Design and preparation of novel nitro-oxide-grafted nanospheres with enhanced hydrogen bonding interaction for O-GlcNAc analysis

Abstract

As an essential modification, O-linked β-N-acetylglucosamine (O-GlcNAc) modulates the functions of many proteins. However, site-specific characterization of O-GlcNAcylated proteins remains challenging. Herein, an innovative material grafted with nitro-oxide (N→O) groups was designed for high affinity enrichment for O-GlcNAc peptides from native proteins. By testing with synthetic O-GlcNAc peptides and standard proteins, the synthesized material exhibited high affinity and selectivity. Based on the material prepared, we developed a workflow for site-specific analysis of O-GlcNAcylated proteins in complex samples. We performed O-GlcNAc proteomics with the PANC-1 cell line, a representative model for pancreatic ductal adenocarcinoma. In total 364 O-GlcNAc peptides from 267 proteins were identified from PANC-1 cells. Among them, 183 proteins were newly found to be O-GlcNAcylated in humans (with 197 O-GlcNAc sites newly reported). The material and method can be facilely applied for site-specific O-GlcNAc proteomics in other complex samples.

GRAPHICAL ABSTRACT

■ INTRODUCTION

As a ubiquitous post-translational modification (PTM), protein glycosylation has critical roles in various biological events.^1,2 Amongst a wide array of glycans, the O-linked β-N-acetylglucosamine (O-GlcNAc) modification (i.e., O-GlcNAcylation) is unique and crucial.^3,4 as a monosaccharide modification, O-GlcNAcylation occurs on numerous proteins in the nucleus, cytosol, and mitochondria. As a nutrient sensor, O-GlcNAcylation integrates major cellular metabolic pathways. It is fundamental in virtually all processes examined (e.g., cell cycle, genome maintenance, epigenetic regulation, cellular metabolism, among others). ^5-9 Despite its functional importance, large-scale identification (especially site-specific characterization) of protein O-GlcNAcylation is challenging, largely due to several shortcomings (such as extremely low stoichiometry and lability in the gas phase). ^{10, 11} To that end, various materials/methods have been proposed for O-GlcNAc enrichment, such as antibody-based immunoprecipitation,^{12, 13} lectin affinity chromatography,^14-17, phenylboronic acid (PBA)-based affinity,¹⁸ hydrophilic interaction liquid chromatography,^19--21 and chemoenzymatic/metabolic labeling.^22-33 Although chemoenzymatic/metabolic labeling methods have shown some success, tedious enrichment steps are often required and substantial sample loss often occurs. In contrast, affinity chromatography provides straightforward enrichment even for native O-GlcNAc proteins/peptides. However, such methods generally suffer from low enrichment efficiency because of the low affinity toward the O-GlcNAc group on proteins/peptides. Clearly there is an unmet need to enrich O-GlcNAc proteins/peptides before the analysis by mass spectrometry.

Hydrophilic interaction is an excellent approach to retain and separate polar/hydrophilic compounds (including glycosides and carbohydrates).^28,34 For hydrophilic retention, three basic interactions (including hydrogen bonding, hydrophilic partitioning, and electrostatic interactions) are involved, with hydrogen bonding being one of the most important driven forces.^35-37 Currently, enhancing hydrophilic interaction is performed mainly by increasing binding sites via tailing the matrix surface or by changing the amount of functional groups on the materials to separate oligosaccharides-modified peptides.^38,39 The carbohydrate modification strategy is one of the most typical examples. Hydrogen bonding interaction between hydroxyl groups in the monosaccharides and hydrophilic materials is regarded as the hallmark of carbohydrate-carbohydrate interactions (CCI)--even though an individual CCI is weak, multiple-site binding between adjacent monosaccharides can substantially enhance this kind of interaction for complex glycans. However, such a strategy is not applicable for O-GlcNAc analysis, in which a peptide is often modified on few sites by only the single monosaccharide (as seen in many cases). Other than increasing binding site numbers, hydrophilic interaction materials designed to increase affinity via other mechanisms are seldom reported. In addition, the actual strength of hydrogen bonding (the key factor that will affect hydrophilic affinity) has not been explored and evaluated yet.

