Introduction

Phosphorylation and glycosylation are two of the most ubiquitous and important protein posttranslational modifications (PTMs) that can significantly alter protein tertiary structure, stability, turnover, and protein-protein interactions [1]. It has been estimated that nearly one third of human proteins are phosphorylated and more than one half of mammalian proteins are glycosylated [2, 3]. They are highly associated with many biological processes, including signal transduction, enzyme activity regulation, protein folding, and cell adhesion [4, 5], and related to various human diseases, including diabetes, Alzheimer’s disease, autoimmunity, congenital disorders of glycosylation, and cancer [6,7,8,9,10,11,12,13,14]. It also has been demonstrated that there is complex crosstalk between the two PTMs in biological regulation [15, 16]. In order to study the PTM-related biological processes and further dissect the molecular mechanisms underlying the diseases involving these two modifications, the ability to simultaneously analyze both phosphorylation and glycosylation at the proteome level is essential.

Bottom-up proteomics (i.e., shotgun proteomics), which couples liquid chromatography (LC) and tandem mass spectrometry to analyze proteolytic peptides, has become a main-stream technology that enables rapid profiling of proteins in complex samples [17,18,19]. The technology has undergone rapid development in the past few decades, and its high-throughput allows the whole proteome from eukaryotic cells to be analyzed within just a few hours [20,21,22]. To better adjust the method for analyzing phosphate or glycan modified peptides, more sophisticated LC-MS experiments have been designed such as the “multistage activation (MSA)” method [23], the electron-driven dissociation [24, 25], and the hybrid fragmentation strategies such as electron-transfer/higher-energy collision dissociation (EThcD) and activated ion electron-transfer dissociation (AI-ETD) that combine orthogonal types of activation on the modified precursor ions [26,27,28,29,30,31,32,33]. However, neither of the above strategies can avoid the increased duty cycle of MS/MS analysis, which will compromise the number of MS/MS events and reduce the large-scale profiling efficiency and overall coverage of modified peptides. Furthermore, MSn and electron-driven fragmentation capabilities are currently only available on a limited number of instruments. On the other hand, stepped-energy of higher-energy collision dissociation (stepped HCD or stepped NCE) approach, which combines the fragment ions from three different collision energies and records these fragments in a single MS/MS spectrum, was found to increase the diversity of the fragment ions without significantly increasing the MS analysis duty cycle or sacrificing the throughput, which benefits large-scale PTM analysis. For instance, improved phospho site localization through increased sequence coverage on HEK 293T cell lysate was reported by applying a stepped HCD approach [34]. A recent application of stepped NCE to the human serum proteome also proved its capability to fully characterize intact glycopeptides on a large scale, while simultaneously yielding structural information of the peptide backbones and the covalently linked glycans via high and low collision energies [35].

Despite the technical improvements to MS instruments, it is still challenging to analyze phosphorylated or glycosylated peptides by direct LC-MS without reducing sample complexity, due to the sub-stoichiometric level abundance of the two PTMs and the poor ionization efficiency in positive ion mode MS derived from additional negative charges of these PTMs. Therefore, numerous enrichment techniques have been developed targeting phosphorylated or glycosylated peptides, including metal oxide affinity chromatography (MOAC) such as TiO2 and ZrO2 enrichment [36, 37], immobilized metal affinity chromatography (IMAC) [38,39,40,41,42], lectin affinity chromatography-based enrichment [43, 44], hydrazide chemistry-based SPE method [45,46,47,48], boronic acid enrichment method [49,50,51], and hydrophilic interaction chromatography (HILIC) [52,53,54,55]. Most of the methods listed above were designed to target phosphorylation or glycosylation alone and each has its own limitations. MOAC and IMAC suffer from nonspecific binding of aspartic acid and glutamic acid-rich peptides. Lectin affinity chromatography exhibits biases towards specific glycan structures. Hydrazide chemistry-based approaches can enrich N-glycopeptides with high specificity and efficiency, but the release of enriched analytes relies on enzymatic digestion that makes intact glycopeptide analysis impossible. In recent years, novel MOAC and IMAC materials were synthesized with optimized morphology, ligand selection, and density of functional groups to increase the selectivity and enrichment efficiency of both phosphorylated and glycosylated peptides [56,57,58]. However, most of the applications of those novel enrichment materials were limited to MALDI-MS validation on mixtures of several phosphorylated and glycosylated peptide standards, while the in-depth analysis of the modified proteome was limited by the availability of the custom synthesized material to biological labs.

