The nature of the polysaccharide molecules of the human enteric pathogen Campylobacter jejuni has been the subject of debate. Previously, C. jejuni 81116 was shown to contain two different polysaccharides, one acidic (polysaccharide A) and the other neutral (polysaccharide B), occurring in a 3 : 1 ratio, respectively. The aim of this study was to determine the molecular origin of these polysaccharides. Using a combination of centrifugation, gel permeation chromatography, chemical assays, and 1H-NMR analysis, polysaccharide B was shown to be derived from lipopolysaccharide and polysaccharide A from capsular polysaccharide. Thus, C. jejuni 81116 produces both lipopolysaccharide-like molecules and capsular polysaccharide.
The Gram-negative bacterium Campylobacter jejuni is recognized as the principal bacterial cause of human gastroenteritis worldwide (Allos, 2001). In addition, C. jejuni is the most frequent antecedent infection in patients with Guillain–Barré syndrome (GBS), a peripheral neuropathy, characterized by progressive paralysis (Nachamkin et al., 1998; Prendergast & Moran, 2000). The core oligosaccharides of C. jejuni lipo-oligosaccharides (LOS) containing N-acetylneuraminic (sialic) acid, and mimicking the saccharide component of human gangliosides, have been posited to elicit an autoimmune response contributing to the development of the syndrome (Prendergast & Moran, 2000; Moran et al., 2002).
In the past, molecular serospecificity in the heat-stable (HS) antigen serotyping scheme of C. jejuni had been attributed to high-molecular-weight (–Mr) lipopolysaccharide (LPS) (Preston & Penner, 1987), but more recent evidence indicates that LPS-independent polysaccharides are produced by C. jejuni, which contribute to serospecificity. Based on mutational analysis of the kpsM, kpsS and kpsC genes of C. jejuni, Karlyshev (2000) suggested that C. jejuni produces low-Mr LOS, and not high-Mr LPS, and that serodominant higher-Mr bands in immunoblotting with antisera were derived from capsular polysaccharide (CPS). Mutational analysis of the waaF gene, encoding a heptosyltransferase involved in synthesis of the core oligosaccharide of LPS, of C. jejuni serotypes HS:2 and HS:6 showed the occurrence of an extracellular polysaccharide independent of the core oligosaccharide and lipid A moieties of LPS (Oldfield et al., 2002). Similar results were obtained by Fry (2000) after the construction of a galE mutant in an HS:6 isolate, C. jejuni 81116. Moreover, in C. jejuni serotypes HS:3 and HS:41 polysaccharides independent of core oligosaccharide and lipid A have been characterized in structural studies (Aspinall et al., 1995; Hanniffy et al., 1999). Likewise, polysaccharides, deduced to be capsular in origin, have been reported in serotypes HS:1 and HS:2 (St Michael et al., 2002; McNally et al., 2005). Also, phase-variability (antigenic variation) of C. jejuni CPSs has been described (Bacon et al., 2001). Furthermore, a capsule-like structure was visible upon electron microscopic examination with Alcian blue staining (Karlyshev et al., 2001) and the lipid anchor of CPS has been identified (Corcoran et al., 2006). Thus, it was inferred that C. jejuni produces LOS and CPS only (Karlyshev et al., 2000).
Campylobacter jejuni 81116 was originally isolated from a water-borne human outbreak of gastroenteritis (Palmer et al., 1983). In our previous study, the acid-degraded phenol–water extract of this strain was found to contain two different polysaccharides, each having a tetrasaccharide repeating unit: one acidic (polysaccharide A) and the other neutral (polysaccharide B), occurring in a 3 : 1 ratio, respectively (Muldoon et al., 2002) (Fig. 1). Besides the possibility that both of these are CPSs, preliminary reports suggest that in addition to CPS, other capsule-independent polysaccharides can occur in some C. jejuni strains (Corcoran & Moran, 2005; McLennan et al., 2005). On the other hand, other Gram-negative bacterial species, such as Escherichia coli, can produce neutral and acidic polysaccharides as an O-chain polysaccharide of LPS and an independent extracellular or CPS, respectively (Knirel & Kochetkov, 1994; Jansson, 1999). Thus, this situation might more closely resemble that of the two polysaccharides in C. jejuni 81116. Importantly, core sugars derived from LPS or LOS, such as various heptoses and 3-deoxy-D-manno-2-octulsonic acid (Kdo), were not detected in the polysaccharide mixture of this C. jejuni strain by sugar analysis when previously analyzed (Muldoon et al., 2002). Nevertheless, in addition to the possibility of these sugars being absent, even if present, their proportions by weight would be low and could potentially be below the detection limits of the gas chromatrographic methodologies previously used.
