All C. jejuni strains tested in this study showed remarkable similarity for the general types of glycan structures that were recognised. Looking globally at the total array, C. jejuni behaves as a species with little variation, each strain bound to both α and β galactose, terminal and subterminal fucosylated structures and to a subset of glycoaminoglycans at all conditions tested. All strains also exhibited binding to a broader range of glycans when placed under environmental stress. Only chitin, a common insect and crustacean glycan, showed major differences when viewed from a global perspective, with one strain, C. jejuni 11168, failing to recognise any chitin molecule. No major difference was observed between C. jejuni strains isolated from different hosts.
The possibility of galactose and fucose being involved in the persistent colonisation of C. jejuni[3, 4] is supported by the interactions observed in this study. All twelve strains, whether isolated from avian or clinical sources, bound broadly to uncapped galactose structures and fucosylated structures. These results were confirmed by inhibition of adherence to cells blocked by competing C. jejuni adherence with UEA-I.
Of the strains tested only one chicken isolate (331) and one clinical isolate (520) showed variability in the galactose structures bound. Of interest is the broad specificity of all the C. jejuni strains for galactose and fucosylated structures. Only strain, C. jejuni 520, showed binding differences based on linkage specificity with Galβ1-3GalNAc (asialo-GM1 1 F) and terminal α-1-4 linked di-galactose (1 K) glycan structures not being recognised.
The fact that C. jejuni recognises a broad range of both α and β linked galactose may offer some explanation for such a broad host range, as might the lack of specificity for linkage and position of fucose in fucosylated structures. α-linked galactose are not common in humans but are common in many other mammals and avian species [13–17]. Some strains of C. jejuni are known to produce the P-antigen, a terminal α-linked galactose, as a part of their LOS structure to mimic the glycans of potential avian and non-human mammalian hosts [13, 18]. β-linked galactose structures are common to all animals known to be infected with C. jejuni. The fact that C. jejuni recognises both α and β linked galactose indicates either a broad specificity galactose binding lectin or two or more lectins with restricted specificity. As binding to these different galactose structures is not preferential under any condition tested, it is likely that a single yet to be identified broad specificity glactose binding lectin is expressed by C. jejuni.
Fucose is a known chemoattractant of C. jejuni but the binding observed in our glycan array analysis is unlikely to be related to the periplasmic receptors for chemotaxis. Fucose surface expression in humans is dependent on a range of fucosyltransferases that can be differentially expressed both throughout tissues and between individuals resulting in differential fucosylation between tissue types or differential fucosylation of the same tissue types when comparing two nonrelated individuals. As C. jejuni has no preference for linkage or location it is likely that either the same protein that recognises galactose is binding fucosylated structures but ignoring the presence of fucose or that C. jejuni has a broad specificity fucose binding lectin.
Binding to N-acetylglucosamine structures was differential between strains with three strains not recognising GlcNAc structures at all (C. jejuni 11168, 019 and 108). Typically among strains that did recognise GlcNAc structures the longer repeats were preferred. Only C. jejuni 331 (under all conditions), 81116 (under all conditions) and 351 (under environmental conditions) recognised the short repeats. Chitin a common glycoconjugate found in insects and crustaceans is comprised of repeating GlcNAc residues. It is possible that C. jejuni strains that recognise GlcNAc structures may use insects as vectors as described by Hald et al.[19], or that strains with GlcNAc recognition can better infect crustaceans to survive and propagate in fresh water ponds and streams [19, 20]. Chitin recognition may therefore be important for environmental survival and spread, also offering advantages for re-infection of more preferred avian or mammalian hosts.
In line with previously reported data [3], mannose was recognised more often after environmental stress by most of the C. jejuni strains tested. C. jejuni 331 and 81116 were the only strains to recognise a wide variety of mannose structures under all growth/maintenance conditions. Several other strains, more common to the chicken isolates tested (Human isolate: C. jejuni 351; Chicken isolates: C. jejuni 108, 434 and 506), also recognised some of the branched mannose structures under all conditions tested. Branched mannose is far more common in complex N-linked glycans found on many different cell surface proteins. These branched mannose structures are typically capped by other sugars including Glc/GlcNAc, Gal/GalNAc and sialic acid implying that either these interactions are through subterminal binding proteins that can recognise capped structures or are not biologically relevant to infection/colonisation. From the binding profile of C. jejuni to the complex sialylated structure, 11D, it appears in all cases but C. jejuni 108 that subterminal recognition of mannose in complex N-linked glycans can be ruled out.
