Characterization of Cell Surface Glycoproteins using Enzymatic Treatment and Mass Spectrometry

Abstract

Almost all proteins on the cell surface are glycosylated. Cell surface glycoproteins participate in various cellular pathways, such as cell adhesion, cell-cell communication, and immune response. Due to their functional importance, glycoproteins on cell surface often serve as potential therapeutic targets. Recent advancements in mass spectrometry (MS) have facilitated the characterization of glycoproteins that are generally expressed on cell surface, secreted to extracellular environment, or found in intracellular organelles such as endoplasmic reticulum, Golgi apparatus, and peroxisome. However, the selective characterization of glycoproteins on cell surface remains challenging. In this study, we applied enzymatic treatment to live cells, followed by MS-based glycoproteomic analysis, to assess changes in protein glycosylation at different treatment time points as a method to identify cell surface glycoproteins. To demonstrate this approach, a renal cell carcinoma cell line, A498, was treated with glycosidases, sialidase and PNGase F, over two treatment time intervals, 2-hour and 24-hour. Glycoproteins were identified as cell surface glycoproteins from A498 cells when enzyme treatment altered the glycosylation of the glycoproteins. The results revealed the effectiveness of integrating enzymatic treatment with MS-based glycoproteomics for analyzing cell surface glycoproteins. Our established method has demonstrated the potential applications for assessing accessibility of therapeutic targets on the cell surface over time and supporting the development of new targeted therapies.

Graphical Abstract

INTRODUCTION

Cell surface glycoproteins play critical roles in various cellular pathways and are essential for maintaining cellular homeostasis^1–4. They participate in many key biological processes, including cell-cell interaction, signal transduction, and immune response modulation^2,5,6. Cells modify carbohydrate structure and function to adapt to different environmental cues, such as changes in nutrient availability and fluctuations in pH levels^7–9. Consequently, abnormal protein glycosylation on the cell surface is a hallmark of diseases and can serve as potential biomarkers for diagnostic and therapeutic purposes^10–14. Therefore, it is not surprising that many current therapeutic targets are cell surface glycoproteins^14–18. Despite the importance of cell surface glycoproteins in numerous physiological processes and clinical applications, our understanding of the cell surface glycoproteome remains limited.

Profiling cell surface glycoproteins poses challenges due to the heterogeneity of glycans and low abundance of many glycoproteins on the cell surface^19,20. In recent years, mass spectrometry (MS)-based glycoproteomics has emerged as a powerful tool to identify glycosylation sites and quantify intact glycopeptides in cells^21–27. However, a key limitation of this method is its difficulty in distinguishing cell surface glycoproteins from the ones inside the cell. Conventionally, cell samples are lysed first, and then the intact glycopeptides are enriched for LC-MS/MS analysis²⁸. As a result, the glycoproteins identified from cell lysate contain a mixture of intracellular and surface components, which complicates the analysis of cell surface glycoproteome.

Several methods have been developed and integrated with MS to specifically characterize surface glycoproteins, such as metabolic labeling, catalytic enzymatic tagging, chemical oxidation, and membrane biotinylation^29–33. However, these approaches may not be applicable for observing the accessibility of cell surface glycoproteins. This highlights the need for new methods that permit the characterization of cell surface glycoproteins under physiological conditions. Such advancements would greatly contribute to the development of cell surface glycoprotein-based therapeutic strategies^15,17,34.

To meet the needs, we developed an integrative method for analyzing cell surface glycoproteins that integrates live-cell enzymatic treatments of glycans with MS-based glycoproteomics. Glycosidases, enzymes that hydrolyze glycosidic bonds in oligosaccharides and glycoconjugates, were used to modify the glycosylation patterns on the external side of the cell. The treatment predominantly affects cell surface glycoproteins while preserving cell integrity in a mild reaction condition. To explore the heterogeneous accessibility of glycans under live-cell conditions, we introduced sialidase and PNGase F to target different glycoprotein compositions. Sialidase cleaves sialic acids at the terminal positions of complex and hybrid glycans, while PNGase F removes the entire N-glycans from glycoproteins^35,36. In addition, cells were treated with the glycosidases for 2 hours and 24 hours to observe the time-dependent response of individual glycoproteins to each treatment. The results showed that both sialidase and PNGase F were able to modify glycoproteins involved in critical processes on the cell surface, with some unique glycoproteins found by each enzyme conditions. Overall, an increase in treatment time led to the identification of more glycoproteins.

