TEXT

Top Abstract Text References

Guillain-Barré syndrome (GBS), an acute polyneuropathy, is now the most common cause of generalized paralysis and in two-thirds of cases is preceded by a respiratory or gastrointestinal infection (8, 12). Campylobacter jejuni, a leading cause of acute gastroenteritis, has been identified as the most frequent infectious agent associated with the development of GBS (up to 66% of patients) (8, 17). C. jejuni can be serotyped based on differences in the polysaccharide structure (O side chain and core oligosaccharide [OS]) of the lipopolysaccharide (LPS) (O antigen) of the bacterium (16). Reports have shown that the C. jejuni isolates obtained from diarrheic patients prior to the onset of GBS belonged to serotype O:19, an uncommon serotype in gastroenteritis patients (6, 9, 23). Other C. jejuni serotypes commonly identified in association with GBS include O:2, O:2/44, O:4/59, O:15, O:18, O:21, O:24, O:30, O:37, and O:53 (6, 10, 12, 17).

Autoreactive antibodies to gangliosides, especially GM₁, are found in 20% of GBS patient sera, particularly after C. jejuni infection (8, 12, 19, 25), and are also found in the sera of 50% of patients with the chronic neuropathy termed multifocal motor neuropathy (14). Neuropathy-associated GM₁ antisera have been shown to cross-react with C. jejuni LPSs (19, 22, 23). It is thus currently hypothesized that antiganglioside antibodies may be induced as a result of molecular mimicry between peripheral nerve gangliosides and structurally similar C. jejuni LPS (19, 23). Although there are indications that anti-GM₁ antibodies may lead to the activation of inflammatory pathways and act by disrupting membrane ion channel function at nodes of Ranvier (20), experimental proof of involvement in disrupting nerve function has been difficult to conclusively demonstrate. However, since anti-GM₁ antibodies in human sera are likely to be a contributory factor in the induction of GBS, an important step in elucidating the pathogenesis of the disease is determining the structure of the immunogenic epitopes in ganglioside-mimicking C. jejuni LPS.

Chemical studies of the LPS extracted from C. jejuni O:19 have shown that the terminal regions of the LPS core mimic human gangliosides GM₁, GD_1a, GT_1a, and GD₃ (2, 9, 24). GM₂-like OS structures occur in LPS from O:1, O:23, and O:36 (4), whereas the core OSs of C. jejuni O:4 and O:41 mimic gangliosides GD_1a and GM₁, respectively (4, 15). Mimicry of C. jejuni O:2 LPS is limited to a disaccharide present in a range of gangliosides (3).

The authors of several studies have previously investigated the reactivities of human and animal anti-GM₁ antisera with C. jejuni LPS and demonstrated the principle of cross-reactivity. However, no information is available on the extent to which antibodies with different fine specificities of epitope recognition for GM₁ are capable of binding GM₁-like LPSs. In this study, we aimed to use a set of human monoclonal antibodies (MAbs) that are reactive with GM₁ and have been characterized as structurally distinct (13), in conjunction with a panel of well-defined LPSs, to determine the degree to which ganglioside GM₁ and C. jejuni LPSs share immunoreactive epitopes.

C. jejuni serostrains O:2 (ATCC 43430), O:3 (ATCC 43431), O:4 (ATCC 43432), O:19 (ATCC 43446), and O:41 (ATCC 43460) were obtained from the American Type Culture Collection (Manassas, Va.). The details concerning three GBS patients and one enteritis patient from whom C. jejuni O:41 strains (16971.94GSH, 28134.94GSH, 260.94RXH, and 176.83, respectively) were isolated have been described previously (7, 15). Isolates and serostrains were routinely cultured on blood agar under microaerobic conditions at 37°C for 48 h, bacterial biomass was harvested, and the bulk extraction of LPS was performed by the phenol-water extraction procedure (11). In addition, LPSs from two GBS isolates, C. jejuni OH4382 and OH4384, which exhibit mimicry of gangliosides GD₃ and GT_1a, respectively (2), were a generous gift from G. O. Aspinall (York University, Toronto, Ontario, Canada). The immunoglobulin M (IgM) anti-GM₁ MAbs termed BO_1-1, BO_3-1, SM_1-8, and WO_1-4 were cloned from peripheral blood lymphocytes of three multifocal motor neuropathy patients, all of whom had abnormally elevated anti-GM₁ antibody titers, and have been described previously (13, 21). The MAbs were purified by the ultrafiltration of culture supernatants and checked for monoclonality by isoelectric focusing (21).

