Cholera intoxication of human enteroids reveals interplay between decoy and functional glycoconjugate ligands

Abstract

Prior research on cholera toxin (CT) binding and intoxication has relied on human colonic cancer derived epithelial cells. While these transformed cell lines have been beneficial, they neither derive from small intestine where intoxication occurs, nor represent the diversity of small intestinal epithelial cells (SI-ECs) and variation in glycoconjugate expression among individuals. Here, we used human enteroids, derived from jejunal biopsies of multipledonors to study CT binding and intoxication of human non-transformed SI-ECs. We modulated surface expression of glycosphingolipids, glycoproteins and specific glycans to distinguish the role of each glycan/glycoconjugate. Cholera-toxin-subunit-B (CTB) mutants were generated to decipher the preference of each glycoconjugate to different binding sites and the correlation between CT binding and intoxication. Human enteroids contain trace amounts of GM1, but other glycosphingolipids may be contributing to CT intoxication. We discovered that inhibition of either fucosylation or O-glycosylation sensitize enteroids to CT-intoxication. This can either be a consequence of the removal of fucosylated "decoy-like-ligands" binding to CTB's non-canonical site and/or increase in the availability of Gal/GalNAc-terminating glycoconjugates binding to the canonical site. Furthermore, simultaneous inhibition of fucosylation and O-glycosylation increased the availability of additional Gal/GalNAc-terminating glycoconjugates but counteracted the sensitization in CT intoxication caused by inhibiting O-glycosylation because of reduction in fucose. This implies a dual role of fucose as a functional glycan and a decoy, the interplay of which influences CT binding and intoxication. Finally, while the results were similar for enteroids from different donors, they were not identical, pointing to a role for human genetic variation in determining sensitivity to CT.

Keywords: O-glycosylation; cholera toxin; decoy like ligands; enteroid monolayers; fucosylation.

Fig. 1

Role of glycosphingolipids in CT intoxication of human enteroids. The CT induced ion efflux (intoxication) is determined by estimating the change in short circuit current from transmembrane voltage and resistance measurements in enteroids cultured with or without 50 μM NB-DGJ, that inhibits glycosphingolipid (including GM1) biosynthesis, for 72 h prior to CT challenge. a) Rate of CT intoxication in enteroid monolayers from jejunal biopsies from four different donors. The data shown is a representative of n ≥ 4 independent experiments of each enteroid donor. b) Pooled data of normalized change in short circuit current (Isc) i.e. Isc normalized to when no NB-DGJ is present, 5 h after apical treatment with CT. Data is reported as mean ± S.E. (n ≥ 4 biological replicates). n.s (p > 0.05).

Fig. 2

Role of fucosylation in CT intoxication of human enteroids. a) Relative fucosylation-dependent binding of lectin UEA-1 and AAL is reduced in the presence of 200 μM 2F-Fuc, fucosylation inhibitor. To statistically analyze the multiple biological replicates, the relative geometric mean fluorescence intensity (rGMFI) is determined by normalizing the GMFI for each condition to the GMFI when no inhibitor is present in the same experiment. Data is reported as mean ± S.E. (n ≥ 6 biological replicates). b) The CT induced ion efflux (intoxication) is determined by estimating the change in short circuit current from transmembrane voltage and resistance measurements in enteroids treated with or without 200 μM 2F-Fuc. The data shown is a representative of n ≥ 7 independent experiments. c) Normalized change in short circuit current (Isc) after 5 h of apical treatment with CT. Inhibition of fucosylation increases the CT intoxication levels. Data is reported as mean ± S.E. (n ≥ 7 biological replicates). d) Gal/GalNAc dependent PNA binding to human enteroids is significantly increased in the presence of 2F-Fuc. Data is reported as mean ± S.E. (n ≥ 5 biological replicates). p value: ≥0.05 (ns), <0.05 (^*), <0.01(^**), <0.001(^***) and <0.0001(^****).

Fig. 3

Binding of site-specific CTB mutants to a) plate-immobilized GM1-HSA (0.05 μg mL⁻¹) as assessed by ELISA, b) plate-immobilized tri-Lewis-x-APE-HSA (1 μg mL⁻¹) as assessed by ELISA, c) murine splenic T cells from wild-type mice (GM1+/−) and B4galnt1 knockout mice (GM1−/−) as assessed by flow cytometry, and d) human granulocytes as assessed by flow cytometry. a, b) One representative of three independent experiments is shown. c, d) Data is reported as mean ± S.E. (n ≥ 3 biological replicates) of geometric mean fluorescence intensity (GMFI). p value: ≥0.05 (ns), <0.05 (^*), <0.01(^**), <0.001(^***), and <0.0001(^****).

Fig. 4

a) Binding of site-specific CTB mutants to human enteroids. Data is reported as mean ± S.E. (n ≥ 4 biological replicates) of GMFIs measured by flow cytometry. b) Binding of CTB mutants to human enteroids with and without inhibitors, 50 μM NB-DGJ (inhibition of glycosphingolipid synthesis) and 200 μM 2F-Fuc (inhibition of fucosylation). Data is reported as mean ± S.E. (n ≥ 5 biological replicates) of relative GMFI percentage (%rGMFI) where GMFI is normalized such that the measured GMFI with each mutant when no inhibitor is present is considered 100%. p value: ≥ 0.05 (ns), <0.05 (^*), <0.01(^**), <0.001(^***), and <0.0001(^****).

Fig. 5

CT binding to and intoxication of human enteroids following inhibition of O-linked (2 mM benzyl-α-GalNAc) or N-linked (10 μM kifunensine) glycosylation. a) Rate of CT-mediated intoxication of enteroid monolayers. The data shown is a representative of n ≥ 4 independent experiments. b) Pooled data of normalized change in short circuit current (Isc) i.e. Isc normalized to when no inhibitor is present, 5 h after apical treatment with CT. Data is reported as mean ± S.E. (n ≥ 4 biological replicates). Binding of c) CTB mutants, d) PNA (galactose binding), e) UEA-1 (fucose binding) to human enteroids with and without benzyl-α-GalNAc and kifunensine measured via flow cytometry. Data is reported as mean ± S.E. (n ≥ 4 biological replicates) of relative GMFI percentage (%rGMFI) where GMFI is normalized such that the measured GMFI with each mutant when no inhibitor is present is considered as 100%. p value: ≥ 0.05 (ns), <0.05 (^*), <0.01(^**), <0.001(^***) and <0.0001(^****).

Fig. 6

Role of glycosphingolipids and fucosylation in the absence of “decoy-like-ligands.” a) CT intoxication, Isc normalized to when only benzyl-α-GalNAc is present, 5 h after apical treatment with CT. Data is reported as mean ± S.E. (n = 3 biological replicates). b) Binding of CTB mutants to human enteroids measured via flow cytometry. Data is reported as mean ± S.E. (n ≥ 4 biological replicates) of relative GMFI percentage (%rGMFI) where GMFI is normalized such that the measured GMFI with each mutant when only Benzyl-α-GalNAc is present is considered as 100%. p value: ≥ 0.05 (ns), <0.05 (^*), and <0.0001(^****).

Fig. 7

A schematic displaying an intricate interplay between different glycoconjugates (glycosphingolipids and glycoproteins) in CT binding and intoxication. The figure was created with biorender.com.