison of CID and EAD Fragmentation \nHaving determined optimal EAD parameters for glycopeptide analysis, we next performed a \ncomparative analysis of four fragmentation schemes : standard CID (beam -type), EAD, and two \nmixed fragmentation methods combining EAD with supplemental CID fragmentation termed \nhereafter as EAciD. The EAciD methods featured two different levels of supplemental CID, both \nsupplied based on a rolling collision energy (CE) equation determined by precursor ion m/z and \ncharge state. The first EAciD method (default EAciD) used a standard rolling CE equation. The \nsecond method (low energy EAciD) used an adjusted rolling CE equation with a 50% reduced \ngradient. We tested these fragmentation schemes on the complex enriched glycopeptide sample \noriginating from proteins secreted from human A549 cells. Across all methods we identified 1,132 \nhuman protein groups and 834 unique glycopeptid es, including identifications of both O-\nglycopeptides and each of the major glycan classes of N-glycopeptides (oligomannose, hybrid and \ncomplex). We performed principal component analysis (PCA) of the highest score attributed to \neach unique GPSMs by Byonic for each method (Figure 2A). We observed tight clustering \nbetween the replicates of each method, and next between EAD and low energy EAciD  methods, \nconsistent with the methods providing robust yet distinct glycopeptide identification performance \n(Figure 2A).  \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\n \nFigure 2. Variance between the performances of CID, EAD and hybrid EAciD fragmentation methods. LC-\nMS/MS analysis of an enriched glycopeptide sample originating from secreted proteins from human A549 cells (n=3) \nunder four fragmentation schemes: CID (blue), default EAciD (red), low energy EAciD (green) and EAD (purple). \n(A) Principal component analysis (PCA) of uniquely identified glycopeptide PSMs and their highest associated scores \n(Byonic™, PMI). PCA was generated using the web-based tool ClustVis. (B) Number of unique glycopeptide PSMs \n(score > 200) stratified into four categories (O-GalNAc, complex N-glycans, oligomannose N-glycans and hybrid N-\nglycans). Relative proportions of each category are graphically represented within the donut chart above each bar. (C) \nProportions of unique glycopeptide PSMs (score > 200) separated by their associated charge state ( z) from +2-6. A \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\ntwo-way ANOVA analysis and Dunnett’s multiple comparison test was performed to test significance of each method \ncompared with CID for each charge state (Prism v9.4.0). Significant differences were observed between CID and \nEAciD (Low Energy) methods at +3 (p = 0.0082) and +4 (0.0081) charge states, and between CID and EAD also at \n+3 (p = 0.0022) and +4 (p = 0.0103). (D) Distribution of Byonic scores of unique glycopeptide PSMs. Values show \nPSM counts in bin widths of 50. Non -linear curves were fit to the dat a using a sum of two gaussian curves model \n(Prism v9.4.0). Dotted blue line corresponds with the peak of this curve under CID fragmentation. \n \nTo assess the performance of each fragmentation method, we considered a range of quantitative \nand qualitative metrics. We first considered whether the different fragmentation methods biased \nthe resulting characterisation of the glycoproteome. For instance, previous work has reported \nreduced efficiency of ExD methods for glycopeptides with large, negatively charged glycans or \nprecursors with lower charge states [27, 41]. We first asked if the various CID/EAD fragmentation \nmethods identified different proportions of O-glycopeptides or the three major classes of N-\nglycopeptides (complex, oligomannose, and hybrid). We found that the relative proportions of \neach