Gangliosides (Sigma Chemical Co., St. Louis, Mo.) and LPSs were analyzed by thin-layer chromatography (TLC) on precoated silica gel 60 glass plates (Merck, Darmstadt, Germany) by using solvent systems of chloroform-methanol-0.22% CaCl₂ · 2H₂O (50:45:10 [vol/vol/vol]) (18) and n-propanol-water-25% NH₄OH (60:30:10 [vol/vol/vol] (19) as developers for gangliosides and LPSs, respectively. Chemical staining was performed with a resorcinol-HCl reagent (19), and immunostaining was performed by using the procedure of Saito et al. (18), as modified by Schwerer et al. (19), with MAbs diluted to a concentration of 10 µg/ml as the primary antibody and peroxidase-conjugated anti-human IgM (Dako, Cambridge, United Kingdom) diluted 1:1,000 as the secondary antibody. Binding experiments with cholera toxin-peroxidase conjugate (CT-HRP; Sigma) and peanut agglutinin (PNA)-HRP conjugate (Kem-En-Tec, Copenhagen, Denmark) were carried out with only one TLC overlay step by using CT-HRP or PNA-HRP at dilutions of 1:1,000 and 1:50, respectively. Inhibition experiments were performed by using the B subunit of CT (Sigma) at 1 µg/ml to overlay separated gangliosides on TLC plates and incubating the plates at room temperature for 1 h before the addition of MAbs. Preliminary PNA blocking experiments, in which unlabeled PNA was used to block PNA-HRP, failed to reproducibly demonstrate a blocking effect, and hence, PNA blocking experiments were not conducted further.

Two experiments were performed to test the reaction of the MAbs with the core OS of C. jejuni O:41 LPS. First, lipid A was liberated from C. jejuni 176.83 LPS by acid hydrolysis (11), the resultant free lipid A (1 to 4 µg) was dotted onto TLC plates and overlaid with the anti-GM₁ MAbs (BO_1-1, BO_3-1, and SM_1-8), and immunoreactants were detected as described for immunostaining. Second, the saccharide mixture from the acid hydrolysis of LPS was fractionated by gel permeation chromatography on Bio-Gel P6 (Bio-Rad Laboratories, Hercules, Calif.) and TSK-40 columns (5, 15), yielding core OS in the second peak which was subsequently freeze-dried. The MAbs (10 µg/ml) were immunoadsorbed with 200 µg of core OS (37°C for 3 h), immunoprecipitates were removed by centrifugation (10,000 × g for 10 min), and the supernatants were reacted with intact C. jejuni 176.83 LPS.

The binding patterns of the four MAbs to ganglioside GM₁ and C. jejuni LPSs are summarized in Table 1. The specificities of the MAbs were confirmed by their binding to GM₁ and are consistent with previous results (13, 21). The MAbs reacted to different degrees with each of the serotype O:41 LPSs tested. BO_1-1 and BO_3-1 both recognized the LPS of the enteritis strain (176.83), whereas BO_1-1 and BO_3-1 separately bound LPSs of two different GBS-associated strains (28134.94GSH and 260.94RXH, respectively). In addition, both MAbs bound serostrain O:2 LPS but did not react with serostrain O:41 LPS or with any of the other C. jejuni LPSs. SM_1-8, which reacted with ganglioside GM₁ only, recognized all C. jejuni O:41 LPSs with different intensities, including those from the three GBS-associated strains (Fig. 1), but did not bind to any of the other C. jejuni LPSs. WO_1-4 recognized three of the four serotype O:41 LPSs (excluding that from the enteritis strain) and the serostrain O:41 LPS but did not react with the other C. jejuni LPSs. Previously, we undertook investigations of the structures of LPSs from the C. jejuni O:41 strains and, in particular, established chemically that the core OS structure of C. jejuni 16971.94GSH shared a tetrasaccharide with ganglioside GM₁ (15). The putative regions of binding of MAbs SM_1-8 and WO_1-4 in C. jejuni 16971.94GSH LPS are shown in Fig. 2. MAbs BO_1-1 and BO_3-1 are not indicated in Fig. 2, since no reaction was observed with this particular C. jejuni O:41 LPS. However, the binding patterns of BO_1-1 and BO_3-1 are more comparable to that of MAb WO_1-4 than that of MAb SM_1-8. Therefore, ganglioside GM₁ and C. jejuni O:41 LPSs share an immunoreactive epitope, but because the patterns of MAb reactivity varied with individual C. jejuni O:41 LPSs (including those MAbs which were specific for GM₁ only), there must be a degree of structural variability present within this epitope(s). Alternatively, there may be differences in the surface topography or density of the GM₁-like molecule(s) in the C. jejuni O:41 LPSs. Furthermore, the influence of an O side chain in C. jejuni O:41 LPS on MAb binding can be excluded, since electrophoretic and immunoblotting analyses have shown that C. jejuni O:41 strains produce low-M_r LPS without an O chain (15) and chemical analyses indicate the presence of only high-M_r extracellular polysaccharides independent of LPS (5, 15).