Herein, we synthesized a novel type of material nitro-oxide-grafted affinity nanospheres for O-GlcNAc analysis. In brief, amine-activated silica nanoparticles were prepared and oxidized to produce nitro-oxide (N→O) groups. The material was systemically characterized and evaluated with standards and complex samples. Our results show that the newly prepared material has strong hydrogen bonding affinity toward the O-H moieties of O-GlcNAc groups. The corresponding enrichment procedure (which minimizes the interfence of other types of glycans) enables facile enrichment of O-GlcNAcyalted peptides and facilitates site-specific O-GlcNAc proteomics in complex samples.

EXPERIMENTAL SECTION

Materials and reagents.

Tetraethyl orthosilicate (98%, TEOS), toluene, ammonium hydroxide solution (28.0-30.0%), formaldehyde solution (ACS reagent, 37 wt. % in H2O), α-crystallin, dithiothreitol (DTT), iodoacetamide, PUGNAC, and ammonium formate solution (NH₄FA, Bioultra, 10 M) and 3-[2-(2-aminoethylamino)ethylamino]propyl-trimethoxysilane (3APMS) were ordered from Sigma Aldrich (St. Louis, MO). Thiamet G was obtained from Cayman Chemicals. Trypsin (sequencing grade) was purchased from Promega (Madison, WI). Dulbecco's Modified Eagle's medium (DMEM), fetal bovine serum (FBS), and hydrogen peroxide solution (30%) were obtained from VWR chemicals. α2-3,6,8 Neuraminidase, O-Glycosidase, and PNGase F were purchased from NEB (Beverly, MA). The midi S-Trap columns were purchased from Protifi (Huntington, NY). Three O-GlcNAc peptides were custom synthesized by AnaSpec, Inc. (Fremont, CA): peptide #1: YSPT(O-GlcNAc)SPSK, MW 1069.5048; peptide #2: TAPTS(O-GlcNAc)TIAPG, MW 1118.5576; peptide #3: YSPT(O-GlcNAc)S(O-GlcNAc)SPSK, MW 1359.6162. Trifluoroacetic Acid (TFA, >99.5%), LC/MS grade formic acid (FA) and acetonitrile (ACN) were obtained from Fisher Scientific (Waltham, MA).

Preparation and characterization of the nitro-oxide-grafted nanospheres.

The scheme for the preparation of N-O based materials is shown in Figure 1. First, silica nanoparticles were synthesized with the Stöber method according to the previous protocol.⁴⁰ In brief, 4.0 mL of TEOS was quickly added to a hydrolysis solution (consisting of 6.7 mL ammonium hydroxide solution, 70 mL absolute ethanol, and 5.1 mL deionized water) and incubated for 40 min at room temperature with stirring. After centrifugation at 3000 g for 5 min, the obtained silica gel (native SiO₂ beads) was washed with water and ethanol, respectively. The nanospheres were then dried under vacuum for 18 h at 70 °C. To 1 g of the dried nanoparticles, 4 mL of 3APMS and 80 mL of dry toluene were added and incubated for 24 h under heating with stirring. After centrifugation and washing with ethanol, the resulting particles (3APMS@SiO₂) were dried under vacuum at 70 °C for 12 h.⁴¹ Dimethyl-3APMS@SiO₂ beads were prepared by reductive demethylation of amino groups on 3APMS@SiO₂ beads with formaldehyde (HCHO) and cyanoborohydride (NaBH₃CN). In brief, several steps were performed: (i) reconstitute the silica beads (50 mg) in 4 ml of 50 mM PBS (pH 8.0); (ii) add 500 μL of 4% (vol/vol) HCHO and 500 μL of 1.0 M NaBH₃CN; (iii) incubate in a fume hood overnight with magnetic stirring; (iv) wash the beads thoroughly with water. Finally, the N, N-dimethylated amino groups on Dimethyl-3APMS@SiO₂ beads were oxidized with H₂O₂ solution to obtain NO-3APMS@SiO₂.

Figure 1.

(A) Structural formula of donors and acceptors for hydrogen bonding; (B) Concentration-variation ¹H NMR spectra of 4-methylmorpholine N-oxide with phenol (400 MHz, CDCl₃); (C) Concentration-variation ¹H NMR spectra of 4-methylmorpholine with phenol (400 MHz, CDCl₃).