glycopeptide class were comparable with each method (Figure 2B), and hence concluded that \nthe EAD -based fragmentation methods did not bias identification of specific classes of \nglycopeptide compared to CID. We next tested if the performance of the fragmentation method s \nwas impacted by precursor charge state. This analysis showed that EAD and low energy EAciD \nmethods tended to identify glycopeptides with higher charge states, compared to CID and higher \nenergy EAciD (Figure 2C). Although a bias for identification of precursors with higher charge \nstates likely contributed to the lower number of identifications by EAD and low energy EAciD, \nthis did not otherwise bias the coverage of the glycoproteome by these methods.  \n \nOptimising methods to obtain the most glycopeptide identifications is a key goal in LC-MS/MS \nglycoproteomics. However, it is also critical to consider  overall glycopeptide spectral quality, as \nhigher quality spectra increase the confidence of glycan compositional characterisation and site \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\nlocalisation. We therefore next considered the quantitative and qualitative performances of EAD \nand EAciD methods  compared to standard CID methods by considering the distribution of \nglycopeptide identifications and their associated spectral scores (Figure 2D). By this metric, CID \noutperformed all other methods by yielding the most unique glycopeptide PSMs, that were also on \naverage associated with higher quality spectra. Our optimised EAD method identified \ncomparatively fewer glycopeptides. However, s upplementing EAD with CID fragmentation  \nimproved the overall quality of glycopeptide spectra, as determined by a shift in the distributions \ntowards higher scores, and in the case of the default EAciD method, also increased the number of \nglycopeptide identifications (Figure 2D). \n \nEAD-type fragmentation can be particularly powerful for determining the precise sites of labile \nmodifications in a peptide. The “delta mod” score provided by Byonic is a metric of the confidence \nof the localisation of peptide modifications, with scores > 10.0 representing a high likelihood that \nmodifications are correctly localised . We tested the performance of the CID, EAD, and EAciD \nmethods as judged by delta mod scores for O-glycopeptides, and found that both EAD and low \nenergy EAciD outperform ed CID, with a significantly higher proportion of PSMs passing this \nconfidence threshold (Figure 3). \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\nFigure 3. Comparison of CID, EAD, and hybrid EAciD fragmentation methods using a localisation metric. \nMean Byonic delta mod score (± SEM) of unique O-GalNAc glycopeptide PSMs (score > 200) from LC -MS/MS \nanalysis of an enriched glycopeptide sample originating from secreted proteins from human A549 cells using four \nfragmentation methods (CID,  high energy EAciD, low energy EAciD and EAD). Delta mod scores > 10.0 (in blue) \nindicate confidence that all peptide modifications are correctly positioned. A Fisher Exact test is performed between \nCID and EAD-based fragmentation methods based on the number of PSMs that pass and fail the confidence threshold \n(Prism v9.4.0). A significant increase in the number of PSMs passing the confidence threshold was observed with low \nenergy EAciD (p = 0.0086) and EAD (p < 0.0001). \n \nDifferent sets of glycopeptides were identified with significantly higher confidence by the CID or \nEAD-based methods. Of the glycopeptides identified across all  methods, 72 had a significantly \ndifferent score between CID and at least one of the EAD-based methods (Figure 4A). The majority \nof these GPSMs (84.7%) had a higher score  with CID, consistent with the overall better \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\nperformance of this fragmentation method. However, some glycopeptides had a higher score with \nat least one of the EAD-based methods than with CID. We therefore examined these glycopeptides \nto identify any features of these glycopeptides that correlated with their improved score with EAD. \nThe glycopeptides that were better identified with an EAD method were significantly enriched in \ncomplex N-glycans (p≤0.01) (Figure 4B), had significantly higher glycan masses (p≤0.01) (Figure \n4C), and had higher charge states (p≤0.01) (Figure 4D). In contrast, peptide centric features such \nas peptide length (Figure 4E), isoelectric point (Figure 4F), and peptide mass (Figure 4G) were \nnot significantly different. \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\nFigure 4. Characterisation of differentially scored glycopeptide PSMs under CID, EAD, and hybrid EAciD \nfragmentation methods (A) Clustered heatmap of Byonic scores of glycopeptide PSMs (score > 200) from a complex \nenriched glycopeptide sample from human A549 secreted proteins, analysed by LC -MS/MS with four methods of \nfragmentation (CID, default EAciD, low energy EAciD and EAD). Scores are represented within the heatmap as their \nZ-scores, or number of standard deviations from the mean of the entire group. H eatmap clustering was performed \nusing a Euclidean model in ClustVis. (B-G) depict information on glycopeptide PSMs that are scored higher in either \nCID or at least one of the EAD or EAciD methods (B) Relative proportion of complex and oligomannose glycans. A \nFisher Exact test is performed for statistical significance (p = 0.0069). (C) Relative proportion of glycopeptide PSMs \nwith charge states ≤ +3 and ≥ +4. A Fisher Exact test is performed for statistical significance (p = 0.0023). (D) Average \nglycan mass (Da) of glycopeptide PSMs. A two-tailed t-test is performed for statistical significance (p = 0.0055). (E) \nAverage peptide length of glycopeptide PSMs. A two-tailed t-test is performed for statistical significance (p = 0.0690). \n(F) The average isoelectric point (pI) of the underlying peptide from each glycopeptide PSM. A two -tailed t-test is \nperformed for statistical significance ( p = 0.471). (G) Average peptide mass (Da) of the underlying peptide from each \nglycopeptide PSM. A two-tailed t-test is performed for statistical significance (p = 0.0549). Statistical tests performed \nin Prism v9.4.0. \n \nAlthough spectral scores provide a good indication of overall quality, direct comparisons between \nthe glycopeptide spectra produced by different fragmentation methods can provide additional \nevidence of the performance of each that may be overlooked by simp le metrics. We assessed the \nperformance of each fragmentation method by examining glycopeptide PSMs for different types \nof glycosylation identified across each dissociation method. Spectra for an Ephrin -A1 (P20827) \nglycopeptide modified with a HexNAc(2)Hex(8) oligomannose glycan varied by method (Figure \n5A-D). Precursor consumption in EAD and low energy EAciD was low compared with default \nEAciD and standard CID (Figure 5A-D). However, the presence of c6 and z6 fragment ions in all \nthree EAD -based methods allowed unambiguous localisation of the N-glycan (Figure 5B-D). \nOxonium ions and B- and Y-ions in CID, default EAciD and, to a lesser extent, low energy EAciD \nallowed characterisation of the glycan (Figure 5A-C). Spectra for a metalloproteinase inhibitor-1 \n(P01033) glycopeptide modified with a HexNAc(4)Hex(5)Fuc(1)NeuAc(2) complex glycan \nrevealed both similarities and differences to the oligomannose glycopeptide. CID and EAciD \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\nspectra were dominated by oxonium ion and B - and Y -fragment ions useful for glycan \ncharacterisation (Figure 6A-C). The CID spectra featured b- and y-ions that could characterise the \npeptide backbone, albeit at relatively low intensity compared to oxonium ions (Figure 6A). Default \nEAciD spectra for this glycopeptide did not generate intact glycopeptide fragment ions that could \nbe used to localise the glycan (Figure 6B), but this was possible with low energy EAciD and EAD, \nwhich produced informative doubly charged c8++ and z8++ ions (Figure 6C & D). Analysis of an \nO-glycopeptide from cathepsin D (P07339) showed informative oxonium and B- and Y-fragments \nwith CID (Figure 7A-C), low precursor consumption in EAD and low energy EAciD (Figure 7C \n& D ), and the presence of the c9 fragment ion in all EAD -based methods that enabled site-\nlocalisation of the O-glycan (Figure 7B-D). \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\n \nFigure 5: Fragmentation patterns of an oligomannose glycopeptide under various fragmentation methods. \nMS/MS analysis of a glycopeptide of a human Ephrin-A1 glycoprotein (P20827) with an oligomannose type N-glycan \nHexNAc(2)Hex(8) attached under four fragmentation methods (A) CID fragmented (observed 1007.078 m/z, 3+ and \nscoring 554.04) (B) default EAciD fragmented (observed 1007.078 m/z, 3+ and scoring 398.60) (C) low energy EAciD \nfragmented (observed 1007.083 m/z, 3+ and scoring 371.64) and  (D) EAD fragmented (observed 1007.078 m/z, 3+ \nand scoring 346.22)   \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\n \nFigure 6: Fragmentation patterns of a complex glycopeptide under various fragmentation methods. MS/MS \nanalysis of a glycopeptide of a human metalloproteinase inhibitor -1 glycoprotein (P01033) with a complex type N-\nglycan HexNAc(4)Hex(5)Fuc(1)NeuAc(2) attached under four fragmentation methods (A) CID fragmented (observed \n1026.691 m/z, 4+ and scoring 677.18) (B) default EAciD fragmented (observed 1026.689 m/z, 4+ and scoring 494.57) \n(C) low energy EAciD fragmented (observed 1026.681 m/z, 4+ and scoring 590.87 ) and (D) EAD fragmented \n(observed 1026.693 m/z, 4+ and scoring 692.49).   \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\n \nFigure 7: Fragmentation patterns of an O-glycopeptide under various fragmentation methods. MS/MS analysis \nof a glycopeptide of a human cathepsin D glycoprotein (P07339) with an O-glycan HexNAc(1)Hex(1)NeuAc(2) \nattached under four fragmentation methods (A) CID fragmented (observed 949.120 m/z, 3+ and scoring 889.29) (B) \ndefault EAciD fragmented (observed 949.122 m/z, 3+ and scoring 615.58) (C) low energy EAciD fragmented \n(observed 949.125 m/z, 3+ and scoring 482.75)  and (D) EAD fragmented (observed 949.120 m/z, 3+ and scoring \n382.25). \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\nDiscussion \nMaking use of the possibility of tuning EAD parameters on the ZenoTOF 7600, we optimized RF, \nelectron KE current, beam current, and reaction times  for analysis of a complex sample of \nmammalian O- and N-glycopeptides. We found that the highest number of glycopeptide \nidentifications were observed using EAD with electron KE values of 8 and 12 eV. This is \nconsistent with previous studies that have reported advantages in glycopeptide analysis in the hot \nECD range [42], particularly for  sialylated glyc opeptides [27]. Optimal KE values for EAD \nfragmentation in glycoproteomic analyses will likely vary depending on the sample glycan \ncomposition, although a hot ECD value of ~8 eV is reasonable for glycopeptide analysis. We found \nthat a moderately increased RF amplitude improved glycopeptide identification, however further \nincreases in RF amplitude began to instead have a detrimental effect. Glycopeptide fragmentation \nproduces ions across a wide mass range, from oxonium ions to large Y-ions, consistent with