View larger version (51K): [in this window] [in a new window]   FIG. 1.   Binding of human IgM MAb SM1-8 to purified C. jejuni O:41 LPSs. Lane 1, C. jejuni 16971.94GSH LPS; lane 2, C. jejuni 260.94RXH LPS; and lane 3, C. jejuni 28134.94GSH LPS. The immunostained chromatogram was overlaid first with MAb and subsequently with anti-human IgM. A sample of 1 µg was applied per lane.

View larger version (10K): [in this window] [in a new window]   FIG. 2.   Regions of gangliosides asialo-GM1 (crosshatched) and GM1 (solid) containing the putative binding sites of MAbs WO1-4 and SM1-8 (13, 21) shown on the proposed structure of C. jejuni 16971.94GSH LPS (15). The MAb WO1-4 binds the terminal dissacharide Gal1-3GalNAc, whereas MAb SM1-8 binds the same disaccharide in the presence of laterally attached N-acetylneuraminic acid (Neu5Ac). MAbs BO1-1 and BO3-1 are not indicated, since no reaction was observed with this LPS. PEtn, phosphoethanolamine; Kdo, 3-deoxy-D-manno-octulosonic acid; Hep, L-glycero-D-manno-heptose; Glc, glucose; Gal, galactose; GalNAc, N-acetylgalactosamine.

Interestingly, none of the anti-GM₁ MAbs reacted with serostrain O:4 or O:19 LPSs. The core OS of serostrain O:4 LPS contains a pentasaccharide in common with ganglioside GD_1a (4). None of the MAbs react with ganglioside GD_1a (13, 21), and so the lack of a reaction with serostrain O:4 LPS is expected. Although GM₁ mimicry, along with mimicry of ganglioside GD_1a, is present in the core OS of serostrain O:19 LPS (2), the lack of MAb reactivity may be explained by the presence of an O side chain in this LPS (2, 10), which affects epitope accessibility. As expected, no reaction occurred when the MAbs were reacted with LPSs of the C. jejuni O:19 GBS-associated strains (OH4382 and OH4384) which have been shown to mimic GD₃ and GT_1a, respectively (2). Furthermore, none of the MAbs reacted with free lipid A, and MAb preparations immunoadsorbed with the isolated core OS of C. jejuni 176.83 did not bind with the intact LPS of this strain. This provides conclusive evidence that it is the core OS region to which the antibodies bind and not lipid A.

CT, a specific ligand for ganglioside GM₁, and PNA, a ligand for the disaccharide moiety Gal1-3GalNAc, were tested for their abilities to react with ganglioside GM₁ and C. jejuni LPSs (Table 1). In addition to binding with gangliosides GM₁ (Table 1) and asialo-GM₁ (data not shown), which is consistent with previous results (13, 21), CT-HRP showed a strong reactivity with serostrain O:4 and O:19 LPSs and all C. jejuni O:41 LPSs, including that of the serostrain. This is consistent with the presence of a GM₁-like structure in these LPSs. A weaker reaction was observed with serostrain O:2 LPS, whose core OS lacks GalNAc and contains a sialylated Gal1-3Gal disaccharide (3). No reaction was detected with the LPS of serostrain O:3, a strain whose LPS is not sialylated, does not mimic gangliosides, and is not associated with the development of GBS (1, 11). PNA-HRP bound to GM₁ and asialo-GM₁ gangliosides and reacted avidly with O:2 LPS, suggesting that PNA can bind to Gal1-3Gal in addition to Gal1-3GalNAc. PNA-HRP did not react with serostrain O:41 LPS but reacted with the other C. jejuni serotype O:41 LPSs (Table 1). The latter suggests the presence of a Gal1-3GalNAc or Gal1-3Gal disaccharide in these serotype O:41 LPSs. PNA-HRP bound weakly to serostrain O:19 LPS but did not react with the LPSs of the two GBS-associated C. jejuni O:19 strains or with serostrain O:4 LPS (Table 1), reflecting the occurrence of terminal sialylation in these latter LPSs. No reaction was observed with the control C. jejuni O:3 LPS.

