Presentation Mode of Glycans Affect Recognition of Human Serum anti-Neu5Gc IgG Antibodies

Abstract

Recognition of carbohydrates by antibodies can be affected by antigen composition and density. This had been investigated in a variety of controllable multivalent systems using synthetic carbohydrate antigens, yet such effects on anticarbohydrate antibodies in circulating human serum have not been fully addressed thus far. All humans develop a polyclonal and diverse response against carbohydrates containing a nonhuman sialic acid form, N-glycolylneuraminic acid (Neu5Gc). This red meat-derived monosaccharide is incorporated into a diverse collection of human glycans resulting in circulating anti-Neu5Gc antibodies in human sera. Such antibodies can cause exacerbation of diseases mediated by chronic inflammation such as cancer and atherosclerosis. We aimed to evaluate how different presentation modes of Neu5Gc-glycans can affect the detection of anti-Neu5Gc IgGs in human serum. Here, we compare serum IgG recognition of Neu5Gc-containing glycoproteins, glycopeptides, and synthetic glycans. First, Neu5Gc-positive or Neu5Gc-deficient mouse strains were used to generate glycopeptides from serum glycoproteins. Then we developed a reproducible ELISA to screen human sera against Neu5Gc-positive glycopeptides for detection of human serum anti-Neu5Gc IgGs. Finally, we evaluated ELISA screens against glycopeptides in comparison with glycoproteins, as well as against elaborated arrays displaying synthetic Neu5Gc-glycans. Our results demonstrate that the presentation mode and diversity of Neu5Gc-glycans are critical for detection of the full collection of human serum anti-Neu5Gc IgGs.

Figure 1

Generating glycopeptides (GP) from mouse serum. Schematic diagram of method. Mouse serum was collected, then digested with Pronase and GPs purified.

Figure 2

Characterization of Pronase digest. Mouse serum with total 80 mg protein was used per digestion. (A) 12.5% gel was loaded with serum digest at 2.5 μg/lane (lanes 1,2,4,6,8,10), 5 μg/lane (lanes 3,5,7,9,11), or 6 μg of Pronase (lane 12), then daily protein digestion was evaluated by silver staining (protein quantified by BCA assay). (B) Total sialic acid content in Pronase digest (15 μL) was analyzed daily by DMB-HPLC, showing no significant loss during digestion (representative of at least two independent experiments; mean ± SEM).

Figure 3

Characterization of Sia in glycopeptides. (A) Pronase digest was separated by size though Amicon 3 kDa filters, and Sia content quantified by DMB-HPLC (top fraction contains most GPs; bottom fraction runthrough, RT, shows minimal Sia loss). (B) Qualitative characterization of Sia in GPs from WT mouse sera by ELISA (coated at 1 μg/well protein) confirmed presence of Neu5Gc (by anti-Neu5Gc-IgY), and demonstrated Siaα2-6 (SNA) and Siaα2-3 (MAL-II) GP-glycoconjugates. Antibody/lectins signal was dramatically reduced after mild periodate treatment (that cleaves C8–C9 in side chain of Sia), demonstrating their Sia specific-recognition (representative of two independent experiments; mean ± SD of triplicates).

Figure 4

Evaluating reproducibility of human serum IgG reactivity against Neu5Gc^pos-GPs by ELISA. (A,B) 96-well Costar plates were coated with Neu5Gc^pos-GPs in several concentrations based on protein (A; BCA) or Sia (B; DMB-HPLC) content, then binding of human serum IgG (SV7) was evaluated by ELISA. (C,D) 96-well Costar plates were coated with optimal Neu5Gc^pos-GPs 0.25 μg protein/well (C) or 150 pmol Sia/well (D) from four different GPs batches, then reproducibility of recognition by human serum IgG was evaluated by ELISA with several samples (serum diluted 1/100). Batch-to-batch variability was minimal when plates were coated according to Sia content (representative of two independent experiments; mean ± SD of triplicates).

Figure 5

Evaluating specific detection of human serum anti-Neu5Gc IgGs by ELISA against Neu5Gc^pos-GPs. (A) 96-well Costar plates were coated with Neu5Gc^pos-GPs, then binding of human serum IgG (diluted 1/100) was evaluated by ELISA, either as is (0), or after preincubation with competing Neu5Gc^pos-GPs (0.03−0.25 mM Sia). This demonstrated specific anti-Neu5Gc IgG reactivity in all samples, and signal was maximally inhibited already at 0.03 mM Sia. (B) 28 human serum samples were examined (at 1/100 dilution) by ELISA against coated Neu5Gc^pos-GPs, either as is (0) or after preincubation with Cmah-KO mouse sera (1/4000) or Neu5Gc^pos-GPs (0.03 mM and 0.06 mM). The Cmah-KO sera did not absorb any human serum reactivity, while Neu5Gc^pos-GPs at both concentrations inhibited most reactivity (One-Way ANOVA). (C) Anti-Neu5Gc IgGs specific reactivity was calculated by deducting the signal obtained after preincubation with 0.03 mM Neu5Gc^pos-GPs from the signal with no inhibition (glycopeptides ELISA inhibition assay; GP-EIA). (D) Comparing the human serum reactivity against Neu5Gc^pos-GPs as is (GP) versus the GP-EIA signal showed strong correlation (Pearson r = 0.977). (Representative of at least two independent experiments for A-D; mean ± SD of triplicates.)

Figure 6

Human sera anti-Neu5Gc IgGs detection against mouse sera sialo-glycoproteins versus sialo-glycopeptides. (A) 96-well Costar plates were coated either with Neu5Gc^pos-GPs (150 pmol/well) or undigested WT mouse serum (1 μg/well; selected to be negative for mouse-anti-human IgG reactivity), then binding of 39 human sera samples was evaluated by ELISA, detected by HRP-anti-human IgG (Representative of two independent experiments; mean ± SD of duplicates). (B) Comparing human sera IgG reactivity against Neu5Gc^pos-GP or undigested WT mouse serum shows no correlation (Pearson r = 0.036). (C) Anti-Neu5Gc IgG reactivity in 11 human sera samples (expected to have high/medium/low reactivity) were evaluated either by EIA or GP-EIA. In EIA assay, undigested WT mouse serum (1 μg/well; selected to be negative for mouse-anti-human IgG reactivity) was coated to 96-well plates, then examined with human serum samples (diluted 1/100) that were preincubated with Cmah-KO mouse sera (1/4000; to absorb human anti-mouse reactivity). In GP-EIA assay, the plates were coated with Neu5Gc^pos-GP, and human sera preincubated with Neu5Gc^pos-GP (0.03 mM). Representative of at least two independent experiments; mean ± SD of duplicates). (D) Comparing specific anti-Neu5Gc IgG signal obtained with EIA versus GP-EIA showed lack of correlation (Pearson r = 0.034).

Figure 7

Comparison of human serum anti-Neu5Gc IgG reactivity analyzed by sialoglycan microarray versus EIA and GP-EIA. (A) Sialoglycan microarray analysis of sera from healthy donors (diluted 1/100). Relative fluorescent units (RFU) are presented as a heatmap. (B−D) Correlation analysis of anti-Neu5Gc IgG reactivity obtained by GP-EIA and EIA versus sialoglycan microarray (average of all Neu5Gc-glycans per serum sample). (C) No correlation between anti-Neu5Gc IgG antibody reactivity between GP-EIA and sialoglycan microarray (Pearson r = 0.098). (D) Strong correlation observed between anti-Neu5Gc IgG antibody levels between EIA and sialoglycan microarray (Pearson r = 0.905).