better \nperformance at moderate to high RF values. Reduced performance at high RF amplitude has been \nattributed to increased electron energy by RF heating, and loss of ion trapping consistent with our \nobservations of a moderate optimal RF value [35, 43]. We found an electron beam current of 5000 \nnA was optimal for glycopeptide analysis, consistent with previous reports [40]. Finally, a reaction \ntime of 20 ms was optimal for glycopeptide analysis, consistent with standard methods for peptide \nanalysis. While we identified the optimum parameters for this particular sample, it is possible that \nother samples, in particular those with different types of glycosylation, may benefit from different \nEAD parameters. \n \nDirectly comparing CID, EAD, and EAciD methods, we found that CID performed best in terms \nof the total number of glycopeptides identified, while EAD and EAciD were superior in providing \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\nlocalisation information in the form of c and z product ions with an intact glycan. EAciD methods \nfeature peptide backbone fragmentation, c and z fragment ions  that can localise the site of \nglycosylation, and oxonium ion profiles that can support glycan composition assignments, \nenabling more complete glycopeptide characterisation. However, we found that EAciD methods \ndid come with a cost to the overall number of glycopeptide identifications. We applied two levels \nof supplemental CID, both based on rolling collision energy. Although both EAciD methods had \nsome beneficial features of CID and EAD, the performance of the two methods was quite different. \nThe default EAciD  method performed more like CID, with poorer localisation metrics but \nincreased identifications. In contrast, reducing the supplemental collision energy in the low energy \nEAciD method improved localisation and generated information rich spectra, but had relatively \nfew glycopeptide identifications. The choice to use CID, EAD, or EAciD for glycoproteomics is \ndependent on whether the desired experimental objectives  favour greater coverage of the \nglycoproteome or in-depth characterisation of the measured glycopeptides.  \n \nConclusion \nWe have developed and demonstrated an optimised workflow for mammalian glycoproteomics \nusing EAD fragmentation on the ZenoTOF 7600. We found that standard beam-type CID could \nidentify more glycopeptid es than EAD, but that EAD could provide more  information on the \nprecise sites of glycosylation. EAD with supplemental collision energy (EAciD) further improved \nglycoproteomic analyses compared to standard EAD.  This EAciD combination of EAD and CID \nfragmentation provides highly informative glycopeptide MS/MS spectra suitable for \ncharacterising the peptide, glycan, and the site of modification for glycopeptides with diverse O- \nor N-glycans.  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\nAcknowledgements \nWe thank The University of Queensland, School of Chemistry and Molecular Biosciences Mass \nSpectrometry Facility for assistance and expertise. This work was supported by a National Health \nand Medical Research Council (NHMRC) Ideas grant APP1186699 to B. L.S. and C. L.P, and \nAustralian Research Council  (ARC) Linkage Infrastructure, Equipment and Facilities  grant \nLE220100068. \n \nReferences \n1. Sarkar A, Wintrode PL. Effects of glycosylation on the stability and flexibility of a \nmetastable protein: the human serpin α(1)-antitrypsin. Int J Mass Spectrom. 2011;302(1-3):69-75. \ndoi: 10.1016/j.ijms.2010.08.003. PubMed PMID: 21765645; PubMed Central PMCID: \nPMCPMC3134971. \n2. Apweiler R, Hermjakob H, Sharon N. On the frequency of protein glycosylation, as \ndeduced from analysis of the SWISS-PROT database11Dedicated to Prof. Akira Kobata and Prof. \nHarry Schachter on the occasion of their 65th birthdays. Biochimica et Biophysic a Acta (BBA) - \nGeneral Subjects. 