A Bruker AVANCE III 600 MHz spectrometer was used for the concentration variation experiments. ¹H NMR spectra of concentration-variation experiments were performed keeping the ─OH donors (phenol) concentrations constant, with varying the proton-acceptors (4-methylmorpholine N-oxide and N, N-dimethyldodecylamine N-oxide 4-methylmorpholine and N, N-dimethyldodecylamine) concentrations. In ¹H NMR, tetramethylsilane (TMS) was used as an internal standard (with chemical shifts shown in ppm). ATR-IR measurement of the catalyst materials was conducted on a Nicolet 6700 instrument equipped with a golden state ATR equipment. The TEM images were recovered by the JEM2010F. The SEM images and EDS mapping images were recorded by Auriga 60. The XPS were recorded on a Thermo-Fisher K-Alpha Plus equipment, with the binding energies corrected by referencing C 1s to 284.6 eV.

Cell Culture.

The pancreatic ductal cell line PANC-1 cells were grown in DMEM medium supplemented with 10% FBS and 1% penicillin/streptomycin. Cells were maintained in a 37 °C incubator with 5% CO₂. Cells were treated by 2 μM Thiamet G for 4 h before harvesting. After treatment with 0.25% (w/v) trypsin/EDTA, cells were washed once with 10 mL cold PBS. The cell suspension was centrifuged at 1200 rpm for 2 min at 4 °C, with the cell pellet stored at −80 °C before analysis.

Protein extraction and digestion.

PANC-1 cells were suspended in 1 mL cell lysis buffer (5% SDS, 2 μM PUGNAc, 1x protease inhibitor cocktail, 50 mM TEABC) by pipetting up and down. Benzonase (500 U) was added, with the lysate incubated on ice for 20 min. The cell suspension was then sonicated with a probe-tip sonicator (Qsonica) at 30% amp for 5 pulses (each with 10 sec on and 20 sec off) on ice. The cell lysates were centrifuged at 13000 g at 4 °C for 15 min, with the supernatant transferred into a new tube. Extracted proteins were processed with the suspension trapping (S-Trap) method as reported previously,⁴² with minor modifications. Prior to tryptic digestion, proteins were treated with a mixture of enzymes (consisting of PNGase F and α2-3,6,8 Neuraminidase as well as O-Glycosidase) at 37 °C overnight.

O-GlcNAc peptide enrichment.

200 μL of loading buffer (ACN/H2O/FA = 90/9/1) was used to suspend NO-3APMS@SiO₂ nanoparticles. To the suspension, standard O-GlcNAc peptides or alpha-crystallin digests were added and incubated for 1 h with shaking. Nanoparticles were centrifuged for 2 min at 6000 g and washed by 100 μL of washing buffer (ACN/H₂O/FA = 90/9/1) for three times. Captured O-GlcNAc peptides were eluted two times with 20 μL of eluting buffer (ACN/H₂O/FA = 30/69/1) for 10 min, with the eluents dried with Speedvac.

For enrichment of O-GlcNAc peptides from complex samples, digests of PANC-1 cells dissolved in 500 μL of loading buffer were mixed with 30 mg NO-3APMS@SiO₂ nanospheres and incubated for 2 h. The nanospheres were washed with 3×500 μL of 90% ACN containing 1% FA. Enriched peptides were then eluted twice with 200 μL of H₂O containing 0.1% FA. The eluents were freeze-dried and fractionated by high pH RPLC as described previously.⁴³

NanoUPLC-MS/MS.

Peptides from alpha-crystallin and complex samples were analyzed by a system integrating Orbitrap Lumos and nanoAcquity UPLC, with similar settings as described previously.⁴² HCD product dependent EThcD (i.e., HCDpdEThcD)⁴⁴ was used for MS/MS acquisition, in which EThcD was triggered with more than one fragment peaks (i.e., 204.0867, 138.0545, 126.055, 186.0761 and 168.0655) observed in the HCD scan. Supplemental activation (SA) collision energy of EThcD was set as 30%. A nanoAcquity UPLC interfaced with TripleTOF 6600 in IDA mode was used for the analysis of standard O-GlcNAc peptides. All settings were the same as described previously,⁴⁵ except that a short gradient of 1 h was used (i.e., 0 min, 1% buffer B; 1 min, 5% buffer B; 35 min, 45% buffer B; 37 min, 99% buffer B; 40 min, 99% buffer B; 40.1 min, 1% buffer B, 60 min, 1% buffer B).

