Integrative Proteomics and N-Glycoproteomics Analyses of Rheumatoid Arthritis Synovium Reveal Immune-Associated Glycopeptides

Abstract

Rheumatoid arthritis (RA) is a typical autoimmune disease characterized by synovial inflammation, synovial tissue hyperplasia, and destruction of bone and cartilage. Protein glycosylation plays key roles in the pathogenesis of RA but in-depth glycoproteomics analysis of synovial tissues is still lacking. Here, by using a strategy to quantify intact N-glycopeptides, we identified 1260 intact N-glycopeptides from 481 N-glycosites on 334 glycoproteins in RA synovium. Bioinformatics analysis revealed that the hyper-glycosylated proteins in RA were closely linked to immune responses. By using DNASTAR software, we identified 20 N-glycopeptides whose prototype peptides were highly immunogenic. We next calculated the enrichment scores of nine types of immune cells using specific gene sets from public single-cell transcriptomics data of RA and revealed that the N-glycosylation levels at some sites, such as IGSF10_N2147, MOXD2P_N404, and PTCH2_N812, were significantly correlated with the enrichment scores of certain immune cell types. Furthermore, we showed that aberrant N-glycosylation in the RA synovium was related to increased expression of glycosylation enzymes. Collectively, this work presents, for the first time, the N-glycoproteome of RA synovium and describes immune-associated glycosylation, providing novel insights into RA pathogenesis.

Keywords: N-glycoproteomics; immune cell infiltration; protein glycosylation; rheumatoid arthritis; synovium.

Graphical abstract

Fig. 1

Quantitative proteomics analysis of OA and RA synovial tissues.A, the workflow of quantitative proteomics and N-glycoproteomics analyses of human synovium. B, Pearson correlation analysis of the proteomics technical duplicates. C, OPLS-DA score plots of proteomics data of RA and OA. D, volcano plot showing 427 upregulated and 241 downregulated proteins in RA (median ratio (RA/OA) >1.5 or <0.67, and Student’s t test, p < 0.05). E, GSEA preranked analysis of identified proteins in RA and OA. NES represents the normalized enrichment score. GSEA, gene set enrichment analysis; OA, osteoarthritis; OPLS-DA, orthogonal partial least squares discriminant analysis; RA, rheumatoid arthritis.

Fig. 2

Characteristics of intact N-glycopeptides in RA and OA synovium.A–C, Venn diagram showing identified N-glycopeptides (A), N-glycosites (B), N-glycoproteins, and (C) in RA and OA samples. D, distribution of all identified N-glycopeptides with N-X-T/S/C tripeptide sequons is shown as the means ± SDs (n = 4 cases per OA or RA group). E, pie charts depict the percentage of N-glycoproteins with the identified number of N-glycosites per protein (upper) and the percentage of N-glycosites with the identified number of N-glycans per glycosite in OA and RA samples (below). F, percentage of the six types of glycans in all N-glycosites. G, number of N-glycosites with different glycan types identified in the OA and RA groups. H, UpSet plot represents the frequency of glycan pairs existing at the same N-glycosite. OA, osteoarthritis; OPLS-DA, orthogonal partial least squares discriminant analysis; RA, rheumatoid arthritis.

Fig. 3

Differences in the N-glycoproteome of RA and OA synovium.A, cumulative number of N-glycopeptides identified in OA (n = 4) and RA (n = 4) samples. B, OPLS-DA score plots of N-glycoproteomics data of OA and RA synovium. C, volcano plot showing 67 upregulated and nine downregulated N-glycopeptides in RA (median ratio (RA/OA) >1.5 or <0.67 and Student’s t test, p < 0.05). D, scatter plot depicting the differences in N-glycopeptides and their corresponding proteins between OA and RA. The blue and green dots represent the median ratio (RA/OA) of each N-glycopeptide greater than 1.5 and less than 0.67, respectively. The dashed lines represent the median fold change of proteins at 1.2 and the median fold change of N-glycopeptides at 1.5. E, scatter plot shows the correlation of N-glycopeptides with fold changes greater than 1.5 in (C) before and after normalization to glycoproteins. The blue and green dots represent the normalized ratio (RA/OA) of each N-glycopeptide greater than 1.2 and less than 0.83, respectively. The gray solid line represents the line with slope 1. F, box plot shows the differences in N-glycopeptides in (C) without proteins identified in proteomics (Student’s t test, ∗p < 0.05; ∗∗p < 0.01). G, comparison of the distribution of N-glycans on the significantly changed N-glycopeptides. H, comparison of N-glycoproteins related to different clusters based on cell compartment (CC), biological process (BP), and molecular function (MF). OA, osteoarthritis; RA, rheumatoid arthritis.

