The N-glycome of human embryonic stem cells

Abstract

Background: Complex carbohydrate structures, glycans, are essential components of glycoproteins, glycolipids, and proteoglycans. While individual glycan structures including the SSEA and Tra antigens are already used to define undifferentiated human embryonic stem cells (hESC), the whole spectrum of stem cell glycans has remained unknown. We undertook a global study of the asparagine-linked glycoprotein glycans (N-glycans) of hESC and their differentiated progeny using MALDI-TOF mass spectrometric and NMR spectroscopic profiling. Structural analyses were performed by specific glycosidase enzymes and mass spectrometric fragmentation analyses.

Results: The data demonstrated that hESC have a characteristic N-glycome which consists of both a constant part and a variable part that changes during hESC differentiation. hESC-associated N-glycans were downregulated and new structures emerged in the differentiated cells. Previously mouse embryonic stem cells have been associated with complex fucosylation by use of SSEA-1 antibody. In the present study we found that complex fucosylation was the most characteristic glycosylation feature also in undifferentiated hESC. The most abundant complex fucosylated structures were Lex and H type 2 antennae in sialylated complex-type N-glycans.

Conclusion: The N-glycan phenotype of hESC was shown to reflect their differentiation stage. During differentiation, hESC-associated N-glycan features were replaced by differentiated cell-associated structures. The results indicated that hESC differentiation stage can be determined by direct analysis of the N-glycan profile. These results provide the first overview of the N-glycan profile of hESC and form the basis for future strategies to target stem cell glycans.

Figure 1

Mass spectrometric neutral N-glycan profile of human embryonic stem cells (hESC). A. MALDI-TOF MS spectrum of neutral N-glycan fraction isolated from a hESC sample. B. Average of relative signal intensities from 40 most abundant neutral N-glycans of four finnish hESC lines (blue columns), embryoid bodies derived from the hESC lines (EB, red columns), and stage 3 differentiated cells (st.3, white columns). The columns indicate the mean abundance of each glycan signal (% of the total detected glycan signals). Error bars represent the standard error of mean. Proposed monosaccharide compositions are indicated on the x-axis and proposed structures for the major N-glycans are shown as schematic drawings. Gray circle/H: hexose, green circle: mannose, yellow circle: galactose, blue circle: glucose, gray square/N: N-acetylhexosamine, blue square: N-acetylglucosamine, red triangle/F: fucose/deoxyhexose, violet diamond/S: N-acetylneuraminic acid, light blue diamond/G: N-glycolylneuraminic acid. Asterisks indicate known polyhexose contamination that was not included in panel B.

Figure 2

Mass spectrometric acidic N-glycan profile of hESC. A. MALDI-TOF MS spectrum of acidic N-glycan fraction isolated from a hESC sample. B. Average of relative signal intensities from 40 most abundant sialylated N-glycans of the four hESC lines, EBs, and stage 3 differentiated cells. N-glycan signals carrying structural features associated with either hESC (F ≥ 2, red circles) or differentiated cells (N = H, gray circles) are indicated below panel B. See the legend of Figure 1 for details and monosaccharide symbols.

Figure 3

Sample-to-sample comparison of relative signal intensities of four major N-glycan signals. Sample-to-sample comparison of relative signal intensities of four major N-glycan signals in the whole dataset of four hESC lines (FES 21, FES 22, FES 29, and FES 30), embryoid bodies derived from them (EB), stage 3 differentiated cells (st.3), and two human fibroblast samples from the same cell line that was used as feeder cells in the propagation of hESC. A. Glycan signal S1H5N4F2 is characteristic to hESC. B. Glycan signal S1H5N5F1 is characteristic to differentiated cells. C. and D. The major N-glycan signals S1H5N4F1 (C.) and H9N2 (D.) are abundantly expressed in hESC and show variable expression in the differentiated cell types. Asterisks mark statistical significant differences between sample groups according to one-way ANOVA (p < 0.05).

Figure 4

Proposed structures for 20 most abundant N-glycan signals detected in the present study. Proposed structures for 20 most abundant N-glycan signals detected in the present study (13 neutral and 7 sialylated N-glycans) are based on combined results from MALDI-TOF mass spectrometry, proton NMR spectroscopy (NMR), and exoglycosidase analyses with α-mannosidase (αMan), β1,4- and β1,3-galactosidase (β4Gal), β-N-acetylglucosaminidase, specific α1,3/4- and α1,2-fucosidases (α3/4Fuc and α2Fuc), and broad-range sialidase (SA). Relative abundances in hESC and EB N-glycan profiles are indicated. Only the positive identifications in ¹H-NMR analyses and sensitivity to specific exoglycosidase digestions have been marked. Monosaccharide symbols are as in Figure 1. Where appropriate, glycosidic bonds have been indicated. Two simplifying assumptions have been made: i) in structures with H ≥ 3 and N ≥ 2, the proposed structures have been assigned a trimannosyl core structure, and ii) all fucosylated structures have been assigned a core fucose residue.

Figure 5

Analysis of major complex fucosylated N-glycans of hESC. A. The major sialylated N-glycan signal with complex fucosylation feature (S1H5N4F2) was subjected to exoglycosidase sequencing with linkage-specific fucosidases and broad-range sialidase. As outlined in the reaction scheme, the signal was shown to be an approximately 1:1 mixture of α1,2- and α1,3-/α1,4-fucosylated biantennary complex-type N-glycans with core fucosylation. B. and C. The acidic N-glycan fraction was permethylated and subjected to MS/MS fragmentation. Both S1H5N4F2 (B.) and S1H5N4F1 (C.) produced fragments that supported the structure assignments as indicated in the schematic drawings. D. The studied hESC lines were shown to express four of the known human fucosyltransferases. Minimal glycan acceptor and product specificities for the encoded enzymes are shown in the schematic drawing.

Figure 6

Lectin staining of hESC colonies grown on mouse feeder cell layers. A. Maackia amurensis agglutinin (MAA) that recognizes α2,3-sialylated glycans, preferably in type 2 LacNAc, stained hESC but not feeder cell surfaces. B. Pisum sativum agglutinin (PSA) that recognizes α-mannosylated glycans and core fucosylated N-glycans, stained only mouse feeder cell surfaces. C. Ulex europaeus agglutinin I (UEA-I) that recognizes α1,2-fucosylated glycans preferably within H type 2, stained the hESC colony. Mouse fibroblasts had complementary staining patterns with the lectins, indicating that their surface glycans are clearly different from hESC. D. Fluorescence-assisted cell sorting (FACS) diagrams of UEA-I selected hESC, showing that the majority of hESC were positive for cell surface UEA-I ligands.