Ion exchange chromatography (IEX) such as strong cation exchange (SCX) or strong anion exchange (SAX) has been another widely used peptide and protein separation method in biological and analytical labs [59]. MS coupled with IEX separation has been successfully implemented in phosphorylated peptide analysis, due to the charge difference introduced by the phosphate group [60,61,62]. Electrostatic repulsion-hydrophilic interaction chromatography (ERLIC) is a novel mode of separation utilizing IEX stationary phase with HILIC mobile phases (i.e., analytes are eluted by decreasing organic solvent gradient), which was first introduced by Alpert [63]. In ERLIC, hydrophilic interaction and electrostatic interaction between analytes and stationary phase are superimposed, and one can manipulate the charge states of analytes by operating at optimal pH to maximize the electrostatic attraction between analytes of interest and charged stationary phase, while the electrostatic repulsion antagonizes the retention of interfering molecules that cannot be separated by conventional HILIC with neutral stationary phase [63]. Anion exchange-ERLIC can selectively separate phosphorylated peptides from nonmodified counterparts because of the increased hydrophilicity and additional negative charge introduced by a phosphate group at the operating pH [64]. Glycosylated peptide enrichment by its hydrophilicity is also compatible with ERLIC, and it has been reported that ERLIC outperforms HILIC by neutral stationary phases on isobaric-labeled glycopeptide enrichment, which facilitated further PTM quantification [65, 66]. Furthermore, ERLIC by both SCX and WAX has been investigated for simultaneous, selective isolation of both phosphopeptides and glycopeptides from tryptic digests [67,68,69]. Despite the plentiful application of ERLIC, the mobile phase constitution for simultaneous enrichment of multiple PTMs has not been thoroughly explored.

In this work, we examined the potential of coupling ERLIC SPE with RPLC-MS to achieve large-scale phosphorylation and N-glycosylation analysis. Different mobile phase compositions, including organic phase proportion, ion pairing reagent, salt, and pH, were tested and evaluated to maximize the enrichment performance of both N-glycosylated and phosphorylated tryptic peptides from MM.1S cell lysate (a model cell line for multiple myeloma) [70].

Experimental Section

Materials

Poly SAX LP™ bulk material (12 μm, pore size 300 Å) was obtained from PolyLC (Columbia, MD). 0.22 μm Millex-GV Hydrophilic Durapore (PVDF) Membrane filter units were purchased from Merck Millipore (Tullagreen, Ireland). Iodoacetamide (IAA), Roche protease inhibitor tablets, and Roche PhosSTOP phosphatase inhibitor tablet were purchased from Sigma-Aldrich (St. Louis, MO). Phosphoric acid, ammonium acetate, Tris base, urea, potassium phosphate monobasic, sodium chloride, and calcium chloride were purchased from Fisher Scientific (Pittsburgh, PA). C18 SepPak cartridges were purchased from Waters (Milford, MA). C18 OMIX tips were purchased from Agilent (Santa Clara, CA). Dithiothreitol (DTT) and sequencing grade trypsin were purchased from Promega (Madison, WI). Protein and peptide BCA assays were purchased from Pierce (Rockford, IL).

Cell Culture

MM.1S cells were cultured in RPMI-1640 medium (Corning) supplemented with 10% FBS, 1% sodium pyruvate, 1% penicillin/streptomycin, and 10 mM HEPES at 37 °C in humidified atmosphere with 5% CO2. When the MM.1S cells were confluent, they were harvested and then washed once with PBS to remove remaining media content. Resulting cell pellets were stored under – 80 °C until use.