This study demonstrated the efficacy of using enzymatic treatment combined with MS to characterize cell surface glycoproteins. The established approach can be applied to analyze accessible glycoproteins on the cell surface and observe changes in surface protein glycosylation over time, which has the potential to contribute to the development of precision drug therapies.

EXPERIMENTAL SECTION

Cell Culture.

A498 cells (ATCC) were cultured using minimal essential medium (MEM) (Gibco), supplemented with 10% heat inactivated fetal bovine serum (FBS) (Gibco) and 1% penicillin-streptomycin (Pen-Strep) (Gibco). All cells were maintained in a humidified incubator with 5% CO² at 37°C. Two days prior to glycosidase treatment, A498 cells were plated at a density of 2.5 to 3.0 million cells per dish. One day before treatment, the serum-containing MEM was removed, followed by two additional PBS (Gibco) washes, and replaced with serum-free MEM supplemented with 1% Pen-Strep.

Removal of Cell Surface N-glycans and Sialic Acids with in vivo Glycosidase Treatment.

To assess whether changes in cell surface glycosylation occur with glycosidase treatments, A498 cells were incubated with glycosidases in serum-free MEM (pH 7.5) at 37°C and collected after 2 hours and 24 hours, respectively. Untreated groups were also collected at the 0-hour, 2-hour, and 24-hour time points. The amount of glycosidase added to each dish was calculated based on the enzyme-to-total protein mass ratio, which was similar to the vendor’s protocol, to account for native cell conditions.

In brief, BCA protein assay was first performed to approximate the total protein mass of 2.5 – 3.0 million A498 cells as 300ug. To cleave surface sialic acids, 300 μL of α2–3, 6, 8, 9 neuraminidase A (20,000 U/mL New England BioLabs) from Arthrobacter ureafaciens was directly added to the cell culture media. The neuraminidase was added at a ratio of 1 μL enzyme per 1 μg of A498 cell protein. To remove surface N-glycans, 80 μL of glycerol-free PNGase F (500,000 U/mL, New England BioLabs) from Flavobacterium meningosepticum were directly added to the culture media based on the ratios of 1 μL PNGase F per 3.75 μg cell protein. To terminate the glycosidase treatment, cell culture media containing the glycosidases were removed, and the cells were washed twice with PBS. All cells were stored frozen at −80°C immediately after harvest until testing.

Cell Sample Preparation.

Frozen cell samples were lysed at 4°C, followed by overnight trypsin digestion and desalting using Sep-Pak C18 Cartridges (Waters). To quantitively analyze the glycopeptides, the samples were adjusted to the same peptide concentration and labeled with 10-plex Tandem Mass Tag reagents (Thermo Fisher Scientific). The TMT-labeled samples were combined and separated into 24 fractions using basic reversed-phase liquid chromatography (bRPLC), and 5% of each fraction were collected as global samples. To capture the intact glycopeptides, the remaining 95% of the fractions were combined into 6 fractions and enriched using Oasis MAX solid-phase extraction (Waters). Both the global and intact glycopeptide-enriched samples were subsequently analyzed by LC-MS/MS. Further details regarding sample preparation can be found in the procedure described in our previously published protocol²⁸.

LC-MS/MS Analysis.

Intact glycopeptides were analyzed by Orbitrap Ascend Tribrid Mass Spectrometer (Thermo Scientific). Cell samples were reconstituted with 0.1% formic acid and loaded onto Evotips (Evosep). Loaded peptides were run using 30 SPD (44min) (Evosep). The ion source of the MS was set up with 1.9 kV for position ion and 600 V for negative ion using 300°C of ion transfer tube temperature. To capture the precursor ions, the Orbitrap master scan was configured with a resolution of 60K, a scan range of 350 to 2000 m/z, and a radio frequency lens setting of 60%. The peptide charge state screening was enabled to include 2 to 8 ions with a dynamic exclusion time of 45s to discriminate against previously analyzed ions between +/− 10 ppm. The intensity threshold was set to be 1.0E4. Data-dependent acquisition was employed, using higher-energy collisional dissociation for fragmentation with collision energies set at 25%, 35%, and 45%. The isolation window was set to 0.7 m/z, and the resulting fragment ions were analyzed with the Orbitrap at a resolution of 30K to provide high-resolution MS/MS spectra, with a maximum injection time of 64 ms and Automatic Gain Control target of 3E5. The global samples were run using similar settings, with the following differences: the samples were analyzed using the Extended Method 15 SPD (88 min) (Evosep). The Orbitrap Master Scan was configured with a resolution of 120K and a scan range of 400 to 1400 m/z. The peptide charge state screening was enabled to include 2 to 6 charge states. The intensity threshold was set to be 2.5E4, HCD for fragmentation was set at a collision energy of 35%, and the maximum injection time was 59 ms.