Data analysis.

Database searching of the raw files for alpha-crystallin and complex samples was performed in Proteome Discoverer version 2.4 (Thermo Fisher), by using similar settings as described previously.⁴² In addition, O-GlcNAc (Ser or Thr, +203.079 Da) was set as variable modification. Only sites on O-GlcNAc peptides with localization probability over 75% were considered as unambiguous sites.

Subcellular location of proteins was performed by WoLF PSORT.⁴⁶ Proteins localized to ER/Golgi and extracellular space were excluded from the O-GlcNAc list. WebLogo^{47, 48} was used to generate sequence logos (±6 residues around the O-GlcNAc modification site). GO and protein-protein interaction enrichment were performed by Metascape.⁴⁹ Significantly enriched GO terms were plotted by using R package ‘tidyverse’.⁵⁰

RESULTS AND DISCUSSION

Rationale and material synthesis.

Efficient enrichment of O-GlcNAcylated peptides has been the bottleneck step for their analysis. We aim to develop a novel type of material with high-affinity towards the O-GlcNAc group. The O-GlcNAc moiety appears to be chemically inert for derivation, although it can be chemoenzymatically activated for further derivatization (e.g., by GalT1 or Endo-M²¹). However, its multiple hydroxyl groups provide an alternative for enrichment. Unfortunately, traditional hydrophilic interaction materials (which contain–NH₂, ─CONH, ─CN, and ─CH(OH)CH₂OH groups)³⁶ have limited affinity for O-GlcNAc, due to its monosaccharide nature (c.f. complex glycans). During this interaction, hydrogen bonding is regarded to be the key binding force. In our recent study,⁵¹ by probing with several model hydroxyl compounds, we investigated several proton acceptors (including ─N⁺─O⁻, S═O, C─O─C) in terms of their ability to form hydrogen bonds, and found that N⁺-O⁻ (N→O) formed the strongest hydrogen bonding. Inspired by this observation, we set out to further compare the H-bonding capacity of N→O type acceptors and non N→O type acceptors.

Specifically, two model N→O compounds (4-methylmorpholine N-oxide and N, N-dimethyldodecylamine N-oxide; and two none N→O compounds (4-methylmorpholine and dodecylamine) containing normal nitrogen elements (which usually serve as the hydrophilic sites in traditional hydrophilic interaction materials) were selected as proton acceptors (Figure 1A). Phenol, a compound containing an inert hydroxyl group, was used as the proton donor. The hydrogen bond strength between this proton donor and these proton acceptors was measured and compared by using concentration-variation ¹H NMR. As shown in Figure 1B, the chemical shift of OH group clearly shifts to the downfield (from 4.6 ppm to 11.1 ppm) with the addition of 4-methylmorpholine N-oxide especially (with a molar ratio of 4-methylmorpholine N-oxide : phenol of 4), illustrating the formation of a hydrogen bond. The Gibbs free energy change of this bond formation (ΔG) is −2.1 kcal/mol (Figure S1), as calculated with the concentration dependent ¹H NMR shift according to the Eq. S1 and S2. In contrast, much less shift change of the active proton occurred after the addition of 4-methylmorpholine (from 4.6 ppm to 6.38 ppm) with a molar ratio of 4-methylmorpholine : phenol of 4 (Figure 1C). The ΔG of this interaction was calculated to be −1.1 kcal/mol (only half than that of N→O) (Figure S2). Similar results were obtained when the pair of N, N-dimethyldodecylamine N-oxide and N, N-dimethyldodecylamine were applied as proton donor. Comparison at the same concentration of proton acceptors, N, N-dimethyldodecylamine N-oxide showed a higher chemical shift change (Δδ) illustrating the formation of a stronger hydrogen bond (Figure S3 and Figure S4). Clearly the hydrogen bonding between N→O with ─OH groups is much stronger than the normal nitrogen element (non N→O) with ─OH groups. Based on these observations, we proposed that N-oxide groups would provide strong affinity toward ─OH groups of the O-GlcNAc group and proposed that nitro-oxide-grafted nanospheres would be an excellent material to enrich O-GlcNAc peptides.

