A single conformational transglutaminase 2 epitope contributed by three domains is critical for celiac antibody binding and effects

Abstract

The multifunctional, protein cross-linking transglutaminase 2 (TG2) is the main autoantigen in celiac disease, an autoimmune disorder with defined etiology. Glutamine-rich gliadin peptides from ingested cereals, after their deamidation by TG2, induce T-lymphocyte activation accompanied by autoantibody production against TG2 in 1-2% of the population. The pathogenic role and exact binding properties of these antibodies to TG2 are still unclear. Here we show that antibodies from different celiac patients target the same conformational TG2 epitope formed by spatially close amino acids of adjacent domains. Glu153 and 154 on the first alpha-helix of the core domain and Arg19 on first alpha-helix of the N-terminal domain determine the celiac epitope that is accessible both in the closed and open conformation of TG2 and dependent on the relative position of these helices. Met659 on the C-terminal domain also can cooperate in antibody binding. This composite epitope is disease-specific, recognized by antibodies derived from celiac tissues and associated with biological effects when passively transferred from celiac mothers into their newborns. These findings suggest that celiac antibodies are produced in a surface-specific way for which certain homology of the central glutamic acid residues of the TG2 epitope with deamidated gliadin peptides could be a structural basis. Monoclonal mouse antibodies with partially overlapping epitope specificity released celiac antibodies from patient tissues and antagonized their harmful effects in cell culture experiments. Such antibodies or similar specific competitors will be useful in further functional studies and in exploring whether interference with celiac antibody actions leads to therapeutic benefits.

Fig. 1.

Binding of celiac IgA serum antibodies (n = 7–28) in ELISA to wild-type (WT) and mutant TG2s. Dash indicates median. ***, p < 0.001, compared to WT. S4 contains the combined mutations D151N-E153Q-E154Q-E155Q-E158Q (ref. 21). Roman numerals indicate presence of domains (I–IV).

Fig. 2.

Three-dimensional view of TG2 in the closed conformation with the N-terminal β-sandwich shown in blue, catalytic (core) domain in red, β-barrel 1 in cyan and β-barrel 2 in pink. The bound GDP, fibronectin binding site (Asp94, Asp97) and catalytic triad (Cys277, His335, Asp358) are represented as ball-and-stick side chains. The amino acids of the putative celiac epitope with their lowest distances in angstroms (Å) are illustrated as surface representation (frame).

Fig. 3.

Binding of serum IgA from 58 celiac children (A) and from 18 celiac adults (B) to TG2 mutants in ELISA. Bound antibody concentrations were calculated from a calibrator curve (Fig. S2E) constructed from the concentration-dependent binding of mouse monoclonal anti-TG2 antibody TG100. Binding to wild-type (WT) is 100%. All samples were examined in duplicates. Dash indicates median. p < 0.0001 represents significant differences between groups by ANOVA test. 433 = R433S-E435S (irrelevant control mutant), R = R19S, E = E153S, M = M659S, RM = R19S-M659S, RE = R19S-E153S, EM = E153S-M659S, REM = R19S-E153S-M659S.

Fig. 4.

Binding to mutant TG2 proteins of serum IgA antibodies from patients with early stage celiac disease (•) without villous atrophy, manifest celiac disease (▴) with small bowel villous atrophy (n = 11) and from patients with other autoimmune diseases (n = 11) (○), measured by ELISA. The binding to wild-type (WT) was set to 100%. Dash indicates median; ***, p < 0.001; ns, not significant. R = R19S, RE = R19S-E153S, REM = R19S-E153S-M659S

Fig. 5.

Competition effect of mAb 885 on the recognition of wild-type TG2 in ELISA by antibodies from celiac disease patients (n = 6) and from nonceliac autoimmune patients with anti-TG2 antibodies (n = 6). (A) Remaining IgA binding in the presence of 18 μg/well mAb 885 if the binding without mAb 885 was 100%. Values represent means ± standard errors. (B) Comparison of the displacing effect of anti-TG2 mAbs 885, CUB7402, TG100, or H23 (18 μg/well) using one of the celiac serum samples. Representative values from three independently performed experiments.

Fig. 6.

Tissue-binding and biological effects of celiac antibodies mediated by the celiac TG2 epitope. (A) Epitope specificity of serum IgA (1, 2), of IgA eluted from celiac tissues (IgA placenta 3, 4) and of passively transferred maternal IgG from newborn serum measured using mutant TG2 proteins in ELISA. R = R19S, E = E153S, M = M659S, RE = R19S-E153S, REM = R19S-E153S-M659S, 433 = R433S-E435S. (B) Celiac IgA (green) in vivo deposited in the placenta and on the surface of chorionic villi (arrows) merge to yellow after incubation of the tissue sections with CUB7402 anti-TG2 mAb and double-stained by anti-mouse antibodies (red). After incubation with mAb 885 the IgA signal is no longer visible and anti-TG2 IgA is detected in the buffer by ELISA. No IgA was released by incubation with isotype control mAb (IgG1) or CUB. (C) IgA from celiac (CD) patients (n = 5), autoimmune patients with nonceliac TG2 antibodies (n = 6) and biopsied antibody-negative controls (n = 3) were administered alone or together with mAb 885 to normal HUVECs in matrigel and vessel formation was measured. Median lengths of endothelial tubules ± SD are shown compared to wells without antibodies set to 100%. The term ns represents not significant; ***, p < 0.001. (D) Morphology of HUVECs in culture from a newborn with prenatally bound maternal celiac IgG (red) on their surface (Top and Middle) compared to HUVECs prepared from newborn with celiac mother on diet and negative for antibodies (Bottom). Bars = 50 um.