¹⁵N Metabolic Labeling - TMT Multiplexing Approach to Facilitate the Quantitation of Glycopeptides Derived from Cell Lines

Abstract

The study of glycoproteomics presents a set of unique challenges, primarily due to the low abundance of glycopeptides and their intricate heterogeneity, which is specific to each site. Glycoproteins play a crucial role in numerous biological functions, including cell signaling, adhesion, and intercellular communication, and are increasingly recognized as vital markers in the diagnosis and study of various diseases. Consequently, a quantitative approach to glycopeptide research is essential. One effective strategy to address this need is the use of multiplex glycopeptide labeling. By harnessing the synergies of ¹⁵N metabolic labeling via the isotopic detection of amino sugars with glutamine (IDAWG) technique for glycan parts and tandem mass tag (TMT)pro labeling for peptide backbones, we've developed a method that allows for the accurate quantification and comparison of multiple samples simultaneously. The adoption of the Liquid Chromatography-Synchronous Precursor Selection (LC-SPS-MS3) technique minimizes fragmentation interference, enhancing data reliability, as shown by a 97% TMT labeling efficiency. This method allows for detailed, high-throughput analysis of 32 diverse samples from 231BR cell lines, using both ¹⁴N and ¹⁵N glycopeptides at a 1:1 ratio. A key component of our methodology was the precise correction for isotope and TMTpro distortions, significantly improving quantification accuracy to less than 5% distortion. This breakthrough enhances the efficiency and accuracy of glycoproteomic studies, increasing our understanding of glycoproteins in health and disease. Its applicability to various cancer cell types sets a new standard in quantitative glycoproteomics, enabling deeper investigation into glycopeptide profiles.

Introduction

Protein glycosylation is the most abundant post-translational modification found in living organisms(1). Aberrant glycosylation, which includes changes in the glycosylation sites of proteins or changes in glycoprotein abundance, has been implicated in various disease conditions (2-5). The accurate quantitative analysis of glycosylation is essential for the evaluation of the role of glycosylation in physiological and pathological processes, as well as for investigation of potential pharmacological targets (6). Advancements in mass spectrometry have significantly improved the quantitative analysis of glycosylation, structural elucidation of glycoforms, and site specification on glycopeptides (7, 8). One method of quantifying glycosylation is based on site-specific glycosylation at the glycopeptide level. Studies of biomarkers and basic mechanistic concepts have demonstrated the potential significance of this level of structural analysis. Similar to peptide quantitation, the most important factors for glycopeptide quantitation are sensitivity, specificity, reproducibility, precision, and throughput (9). While these factors are essential for both protein and glycopeptide analysis, the complexity of glycopeptide structures can introduce unique challenges. These include the diversity of glycan moieties and the requirement for more specialized analytical techniques, which may affect the optimization process differently compared to general protein analysis.

The ability to detect and measure even low-abundance glycopeptides in intricate biological mixtures has been made possible by recent developments in mass spectrometry, which now allow for the measurement of site-specific glycosylation alterations between disease states. MS1-based quantification, the most common approach for glycopeptide quantification, involves quantifying molecules by analyzing ions in the initial stage of mass spectrometry (MS1). In this stage, all ions in the sample are examined to measure their mass-to-charge (m/z) ratios. In MS2-based quantification, after the initial mass spectrometry stage (MS1), selected ions are further isolated and then fragmented. MS3 involves an additional stage of fragmentation and analysis beyond MS2. In MS3, fragments produced in the MS2 stage are isolated and further fragmented to provide even more detailed structural information. Quantification can either be a label-free or label-based approach. Label-free quantification of glycopeptides is based on the peak area of the extracted ion chromatogram of the precursor ion from the MS1 scan. The peak area of the given glycopeptide is quantified and compared in different biological samples (10). The throughput of label-free quantification is low and technical variation could be introduced. On the other hand, the label-based approach utilizes several labeling techniques where the glycan , the peptide backbone (11), or both the glycan and peptide (12) are labeled prior to LC/MS analysis. It is possible to isotope label, mix, and combine glycopeptides from various samples for a single MS run. By comparing the isotope abundances, quantitative findings can be achieved concurrently (10). High throughput quantification of glycopeptides can be achieved with the label-based method.