Identification and Quantification of Intact Glycopeptides (MAX-enriched) and Global Proteomics.

The raw files of the MAX-enriched TMT-labeled cell samples were searched using GPQuest with MS-PyCloud (v2.10.1) against the UniProt/Swiss-Prot human protein database (released in May 2018) with the following constraints: tryptic peptides with a maximum of two missed cleavages length were allowed; carbamidomethylation (C, +57.021 Da) and TMT of lysine and N-terminus (+229.162932 Da) were set as static modifications; oxidation (M, +15.995 Da) was selected as dynamic modifications^37,38. The raw files of the TMT-labeled global cell samples were searched using MS-GF+ with similar search constraints except that de-amidation (N, +0.984 Da) was additionally selected as dynamic modifications³⁹. The search results of the enriched global cell samples were evaluated at a peptide-spectrum match (PSM)-level false discovery rate of less than 1%, with a minimum of one PSM per peptide and one peptide per protein. Each TMT channel was normalized against the reference channel to quantitate the abundance of intact glycopeptide and de-glycopeptides (de-amidated peptides containing the glycosylation motif). Intact glycopeptides and de-glycopeptides were excluded from subsequent data analyses if any of the corresponding channels were missing an abundance value.

The MAX-enriched and global data were analyzed, which includes the calculations of fold change (FC) ratio, false discovery rate (FDR) for sialidase and PNGase F treatments, as well as the enrichment of protein molecular functions and biological processes.

Selection of Candidate Intact Glycopeptides and De-glycopeptides Derived from Cell Surface Glycoproteins.

To determine the selection parameter for sialidase-treated samples, intact glycopeptides with identical sequences but varying numbers of sialic acids were first isolated. These sequences were then divided into two groups based on FC: those showing an increase and those showing a decrease. Intact glycopeptides with FC > 1 were ranked in descending order from the highest FC. Given that sialidase reduces glycoprotein sialylation, we anticipated an increased FC in non-sialylated glycopeptides affected by the enzymatic treatment. Consequently, sialylated glycopeptides in this group were designated as false discoveries. Conversely, glycopeptides with FC < 1 were ranked in ascending order from the lowest FC. For glycopeptides impacted by sialidase, we expected to see a decreased FC in sialylated forms. Thus, non-sialylated glycopeptides in this group were considered false discoveries when calculating the FDR.

The selection parameter of PNGase F-treated samples were calculated using a similar approach. We first isolated de-glycopeptides from global proteomics data with matching MAX-enriched intact glycopeptides. Those peptides with FC > 1 were then ranked in descending order, and we determined the threshold parameters by calculating the FDR. For glycopeptides impacted by PNGase F treatment, we anticipated an increased FC in their de-glycosylated forms. Therefore, peptides that did not undergo de-glycosylation in this category were considered false discoveries in the FDR calculation.

In this study, the FC thresholds were determined using an FDR of 0.05 to select candidate intact glycopeptides and de-glycopeptides derived from cell surface glycoproteins, The calculation of FDR for intact glycopeptide and de-glycopeptide sequences within a specific FC range is estimated as:

$$FDR=\frac{Numberoffalsediscoveries}{Totalnumberofsequences}$$

Enrichment Parameters.

To determine the molecular functions and biological processes of the glycoproteins identified by the sialidase and PNGase F treatments, their corresponding genes were analyzed using Metascape (https://metascape.org) with the following settings: minimum overlap = 3, p-value cutoff = 0.01, and minimum enrichment score = 1.5⁴⁰.