1999;1473(1):4-8. doi: 10.1016/S0304-4165(99)00165-8. \n3. Sun F, Suttapitugsakul S, Wu R. Unraveling the surface glycoprotein interaction network \nby integrating chemical crosslinking with MS-based proteomics. Chem Sci. 2021;12(6):2146-55. \nEpub 20210104. doi: 10.1039/d0sc06327d. PubMed PMID: 34163979; PubMed Central PMCID: \nPMCPMC8179341. \n4. Stanley P, Moremen KW, Lewis NE, Taniguchi N, Aebi M. N -Glycans. In: Varki A, \nCummings RD, Esko JD, Stanley P, Hart GW, Aebi M, et al., editors. Essentials of Glycobiology. \nCold Spring Harbor (NY): Cold Spring Harbor Laboratory Press  Copyright © 2022 The \nConsortium of Glycobiology Editors, La Jolla, California; published by Cold Spring Harbor \nLaboratory Press; doi:10.1101/glycobiology.4e.9. All rights reserved.; 2022. p. 103-16. \n5. Haltiwanger RS, Wells L, Freeze HH, Jafar-Nejad H, Okajima T, Stanley P. Other Classes \nof Eukaryotic Glycans. In: Varki A, Cummings RD, Esko JD, Stanley P, Hart GW, Aebi M, et al., \neditors. Essentials of Glycobiology. Cold Spring Harbor (NY): Cold Sprin g Harbor Laboratory \nPress Copyright © 2022 The Consortium of Glycobiology Editors, La Jolla, California; published \nby Cold Spring Harbor Laboratory Press; doi:10.1101/glycobiology.4e.13. All rights reserved.; \n2022. p. 155-64. \n6. Gagneux P, Panin V, Hennet T, Aebi M, Varki A. Evolution of Glycan Diversity. In: Varki \nA, Cummings RD, Esko JD, Stanley P, Hart GW, Aebi M, et al., editors. Essentials of \nGlycobiology. Cold Spring Harbor (NY): Cold Spring Harbor Laboratory Press Copyright © 2022 \nThe Consortium of Glycobiology Editors, La Jolla, California; published by Cold Spring Harbor \nLaboratory Press; doi:10.1101/glycobiology.4e.20. All rights reserved.; 2022. p. 265-78. \n7. Schjoldager KT, Narimatsu Y, Joshi HJ, Clausen H. Global view of human protein \nglycosylation pathways and functions. Nat Rev Mol Cell Biol. 2020;21(12):729 -49. Epub \n20201021. doi: 10.1038/s41580-020-00294-x. PubMed PMID: 33087899. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\n8. Aebi M. N -linked protein glycosylation in the ER. Biochim Biophys Acta. \n2013;1833(11):2430-7. Epub 20130410. doi: 10.1016/j.bbamcr.2013.04.001. PubMed PMID: \n23583305. \n9. Lizak C, Gerber S, Numao S, Aebi M, Locher KP. X -ray structure of a bacterial \noligosaccharyltransferase. Nature. 2011;474(7351):350 -5. Epub 20110615. doi: \n10.1038/nature10151. PubMed PMID: 21677752. \n10. Wild R, Kowal J, Eyring J, Ngwa EM, Aebi M, Locher KP. Structure of the yeast \noligosaccharyltransferase complex gives insight into eukaryotic N -glycosylation. Science. \n2018;359(6375):545-50. Epub 20180104. doi: 10.1126/science.aar5140. PubMed PMID: \n29301962. \n11. Ramírez AS, Kowal J, Locher KP. Cryo -electron microscopy structures of human \noligosaccharyltransferase complexes OST -A and OST -B. Science. 2019;366(6471):1372 -5. doi: \n10.1126/science.aaz3505. PubMed PMID: 31831667. \n12. Ince D, Lucas TM, Malaker SA. Current strategies for characterization of mucin -domain \nglycoproteins. Curr Opin Chem Biol. 2022;69:102174. Epub 20220622. doi: \n10.1016/j.cbpa.2022.102174. PubMed PMID: 35752002. \n13. Hülsmeier AJ, Tobler M, Burda P, Hennet T. Glycosylation site occupancy in health, \ncongenital disorder of glycosylation and fatty liver disease. Sci Rep. 2016;6:33927. Epub \n20161011. doi: 10.1038/srep33927. PubMed PMID: 27725718; PubMed Central PMCID: \nPMCPMC5057071. \n14. Villacrés C, Tayi VS, Lattová E, Perreault H, Butler M. Low glucose depletes glycan \nprecursors, reduces site occupancy and galactosylation of a monoclonal antibody in CHO cell \nculture. Biotechnology Journal. 