In the glycomics field isotopic detection of amino sugars with glutamine (IDAWG) was introduced for glycan analysis by Orlando group(13). This strategy specifically targets the amide side chain of glutamine, which is a key nitrogen source for the biosynthesis of GlcNAc, GalNAc, and sialic acid through the hexosamine biosynthetic pathway. It facilitates mixing differentially labeled cells at the start of analysis to reduce sample variation. However, its limitation lies in its applicability solely to cell culture samples, not to biological fluids or tissues, and the complex interpretation of mass spectra due to the uncertain amount of heavy labeled amino sugars (14, 15). Therefore, this technique can be applied in N-glycan and N-glycopeptide analysis since their consensus core glycans contain amino sugars.

One common label-based strategy is the use of isobaric reagents such as tandem mass tags (TMTs) for labeling peptide backbone. TMT labels primary amines in digested peptides by reacting with the NHS ester-based functional group. After TMT labeling of peptides, all samples can be pooled and processed simultaneously, which minimizes technical variation (16). TMT labeling combined with MS2 and MS3 scans yields more glycopeptides compared to MS2 HCD alone. The MS3 approach for quantifying glycopeptides not only makes it easier to identify glycopeptides, but also produces reporter ions for more accurate quantification (17). In addition, the method for measuring glycopeptides via metabolic labeling is also accessible. Glycopeptide quantitation methods involve two main approaches: labeling on the glycan part or on the peptide backbone. Isotope-targeted glycoproteomics (IsoTaG) (18, 19) and isotope-tagged cleavable linker (isoTCL) (19, 20) isotopic labeling of O-GlcNAc modifications through bioorthogonal conjugation with affinity tags, enabling efficient enrichment and mass spectrometry-based quantification. To label the peptide backbone for glycopeptide analysis, the study employed a strategy combining SILAC with PNGase F for quantification at the MS1 level. (20, 21). By synergistically combining ¹⁵N metabolic labeling (IDAWG) for glycan components with Tandem Mass Tag (TMTpro) labeling for the peptide sections, we present a groundbreaking strategy that significantly enhances the field of glycopeptide analysis. Our dual-labeling technique using IDAWG and TMT allows for precise glycopeptide quantification and simultaneous analysis of up to 32 samples, enhancing accuracy and efficiency. This method uses isotope and TMTpro corrections to improve quantification accuracy across different cell lines, setting a new standard in the field. It meets the need for high-throughput, accurate glycopeptide analysis, deepening our understanding of glycoprotein functions in health and disease.

Experimental Section

Chemicals and reagents

Ammonium bicarbonate (ABC), dithiothreitol (DTT), and iodoacetamide (IAA) were purchased from Sigma-Aldrich (St. Louis, MO, USA). HPLC-grade water, acetonitrile (ACN), and LC-MS grade formic acid (FA) were bought from Fisher Scientific (Fair Lawn, NJ, USA). Mass spectrometry grade trypsin/lys-C was acquired from Promega (Madison, WI, USA). Spectrometric grade trifluoroacetic acid (TFA) 99% was obtained from Aldrich Chemical Co., Inc. (Milwaukee, WI, USA). Micro BCA Protein Assay Kit, tandem mass tag (TMT) isobaric reagents, and HEPES buffer were purchased from Thermo Scientific, Rockford, IL. MDA-MB-231BR (231BR) was provided by Dr. Paul Lockman from West Virginia University, School of Pharmacy, Morgantown, WV. Dulbecco's Modified Eagle's Medium (DMEM), fetal bovine serum (FBS), penicillin-streptomycin solution (100X), and dialyzed FBS were procured from the American Type Culture Collection (ATCC). Corning trypsin EDTA 1× was sourced from Fisher, while phosphate-buffered saline (PBS, 1×), cell culturing plates, and flasks were acquired from Corning.