RESULTS

Analysis of Cell Surface Glycoproteins Using Enzymatic Treatments and MS-Based Glycoproteomics.

To selectively analyze the cell surface glycoproteins, it is crucial to employ a method capable of distinguishing cell surface glycoproteins from secreted or intracellular ones before cell lysis. To achieve this, we treated cells with glycosidases to selectively modify their surface glycosylation under live conditions. This approach distinguishes cell surface glycoproteins from those residing intracellularly, as the plasma membrane restricts these glycosidases from interacting with intracellular components. Since the cell viability could be maintained after enzymatic treatment, this method enables us to observe changes in cell surface glycosylation over different treatment time intervals.

The workflow of the experiment is illustrated in Figure 1. In brief, first, cells were first treated with enzymes (PNGase F or sialidase) in serum-free media to alter cell surface glycans. Second, treated cells and untreated cells were collected at 0-hour, 2-hour, and 24-hour time points after the treatments. Third, the cells were lysed, and cell proteins were digested peptides and labeled with TMT reagents. Fourth, the intact glycopeptides were further enriched by MAX. Fifth, the global and MAX-enriched samples underwent LC-MS/MS analysis for global proteomics for identification of de-glycosylated glycopeptides and glycoproteomics for identification of intact glycopeptides, respectively. Lastly, the effects of enzymatic treatments were evaluated by examining changes in the glycosylation levels of glycoproteins. The effect of sialidase treatment was examined by the de-sialylation of intact glycopeptides and PNGase F treatment was assessed by analyzing both intact glycopeptides and de-glycopeptides.

Figure 1.

Workflow of cell surface glycoprotein identification using enzymatic treatment and mass spectrometry. Created with BioRender.com

The use of two different glycosidase could address the heterogeneous accessibility of glycans. PNGase F removes the entire N-glycan and introduces de-amination at the asparagine of the N-X-(S/T) motif, where X can be any amino acid except proline. This process generates peptides containing de-glycosylation sites that can be identified from the global proteomic data. On the other hand, sialidase cleaves the sialic acids at the terminal position of glycans and leaves the remaining structure still attached to the peptides. Importantly, this dual approach allows for the specific identification of cell surface glycoproteins based on selective enzymatic modifications applied under live conditions. While PNGase F cleaves N-glycans at their core, its activity can be impeded by certain bulky or highly branched glycan structures, which may sterically hinder the enzyme’s access to the innermost N-acetylglucosamine and asparagine residues at specific glycosylation sites. By using sialidase treatment, we targeted sialylated glycoproteins that PNGase F could not access. However, while sialidase can alter the glycosylation of sialylated glycoproteins, it cannot modify the glycosylation pattern of high-mannose N-glycans, as these do not contain sialic acids. Through PNGase F treatment, we were able to remove those glycans that sialidase could not act on. Therefore, the use of different enzymes provides more comprehensive characterization of cell surface glycoproteins based on glycosidase accessibility.

Sialidase-Induced Sialylation Changes in Intact Glycopeptides Over Time as an Indicative Measure for the Identification of Cell Surface Glycoproteins.

Sialic acids are a family of nine-carbon carbohydrates often found at the terminal positions of glycans. For N-linked glycosylation, the sialylation of proteins primarily occurs in the Golgi apparatus, where sialyltransferases add sialic acids to galactose residues of N-glycans. Sialic acids can be removed through enzymatic hydrolysis by sialidase (neuraminidase)⁴¹. In this study, a broad-specificity α2–3,6,8,9 neuraminidase was used to cleave sialic acids to identify cell surface proteins according to the observed changes in sialylation in response to sialidase treatment over time.

Although high mannose intact glycopeptides were detected from our MS-based glycoproteomic data, they were excluded from the downstream analysis because of lack of sialylation. Only intact glycopeptides with complex and hybrid glycoforms were used for subsequent data analysis. To identify intact glycopeptides affected by the treatment, we ranked intact glycopeptide expression between treated and untreated groups at 2-hour and 24-hour time points based on their FC ratios, where FC ratio > 1 should indicate an increase of the expression of non-sialylated glycopeptides in sialidase-treated cells compared to the untreated ones. Indeed, most glycopeptides with an FC greater than one were non-sialylated, while most glycopeptides with an FC less than one were sialylated in the treated group (Figure 2A and Figure S1). These results indicated that sialidase treatment successfully cleaved the sialic acids at the terminal of N-glycans and altered glycoprotein sialylation.