In view of this theory, we chose to develop a novel type of nitro-oxide-grafted material, which would provide strong affinity toward the O-H moiety on GlcNAc groups (via enhanced hydrogen bonding). The synthesis procedure of NO-3AMPS@SiO₂ nanoparticles is displayed in Figure 2A. Firstly, amino-functionalized silica (3APMS@SiO₂) was prepared by modification with 3-[2-(2-aminoethylamino)ethylamino]-propyl-trimethoxysilane (3APMS). Then the primary and secondary amines were reductively methylated with formaldehyde followed by reduction with cyanoborohydride to synthesize Dimethyl-3APMS@SiO₂ beads.⁵² The NO-3AMPS@SiO₂ nanoparticles were obtained by oxidation of Dimethyl-3APMS@SiO₂ with H₂O₂. The material was then systematically characterized, optimized, and evaluated with standards and complex samples.

Figure 2.

A) The preparation scheme of NO-3APMS@SiO₂ nanoparticles; B) Chemical structure of the purposed hydrogen bonding mode between O-GlcNAc peptide and NO-3APMS@SiO₂ material; C) Procedure for the analysis of O-GlcNAc peptides.

Characterization of the nanospheres.

TEM and SEM were utilized to characterize the morphology of the silica nanoparticles from each step. The TEM (Figure 3A) and SEM (Figure S5) images show that the materials are uniform nanospheres with an average size of 330 nm, and that the modification process did not affect their morphology. SEM mapping images of NO-3AMPS@SiO₂ (Figure 3B) clearly show the nitrogen element distribution on the silica particles, suggesting the successful incorporation of N-containing function groups on the silica spheres. FT-IR and XPS were used for the further confirmation of the element N containing function groups belonging to N-O oxide. As seen in Figure S7, a vibration band at 3330 cm⁻¹ indicates the presence of OH groups.⁵³ The broad peaks ranging from 950 cm⁻¹ to 1100 cm⁻¹ are probably resulted from the siloxane vibrations of (SiO)_n. Compared with the native silica material, the 3APMS@SiO2 material shows absorption bands at and 2820 cm⁻¹ and 2870 cm⁻¹ (which may be ascribed to the vibrations of CH₂ groups) and 1650 cm⁻¹ (bending; due to the vibration of N-H⁵⁴), demonstrating the successful immobilization of 3APMS on the surface of silica gel. Furthermore, the weakened peak at 1650 cm⁻¹ of Dimethyl-3APMS@SiO₂ indicates the dimethylation of the amine group. Moreover, as shown in Figure 3C, the band at 946 cm⁻¹ reflects the stretching vibration of N-O,⁵⁵ indicating amine oxide formation in the NO-3APMS@SiO₂ material. XPS (Figure 3D) was then used to confirmed the formation of N-O oxide. After the oxidation of Dimethyl-3APMS@SiO₂ to NO-3APMS@SiO₂, the N 1s peak shifted from 398.8 eV to 400.2 eV, suggesting that the nitrogen is more positively charged (in accordance with the structure N⁺-O⁻ structure), also illustrating the existence of N-Oxide groups. These densely anchored nitro-oxide groups on the nanoparticle surface would provide strong hydrogen-bonding interactions for the retention of O-GlcNAc peptides (Figure 2B).

Figure 3.

(A) TEM images and particle size distributions of the a) Native SiO₂, b) 3AMPS@SiO₂, c) Dimethyl-3APMS@SiO₂ and d) NO-3AMPS@SiO₂ nanoparticles; (B) Elemental mapping images of NO-3AMPS@SiO₂ nanoparticles; (C) XPS N 1s spectra of 3AMPS@SiO₂ and NO-3AMPS@SiO₂ nanoparticles; (D) IR characterizations of Native SiO₂, 3AMPS@SiO₂, Dimethyl-3APMS@SiO2 and NO-3AMPS@SiO₂ nanoparticles (wavenumber from 1100~700 cm⁻¹).

Enrichment performance of the nanospheres.

