Increasing the Productivity of Glycopeptides Analysis by Using Higher-Energy Collision Dissociation-Accurate Mass-Product-Dependent Electron Transfer Dissociation

Currently, glycans are attracting attention from the scientific community as potential biomarkers or as posttranslational modifications (PTMs) of therapeutic proteins. However, structural characterization of glycoproteins and glycopeptides remains analytically challenging. Here, we report on the implementation of a novel acquisition strategy termed higher-energy collision dissociation-accurate mass-product-dependent electron transfer dissociation (HCD-PD-ETD) on a hybrid linear ion trap-orbitrap mass spectrometer. This acquisition strategy uses the complementary fragmentations of ETD and HCD for glycopeptides analysis in an intelligent fashion. Furthermore, the approach minimizes user input for optimizing instrumental parameters and enables straightforward detection of glycopeptides. ETD spectra are only acquired when glycan oxonium ions from MS/MS HCD are detected. The advantage of this approach is that it streamlines data analysis and improves dynamic range and duty cycle. Here, we present the benefits of HCD-PD-ETD relative to the traditional alternating HCD/ETD for a trainer set containing twelve-protein mixture with two glycoproteins: human serotransferrin, ovalbumin and contaminations of two other: bovine alpha 1 acid glycoprotein (bAGP) and bovine fetuin.


Introduction
Glycosylation is an important PTM that plays crucial roles in various biochemical processes, ranging from mediation of interactions between cells to defining cellular identities within complex tissues [1,2]. In addition, glycan structures are unique to the proteins which they are attached to and to the site of attachment and, thus, play crucial roles in controlling the activities of the proteins. Many glycans, also, show disease-related expression level changes [3,4]. For example, changes in glycosylation patterns have been used as a means to monitor progression of cancer [3][4][5][6]. In many instances, it is essential to characterize the exact glycan structure and site of glycosylation to better understand the protein-mediated interaction that these glycans undergo.
Mass spectrometry (MS) has emerged as one of the most powerful tools for proteomics due to its sensitivity of detection and its ability to analyze complex mixtures derived from a variety of organisms and cell lines. However, structural characterization of glycoproteins/glycopeptides remains analytically challenging due to the reliance on traditional acquisition strategies and MS/MS fragmentation techniques.
Conventional proteomics has benefitted tremendously from collision-activated dissociation (CAD) due to the ease of implementation of the technique on most commercial mass spectrometers and the abundant peptide bond cleavages that this technique generates, resulting in large number of peptides and proteins identified. Unfortunately for glycoproteomics, CAD does not provide the necessary fragment ions to thoroughly characterize intact glycopeptide structures [7]. Depending upon the mass spectrometers used, CAD provides varying degrees of structural information. For example, low-energy CAD on most mass spectrometers predominantly generates glycosidic bond cleavages with minimal fragmentation occurring along the peptide backbone [8][9][10][11]. Additionally, cleavages also tend to occur between the peptide-glycan bond, resulting in loss of information about the site of glycosylation. The increase of collision energy can 2 International Journal of Proteomics result in more efficient fragmentation of the peptide backbone, but this strategy can result in mixed MS/MS spectra where both the glycan and peptide fragment ions are present in the same spectra, complicating spectral interpretation [12]. Regardless of whether high-or low-energy CAD is employed, fragmentation of the peptide-glycan bond still occurs limiting the ability to derive information about the site of glycosylation. Majority of the current approaches have foregone the strategy of examining intact glycopeptides and have focused on obtaining partial information, such as sequencing peptide backbone and identifying sites of glycosylation. For example, the use of N-glycosidase F or A (PNGase F/A) enzymes results in the removal of glycans and conversion of asparagines, the site of glycosylation to aspartic acid. This conversion process can then be monitored by highresolution MS due to a mass shift of 0.9840 Da to identify the site of glycosylation. Additionally, one can increase the confidence in the glycosylation site assignment by incorporation of stable isotope labeling by performing PNGaseF/A digestion in the presence of H 2 O 18 . Such approach involves the release of glycans by a deamidation reaction, and the incorporation of H 2 O 18 will lead to a mass shift of 2.9890 Da on the asparagine residue. But, studies have shown that these types of chemical deamidations can occur spontaneously during sample preparation for release of glycans in the presence and absence of H 2 O 18 leading to number of false positives [13][14][15][16][17]. These issues underline the importance of intact glycopeptides structural analysis, and only this approach enables comprehensive structural characterization. Though CAD is limiting for glycopeptides analysis, alternative fragmentation techniques such as electron-capture dissociation (ECD) [18] and ETD [19,20] are better suited for glycopeptides analysis due to their nonergodic type of dissociation. These dissociation techniques induce extensive fragmentation of the peptide backbone enabling sequencing of the peptide but preserving glycans on the peptide backbone allowing unambiguous assignment of the glycosylation sites, thus providing complementary information to CAD.
Previously, several studies have demonstrated the utility of combining CAD and ETD fragmentation for glycopeptides characterization [19][20][21][22]. For example, Catalina et al. have performed detailed analysis of horse peroxidase using alternating CID, ETD fragmentation [20]. We have shown in the past the importance of combining ETD and CAD, specifically HCD, for full structural characterization of glycopeptides [22]. However, all of these studies have used both types of fragmentation in an unselective fashion. Here, we expand on this approach to report a novel acquisition strategy termed HCD-PD-ETD that enables on the fly identification of glycopeptides and improves overall productivity of glycopeptides analysis. In this approach, the LTQ Orbitrap Velos acquires HCD spectra in a data-dependent fashion. The instrument identifies glycan oxonium ions on the fly in the HCD spectra and triggers ETD spectra on the glycopeptides precursor only. The advantage of this approach is that it streamlines data analysis and improves dynamic range and duty cycle. The benefits of this novel acquisition strategy are examined for a trainer set containing twelve-protein mixture with two glycoproteins and contaminants of two others and compared against the traditional alternating HCD/ETD approach.