Figure 2.

Characterization of cell surface glycoproteins by sialidase treatment. (A) FC analysis of intact glycopeptide levels at the 2-hour and 24-hour time points, ranked in descending order. (B) Changes in the level of individual intact glycopeptides with matching glycosylation sequence, differing only in the level of sialylation, following sialidase treatment at different time points. (C) Comparison of identified intact glycopeptides between samples treated with sialidase at the 2-hour and 24-hour time points. (D) Comparison of identified glycoproteins between samples treated with sialidase at the 2-hour and 24-hour time points. (E) Molecular functions enriched in glycoproteins identified by sialidase treatment.

To examine how the treatment changes individual glycopeptide sialylation over time, intact glycopeptides with identical peptide sequence, differing only in the number of sialic acids, were grouped together to compare their expressions at the 0-hour, 2-hour, and 24-hour time points. In the sialidase-treated group, intact glycopeptides exhibited increased levels of their non-sialylated forms and decreased levels of their sialylated forms at 2-hour and 24-hour after enzymatic treatment compared to the 0-hour -time point (Figure 2B). This pattern is exemplified by a glycopeptide, LNPTVTYGNDSFSAK-N6H7F1S(0–4), originated from intercellular adhesion molecule 1 (ICAM1), a glycosylated adhesion receptor on the cell surface⁴². The glycopeptide presents in both sialylated and non-sialylated (S0) forms at the 0-hour time point, with the sialylated forms containing up to four sialic acids (S1-S4). All sialylated forms of the glycopeptide displayed reduced expression levels in the sialidase-treated group, as shown at both the 2-hour and 24-hour time points. These de-sialylated glycopeptides contributed to an increased expression of the non-sialylated form at the same time points. Similar trends were observed for other intact glycopeptides (Figure 2B and Table S1).

Furthermore, we considered intact glycopeptides with FC ≥ 1.06824 or ≤ 0.96673 (FDR ≤ 0.05) for the 2-hour time point and intact glycopeptides with FC ≥ 1.01118 or ≤ 0.96886 (FDR ≤ 0.05) for the 24-hour time point as candidate intact glycopeptides derived from cell surface glycoproteins (Figure S2A). Using these parameters, we selected 189 intact glycopeptides from the 2-hour time point and 250 from the 24-hour time point, with 175 overlapping between the two time points, resulting in a total of 264 unique glycopeptides identified (Figure 2C). Based on these altered intact glycopeptides, we identified 36 glycoproteins and 44 glycoproteins at the 2-hour and 24-hour time points, respectively. Notably, all glycoproteins identified at the 2-hour time point were also found at the 24-hour time point (Figure 2D).

The molecular functions of the 44 glycoproteins were examined using Metascape⁴⁰. We found that many glycoproteins were related to extracellular matrix binding (p-value = 4.64E-10). Additionally, cell adhesion molecule binding, exogenous protein binding, and collagen binding were also highly enriched terms (Figure 2E). By further investigating the biological processes of these glycoproteins, we observed that cell-cell adhesion was the most enriched. We also found glycoproteins involved in the processes of cell adhesion mediated by integrin, symbiont entry into host cells, and cell junction organization (Figure S2B). The enriched molecular functions and biological processes suggested that the selected glycoproteins mainly participate in activities occurring on the cell surface. This underscored the effectiveness of sialidase treatment in identifying functionally important cell surface glycoproteins.

PNGase F-Induced Changes in Intact Glycopeptide and De-glycopeptide Expression Over Time for the Identification of Cell Surface Glycoproteins.

N-glycans are carbohydrate structures attached to proteins at asparagine residues. These glycans share a common core structure consisting of two N-acetylglucosamine (GlcNAc) and three mannose residues. De-glycosylation, the removal of N-glycans, can be performed by PNGase. This process involves the cleavage of the innermost GlcNAc and the asparagine residue, resulting in the de-amidation of asparagine to aspartic acid⁴³. In this study, PNGase F was used to cleave the N-glycans from glycoproteins to identify cell surface proteins according to the observed changes in protein glycosylation and de-glycosylation in response to PNGase F treatment over time.