The recognition of ganglioside GM₁ by all of the MAbs was completely inhibited by the CT B subunit (Table 1). Thus, CT binds to ganglioside GM₁ and thereby excludes the binding of the MAbs, inferring that CT and the MAbs must recognize closely related structures in ganglioside GM₁. Furthermore, the binding of all the MAbs to the majority of C. jejuni O:41 LPSs was blocked by CT, again confirming that the MAbs and CT recognize a similar epitope in the C. jejuni O:41 LPSs. However, CT did not inhibit MAb WO_1-4 from binding to C. jejuni 28134.94GSH LPS. This demonstrates that CT and MAb WO_1-4 do not recognize the same epitope in this LPS, a further indication that differences occur in the C. jejuni O:41 LPSs. Likewise, CT did not inhibit the binding of BO_3-1 to serostrain O:2 LPS.

This study demonstrates the cross-reactivity of human monoclonal IgM anti-GM₁ antibodies with C. jejuni LPSs associated with GBS. The MAbs are known to have different specificities for ganglioside GM₁ (13, 21); only some of them cross-react to varying degrees with the structurally related glycolipids asialo-GM₁ and GD_1b (21). Thus, each MAb recognizes a slightly different epitope within GM₁, some of which may also occur in asialo-GM₁ and GD_1b. Our results are in accordance with those of others who demonstrated that IgM anti-GM₁ MAbs from patients with chronic motor neuropathy reacted with LPSs of C. jejuni O:4, O:19, and O:50 serostrains (22). The MAbs also differed in their relative reactivities with the C. jejuni LPSs tested, in particular with regard to serotype O:41 LPSs. The results not only show the existence of a GM₁-like epitope(s) in C. jejuni O:41 LPSs but also reveal the existence of differences within the serotype O:41 LPSs. The pattern of binding of the IgM anti-GM₁ MAbs to C. jejuni O:41 LPSs indicates that either slight differences in sugar substitution occur in the core OS of the various C. jejuni O:41 LPSs or variation occurs in how the LPSs present these structures to the antibodies, possibly as a function of antigen density, and hence affect antibody recognition.

Furthermore, as in other studies, CT recognized serostrain O:2, O:4, and O:19 LPSs (19, 22, 23) and CT bound avidly to all C. jejuni O:41 LPSs, including that of the serostrain and the enteritis isolate (C. jejuni 176.83). Yuki et al. observed the binding of CT to serostrain O:19 LPS and deduced that it had a GM₁-like structure (23), which has been confirmed by structural analyses (2). However, LPSs of two C. jejuni O:19 strains from GBS patients (OH4382 and OH4384) which mimic GD₃ and GT_1a, respectively (2), did not react with CT or any other ligands or antibodies used in this study. The binding of IgM MAbs to ganglioside GM₁ and C. jejuni LPSs was inhibited by the CT B subunit; thus, CT and the MAbs recognize the same or a structurally overlapping epitope in ganglioside GM₁ and GBS-associated C. jejuni LPSs. This concurs with the results in a previous report where the binding of GBS sera to O:19 LPS was blocked with CT (22). However, in this study, CT did not inhibit the binding of one of the MAbs to one serotype O:41 LPS, a further indication that slight differences in structure occur within the core OS of C. jejuni O:41 LPSs. Since our inhibition experiments using PNA were considered technically unreliable, we cannot confirm the evidence of others (22) that PNA does not inhibit the binding of anti-GM₁ antibodies to either ganglioside GM₁ or C. jejuni LPSs.

In conclusion, the evident mimicry between C. jejuni LPSs and gangliosides may act as a trigger to stimulate the production of antiganglioside antibodies which may play a role in the pathogenesis of GBS. The mimicry of gangliosides is not limited to those strains associated with GBS, as LPS from the C. jejuni O:41 enteritis isolate reacted in a way similar to that seen with LPSs from the GBS-associated strains. This phenomenon has previously been observed by us with C. jejuni O:19 LPS, whereby an enteritis isolate mimics both gangliosides GM₁ and GD_1a (9), suggesting that, in addition to mimicry, other host or bacterial factors are involved in disease pathogenesis and require further investigation.