Cell Lysis

MM.1S cell pellets were homogenized in lysis buffer (8 M urea, 50 mM Tris, pH = 8, 5 mM CaCl2, 20 mM NaCl, 1 EDTA-free Roche protease inhibitor tablet and 1 Roche PhosSTOP phosphatase inhibitor tablet) with a probe sonicator for 3 pulses at 60 W, 20 kHz for 15 s, each followed by a 30-s pause period for cooling at 4 °C. Crude lysates were then centrifuged at 14000×g for 5 min, after which the supernatant was collected and protein concentrations was measured by Pierce BCA Protein Assay according to the manufacturer’s protocol.

Trypsin Digestion

Lysate containing 2 mg protein was reduced in 5 mM DTT at room temperature for 1 h, followed by alkylation in 15 mM IAA for 30 min in the dark. Alkylation was quenched by adding DTT to 5 mM. The resulting solution was then diluted with Tris buffer (pH = 8) to 0.9 M urea and proteins were digested with trypsin at 1:50 enzyme to protein ratio at 37 °C for 18 h. Digestion was quenched by adding trifluoroacetic acid (TFA) to a final concentration of 0.3% and desalted with C18 SepPak cartridges. Digested peptide concentration was measured by Pierce Peptide BCA assay, and peptides were aliquoted to 100 μg and dried under vacuum before enrichment.

ERLIC Enrichment

ERLIC enrichment of phospho- and N-glycopeptides from MM.1S cell digest was performed in the custom-packed spin-tips [71, 72]. As Fig. 1b illustrates, 3 mg of cotton wool was inserted into a 200-μL pipette tip. SAX LP bulk material was dispersed in 0.1% TFA as a 10 mg/ 200 μL slurry and activated for 15 min under vigorous vortexing. After activation, 60 μL slurry was added to the spin-tip. Solvent was removed by centrifugation at 1200 rpm for 2 min, leaving the SAX material packed above the cotton wool. The stationary phase was then conditioned by 150 μL ACN, 100 mM ammonium acetate, 1% TFA, and the loading buffers, each repeated 3 times. One hundred microgram MM.1S cell digest was dissolved in 150 μL of the corresponding loading buffer and loaded onto the tips by centrifugation at 1200 rpm for 2 min, and the flow-through was collected and loaded again to ensure complete retention. The tips were then washed with 150 μL loading buffers 6 times, after which four fractions were eluted with 300 μL 50/50 ACN/water with 0.1% formic acid (FA), 0.1% FA in water, 0.1% TFA in water, and 300 mM KH2PO4 (pH = 2) and were collected in four separate centrifuge tubes. Each fraction was filtered by a 0.22-μm Millex-GV Hydrophilic Durapore (PVDF) Membrane filter unit, and the KH2PO4 fractions were desalted by C18 OMIX tips according to manufacturer’s protocol, after which all fractions were dried under vacuum before MS analysis.

Figure 1
figure 1

(a) Illustration of ERLIC enrichment and separation mechanism of phosphopeptides and glycopeptides; (b) Custom-made spin-tip composition; (c) Schematic workflow of spin-tip-based ERLIC SPE coupled with LC-MS/MS analysis on MM.1S cell lysate digest

Full size image

NanoLC-MS/MS Analysis

Peptides in each fraction were reconstituted in 3% ACN with 0.1% FA and subjected to reversed phase LC-MS/MS analysis with a Q-Exactive HF orbitrap mass spectrometer (Thermo Fisher Scientific, San Jose, CA) interfaced with a Dionex Ultimate 3000 UPLC system (Thermo Fisher Scientific, San Jose, CA). Peptides were loaded onto a 75-μm i.d. microcapillary column custom-packed with 15 cm of BEH C18 particles (1.7 μm, 130 Å, Waters). Peptides were separated with a 90-min gradient from 3 to 30% ACN with 0.1% FA, followed by 10 min to 75% ACN and then 10 min to 95% ACN. After that, the column was re-equilibrated with 3% ACN for 15 min to prepare for the next injection.