Sample Preparation

Breast cancer cell lines MDA-MB-231BR (231BR) were sub-cultured according to the protocol of our previous publication (12). The ¹⁵N-labeled 231BR was cultured using a specially formulated DMEM medium (heavy medium), incorporating 10 mM of amide-¹⁵N-Gln and 10 mM of asparagine. The heavy medium also contains 10% dialyzed FBS and 2% penicillin–streptomycin. The metabolic labeling procedure adhered to the method presented in our previous paper (12). Briefly, the custom medium was refreshed daily, and after reaching the 4^th generation, the labeled cells were harvested and subjected to cell lysis. For proteomics analysis sample was dissolved in ABC buffer and the protein content in the final sample was measured using the Micro BCATM Protein Assay Kit (Thermo Sc, Rockford, IL) and 500 μg of proteins denatured in a 90°C water bath for 15 min. The denatured proteins were reduced after 45 minutes of incubation at 60°C and 200 mM DTT was added to a sample volume ratio of 1:40. Next, IAA was added with a ratio of 1:10 (v/v) to the sample followed by incubation in darkness at 37°C for 45 minutes to undergo alkylation. Following a 1:40 (v/v) DTT treatment to remove excess IAA, the samples were again incubated at 37°C for 30 minutes. The reduced and alkylated materials should have a pH of close to 8 before tryptic digestion. The samples were incubated at 37°C for eighteen hours to carry out the enzymatic digestion after adding trypsin/lys-C with a protein mass ratio of 1:25 (w/w). After treating in this step, chemically pure, concentrated formic acid was used. The volume of FA added was calculated based on the total volume of the digestion mixture to halt tryptic activity effectively. The final FA content was 0.5%. The mixture was spun in a centrifuge at 14.8 K rpm for 10 minutes. To collect and dry the glycopeptide-containing supernatant, a speed vacuum evaporator was employed.

HILIC enrichment of the tryptic digested cell lines

Enrichment of glycopeptides prior to MS analysis has become a common practice, allowing for the selective capture of glycopeptides by removing most interfering substances like tryptic peptides. Popular enrichment methods include immunoprecipitation(22), hydrazide chemistry(23), lectin affinity chromatography (LAC)(24), and hydrophilic interaction liquid chromatography (HILIC)(25). HILIC is particularly effective due to its ability to bind polar glycopeptides more tightly than tryptic peptides, leveraging the change in glycan polarity(26). HILIC enrichment is used to selectively enrich the low-abundance glycopeptides while removing any non-glycopeptides that would contaminate the samples(26). After the glycopeptide samples were tryptic digested, they were improved with TopTip polyhydroxyethyl A (HILIC) spin columns from Glygen (Columbia, MD, USA). A loading buffer containing 80:20:1 (v/v/v) mixture of ACN, H2O, and TFA was created, and an elution buffer containing 100: 0.1 (v/v) mixture of H₂O and TFA was also made. Following a 100 μL loading buffer conditioning and 100 μL HPLC-grade H₂O washing, the TopTip HILIC spin columns were subjected to a one-minute spin down at 1000× g. Three cycles of washing and conditioning were performed. The samples were then spun on the spin columns at 1000 × g for one minute. Twice more, the flow-through was spun down after being introduced into the column. In the subsequent three washes, the spin columns holding the samples were rinsed with 100 μL of loading buffer. Finally, the glycopeptides were eluted from the samples using 50 μL of elution buffer, and collected in a separate tube, and dried.

Preparation of TMT-labeled glycopeptides

TMT labeling procedure (Thermo Scientific) was performed according to the manufacturer’s instruction. Briefly, glycoproteins were resuspended in 200 mM HEPES at a pH of 8.5, then divided into 16 equal samples of ¹⁴N glycopeptides and 16 of ¹⁵N glycopeptides. TMT reagents, previously dissolved in anhydrous acetonitrile, were then added to each sample at a 1:1 ratio. After an hour, the reaction was halted using 5% hydroxylamine and subsequently acidified with formic acid. Finally, the peptide mixture was desalted using C18 tips.

C18 desalting

Initially, three different buffer solutions were made: one with 100:0.1 (v/v) mixture of ACN and formic acid (FA), another with only water and 0.1% FA, and a third with 60:40:0.1 (v/v/v) mixture of ACN, H₂O and FA. To begin preconditioning, 50 μL of the with 60:40:0.1 (v/v/v) ACN, H₂O and FA mixture was applied over the C18 column three times. Subsequently, the column was loaded three times with 50 μL of a mixture consisting of water and FA solution in a 100: 0.1 (v/v) ratio, using a centrifugal force of 1000 × g for one minute each time. Before adding the sample to the column for centrifugation twice, at 2 minutes each under a 500 × g force, it's important to ensure the sample has minimal organic solvent presence. The column was then washed three times with 50 μL of a mixture consisting of water and FA solution in a 100: 0.1 (v/v) ratio, each time spinning at 1000 × g force for a minute. For elution, after switching the collecting vial, 50 μL of the with 60:40:0.1 (v/v/v) ACN, H2O and FA solution was added three times, followed by two additions of 50 μL of a mixture consisting of ACN and FA solution in a 100: 0.1 (v/v) ratio. The final step involved drying the samples. Completing the desalting process, which is crucial for eliminating salts and other contaminants, prepares the peptides for subsequent analyses like LC-MS.