Similar to C. jejuni binding to mannose, sialic acid recognition was only observed following a period of environmental stress, with all the C. jejuni strains tested exhibiting significantly more binding to sialylated glycans when maintained under normal atmosphere and at room temperature. This indicates that an adhesion/lectin able to bind sialylated glycans is regulated by the exposure of C. jejuni to environmental stress. As yet, no such protein has been elucidated in C. jejuni. Sialic acid is a common glycan present on multiple cell types and is typically the terminal sugar presented. In the intestines MUC1 is the most heavily sialylated protein present, however, MUC1 acts as a decoy receptor for bacteria and other viral and microbial infecting agents [10]. When MUC1 is bound by pathogens it is released from the cell surface and allows the pathogen to be excreted into the environment through the lumen [10]. A number of pathogens, including C. jejuni, are more infectious, have a lower infectious dose or get into deeper tissues faster when administered to MUC1−/− mice [10].
Of the few sialylated structures that were bound more broadly by C. jejuni, 10A (C. jejuni strains 351, 375, 520, 331, 434, 506), 10B (C. jejuni strains 351, 375, 520, 331, 434, 506) and 10D (all strains tested), are all fucosylated, indicating that the binding to these glycans may be more due to fucose than to sialic acid. C. jejuni 81116, once again, recognised a wider variety of sialic acid containing structures than the other C. jejuni strains tested, binding to α2-3 linked sialylactosamine structures. C. jejuni 81116 has a vastly different cell surface glycosylation profile than other C. jejuni producing larger non-sialylated LPS like molecule rather than the traditional LOS seen for other C. jejuni[21]. It may be interesting to speculate that surface glycosylation can play a role in the inhibition of the binding of C. jejuni to sialylated glycans, particularly through charge-charge repulsion. Sialic acid is a negatively charged sugar and C. jejuni strains such as 11168 are known to have surface glycosylation that contains sialic acid [22, 23]. Of the strains that bound to sialyllewis structures (10A and B), we have recently shown that, C. jejuni 351, 375 and 331, do not have surface sialylation [24], indicating these strains may be able to recognise the underlying fucose. We are yet to confirm the sialylation levels of C. jejuni strains 434 and 506. C. jejuni 520 seems to be a special case as the LOS it produces appears to be very heterogenous [24]. We have shown using lectin array and surface plasmon resonance that a proportion of the LOS produced by this strain is completely non-sialylated at all growth conditions tested [24]. It is therefore possible that sufficient C. jejuni 520 was present in the assay with low or no surface sialylation allowing for recognition of the underlying branched fucose.
Glycoaminoglycan binding by C. jejuni on glycan arrays has not previously been reported. C. jejuni in general preferred larger GAG fragments, with the most consistent binding observed to full length GAGs of up to 1.6MDa. GAGs are common extracellular matrix components and are expressed in on the surface of a broad range of cells [25–27]. GAGs are also known to associate with known cell surface targets of C. jejuni including fibronectin [25–27]. Once more 81116 had the broadest recognition for GAG and related structures recognising all the structures present on our array.
The non-invasive C. jejuni strain 331 had a preference for longer, branched galactose structures and was less likely to associate with disaccharides or terminal N-Acetylgalactosamine structures. This is of interest as C. jejuni 331 is known to be a strong chicken coloniser, capable of out competing other C. jejuni strains in co-infection studies and has been proposed as a potential non-virulent bioreplacement bacteria [28, 29]. It is possible that the lack of binding to disaccharide and small sugar subunits by C. jejuni 331 may offer a competitive advantage, allowing 331 to better colonise the intestinal crypts by ignoring smaller sugars in the lumen. Mono- and di-saccharides are common products from the activity of glycosidases in the intestinal tract of animals. This makes mono- and di-saccharides potential decoy receptors for C. jejuni in the chicken gut and as such, bacteria that do not bind to smaller sugars would potentially have a competitive advantage.