Materials and Methods
All protein and glycoprotein standards used in these experiments were purchased from Sigma-Aldrich Chemical Co. (St. Louis, MO, USA).

TMT Labeling.
Isobaric labeling was accomplished with TMT 6 reagent (Thermo Scientific, Rockford, IL, USA) per manufacturer's suggestion. Briefly, the reagent was dissolved in 41 µL of anhydrous acetonitrile and added to 100 µg of tryptically digested twelve-protein mixture. The sample was then incubated at room temperature for 1 hour and quenched with the addition of 8 µL of 5% hydroxylamine and concentrated to 10 µL using Speed Vac (Thermo Scientific, Waltham, MA, USA).  400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600       tolerance and 0.02 Da HCD tolerance, while allowing up to one missed cleavage. For peptide searches, carboxymethylation of cysteine residues (+58.005478 Da) was set as static modification. TMT 6 tag on lysine residues and peptide N termini (+229.16293 Da) were set as variable modification. For glycan identification, the following possible monosaccharide combinations were searched: Hexose, HexNAc, Deoxyhexose, Neu5Gc, and Neu5Ac. Both O-and N-linked glycosylations were considered with H, Na, and K as potential adducts. To minimize on false hits for site of glycosylation, GlycoMaster considered N-linked glycosylation to occur at the amide residue of asparagine in a consensus sequence of Asn-Xaa-Ser/Thr (X can be any amino acid except proline)

LC/MS. A Thermo
and O-linked glycosylation to occur at serine or threonine residues. All results were manually validated.