Instrumentation

The enriched and labeled glycopeptides were resuspended in 2% ACN with 0.1% FA for glycoproteomic analysis. The LC-MS/MS analysis was performed using a Dionex 3000 UltiMate Nano LC system (Thermo Scientific, Sunnyvale, CA, USA) in conjunction with an Orbitrap Fusion Lumos Tribrid Mass Spectrometer (Thermo Scientific, San Jose, CA, USA). Considering several steps and losing samples and peptides during cleaning and enrichment(27, 28), 100 μg of each glycopeptide sample was injected into the LC apparatus for separation. Cleaning and online purification were performed in a C18 Acclaim PepMap trap column (75 μm 20 mm, 3 μm, 100 A°, Thermo Scientific, Sunnyvale, CA, USA) before separation in a C18 Acclaim PepMap RSLC column (75 μm 150 mm, 2 μm, 100 A°, Thermo Scientific, Sunnyvale, CA, USA). A multistage gradient flow separated the glycopeptides for 135 minutes at a rate of 0.3 μL/min. Mobile phase A consisted of a 98:2:0.1 (v/v/v) mixture of H₂O, ACN, and FA, and mobile phase B consisted of a 100:0.1 (v/v) mixture of ACN, and FA. The column oven temperature was set to 60°C. Mobile phase B was kept at 2% for the first 10 minutes of the run, then increased to 30% for the remaining 95 minutes. The rise in mobile phase B was from 30% to 50% in 95-110 minutes, 50% to 90% in 110-113 minutes, and then sustained at 90% for 118 minutes. The mobile phase was then reduced to the starting position by 2%, remaining until the run was completed. Using a nano electrospray ionization source, the separated N-glycopeptides were supplied in positive ion mode at 2.0 kV to the Orbitrap Fusion Lumos Tribrid Mass Spectrometer. The study was carried out in data-dependent acquisition (DDA) mode. The scanning process started with an MS1 spectrum, using the Orbitrap for analysis (resolution of 120k, range 370-1800 m/z). The most prominent precursors were then chosen for MS2/MS3 analysis. MS2 was performed using HCD with orbitrap analysis (NCE of 30, isolation window of 3). After each MS2 spectrum, an MS3 spectrum was gathered using the SPS-MS3 technique. This method involves capturing the ten highest MS2 fragment ions in the MS3 precursor group using isolation waveforms with various frequency notches. MS1 data was collected using 120k resolution, maximum injection time 50 ms, and AGC 1 × 10⁶, MS2 data was collected using maximum injection time 50 ms, and AGC of 5 × 10⁴. Each MS3 precursor group was fragmented by HCD and analyzed using the Orbitrap (NCE of 55, Max AGC of 1 × 10⁶, max injection time of 120 ms, and Orbitrap resolution of 120k).

Data processing

The glycopeptides were identified using Byonic version v4.1.10. The parameters were as follows: 2 max missed cleavages; carbamidomethyl (C) as a fixed modification; oxidation (M) and acetyl (protein N-term) as common modifications; TMT16plex as common modification; in fragmentation part QTOF/HCD fragmentation; fragment mass tolerance 20 ppm; precursor mass tolerance 10 ppm; and the glycan list was based on the 82 glycan list specifically for human breast cancer cells (29). The mass of 304.2071 in the Byonic modification list was used to identify TMT labeled peptides. The mass of 304.2071 corresponds to the monoisotopic mass of the TMTpro 16plex label used for peptide quantification. This mass is utilized by Thermo Fisher's software tools for database searching, specifically tailored for identifying peptides labeled with TMTpro 16plex in mass spectrometry data. The identified labeled glycopeptides were presented in Table S1 as an excel sheet. The masses m/z 205.0836 and m/z 377.1936 are identified as the ¹⁵N-labeled fragment ions of N-acetylglucosamine (GlcNAc) and N-acetylneuraminic acid (NeuAc), respectively. These ions are crucial for confirming the presence of these sugar residues in the glycans attached to peptides, as the ¹⁵N labeling results in a 1 Da increase in the mass of typical fragments, aligning with the observed masses(12, 27, 30). Later, glycopeptides labeled with both TMT and ¹⁵N were analyzed in Xcaliber software manually to obtain reporter ions in MS3 at m/z 126.1277, 127.1247, 127.1310, 128.1281, 128.1344, 129.1314, 129.1377, 130.1348, 130.1411, 131.1381, 131.1445, 132.1415, 132.1478, 133.1448, 133.1512, and 134.1482. The ratios of the reporter ion peak intensities were used to show relative abundance ratios of the corresponding N-glycopeptides. The ¹⁵N glycan labeling efficiency was calculated using the method in our previous paper(12). The deconvolution of overlapped isotopes was achieved by subtracting the theoretical isotope distribution of unlabeled species. The remaining isotopes were from labeled signals. The intensity of the TMT reporter ions was corrected according to the product data sheet (ThermoFisher Scientific). The correction of glycan labeling efficiency and deconvolution of isotope overlapping were performed on Python 3.10 with NumPy and pyOpenMS packages(31, 32).