The mass spectrometer was operated in a top 15 data-dependent acquisition mode. Survey scans of peptide precursors from m/z 300 to 2000 were performed at a resolving power of 60 K and an AGC target of 2 × 105 with a maximum injection time of 150 ms. The top 15 intense precursor ions were selected and subjected to the stepped HCD fragmentation at normalized collision energy of 22, 30, and 38% followed by tandem MS acquisition at a resolving power of 30 K and an AGC target of 5 × 104, with a maximum injection time of 250 ms. Precursors were subjected to a dynamic exclusion of 15 s with a 10-ppm mass tolerance.

Data Analysis

Raw files were processed with the Byonic search engine (Protein Metrics Inc., San Carlos, CA) embedded within Proteome Discoverer 2.1 (Thermo Fisher Scientific, San Jose, CA). Spectra were searched against the UniProt Homo sapiens proteome database (April 12, 2016; 16,764 entries) with trypsin as the specific digestion enzyme and maximum two missed cleavages. The parent mass error tolerance was set to be 50 ppm and fragment mass tolerance was 0.02 Da. Fixed modifications are specified as carbamidomethylation on cysteine residues (+ 57.02146 Da). Dynamic modifications included oxidation of methionine residues (+ 15.99492 Da), phosphorylation on S, Y, and T, and N-glycosylation. It is reported that ERLIC can enrich for O- glycopeptides [73], but in this work, we only analyzed N-glycopeptides from each enrichment. Glycan modifications were specified as the common mammalian N-glycome (default N-glycome database in Byonic with removal of sodium mass, which contains overall 309 N-glycans), expanded with typical mannose-6-phosphate glycans including HexNAc (2)-Hex (4–9)-Phospho (1–2), HexNAc (3–4)-Hex (4–9)-Phospho (1–2), HexNAc (2) Hex (3–4) Phospho (1), and HexNAc (3) Hex (3–4) Phospho (1). Identifications were filtered at 1% false discovery rate (FDR).

Results and Discussion

Because multiple interactions such as hydrophilic interaction and electrostatic interaction are superimposed while operating ERLIC, selecting the best mobile phase is not straightforward. Here, we aim to examine the impact of organic phase proportion, ion pairing reagent, pH, and salt content on the retention and elution of phospho- and N-glycopeptides in ERLIC-RP-MS utilizing the new PolySAX LP material [64]. This work will benefit studies of the mechanism of ERLIC and broaden its application to simultaneous phosphorylation and N-glycosylation analysis at a large scale.

Loading Buffer Affects Enrichment Efficiency

To evaluate the impact of mobile phase composition on the retention of modified peptides, 10 different loading buffers were compared according to Table 1. Since HILIC was named and defined in 1990 [74], its mechanism of retention has aroused the interest of many scientists, among whom the partitioning theory is the most supported. In this theory, a water-rich layer is immobilized on the stationary phase due to the high density of hydrophilic functional groups, then the retention of solutes was equilibrated by the partition between the bulk eluent and the water-rich layer, which is different from conventional normal phase chromatography where the retention is predominantly governed by surface adsorption [75]. According to the empirical observations, mobile phase with a higher content of aqueous phase serves as a stronger eluent. Most of the pioneering studies utilizing ERLIC SPE for N-glycopeptide enrichment used 95% ACN for loading and washing without thorough justification [65, 66]. However, we found this condition to be too retentive in that many nonmodified peptides with minimal hydrophilicity were co-retained with the modified peptides of interest. In Fig. 2, the loading buffers with high (95%) ACN groups c, d, and e) generally retain more total peptides than the low (80%) ACN buffer conditions, where the difference in peptide retention is as high as 10 times greater with the 95% ACN combinations (except for combination f, with 1% FA added), but the LC-MS identified significantly fewer modified peptides in high organic phase loading conditions. The identification of phosphorylated or N-glycosylated peptides was largely suppressed by the numerous co-eluting and co-fragmenting nonmodified tryptic peptides that ionized better or were more abundant. On the other hand, mobile phase with 80% ACN was adequate to retain the majority of the modified peptides while the nonmodified peptides were removed during the washing step. By modulating the organic phase proportion in the mobile phase, the ERLIC enrichment specificity was greatly increased, benefiting the MS identification of peptides with phosphorylation and N-glycosylation.

