IgA displays site- and subclass-specific glycoform differences despite equal glycoenzyme expression

Abstract

Background: Glycosylation is an important posttranslational modification of proteins and in most cases indispensable for proper protein function. Like most soluble proteins, IgA, the second most prevalent antibody in human serum, contains several N- and O-glycosylation sites. While for IgG the impact of Fc glycosylation on effector functions and inflammatory potential has been studied intensively, only little is known for IgA. In addition, only glimpses exist regarding the regulation of IgA glycosylation. We have previously shown that IgA1 and IgA2 differ functionally and also show differences in their glycosylation pattern. The more pro-inflammatory IgA2 which is linked to autoimmune diseases displays decreased sialylation, galactosylation, fucosylation and bisection as compared to IgA1. In the present study, we aimed to investigate these differences in glycosylation in detail and to explore the mechanisms underlying them.

Methods: IgA1 and IgA2 was isolated from serum of 12 healthy donors. Site specific glycosylation was analyzed by mass spectrometry. In addition, human bone marrow plasma cells were investigated using single cell mRNA sequencing, flow cytometry and ELISpot.

Results: We found that certain glycoforms greatly differ in their abundance between IgA1 and IgA2 while others are equally abundant. Overall, the IgA2 glycans displayed a more immature phenotype with a higher prevalence of oligomannose and fewer fully processed glycans. Of note, these differences can't be explained by differences in the glycosylation enzyme machinery as mRNA sequencing and flow cytometry analysis showed equal enzyme expression in IgA1 and IgA2 producing plasma cells. ELISpot analysis suggested a slightly increased antibody production rate in IgA2 producing plasma cells which might contribute to its lower glycan processing rates. But this difference was only minor, suggesting that further factors such as steric accessibility determine glycan processing. This is supported by the fact that glycans at different positions on the same IgA chain differ dramatically in fucosylation, sialylation and bisection.

Conclusion: In summary, our detailed overview of IgA1 and IgA2 glycosylation shows a class, subclass, and site-specific glycosylation fingerprint, most likely due to structural differences of the protein backbones.

Keywords: Glycoenzyme; Glycoform; Glycosylation; IgA; Plasma cells.

Fig. 1

Schematic illustration of IgA N- and O-glycosylation sites

Fig. 2

Major differences between glycoforms at distinct glycosylation sites of IgA and between Ig (sub)classes. (A) Pie charts of the relative abundances of glycoforms decorating IgA1 and IgA2 glycosylation sites (IgG for reference). Shown are the means of 12 donors. N340/327 C and N340/327Y refer to the truncated peptide, lacking terminal tyrosine, and the full peptide, respectively. ^# For the IgG glycosylation site, the conventional amino acid number according to literature has been used (e.g [39]), while glycosylation sites of IgA1 and 2 are indicated according to UniProt numbering (P01876-1 and P01877-1). The presence of fucose is depicted as inner red circles; the degree of sialylation is depicted as outer pink circles. (B) Simplified presentation of the abundance of glycosylation traits for the different glycosylation sites. For sites conserved in both IgA subclasses, the average value of IgA1 and IgA2 was used. (C) Simplified presentation of the abundance of glycosylation traits for the different Ig (sub)classes. For better comparability, the averages of the conserved glycosylation sites were used for IgA1 and IgA2

Fig. 3

Distinct glycoforms differ in their abundance between IgA1 and IgA2. Percentage and structure of the 5 glycoforms that differ most in their abundance between IgA1 and IgA2 at the conserved glycosylation sites. Significance was tested with paired t-test. *** = p < 0.001; glycan code: H = hexose (mannose or galactose); N = N-acetylglucosamine; F = fucose; S = sialic acid

Fig. 4

IgA1 and IgA2 producing plasma cells show equal glycosyltransferase amounts, but different antibody release rates. (A) Single cell RNA sequencing analysis UMAP representation of plasma cells sorted from bone marrow of healthy donors. Shown are combined data from 3 donors. Heavy chain subclass information was extracted from VDJ sequencing annotations. Here, only IGHA1/2 are highlighted. Other IGHC isotypes can be seen in Supplementary Fig. 3. (B) Dot plot for expression of glycosyltransferases in IgHA1 and IgHA2 expressing plasma cells. Dot colors represent mean expression of the genes in each cell group and dot sizes indicate the percentage of cells expressing the respective genes. (C) Schematic overview of the contribution of selected enzymes to glycan processing. MAN1A2 is involved in trimming of the outer-arm mannose residues, B4GALT1 and ST6GAL1 add galactose and terminal sialic acid residues, respectively. (D) Flow cytometry analysis of IgA1 and IgA2 producing bone marrow plasma cells. Shown is the mean fluorescence intensity (MFI) for surface 2,6-sialic acid (measured by binding of sambuccus nigra agglutinine) and intracellular MAN1A2 and B4GALT1 (measured by binding of specific antibodies). Data are normalized on the MFI of IgA1 producing plasma cells. Every dot represents one donor. N = 5–6. E) ELISpot analysis of IgA1 and IgA2 producing bone marrow plasma cells. Shown is the median and the 75 percentile of the spot size. Significances were tested with Wilcoxon rank test. *– p < 0.05; ns– not significant. (F) Representative ELISpot images. Scale bar = 2 mm