Results and Discussion

The reaction workflow is shown in Scheme 1, where 16 samples from the 231BR cell line were labeled by the IDAWG method with ¹⁵N medium, and another 16 samples were unlabeled with normal medium. After cell lysis, tryptic digestion, and enrichment, each of the 32 samples were labeled with TMT16 pro, combined as one sample, and analyzed by LC-MS/MS. Thus, there is two-step labeling in this reaction: first, labeling glycans with ¹⁵N-Glu and second, labeling peptides with TMTpro. Each step should be considered when evaluating labeling efficiency.

Scheme 1:

Integrated workflow for LC-MS/MS analysis of 32 IDAWG-labeled samples with subsequent TMTpro tagging.

¹⁵N labeling efficiency of glycans

According to the biosynthetic of IDAWG, glycans GlcNAc, GalNAc, and sialic acid containing molecules are targets for isotopic labeling by supplementation of cell culture medium with amide-¹⁵N-Gln . As shown in Figure 1, glycan structure Hex₄HexNAc₅Fuc₁NeuAc₂ with 6 labeling sites (4 HexNAc and 2 NeuAc) attached to peptide backbone and all 6 sites were identified in the mixed sample. ¹⁴N and ¹⁵N sites were resolved clearly (Figure 1b, c), indicating the labeling efficiency was high; this was also confirmed via no retention time shift of light and heavy glycopeptides mix (Figure 1a). Fragment ions including GlcNAc and NeuAc were detected with 1 Da increase, indicating that most of these monosaccharides were labeled with ¹⁵N (Figure 1c, 1d). To obtain accurate results for glycan labeling efficiency, it is necessary to consider reagent impurity and isotope overlap. In electronic supplementary material Fig S1a, the m/z 878.75 was assigned to ¹⁴N glycopeptide ISNLTLYK HexNAc₂Hex₈ and m/z 879.42 was assigned to ¹⁵N glycopeptide ISNLTLYK HexNAc₂Hex₈, theoretically. However, due to the existence of only two ¹⁵N labeling sites for high mannose structure, the isotopes of the ¹⁴N glycopeptide overlapped with isotopes of the ¹⁵N glycopeptide. The overlap was demonstrated through the fragment patterns of precursor m/z 879.42, showcasing that the fragment ions had both unmodified GlcNAc and GlcNAc labeled with ¹⁵N (supplementary material Figure S1c). Therefore, the correction of labeling efficiency was applied to obtain more consistent and accurate results. We used Python software, as was introduced for the correction of labeling efficiency of ¹⁵N-labeled glycans in our previous work (12). Consistent with our published research, we again achieved high labeling efficiency of glycans here, as shown in supplementary material Figure S1a. In addition to glycan labeling, it is also necessary to evaluate the labeling efficiency of TMT in peptide backbone.

Figure 1:

IDAWG labeling of glycopeptide NTILER HexNAc₄Hex₅Fuc₁NeuAc₂ in mixed light (¹⁴N) and heavy (¹⁵N) + TMT in 1:1 ratio. a) EIC of light (¹⁴N) +TMT in black color, and heavy (¹⁵N) +TMT glycopeptide in red color indicating no retention time shift. b) MS spectra of mixed glycopeptides (isomers). First isotopes for retention time 95.3 min (isomer 1), and c) second isotopes for retention time 93.4 (isomer 2). ¹⁴N arrow pointed light+ TMT monoisotope while ¹⁵N arrow pointed heavy +TMT labeled monoisotope. The corrected experimental ratio was 1:1 between light and heavy glycopeptides. MS/MS fragmentation spectrum of d) ¹⁴N (light) glycopeptide and e) ¹⁵N (heavy) glycopeptide, showing that all residues were labeled with ¹⁵N.