Table 1 ERLIC Loading Buffer Composition
Full size table
Figure 2
figure 2

Identification results of peptide modifications in 100 μg of MM.1S cell lysate by ERLIC fractionation. ERLIC loading buffer conditions are provided in Table 1. (a) Total peptide identification; (b) Glycopeptide and phosphopeptide identifications (c) % Specificity of ERLIC enrichment for each condition, calculated via dividing the number of modified peptides identified by the total number of peptides

Full size image

Ion pairing has been reported as an effective strategy to increase the specificity in HILIC based enrichment of glycosylated peptides, and ion pairing HILIC has been successfully applied to mammalian and plant glycoproteomics in various forms including SPE and HPLC [52, 76,77,78]. Wimley-White water/octanol free energy scale of amino acids has shown theoretical evidence that peptides with charged moieties have lower hydrophobicity than their uncharged forms [77, 79], which is consistent with the experimental results in ion-pairing normal phase LC that the hydrophobicity of charged nonglycopeptides increases and the separation between glycopeptides and nonglycopeptides was achieved when efficient ion pairing reagents were added [77]. Ion pairing reagents such as monovalent salt ions including NaCl, LiCl, and KCl or acids including acetic acid, formic acid, and TFA will “neutralize” the charges of peptides by forming ion-pairs with the positively charged groups such as the N-termini, lysine, and arginine residues, and the negatively charged groups including C-termini, aspartate, and glutamate residues. With the nonmodified peptides’ charges neutralized, their hydrophilicity is largely reduced, and the glycopeptide enrichment specificity is improved because the hydrophilic retention of analytes will predominantly depend on the existing glycans whose hydrophilicity is moderately affected by ion pairing reagents [52, 77, 78]. Moreover, a recent study systematically evaluated the impact of different salts on the retention of HILIC and ERLIC, where they found well-hydrated counterions of charged analytes usually increase the hydrophilic retention while poorly hydrated counterions decrease it [80]. The similar concept of counterion pairing was adapted to ERLIC here, with the choice of the poorly hydrated TFA and weakly hydrated FA as the ion pairing reagents being compared. On the other hand, the two acid modifiers resulted in different pH of the mobile phase and shifted the protonation-dissociation equilibrium of phosphate and other functional groups with acidic protons in the analytes, which would potentially influence the electrostatic interaction-based retention. As the most hydrophilic group in peptides except for basic residues, charged phosphate groups can increase the hydrophilicity of tryptic peptides [74]. However, the phosphate group alone does not suffice to permit the separation of phosphopeptides from nonphosphopeptides as a set in HILIC. In ERLIC, the concurrent electrostatic interaction plays a significant role in phosphopeptide retention. Phosphate has a pKa1 of 2.15, which is lower than aspartic and glutamic residues and the C-terminus. With an appropriate pH in between them, the majority of peptides with acidic residues can be uncharged while the charged phosphopeptides can be strongly attracted by the positively charged SAX stationary phase. Thus, the enrichment selectivity of phosphopeptides and highly hydrophilic glycopeptides over nonmodified peptides can be enhanced with the selection of pH at 2–3. As Fig. 2b indicates, TFA serves as the best ion pairing reagent for selective N-glycosylated peptide retention, where use of loading buffer combinations of a, b, i, and j containing either 1% or 0.1% TFA resulted in more than 300 N-glycopeptide identifications. For phosphorylated peptides, the retention depends on hydrophilic interaction and electrostatic interaction synergistically. At high organic phase loading conditions, most of the phosphorylated peptides can be retained and identified regardless of the TFA content, because the strong hydrophilic interaction plays a predominate role. At low organic phase and stronger hydrophilic eluent loading conditions as in groups a, b, and g–j, the electrostatic interaction became a more significant component of the mechanism of phosphopeptide retention. At lower TFA or FA concentration, pH was higher and phosphate groups were more highly deprotonated and better attracted by the SAX resin. For example, in group b, 165 phosphopeptides were identified with high confidence whereas there were only 54 phosphopeptides identified in group a. Group h with even higher pH of 3 of 0.1% FA also showed great phosphopeptide retention where 171 phosphopeptides were identified in the total four fractions.