Three standard O-GlcNAc peptides, including one peptide with two O-GlcNAc sites (i.e., peptide #1: YSPT(O-GlcNAc)S(O-GlcNAc)SPSK) and two peptides with one O-GlcNAc site (i.e., peptide #2: YSPT(O-GlcNAc)SPSK and peptide #3: TAPTS(O-GlcNAc)TIAPG), were used as to assess the capture performance of NO-3AMPS@SiO₂ nanoparticles. As a test run, the standard O-GlcNAc peptide mixture (100 femtomole of each) was submitted to nitro-oxide affinity nanospheres-based enrichment (with the procedure shown in Figure 2C). Enriched peptides were eluted by 0.1% FA. The bare silica and silica chemically modified with aminopropyl groups were regarded as the first generation of stationary phases previously.^34,37 We found that native SiO₂ showed little capture for peptides #2 and #3, but a small amount of peptide #1 (which might be ascribed to the higher hydrophilicity due to the presence of two O-GlcNAc groups) (Figure 4A). 3APMS@SiO₂ and Dimethyl-3APMS@SiO₂ nanoparticles did not capture any O-GlcNAc peptides (Figure 4B and 4C). However, NO-3AMPS@SiO₂ nanoparticles, which resulted from oxidation of the dimethylated aminopropyl groups, nicely captured all three O-GlcNAc peptides (Figure 4D), with good MS² fragments observed for each peptide (Figure 4E-4G). These results demonstrate excellent performance of the NO-3AMPS@SiO₂ nanospheres for the high affinity enrichment of O-GlcNAcylated peptides. Subsequently, parameters that might affect the performance of hydrophilic interaction were investigated to obtain the best enrichment efficiency for O-GlcNAc peptides. Firstly, the influence of oxidation conditions of Dimethyl-3APMS@SiO₂ material to NO-3AMPS@SiO₂ was evaluated. It appeared that the concentration of H₂O₂ and oxidation temperature and time have significant effects in the enrichment efficiency of target peptides (Figure S7A). With a lower concentration of H₂O₂ and lower oxidation temperature, or a shorter time, the oxidation of the tertiary amine was not complete, leading to low capture efficiency for certain O-GlcNAc peptides. The optimum performance was achieved by using the concentration (30% v/v) of H₂O₂ as the oxide reagent with heating at 70 °C for 12 h. The loading buffer composition (such as ACN percentage, ionic additive type, and concentration) was investigated. As shown in Figure S7B, the capture efficiency of O-GlcNAcylated peptides was increased with the increase of ACN percentage, suggesting a hydrophilic interaction mechanism. Since hydrophilic interaction-based enrichment is commonly performed by adding FA, TFA or NH₄FA in the buffer to increase the retention interaction of peptides with complex glycans onto the materials.^[18] Among them, FA turned out to be the best additive for enriching O-GlcNAc peptides, with little influence observed for different concentrations. The ratio between the amount of NO-3APMS@SiO₂ (ranging from 50 μg to 10 mg) and O-GlcNAc peptides was also investigated. As illustrated in Figure S7C, the optimum enrichment efficiency for all three O-GlcNAc peptides was obtained with an amount of enrichment material of 8 mg.

Figure 4.

The comparison of A) Native SiO₂ nanoparticles with B) 3AMPS@SiO₂, C) Dimethyl-3APMS@SiO₂ and D) NO-3AMPS@SiO₂ nanoparticles for the capture of three standard O-GlcNAc peptides, and the corresponding MS² spectra of the (E) peptide #1, (F) peptide #2, (G) peptide #3 from the original standard peptides and enriched peptides by NO-3AMPS@SiO₂.

To evaluate the enrichment reproducibility and preparation reproducibility, quadruplicates were carried out with the mixture of standard O-GlcNAcylated peptides. Three peptides were all detected in the four experiments. Run-to-run RSDs were 8.6% for peptide #1, 8.2% for peptide #2, and 1.5% for peptide #3, respectively. And batch-to-batch RSDs were 16.6% for peptide #1, 7.2% for peptide #2, and 9.1% for peptide #3, respectively (Figure S7D). These results suggest great reproducibility of the preparation of the nanospheres and the enrichment method developed.