TMT labeling efficiency of glycopeptides

The mass matching strategy was used to identify glycopeptides that were labeled with TMT. The labeled glycopeptides mass found on Byonic was utilized to search for the glycopeptides peak (supplementary material Figure S2). Next, the mass of glycopeptides without TMT was searched to assess labeling efficiency. In labeling evaluation, both the masses of glycopeptides labeled with TMT+ ¹⁵N and TMT+¹⁴N were considered. Figure 2 explains the TMT labeling efficiency. In Figure 2a, a ¹⁴N glycopeptide labeled with TMT at m/z 1134.1602 was clearly identified, while m/z 1032.70 related to light (¹⁴N) glycopeptides without TMT did not find. In Figures 2b and 2c, a mass of ¹⁵N glycopeptides with TMT in ¹⁵N labeled glycopeptides and mix of ¹⁴N+¹⁵N TMT were identified, while in the left side, glycopeptides without TMT were not detected. The labeling efficiency is 100% when there is no unlabeled or partial labeled peak of glycopeptides like the example in Figure 2. In Figure 3, we present the TMT labeling efficiencies of distinct glycopeptide classes. Panel a delineates the mean labeling efficiencies for three glycoform subclasses: high mannose (96%), fuco-sialylated (97%), and complex (81%) each with the average of standard deviation (SD) of ±5%. Several studies have demonstrated the prevalence of high mannose glycopeptides in breast cancer cell lines(33, 34). The abundance of high mannose glycopeptides, coupled with the accessibility of amino groups, could be another key factor influencing the labeling efficiency of TMT. Chen et al. has indicated higher yields of TMT labeling across different glycoforms(35). Thus, considering the aforementioned factors, TMT labeling may be biased towards glycopeptides that are both abundant and have more accessibility for amino groups for labeling. The error bars represent the standard deviation, indicating the precision of our labeling efficiency measurements within each subclass. Figure 3b complements these findings by providing a box plot that visualizes the distribution of labeling efficiencies for the entire set of glycopeptides identified in the experiment, prior to any exclusion based on efficiency thresholds.

Figure 2:

Efficiency evaluation of sialylated-fucosylated glycopeptides labelling with TMT, a) ¹⁴N (light) glycopeptide, b) ¹⁵N (heavy) glycopeptide, and c) mixture of light and heavy glycopeptides. In the left side, the precursor peak of species without TMT was not detected, while on the right side the glycopeptide labeled with TMT is presented, indicating the efficiency of TMT labeling.

Figure 3:

a) Bar graph of TMT labeling efficiency for the main types of glycopeptides: this graph displays the labeling efficiency, calculated as: labeling efficiency = (peak intensity of fully labeled species)/ (peak intensity of partial + fully labeled species) b) Box plot of labeling efficiency of all identified glycopeptides.

Quantitation accuracy of glycopeptides with TMT

The glycopeptide samples obtained from 32 individual 231BR samples (16 ¹⁴N and 16 ¹⁵N cell lines) were labeled with 32 TMT channels and mixed with a ratio of 1:1. The measured ratio between TMT channels in the 32 samples should be approximately 1:1 if the quantitation was correct. In complicated mixture samples like 16 cell lines, similar m/z peptides are also co-isolated when a peptide is extracted in the MS1 spectrum for MS2 analysis (36). After MS2 fragmentation, the reporter ion signal for that isolation will be a mix of reporter ions from the target peptide and all other contaminating peptides. Thus, nearly all measurements are frequently skewed significantly (37). Various approaches have been documented to tackle the problem of ratio compression. These methods encompass a wide range of techniques such as in-depth fractionation (38), using a slim ion-isolation window (39), separation in the gas phase (40), the MultiNotch MS3 technique (17, 37, 41), and adjustments made computationally. Several papers have compared MS2 and MS3 to determine the appropriate method for their TMTpro analysis and to provide a more accurate and efficient approach for quantitative analysis of glycoproteomics in MS3 (42). Thompson and coworkers observed that using TMTpro not only reduces the overall amount of samples for each of the TMT channels, but also improves efficiency in multiplexed experiments (42-44).