Conditioning is an essential preparation step before operating SAX resins. A highly concentrated salt buffer such as triethylammonium acetate is usually used to flush SAX resin for various reasons [65, 66]. Some of the main reasons include (1) it can wash off the byproducts that came from SAX resin synthesis before the first time usage and (2) well-hydrated counter ions can partition into the immobilized aqueous layer on the stationary phase and pulls the charged solutes into the aqueous layer thus increase the HILIC retention, while poorly hydrated counter ions have the opposite effect [80]. To prove the concept, we incorporated a control group i without flushing with salt in the experimental procedure before sample loading. Surprisingly, we found both N-glycosylated and phosphorylated peptide identifications from group i actually had a moderate increase over group a, which could arise from the more well-hydrated counterions that PolySAX LP has from manufacturer actually outperform acetate. It is worth mentioning that a large amount of 1% FA as the mobile phase modifier in groups f and g resulted in largely reduced retention of both modified and nonmodified peptides comparing to groups e and h. We hypothesize that high concentration of FA partitioned in the immobilized aqueous layer can compete the electrostatic interaction between charged analytes and stationary phase to a large extent, while the retention of hydrophilic peptides is reduced by “salt out” effect [80]. In contrast, these effects are not observed with 1% TFA groups a, c, and j, which is because TFA tends to partition mostly into the organic mobile phase of ERLIC. Figure 2c shows the proportion of identified N-glycosylated or phosphorylated peptides within the total identified peptides for each combination. Group b is the optimal combination among the ten tested, with enrichment specificity of 24.8% and 11.0% for N-glycosylated and phosphorylated peptides, respectively.

Eluting Profile Reveals the Key Retention Mechanism

To further separate the modified peptides and simplify the MS-based analysis, four crude elution fractions were collected with mobile phase conditions as indicated in Table 2. The peptide spectral matches (PSMs) of each modification-containing peptides are counted and the distributions of the PSMs among the eluting fractions are shown in Fig. 3. Figure 3(a) summarizes total PSM distribution in four fractions. Figure 3b shows that the majority of N-glycosylated peptides elute in the first two fractions with the decreasing gradient of organic phase, indicating the retention mechanism for N-glycopeptides is predominately by hydrophilic interaction. On the other hand, in Fig. 3c, phosphopeptides aggregate in the fractions 3 and 4, where the pH is lowered and the competitive salt is added to reduce the column interaction with the SAX resin. The results show that ERLIC can separate peptides with the two types of PTMs, which improves MS analysis and minimizes interference between peptides with the two modifications. Sialic acid containing N-glycopeptides have additional acidic groups of which the pKa is 2.6. According to Fig. 3d, the elution profiles of sialylated N-glycopeptides are consistent with those of the bulk N-glycopeptides from each condition in Fig. 3b. The results indicate the predominant mechanism for ERLIC to retain sialic acid containing N-glycopeptides at pH 1~3 is through hydrophilic interaction. We hypothesize that if pH were further increased to > 3, the electrostatic interaction would have more contribution.