2015;10(7):1051-66. doi: 10.1002/biot.201400662. \n15. Zacchi LF, Schulz BL. N-glycoprotein macroheterogeneity: biological implications and \nproteomic characterization. Glycoconj J. 2016;33(3):359 -76. Epub 20151205. doi: \n10.1007/s10719-015-9641-3. PubMed PMID: 26638212. \n16. Bagdonaite I, Malaker SA, Polasky DA, Riley NM, Schjoldager K, Vakhrushev SY, et al. \nGlycoproteomics. Nature Reviews Methods Primers. 2022;2(1):48. doi: 10.1038/s43586 -022-\n00128-4. \n17. Bergstrom KS, Xia L. Mucin -type O -glycans and their roles in intestinal homeostasis. \nGlycobiology. 2013;23(9):1026-37. Epub 20130610. doi: 10.1093/glycob/cwt045. PubMed PMID: \n23752712; PubMed Central PMCID: PMCPMC3858029. \n18. Stavenhagen K, Hinneburg H, Thaysen-Andersen M, Hartmann L, Silva DV, Fuchser J, et \nal. Quantitative mapping of glycoprotein micro -heterogeneity and macro -heterogeneity: an \nevaluation of mass spectrometry signal strengths using synthetic peptides and g lycopeptides. \nJournal of Mass Spectrometry. 2013;48(6):627-39. doi: 10.1002/jms.3210. \n19. Piovesana S, Cavaliere C, Cerrato A, Laganà A, Montone CM, Capriotti AL. Recent trends \nin glycoproteomics by characterization of intact glycopeptides. Analytical and Bioanalytical \nChemistry. 2023. doi: 10.1007/s00216-023-04592-z. \n20. Cao L, Tolić N, Qu Y, Meng D, Zhao R, Zhang Q, et al. Characterization of intact N- and \nO-linked glycopeptides using higher energy collisional dissociation. Analytical Biochemistry. \n2014;452:96-102. doi: https://doi.org/10.1016/j.ab.2014.01.003. \n21. Mitchell Wells J, McLuckey SA. Collision‐Induced Dissociation (CID) of Peptides and \nProteins.  Methods in Enzymology. 402: Academic Press; 2005. p. 148-85. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\n22. Domon B, Costello CE. A systematic nomenclature for carbohydrate fragmentations in \nFAB-MS/MS spectra of glycoconjugates. Glycoconjugate Journal. 1988;5(4):397 -409. doi: \n10.1007/BF01049915. \n23. Hoffmann M, Pioch M, Pralow A, Hennig R, Kottler R, Reichl U, et al. The Fine Art of \nDestruction: A Guide to In-Depth Glycoproteomic Analyses—Exploiting the Diagnostic Potential \nof Fragment Ions. PROTEOMICS. 2018;18(24):1800282. doi: 10.1002/pmic.201800282. \n24. Khatri K, Pu Y, Klein JA, Wei J, Costello CE, Lin C, et al. Comparison of Collisional and \nElectron-Based Dissociation Modes for Middle-Down Analysis of Multiply Glycosylated Peptides. \nJ Am Soc Mass Spectrom. 2018;29(6):1075-85. Epub 20180416. doi: 10.1007/s13361-018-1909-\ny. PubMed PMID: 29663256; PubMed Central PMCID: PMCPMC6004259. \n25. Thaysen-Andersen M, Wilkinson BL, Payne RJ, Packer NH. Site-specific characterisation \nof densely O-glycosylated mucin-type peptides using electron transfer dissociation ESI -MS/MS. \nELECTROPHORESIS. 2011;32(24):3536-45. doi: 10.1002/elps.201100294. \n26. Mirgorodskaya E, Roepstorff P, Zubarev RA. Localization of O -Glycosylation Sites in \nPeptides by Electron Capture Dissociation in a Fourier Transform Mass Spectrometer. Analytical \nChemistry. 1999;71(20):4431-6. doi: 10.1021/ac990578v. \n27. Manri N, Satake H, Kaneko A, Hirabayashi A, Baba T, Sakamoto T. Glycopeptide \nIdentification Using Liquid-Chromatography-Compatible Hot Electron Capture Dissociation in a \nRadio-Frequency-Quadrupole Ion Trap. Analytical Chemistry. 2013;85(4):2056 -63. doi : \n10.1021/ac301834t. \n28. Zhurov KO, Fornelli L, Wodrich MD, Laskay ÜA, Tsybin YO. Principles of electron \ncapture and transfer dissociation mass spectrometry applied to peptide and protein structure \nanalysis. Chemical Society Reviews. 2013;42(12):5014-30. doi: 10.1039/C3CS35477F. \n29. Zhao C, Xie B, Chan S-Y, Costello CE, O’Connor PB. Collisionally Activated Dissociation \nand Electron Capture Dissociation Provide Complementary Structural Information for Branched \nPermethylated Oligosaccharides. Journal of the American Society for Mass  Spectrometry. \n2008;19(1):138-50. doi: 10.1016/j.jasms.2007.10.022. \n30. Budnik BA, Nielsen ML, Olsen JV, Haselmann KF, Hörth P, Haehnel W, et al. Can relative \ncleavage frequencies in peptides provide additional sequence information? International Journal \nof Mass Spectrometry. 