Enhanced mass spectrometric mapping of the human GalNAc-type O-glycoproteome with SimpleCells

Abstract

Characterizing protein GalNAc-type O-glycosylation has long been a major challenge, and as a result, our understanding of this glycoproteome is particularly poor. Recently, we presented a novel strategy for high throughput identification of O-GalNAc glycosites using zinc finger nuclease gene-engineered "SimpleCell" lines producing homogeneous truncated O-glycosylation. Total lysates of cells were trypsinized and subjected to lectin affinity chromatography enrichment, followed by identification of GalNAc O-glycopeptides by nLC-MS/MS, with electron transfer dissociation employed to specify sites of O-glycosylation. Here, we demonstrate a substantial improvement in the SimpleCell strategy by including an additional stage of lectin affinity chromatography on secreted glycoproteins from culture media (secretome) and by incorporating pre-fractionation of affinity-enriched glycopeptides via IEF before nLC-MS/MS. We applied these improvements to three human SimpleCells studied previously, and each yielded a substantial increase in the number of O-glycoproteins and O-glycosites identified. We found that analysis of the secretome was an important independent factor for increasing identifications, suggesting that further substantial improvements can also be sought through analysis of subcellular organelle fractions. In addition, we uncovered a substantial nonoverlapping set of O-glycoproteins and O-glycosites using an alternative protease digestion (chymotrypsin). In total, the improvements led to identification of 259 glycoproteins, of which 152 (59%) were novel compared with our previous strategy using the same three cell lines. With respect to individual glycosites, we identified a total of 856 sites, of which 508 (59%) were novel compared with our previous strategy; this includes four new identifications of O-GalNAc attached to tyrosine. Furthermore, we uncovered ≈ 220 O-glycosites wherein the peptides were clearly identified, but the glycosites could not be unambiguously assigned to specific positions. The improved strategy should greatly facilitate high throughput characterization of the human GalNAc-type O-glycoproteome as well as be applicable to analysis of other O-glycoproteomes.

Fig. 1.

Outline of modified workflow. SimpleCell pellet is subjected to lysis, followed by protease and neuraminidase digestion; this is followed by O-glycopeptide enrichment by LWAC on a long column of immobilized VVA. In parallel, culture medium is treated with neuraminidase and then subjected to LWAC on a short column of VVA; this is followed by protease and neuraminidase digestion and LWAC enrichment on a long VVA column as for the cell lysate. After a pre-screening nLC-MS run to determine which LWAC fractions contained the most O-glycopeptides, these were divided into portions, with one part of each submitted directly to nLC-MS and the other parts pooled and submitted to IEF fractionation. IEF fractions (12) were then submitted to nLC-MS. Following computational processing of nLC-MS data as described, all spectral identifications were validated by inspection.

Fig. 2.

Venn diagrams showing uniqueness and overlap between data sets obtained using workflows incorporating LC/MS only versus IEF-LC/MS, with respect to number of O-glycoproteins (left) and O-glycosites (right) identified in equal portions of a Capan-1 SimpleCell lysate.

Fig. 3.

A, monitoring of fractions from K562 secretome preparation, visualized by Coomassie and lectin blot staining (left and right panels, respectively). Conditioned medium was dialyzed (lane 1) and neuraminidase-treated (lane 2). The medium was then filtered and diluted with LAC A buffer (lane 3) and passed twice over a short VVA column collecting a sample from the first (lane 4) and second flow-through (lane 5). The column was subsequently washed (lane 6) and eluted four times with 1 ml of 0.15 m GalNAc (lanes 7–10) and two times with 1.5 ml of 0.4 m GalNAc (lanes 11 and 12). B, Venn diagrams showing uniqueness and overlap between O-glycosite data sets obtained from secretomes versus cell lysates of the three SimpleCell lines studied, presented individually and as cumulative (nonredundant) totals.

Fig. 4.

ESI-Orbitrap-MS2 spectra of selected O-glycopeptides identified in this study. HCD-MS2 (A) and ETD-MS2 (B) of O-glycopeptide from agrin (residues 1288–1300; precursor m/z 626.3097, z = 3⁺) data consistent only with glycosylation of Thr-1294 and Ser-1295; HCD-MS2 (C) and ETD-MS2 (D) of O-glycopeptide from CD44 (residues 514–532; precursor m/z 763.0129, z = 3⁺) data consistent only with glycosylation of Thr-528; HCD-MS2 (E) and ETD-MS2 (F) of O-glycopeptide from GalNAc-T5 (residues 290–303; precursor m/z 704.3594, z = 3⁺) data consistent only with glycosylation of Ser-292, Thr-296, and Ser-300. Precursors selected are indicated by an asterisk in the ETD-MS2 spectra. Deduced glycosylated residues are denoted by Consortium for Functional Glycomics standard yellow square in the peptide sequence heading each panel.

Fig. 5.

ESI-ETD-Orbitrap-MS2 spectra of selected tyrosine O-glycopeptides identified in this study. ETD-MS2 spectra of O-glycopeptides from CD44 (residues 545–568; precursor m/z 871.1893, z = 5⁺) are consistent with glycosylation of Tyr-560 along with Ser-653, Thr-654, Thr-655, Thr-661, Ser-662, Thr-667 (A); NUCB1 (residues 32–53; precursor m/z 761.3487, z = 4⁺) is consistent with glycosylation of Tyr-50 along with Thr-39 and Thr-42 (B); ECM1 (residues 32–53; precursor m/z 721.3652, z = 4⁺) is consistent with glycosylation of Tyr-43 along with Ser-48 (C); and PRAP1 (residues 102–127; precursor m/z 858.6449, z = 4⁺) is consistent with glycosylation of Tyr-113 along with Ser-104 (D).

Fig. 6.

Venn diagrams showing uniqueness and overlap between data sets obtained from three SimpleCell lines studied with respect to identified O-glycoproteins (A) and O-glycosites (B). Cell line totals are presented individually versus each other (top) and as cumulative totals compared with data obtained previously by Steentoft et al. (5) (bottom).

Fig. 7.

Analysis of cumulative results to date (this paper and Steentoft et al. (5)) with respect to O-glycosite distributions. Histograms show O-glycosite distributions by number of Tn identified per number of proteins (A) and per number of glycopeptides (B).