Results and Discussion
Current MS/MS acquisition strategies for glycopeptides analysis rely on acquiring MS/MS spectra for all precursors in a mass spectrum regardless of whether the precursor belongs to a glycopeptide. The resultant spectra are then interrogated after acquisition for characteristic glycan oxonium ions (which are dominant in the MS/MS CAD spectra) for detection of glycopeptides, and the corresponding spectra are then used in the identification [19][20][21][22][23][24][25]. The limitation  of this approach is that both data acquisition and data interrogation are not performed as efficiently as possible. By triggering MS/MS spectra on all precursors, both duty cycle and dynamic range of analysis are decreased. Furthermore, postacquisition extraction of glycan oxonium ions requires additional tools to ascertain information about detected glycopeptides: their charge state and m/z. Further adding to the inefficiency of the current MS/MS acquisition strategies is that if a mass spectrometer with nominal mass accuracy is used, then more than one glycan oxonium ion must be extracted to minimize on false positives due to near mass isobaric ions [26]. However, in our approach, we employ HCD to generate glycan oxonium ions. The advantage of using HCD is that the generated fragment ions are measured within the orbitrap mass analyzer with high resolution and mass accuracy (HR/AM) allowing for unambiguous assignment of glycan oxonium ions [23]. Using orbitrap detection to our advantage, we have implanted a novel instrumental control within the hybrid linear ion trap-orbitrap mass spectrometer termed HCD-PD-ETD ( Figure 1). This instrumental control enables the mass spectrometer to identify glycopeptides on the fly and acquire data in an intelligent fashion. In this approach, we specify certain glycan oxonium ions for the instruments to monitor within the HCD spectrum (up to ten different product ions can be specified).
We typically use m/z 204.0864 HexNAc oxonium ion and its fragments 168.0653 and 138.0550 as diagnostic ions to be monitored in HCD fragmentation, and, by method definition, the monitored ions should be in the top 20-30 most intense fragment ions. However, if glycopeptides of interest do not contain terminal HexNAc, other diagnostic oxonium/product ions can be selected and collision energy optimized accordingly. The hybrid linear ion trap-orbitrap mass spectrometer equipped with ETD then acquires HCD spectra in a data-dependent fashion. Each HCD spectrum is interrogated by the instrument on the fly for the presence of diagnostic glycan oxonium (product) ions at ppm mass accuracy. If a diagnostic glycan oxonium (product) ion is detected, an ETD spectrum for the glycopeptide precursor is acquired. It should be noted that we can specify mass accuracy as low as 5 ppm for high selectivity. This novel acquisition strategy provides numerous advantages. Primarily, it increases overall productivity for MS analysis of glycopeptides by acquiring ETD spectra only when a glycopeptide is detected, enabling improvements in both duty cycle and dynamic range of analysis. Additionally, this approach improves data analysis by decreasing the rate of false positives, the overall file size, and the number of ETD spectra that are extrapolated to characterize glycopeptides. To demonstrate the advantage of our novel acquisition strategy, we selected a twelve-protein mixture containing two glycoproteins: human serotransferrin and ovalbumin and contaminants of two others: bAGP and bovine fetuin. The mixture was digested with trypsin and analyzed by the traditional alternating HCD/ETD approach and the novel HCD-PD-ETD approach. Enrichment steps were avoided prior to analysis to ensure that the glycopeptides would be very low in abundances; this gave us the opportunity to evaluate the dynamic range of analysis for both acquisition strategies. Of the two, only HCD-PD-ETD approach identified glycopeptides from the contaminant glycoproteins (bAGP and bovine fetuin) present in the sample (Figure 2). While the HCD/ETD acquisition strategy identified numerous peptides and expected glycopeptides (eluted at 34, 40, and 48 min as shown in Figure 2 for HCD-PD-ETD approach) from serotransferrin and ovalbumin, it was unable to identify glycopeptides from the contaminant glycoproteins. Manual inspection of the RAW data confirmed this, as precursors associated with the contaminant glycopeptides were not targeted for MS/MS by HCD or ETD in the alternating approach. Analysis of the HCD-PD-ETD data revealed the identification of glycopeptides from expected glycoproteins and the contaminant glycoproteins. Identified minor glycopeptides from the contaminant glycoproteins by HCD-PD-ETD acquisition method are presented in Table 1, and representative ETD MS/MS spectra used in identification are shown in Figures 3 and 4. Because HCD-PD-ETD acquisition strategy did not waste time acquiring ETD data for nonglycosylated peptides, it was able to dig deeper into the sample and identify this low abundant glycopeptides that the alternating HCD/ETD acquisition strategy could not target.
In summary, we have been able to develop a novel acquisition strategy that minimizes some of the pain points associated with glycopeptide analysis. By combining the complementary information provided by HCD and ETD in an intelligent acquisition control, the overall productivity of glycopeptides analysis is improved.