In the current study, we utilized a computational technique to obtain accurate quantification analysis. To provide accurate data, the SPS-MS3 technique was utilized, and then an algorithm was developed for normalization and correction of quantitative values. In the MS2 spectrum of glycopeptides, all 16 reporter ions could be identified but with ratio distortion and lower accuracy (45). As shown in supplementary material Figure S3, fragment ions at m/z value of 126.0543 and 127.0513 appeared in both MS2 and MS3-based analysis. These oxonium ions originating from a HexNAc moiety are highlighted in the color red. They interfered with TMT reporter ions; a sufficient resolution should be utilized for adequate resolution of the close isotopes, specifically for TMTpro where the isotopes have only 0.001 Da differences. Here, a resolution of 120k was used to clearly identify all 16 TMT channels. High resolution makes stronger signals from isobaric mass tags, better detection of N-glycopeptide fragments, and significantly reduced interference in multiplex quantification. With isotopic peaks resolved at a 120k resolution, interference with adjacent primary reporter ion peaks is minimized, enabling the most accurate quantitation after correcting for isotopic interference(46-48), and it was recommended by thermosfisher, vendor of both reagent and the instrument. However, in the MS3-based analysis, the intensity of the interference ions was much lower than the reporter ions, which was not observed in the MS2-based analysis. Using MS3, which includes an additional fragmentation step, is crucial for ensuring that quantification remains unaffected by co-isolated precursor ions—a prevalent issue in MS2-based quantification that leads to ratio compression and quantitative inaccuracies. Thus, the MS3- based TMT quantification was analyzed, and reporter ions obtained from SPS-MS3 were used for quantification of glycopeptides.

Therefore, the precursor ion m/z was used to find TMT reporter ions of glycopeptides in MS3 fragments. Figure 4a represents the EICs of ¹⁴N+TMT (black) and ¹⁵N+ TMT (red) glycopeptide LVNISVGRLDLEK HexNAc₂Hex₉, acquired by using their corresponding monoisotopic m/z. It is clear that there was no retention time shift on the mentioned glycopeptide labeled with TMT extracted from the ¹⁴N and ¹⁵N cell lines. Since there was no retention time shift, there was no change in the instruments over time. Figures 4b and 4c show TMT reporter ions obtained from ¹⁴N and ¹⁵N labeled glycopeptides in black and red colors, respectively. The ratio used for quantification shows less than 10% distortion between the reporter ions. Similarly, in supplementary material Figure S4 the glycopeptide TVLENSTSYEEAK HexNAc₂Hex₈ reporter ions in ¹⁴N and ¹⁵N are presented. This glycopeptide only has 2 HexNAcs, which means the mass difference between light and heavy glycopeptide was 2 Da, resulting in isotope overlapping. We deconvoluted the overlapped isotopes based on the method reported before(12) and only utilized the isotope intensity from heavy glycopeptide for quantification.

Figure 4:

Quantitation validation of TMT labeled in mixed ¹⁴N and ¹⁵N in ratio 1:1 glycopeptides (LVNISVGRLDLEK HexNAc₂Hex₉). a) EICs of ¹⁴N glycopeptide labeled with TMT glycopeptide (black), ¹⁵N labeled with TMT glycopeptide (red), indicating no retention time shift. b) SPS-MS3 TMT reporter ions of glycopeptide in ¹⁴N glycopeptide (black), c) ¹⁵N glycopeptide (red).

The correction considered the reporter ion isotope distribution for each of the TMTpro reagents. According to the product data sheet, M-2, M-1, M, M+1 and M+2 may be present in each of the label reagents, so that the intensity of an individual m/z peak could consist of multiple resources (i.e., natural carbon isotopes and incomplete stable isotope incorporation). https://assets.thermofisher.com/TFS-Assets/LSG/certificate/Certificates-of-Analysis/A44520_UI292951.PDF. Thus, the correction calculated the theoretical correction factors for all the reporter ion intensities.