Fig. 4

RA- and OA-specific intact N-glycopeptides in synovium.A, heatmap showing the 29 RA- and 1 OA-specific N-glycopeptides with glycan compositions in synovium. The value for each N-glycopeptide is the adjusted intensity with a row-scaled Z score. B, GO analysis of the 16 glycoproteins derived from 29 RA-specific N-glycopeptides in terms of cellular component (CC), biological process (BP), and molecular function (MF). The top ten biological processes and cellular components are shown. C, Sankey diagram showing the 11 glycoproteins in the top ten biological processes of (B) and their interlinkages. D, heatmap representing the appearance of different glycan compositions at N-glycosites of the 11 glycoproteins in (C). OA, osteoarthritis; RA, rheumatoid arthritis.

Fig. 5

Antigenic analysis of N-glycopeptides.A, a workflow of antigenicity prediction of 146 prototype peptides derived from differentially expressed N-glycopeptides between OA and RA by using DNASTAR software. B, 3D scatter plot showing the hydrophilicity, flexible regions, β-turns, surface probability, and antigenic index of upregulated and downregulated N-glycopeptides in the RA and OA groups. C, heatmap showing the abundances of 20 N-glycopeptides whose prototype peptides were predicted to have good antigenicity. The protein changes are shown on the right of the figure (Student’s t test, ∗p < 0.05; ∗∗p < 0.01). OA, osteoarthritis; RA, rheumatoid arthritis.

Fig. 6

Integrative multiomics analyses to identify immune-associated N-glycopeptides in RA.A, comparison of estimated immune scores between the RA and OA groups using the ESTIMATE algorithm based on the proteomics data. The box represents the median (thick line) and the quartiles (line). B, heatmap for row-scaled enrichment scores of nine immune cell types in synovial tissues. The enrichment scores were calculated by ssGSEA using specific gene sets for these immune cell types. Rows represent infiltrating immune cells and columns represent samples (Wilcoxon rank-sum test, ∗p < 0.05). C, representative differentially expressed N-glycopeptides show a significant linear correlation with the infiltration of immune cells. R represents the regression coefficient and the p value represents the significance of the linear correlation. D, H&E staining of synovial tissue sections and immunofluorescence staining of CD3, CD4, CD8, CD19, CD56, CD68, CD83, and the percentage of CD3, CD4, CD8, CD19, CD56, and CD68, CD83-positive cells in synovial tissues from OA (n = 4) and RA (n = 4) patients. The statistics of immune-positive cells were calculated using Fiji/ImageJ. Each value represents the average obtained from four independent experiments. Data are presented as the mean ± SD. (Student’s t test, ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001). H&E staining of synovial tissue sections was used to observe biopsy pathological morphology. ESTIMATE, Estimation of STromal and Immune cells in MAlignant Tumor tissues using Expression data; OA, osteoarthritis; RA, rheumatoid arthritis; ssGSEA, single sample gene set enrichment analysis.

Fig. 7

Correlation analysis of N-glycopeptides and glycosylation enzymes.A, heatmap showing the expression of glycosylation-associated transferase, transporter, glucosidase, and hydrolase in RA and OA synovial tissues based on the proteomics data (Student’s t test, ∗p < 0.05; ∗∗p < 0.01). B, bubble chart shows differentially expressed N-glycopeptides and glycosylation enzymes with significant linear correlation (Spearman's rank correlation coefficient, p < 0.05). C, representative differentially expressed N-glycopeptides linearly correlated with glycosylation enzymes. R represents the regression coefficient and the p value represents the significance of the linear correlation. OA, osteoarthritis; RA, rheumatoid arthritis.