The MAX-enriched data were used for intact glycopeptide analysis, while the de-glycosylated peptides were identified and quantified from global proteomic data. To observe the peptides affected by the PNGase F treatment, we ranked the expression of intact glycopeptides and de-glycopeptides between treated and untreated groups at the 2-hour and 24-hour time points based on their FC ratios. Since PNGase F facilitates de-glycosylation, it would increase the expression of de-glycopeptides and decrease that of intact glycopeptides with the same peptide sequence. Indeed, several intact glycopeptides with decreased FC had corresponding de-glycosylated forms with increased FC in the treated group (Figure 3A). This result indicated that PNGase F treatment successfully removed N-glycans from these glycoproteins and modified protein glycosylation.

Figure 3.

Characterization of cell surface glycoproteins by PNGase F treatment. (A) De-glycopeptide sequences with FC > 1 and intact glycopeptide sequences with FC < 1 at the 2-hour and 24-hour time points, ranked in descending order. (B) Changes in the level of individual peptides with matching de-glycosylated and glycosylated forms following PNGase F treatment at different time points. (C) Comparison of identified de-glycopeptides between samples treated with PNGase F at the 2-hour and 24-hour time points, (D) Comparison of identified glycoproteins between samples treated with PNGase F at the 2-hour and 24-hour time points (E) Molecular functions enriched in glycoproteins identified by PNGase F treatment.

To observe how the treatment changes individual peptide glycosylation over time, individual intact glycopeptides and de-glycopeptides with the same peptide sequences were grouped together to compare their expressions at the 2-hour and 24-hour time points. In the PNGase F-treated group, several intact glycopeptides had decreased expression levels, while their de-glycosylated forms revealed increased expression compared to the untreated group (Figure 3B). This pattern is exemplified by a peptide, TASN*LTVSVLEAEGVFEK, originated from the CD109 gene, which codes for a GPI-linked glycoprotein located on the cell surface⁴⁴. The peptide had a de-glycosylated form and 7 different glycosylated forms. Notably, most of the intact glycopeptides displayed reduced expression levels at both the 2-hour and 24-hour time points, contributing to an increased level of the de-glycopeptides at the same time points. Similar trends were also observed for additional peptides with matching glycosylated and de-glycosylated sequences (Figure 3B and Table S2).

Furthermore, we considered de-glycopeptides with FC ≥ 1.16416 (FDR ≤ 0.05) for the 2-hour time point and de-glycopeptides with FC ≥ 1.02382 (FDR ≤ 0.05) for the 24-hour time point as candidate de-glycopeptides derived from cell surface glycoproteins (Figure S3A). Using these parameters, we selected 32 de-glycopeptides from the 2-hour time point and 110 from the 24-hour time point, with 31 overlapping ones between the two time points, resulting in a total of 111 unique de-glycopeptides identified (Figure 3C). Based on these de-glycopeptides, we identified 26 and 72 glycoproteins at the 2-hour and 24-hour time points, respectively. Interestingly, all except one glycoprotein identified at the 2-hour time point were also found at the 24-hour time point (Figure 3D).

The molecular functions of the 44 glycoproteins were examined using Metascape⁴⁰. We found that many proteins were related to collagen binding (p-value = 5.05E-12). Additionally, exogenous protein binding, fibronectin binding, and cell adhesion molecule binding were also highly enriched terms (Figure 3E). By further investigating the biological processes of these glycoproteins, we observed that symbiont entry into host cell was the most enriched term. We also found proteins involved in collagen fibril organization, cell-cell adhesion, and extracellular matrix organization (Figure S3B). The enriched molecular functions and biological processes suggested that the selected glycoproteins mainly participate in activities occurring on the cell surface. This highlighted the effectiveness of PNGase F treatment in identifying functionally important cell surface glycoproteins.