Figure 5 illustrates the comparison of TMT labeling intensities in samples at a 1:1 labeling ratio, prior to and following correction, for both ¹⁴N glycopeptides (Figures 5a and 5c) and 15N glycopeptides (Figures 5b and 5d). Initially, there was an average distortion of 18% across the reporter ions. However, after applying corrections, the distortion reduced to under 5%. The figure presents this data using bars, where the top and bottom parts represent uncorrected and corrected normalized intensities, respectively, for each TMT label in the samples. Error bars indicate the variability in these measurements. The average ratio of experimental data shifted from 0.74 ± 0.13 before correction to 0.95 ± 0.03 post-correction, demonstrating a significant enhancement in the precision of sample quantification.

Figure 5:

TMT reporter ions in all glycopeptides including high mannose, sialylated, and hybrid glycopeptides. Before correction: a) ¹⁴N glycopeptide labeled with TMT, and b) ¹⁵N glycopeptide labeled with TMT. After correction: c) ¹⁴N glycopeptide labeled with TMT, and d) ¹⁵N glycopeptide labeled with TMT.

Conclusion

In this study we presented a novel and efficient approach for the quantitative analysis of glycopeptides derived from cell lines using the combined techniques of ¹⁵N metabolic labeling and TMT multiplexing. The integration of these methodologies presented a significant advancement in the field of glycoproteomics, particularly in addressing the challenges posed by the low abundance and complex heterogeneity of glycopeptides. The use of ¹⁵N metabolic labeling via the IDAWG method for the glycan component and TMTpro labeling for the peptide backbone of glycopeptides allowed for precise quantification and comparison across 32 samples in a single LC-MS run. This dual-labeling approach enabled the accurate analysis of glycopeptides in a high-throughput manner, effectively increasing the capability to study glycoproteins' role in various biological and pathological processes. Moreover, the incorporation of LC-SPS-MS3 technology in our study has substantially diminished fragmentation interference, enhancing the reliability of the data obtained. The TMT labeling's efficiency, demonstrated at 97%, underscores the specificity of the SPS-MS3 analysis when applied to samples with dual labeling, confirming the method's precision in distinguishing and quantifying dual-labeled glycopeptides. The study's implementation of specific corrections for isotope and TMTpro distortion further enhanced the accuracy of the quantitative results in 32 different samples of 231BR cell lines, including both ¹⁴N and ¹⁵N glycopeptides in a 1 to 1 ratio of a pooled sample. Additionally, the signal-to-noise ratio greater than 10, coupled with mass accuracy under 5%, and an average relative standard deviation of 5% attests to the method's robustness and precision. In future development of ¹⁵N metabolic and informatics tools for quantification, this strategy could be applied to several different types of cancer cell lines to provide a broader understanding of glycopeptide profiles across various cell types.

Supplementary Material

SI

Table S1: All identiifed labeled glycopeptides in 231BR cell line.

SI Table

Figure S1: IDAWG labeling of glycopeptide ISNLTLYK HexNAc₂Hex₉ in mixed light (¹⁴N) + TMT and heavy (¹⁵N) + TMT in 1:1 ratio. a) EIC of light (¹⁴N) +TMT in black color, and heavy (¹⁵N) +TMT glycopeptide in red color indicating no retention time shift. MS spectra of mixed glycopeptides. ¹⁴N isotopic peaks (blue) were overlapped with ¹⁵N isotopic peaks (red) because of 2 Da mass diference. b) MS/MS fragmentation spectrum of ¹⁴N (light) glycopeptide and c) ¹⁵N (heavy) glycopeptide, showing that all residues were labeled with ¹⁵N. The corrected experimental ratio was 1:1.02 between light and heavy glycans.

Figure S2: a) Byonic view of high mannose glycopeptide with different fragments including y, b ions. b) MS spectra of precursor ions in high mannose glycopeptide. c) Reporter ions from 126 to 134, indicating how to use software for identification of glycopeptides labeled with TMT.

Figure S3: TMT reporter ions a) SPS-MS3-based and b) MS2-based. The red color peaks indicate interferences which appear in in higher intensity in MS2 analysis, compared to SPS-MS3 analysis.

Figure S4: Quantitation validation of TMT labeled in mixed ¹⁴N and ¹⁵N in ratio 1:1 glycopeptides (TVLENSTSYEEAK HexNAc₂Hex₈). a) EICs of ¹⁴N glycopeptide with TMT glycopeptide (black), ¹⁵N labeled with TMT glycopeptide (red), indicating no retention time shift. b) SPS-MS3 TMT reporter ions of glycopeptide in ¹⁴N glycopeptide (black). c) SPS-MS3 TMT reporter ions of glycopeptide in ¹⁵N glycopeptide (red).