2002;219(1):283-94. doi: 10.1016/S1387-3806(01)00579-6. \n31. Riley NM, Coon JJ. The Role of Electron Transfer Dissociation in Modern Proteomics. \nAnalytical Chemistry. 2018;90(1):40-64. doi: 10.1021/acs.analchem.7b04810. \n32. Zubarev RA. Reactions of polypeptide ions with electrons in the gas phase. Mass Spectrom \nRev. 2003;22(1):57-77. doi: 10.1002/mas.10042. PubMed PMID: 12768604. \n33. Alagesan K, Hinneburg H, Seeberger PH, Silva DV, Kolarich D. Glycan size and \nattachment site location affect electron transfer dissociation (ETD) fragmentation and automated \nglycopeptide identification. Glycoconjugate Journal. 2019;36(6):487 -93. doi: 1 0.1007/s10719-\n019-09888-w. \n34. Swaney DL, McAlister GC, Wirtala M, Schwartz JC, Syka JEP, Coon JJ. Supplemental \nActivation Method for High -Efficiency Electron -Transfer Dissociation of Doubly Protonated \nPeptide Precursors. Analytical Chemistry. 2007;79(2):477-85. doi: 10.1021/ac061457f. \n35. Baba T, Ryumin P, Duchoslav E, Chen K, Chelur A, Loyd B, et al. Dissociation of \nBiomolecules by an Intense Low-Energy Electron Beam in a High Sensitivity Time-of-Flight Mass \nSpectrometer. Journal of the American Society for Mass Spectrometry. 2021;32(8 ):1964-75. doi: \n10.1021/jasms.0c00425. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint \n\n36. Alagesan K, Khilji SK, Kolarich D. It is all about the solvent: on the importance of the \nmobile phase for ZIC-HILIC glycopeptide enrichment. Anal Bioanal Chem. 2017;409(2):529-38. \nEpub 20161201. doi: 10.1007/s00216 -016-0051-6. PubMed PMID: 27909778; Pu bMed Central \nPMCID: PMCPMC5203826. \n37. Shin J, Kim G, Kabir MH, Park SJ, Lee ST, Lee C. Use of composite protein database \nincluding search result sequences for mass spectrometric analysis of cell secretome. PLoS One. \n2015;10(3):e0121692. Epub 20150330. doi: 10.1371/journal.pone.0121692. Pub Med PMID: \n25822838; PubMed Central PMCID: PMCPMC4378925. \n38. Metsalu T, Vilo J. ClustVis: a web tool for visualizing clustering of multivariate data using \nPrincipal Component Analysis and heatmap. Nucleic Acids Research. 2015;43(W1):W566 -W70. \ndoi: 10.1093/nar/gkv468. \n39. Baba T, Campbell JL, Le Blanc JCY, Hager JW, Thomson BA. Electron Capture \nDissociation in a Branched Radio-Frequency Ion Trap. Analytical Chemistry. 2015;87(1):785-92. \ndoi: 10.1021/ac503773y. \n40. Baba T, Rajabi K, Liu S, Ryumin P, Zhang Z, Pohl K, et al. Electron Impact Excitation of \nIons from Organics on Singly Protonated Peptides with and without Post -Translational \nModifications. Journal of the American Society for Mass Spectrometry. 2022;33(9):1723-32. doi: \n10.1021/jasms.2c00146. \n41. Iavarone AT, Paech K, Williams ER. Effects of charge state and cationizing agent on the \nelectron capture dissociation of a peptide. Anal Chem. 2004;76(8):2231 -8. doi: \n10.1021/ac035431p. PubMed PMID: 15080732; PubMed Central PMCID: PMCPMC1343469. \n42. Baba T, Zhang Z, Liu S, Burton L, Ryumin P, Le Blanc JCY. Localization of Multiple O-\nLinked Glycans Exhibited in Isomeric Glycopeptides by Hot Electron Capture Dissociation. \nJournal of Proteome Research. 2022;21(10):2462-71. doi: 10.1021/acs.jproteome.2c00378. \n43. Baba T, Hashimoto Y, Hasegawa H, Hirabayashi A, Waki I. Electron Capture Dissociation \nin a Radio Frequency Ion Trap. Analytical Chemistry. 2004;76(15):4263 -6. doi: \n10.1021/ac049309h. \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 23, 2024. ; https://doi.org/10.1101/2024.02.22.581095doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}