Evaluation of the Cell Surface Glycoproteins by the Two Enzyme Treatments, Their Potential to Characterize Cell Surface Therapeutic Targets, and Method Repeatability

In this study, we successfully identified a total of 98 unique glycoproteins using sialidase or PNGase F treatments across two distinct time points. Of these, 19 proteins were identified by both sialidase and PNGase F treatments, indicating their ubiquitous accessibility regardless of the type of enzymatic treatment. Additionally, 25 cell surface glycoproteins were only identified with sialidase treatment, while 54 cell surface glycoproteins were only identified with PNGase F. Overall, a greater number of glycoproteins were identified by the PNGase F treatment (Figure 4A). These findings underscored the effectiveness of the two enzyme treatments in addressing heterogeneity of protein glycosylation and in characterizing glycosidase-accessible cell surface glycoproteins.

Figure 4.

Assessment of cell surface glycoprotein identification and potential therapeutic targets. (A) Number of cell surface glycoproteins only identified using sialidase, PNGase, or both enzymes. (B) Heatmap of the therapeutic relevance of identified glycoproteins, clustered by treatment conditions.

A498 cells are a human kidney cancer cell line commonly used in cancer research to study renal cell carcinoma (RCC) and test new therapies^45,46. To assess whether the identified glycoproteins are potential therapeutic targets for RCC, we categorized the glycoproteins based on annotations from The Human Protein Atlas (proteinatlas.org)⁴⁷. The results showed that several identified glycoproteins are either FDA-approved drug targets or have the potential to become such targets (Figure 4B). For example, lysosome-associated membrane protein 2 (LAMP2) was identified by both sialidase and PNGase F treatments (Figure 4B). It has been reported that LAMP2 translocates from the lysosomal membrane to the plasma membrane in several types of cancer cells, though the current mechanism is unclear⁴⁸. Notably, LAMP2 is associated with the development of resistance to sunitinib, a tyrosine kinase inhibitor used as a primary treatment for metastatic RCC⁴⁹. Transferrin receptor (TFRC) was identified only by the sialidase treatment at both timepoints (Figure 4B). TFRC is involved in iron uptake by cells, and its overexpression is often seen in rapidly proliferating cancer cells⁵⁰. Importantly, TFRC supports the progression of RCC and was identified as a RCC biomarker and therapeutic target⁵¹. Last but not least, epidermal growth factor receptor (EGFR) was identified only by the PNGase F treatment (Figure 4B) EGFR is a cell surface receptor tyrosine kinase that, when bound to its ligands, initiates a signaling cascade leading to cell proliferation⁵². Studies have revealed that N-glycosylation affects the structure and function of the EGFR ligand-binding domain⁵³. Significantly, EGFR is an FDA-approved drug target for various cancers⁵⁴. Ongoing research studies have suggested that it could potentially be targeted in RCC in the future^55–57.

The repeatability of the method was evaluated through a repeat experiment conducted under identical glycosidase treatment conditions. The biological replicates of the 0-hour controls from this experiment exhibited a Spearman’s rho of 0.99 (p-values < 1E-50) for both intact glycopeptide and global peptide expressions. Additionally, the pooled samples from the two experiments showed strong Spearman’s rho of 0.91 and 0.93 (p-values < 1E-50) for intact glycopeptides and global peptides, respectively. These results revealed high data correlation both within an individual set and across independent experiments. To assess the repeatability while ensuring consistent comparison of cell surface glycoprotein identification across the two MS results, we used the matching intact glycopeptide and de-glycopeptide sequences from the two experiments to select cell surface glycoproteins in the repeat experiment. In the repeat experiment, we successfully identified 70 out of 98 glycoproteins initially detected in experiment 1 (Figure S4E and Table S4). Overall, these consistent findings supported the repeatability of our method across two experiments.

DISCUSSION

Nearly all cell surface proteins are glycosylated, with many playing essential roles in cellular processes such as cell-cell communication, molecular signaling, and immunoregulation. In addition to their significance in maintaining normal physiological functions, these proteins also represent valuable pathological targets for therapeutic intervention in various diseases. Despite their importance, characterization of cell surface glycoproteins has been difficult due to challenges in distinguishing surface glycoproteins from secreted or intracellular ones. To overcome this limitation, we employed enzymatic treatment combined with MS to specifically analyze changes in cell surface glycoproteome at different enzymatic treatments and time intervals. By doing so, we identified glycoproteins on the surface of A498 cells and evaluated their potential as therapeutic targets using existing databases.