Table 2 ERLIC Eluting Buffer Composition
Full size table
Figure 3
figure 3

PSM distribution of modifications among fractions. ERLIC loading buffer conditions are given in Table 1. Eluting buffer conditions are given in Table 2. (a) Total PSMs; (b) glycopeptide PSMs; (c) phosphopeptide PSMs; (d) sialic acid containing glycopeptide PSMs

Full size image

Furthermore, we manually checked some PSMs of N-glyco-, phospho-, and sialylated N-glycopeptides from the Byonic search engine output. Figure 4 shows three representative spectra, where the HCD with stepped collisional energy of 22%, 30%, and 38% resulted in sufficient peptide backbone fragments with clear modification site information and glycan fragments cleaved at the glycosidic linkage. The results support stepped HCD as a highly efficient and simple MS fragmentation method for simultaneous profiling of multiple PTMs without tedious optimization when coupled with ERLIC fractionation and separation.

Figure 4
figure 4

Representative stepped HCD MS/MS spectra with site-specific identifications of glycopeptides, phosphopeptides and those with sialic acid containing glycopeptides. (a) MS/MS of N-glycopeptide TN[+876.32]STFVQALVEHVKEECDR (AA 214–232) from human prosaposin, spectrum of +4 charged precursor at m/z 785.36; (b) MS/MS of phosphopeptide LQQGAGLESPQGQPEPGAAS[+79.97]PQR (AA 72–94) from human coiled-coil domain containing protein 86, spectrum of +3 charged precursor at m/z 795.04; (c) MS/MS of sialylated glycopeptide AN[+2204.77]C[+57.02]SVYESC[+57.02]VDC[+57.02]VLAR (AA 495–510) from human semaphorin-4A, spectrum of +4 charged precursor at m/z 1027.65; (I) m/z 150 to 410 of (c) with oxonium ions and NeuAc reporter ions resolved and annotated; (II) 7-fold zoom-in of (c) at m/z 410 to 2000 with most of the peptide backbone fragment ions being detected and resolved

Full size image

Conclusions

In this study, we demonstrated the great potential of ERLIC for simultaneous enrichment of phosphorylated and N-glycosylated peptides. Ten different loading buffer conditions were examined, thoroughly considering the factors of organic phase proportion, ion pairing reagents, pH, and salt. While a high concentration of organic phase (95% ACN) has commonly been used in previous studies, it is outperformed here by 80% ACN in terms of hydrophilic interaction-based enrichment’s selectivity and specificity for modified versus nonmodified peptides. The use of a less well-hydrated ion pairing reagent benefits N-glycosylation enrichment by reducing nonspecific retention. With weak hydrophilic eluent, phosphopeptides’ retention is independent of ion-pairing reagents; with intermediate hydrophilic eluent, the pH of mobile phase has a stronger impact on electrostatic interaction-based retention of phosphorylated peptides, where pH between 2 and 3 ensures a decent extent of phosphate group dissociation and phosphopeptide retention, while the majority of acidic peptides remain to be uncharged, which enhances the enrichment selectivity of phosphopeptides by ERLIC. With an inappropriate selection of counterions before or during the enrichment, the retention of modified peptides of interest can be largely hampered by competitive binding to charged resins and “salting out” effect. In the future, it is worth exploring the types of salt modifier that benefit the selective retention of N-glyco- and phosphopeptides in ERLIC.

The PSMs in the elution profiles of phosphorylated and N-glycosylated peptides showed a decent level of separation, where the majority of N-glycopeptides were eluted by disrupting hydrophilic interaction with decreasing organic phase gradient and most phosphopeptides were eluted by disrupting coulombic interaction with increased protonation and competitive salt content. The elution by multiple steps in the ERLIC SPE should be applicable to an HPLC system. We anticipate our optimized ERLIC SPE method will produce great resolution when adapted to HPLC and enable simultaneous enrichment and separation of N-glycosylated and phosphorylated peptides in complex mixtures, with improved coverage.