The affinity of the FimH fimbrial adhesin is receptor-driven and quasi-independent of Escherichia coli pathotypes

Abstract

Type-1 fimbriae are important virulence factors for the establishment of Escherichia coli urinary tract infections. Bacterial adhesion to the high-mannosylated uroplakin Ia glycoprotein receptors of bladder epithelium is mediated by the FimH adhesin. Previous studies have attributed differences in mannose-sensitive adhesion phenotypes between faecal and uropathogenic E. coli to sequence variation in the FimH receptor-binding domain. We find that FimH variants from uropathogenic, faecal and enterohaemorrhagic isolates express the same specificities and affinities for high-mannose structures. The only exceptions are FimHs from O157 strains that carry a mutation (Asn135Lys) in the mannose-binding pocket that abolishes all binding. A high-mannose microarray shows that all substructures are bound by FimH and that the largest oligomannose is not necessarily the best binder. Affinity measurements demonstrate a strong preference towards oligomannosides exposing Manalpha1-3Man at their non-reducing end. Binding is further enhanced by the beta1-4-linkage to GlcNAc, where binding is 100-fold better than that of alpha-d-mannose. Manalpha1-3Manbeta1-4GlcNAc, a major oligosaccharide present in the urine of alpha-mannosidosis patients, thus constitutes a well-defined FimH epitope. Differences in affinities for high-mannose structures are at least 10-fold larger than differences in numbers of adherent bacteria between faecal and uropathogenic strains. Our results imply that the carbohydrate expression profile of targeted host tissues and of natural inhibitors in urine, such as Tamm-Horsfall protein, are stronger determinants of adhesion than FimH variation.

Fig. 1

Structure of the largest high-mannose glycan, oligomannose 9, indicating the D1, D2 and D3 arms.

Fig. 2

Fluorescence signal observed for high-mannose microarray binding by FimHrb_J96 at 0.5 mM. The immobilized oligosaccharides are linked at their reducing end to a thiol-terminated triethylene glycol (insert). Compounds 1-3 are oligomannose 9, 6 and 3, respectively, minus the chitobiose (GlcNAc)₂ (Fig. S1), β-linked like in high-mannose structures (Fig. 1). Manα1-2Manα1-2Man (4), Manα1-2Manα1-6Manα1-6Man (5), d-mannose (6) and Manα1-6Manα1-6Man (7) are all α-linked on the array (Fig. S1), as in natural high-mannose (Fig. 1). d-galactose (8) is β-linked (Fig. S1). Numbers above the bars are ΔG values (kcal mole⁻¹), calculated from the affinities measured by SPR (Table 2) for identical (4, 6 and 7) or similar (1-3) compounds. nb, no binding; nd, not determined.

Fig. 3

Sequence (A) and location (B) of the variations (raspberry red) in the FimH receptor-binding domains of studied E. coli strains. A bound butyl α-d-mannoside (red ball-and-stick model) indicates the location of the binding site (Bouckaert et al., 2005).

Fig. 4

Comparison of the binding profiles of FimH variants with high-mannose substructures. The Gibbs free energy of binding was calculated from the measured affinities of FimHrb_J96, FimHrb_F-18 and FimHrb_CI#4 for the trimannosides used for displacement of [³H]d-mannose (Table 1), and of FimHrb_EH485, FimHrb_EH12 and FimHrb_K514 for the oligomannosides used in the SPR assay (Table 2). The x-axis displays their common mannoside moieties.

Fig. 5

Predicted (AutoDock3) mode of binding of Manα1-3Manβ1-4GlcNAc (ball-and-stick representation) to FimH (PDB entry code 1TR7) [molecular surface presentation using GRASS (Nayal et al., 1999)]. The figure illustrates how the non-reducing mannose is buried in the cavity and the middle mannose and GlcNAc at the reducing end insert in the tyrosine gate between Tyr48 and Tyr137.