The results showed that sialidase and PNGase F treatments successfully identified cell surface glycoproteins at both the 2-hour and 24-hour time points. Examples of glycoproteins involving in cell surface interactions include cadherins (CDH2 and CDH6), integrins (ITGA1, ITGA2, ITGA3, ITGAV, and ITGB1), and members of the cell adhesion molecule family (ALCAM, BCAM, CADM1, ICAM1, and NCAM1). Unique glycoproteins were identified across these four treatment conditions, with some glycoproteins overlapping and others unique to specific enzymes. This underscores our methods’ ability to address the heterogeneous nature of cell surface protein glycosylation for identification purposes. Notably, several identified glycoproteins are FDA-approved cancer drug targets or have potential therapeutic relevance in the context of A498 cells, a cell line for kidney cancer, such as LAMP2, TFRC, and EGFR. Our findings also reveal method repeatability and the potential for this methodology to be applied to other enzymatic treatments or cell types.

Although other approaches, such as differential centrifugation, chemical or biotin labeling of the plasma membrane, are also able to identify cell surface glycoproteins, they can introduce non-specific proteins through protein complexes or toxicity to cells. As a result, these methods are less practical for observing the differential accessibility of cell surface glycoproteins over time. In contrast, enzymatic treatments via glycosidases used in this study not only selectively modify glycosylation levels on the cell surface but also maintains cell viability over the course of the treatment. This is important as it can potentially contribute to the study of glycan remodeling, trafficking, and turnover of glycoproteins on the cell surface. For example, research has shown that LAMP2 translocates from the lysosomal membrane to the plasma membrane in cancer, and its overexpression causes drug resistance in metastatic RCC^48,49. By applying enzymatic treatment, we can potentially observe changes in LAMP2 glycosylation to determine the rate at which glycosylated LAMP2 translocates to the plasma membrane in RCC cells. Understanding the turnover rate of LAMP2 on the cell surface may provide valuable insights into its role in drug resistance mechanisms, potentially guiding the development of new strategies targeting LAMP2-mediated autophagy to overcome drug resistance in the future⁵⁸. Similarly, this method could be applied to functional studies of other therapeutic targets for different disease cell lines.

It is important to note that, while cell surface glycoproteins can be identified using the current approach, the enzymatic treatment conditions need to be optimized. Further efforts to optimize enzyme reaction time could improve enzyme efficiency and lead to the identification of more glycoproteins with greater accuracy. Additionally, exploring the use of other glycosyltransferases or glycosidases or combinations of enzymes could also help identify a broader range of glycoproteins on the cell surface. Functional studies to elucidate the specific roles of each identified glycoprotein are beyond the scope of this study but represent a potential direction for future research. Nonetheless, the results of our study demonstrate the effectiveness of the concept of enzymatic treatment-based identification of cell surface glycoproteins and its potential application to therapeutic target studies.

CONCLUSION

In this study, we demonstrated a method for selectively profiling cell surface glycoproteins using enzymatic treatments combined with MS analysis. This approach overcomes the challenge of distinguishing cell surface glycoproteins from intracellular ones, as the mild enzymatic treatments specifically target external glycoproteins while maintaining cell integrity.

The use of different glycosidases, sialidase and PNGase F, addresses the heterogeneity of cell surface glycan structures in live conditions. By treating the cells with these two enzymes, we identified cell surface glycoproteins accessible at different treatment time points and specific to each enzyme. This approach led to the identification of unique glycoproteins, some common across different conditions, while others were unique to specific treatments. Notable examples of identified glycoproteins include integrins, cadherins, and cell adhesion molecules located on the cell surface. Additionally, several potential and FDA-approved drug targets on the cell surface were also identified. Lastly, our method demonstrated repeatability across two experiment results.

Our findings underscore the effectiveness of enzymatic degradation in identifying cell surface glycoproteins. This method offers a non-invasive alternative to other techniques that often compromise cell health during treatment. Therefore, it has potential applications in analyzing changes in cell surface protein glycosylation over time and providing insights into glycoprotein turnover rates. Overall, our method successfully identified cell surface glycoproteins on A498 cells. With further optimization, it holds promise for characterizing the surface glycoproteome of other cell types and contributing to therapeutic target studies in the future.