The enrichment selectivity of NO-3APMS@SiO₂ was evaluated with a mixture of standard O-GlcNAc peptides and tryptic digest of BSA as a model system. The three standard O-GlcNAc peptides were successfully detected even with the presence of a high amount of BSA digest (1:10000 wt/wt)), indicating great selectivity of the nitro-oxide-grafted nanospheres. Additionally, its performance was validated with α-crystallin (a standard protein of low O-GlcNAc stoichiometry). Two O-GlcNAc sites, (located on three peptides AIPVS(O-GlcNAc)REEKPSSAPSS, TIPITREEKPAVT(O-GlcNAc)AAPKK and TIPITREEKPAVT(O-GlcNAc)AAPK) were identified with high confidence (with the EThcD spectra shown in Figure S8).

Application of nanospheres for native O-GlcNAc proteome analysis.

To further demonstrate the enrichment performance of the nitro-oxide-grafted nanospheres, a complex sample (i.e., lysates from PANC-1 cells) was analyzed. As peptides modified by oligosaccharides are more hydrophilic and abundant, they may also be retained during enrichment. To that end, several glycosidases were used to minimize the potential interfence from N- and O-linked glycans. O-GlcNAc peptides were then enriched with the nitro-oxide-grafted material and analyzed with tandem mass spectrometry in HCD product dependent EThcD (HCDpdEThcD) mode (Figure 5A).

Figure 5.

A) Schematic overview of the strategy integratingr nitro-oxide-grafted nanospheres-based O-GlcNAc enrichment and HCD product dependent EThcD (HCDpdEThcD)-based identification; B) The distribution of newly identified and reported O-GlcNAc proteins; C) The distribution of newly identified and reported O-GlcNAc sites; and D) Biological process of the identified O-GlcNAc proteins.

With such an approach, a total of 364 O-GlcNAcylated peptides from 267 proteins were identified (Table S1). Out of all the O-GlcNAc-containing peptides, 159 peptides with 230 O-GlcNAc sites were unambiguously identified (>75% localization probability). Unsurprisingly, a number of peptides were identified with more than one O-GlcNAc sites (Figure S9; Table S1). But strikingly, a pretty high percentage of peptides contaning single O-GlcNAc sites (~80%) were also enriched and mapped (Figure S9; Table S1), which could be ascribed to the excellent enrichment by using the new meterial developed. Very remarkably, up to 97% (222 out of 230) of all the O-GlcNAc sites were newly identified, given that only 8 sites on a few proteins were reported from PANC-1 cell line samples previously.^56,57 Amongst the proteins, 183 were newly identified as O-GlcNAc proteins and 197 O-GlcNAc sites were distinguished in comparison to all the human O-GlcNAc sites/proteins compiled in O-GlcNAcAtlas⁵⁸ (version O-GlcNAcAtlas_2.0; Figure 5B and 5C; Table S1). GO analysis shows that O-GlcNAcylated proteins are prominently enriched in nucleotide biosynthetic and metabolic processes, with a high portion of proteins involved in nuclear transport and mRNA metabolism (Figure 5D). The molecular functions of O-GlcNAc proteins appear to be highly involved in nuclear pore composition and kinase binding, among others (Figure S10A). Furthermore, sequence logo analysis shows a modification pattern of VVTgS/gTTAA (Figure S10B). In addition, protein intearction network indicates that O-GlcNAcylated proteins are highly clustered in focal adhesion, intracellular protein transport, cytoplasmic ribonucleoprotein granule, and kinase binding (Figure S10C). As the first large-scale study on the mapping of O-GlcNAc sites on proteins in PANC-1 cells, our dataset will help revelation of site-specific functions of O-GlcNAcylated proteins in pancreatic cancer (one of the most devastating cancers).

CONCLUSIONS

In summary, a novel type of nanosphere grafted with nitro-oxide (N→O) functional groups was developed to enrich native O-GlcNAc peptides. It provides strong hydrogen bonding affinity toward the O-H moiety on GlcNAc groups, enabling the highly efficient capture of O-GlcNAc peptides. After testing with standard peptides and proteins, the material was used to analyze O-GlcNAc proteins from PANC-1 cells. The simplified enrichment procedure also allows its application for direct analysis of other complex samples (e.g., tissue samples) in a site-specific manner. Of note, besides O-GlcNAc, other glycans in complex samples may also get enriched, thus prior treatment (e.g., by using PNGase F/O-glycocidases and HPLC fractionation) would facilitate the selective enrichment of O-GlcNAc peptides. Collectively, we provide a promising alternative for site-specific and facile O-GlcNAc proteomics.