In the present study, the two polysaccharides of C. jejuni 81116 were reinvestigated. The aim of this study was to attempt the chemical separation of these polysaccharides and to determine the molecular origins of the two polysaccharides: whether this strain produces two different types of extracellular polysaccharides or whether both a capsular and an LPS-related polysaccharide are present.
Materials and methods
Bacterial strain, growth and isolation of polysaccharides
Large quantities of C. jejuni 81116 were grown on blood agar under microaerobic conditions and harvested to yield biomass (355 mg dry weight) as described (Moran et al., 1991). Material was extracted from the freeze-dried biomass using the hot phenol–water method and the aqueous phase was recovered (Westphal & Jann, 1965). The recovered material was treated with RNase A, DNase II and proteinase K (all from Sigma Chemical Co., St Louis, MO), to remove any contaminating nucleic acids or proteins, as described previously (Moran et al., 1991), and freeze-dried.
Gel electrophoresis and Western blotting
Sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis of the extract was performed using a constant current of 35 mA, a stacking gel of 5% acrylamide and a separating gel of 15% acrylamide as described elsewhere (Moran et al., 1992). Gels were stained with 0.25% (w/v) Coomassie blue (Bio-Rad Laboratories, Hercules, CA), with Alcian blue as described previously (Karlyshev & Wren, 2001) or, after gel fixation and periodate oxidation, with silver stain (Tsai & Frasch, 1982) or with a commercial silver staining kit (Silver Stain Plus Kit, Bio-Rad). A prestained protein molecular weight marker (Bio-Rad) was also included. Alternatively, samples fractionated by SDS-PAGE were electrotransferred onto nitrocellulose membranes by the procedure of Towbin (1979). Western blotting was carried out with HS:6 or HS:7 rabbit antiserum (1 : 100), as the primary antibody, and peroxidase-conjugated goat antirabbit immunoglobulin G (Bio-Rad) (1 : 500) as the secondary antibody. Immunoreactants were visualized using a horseradish-peroxidase development system (Bio-Rad).
Separation of polysaccharides
Polysaccharide B was isolated by the sequence of procedures outlined in Fig. 2. Specifically, the steps involved centrifugation and gel permeation chromatography (GPC) on a variety of columns. Either BioGel TSK-60 (Bio-Rad) or Sephadex G-50 (Amersham Biosciences, Little Chalfont, Bucks, UK) columns (90 × 2.5 and 56 × 2.6 cm, respectively) were used. For GPC, freeze-dried samples were reconstituted in 2 mL H2O, applied to the respective columns and eluted with 0.04 M pyridinium acetate (pH 4.2). The eluate was monitored using a Knauer differential refractometer, and 2-mL fractions were collected. The sequence of steps in the procedure, and the products obtained, are described in detail in the Results section.
Polysaccharide A-enriched fractions were obtained from a phenol–water extract by GPC on a Sephacryl S-300 HR (Amersham Biosciences) column (70.7 × 1.2 cm) eluted with a detergent-containing buffer (0.2 M NaCl, 0.25% sodium deoxycholate, 1 mM EDTA, 0.02% sodium azide, 10 mM Trizma, pH 8.0) (Peterson & McGroarty, 1985; Rivera et al., 1988; Isshiki et al., 2001). Fractions (2 mL) were collected and monitored by colorimetric assays.
Capsular polysaccharide extraction
Extraction of CPS was performed as described previously (Corcoran et al., 2006) using the method of Liu (1971). Essentially, after bacterial biomass was subjected to precipitation with CaCl2 and absolute alcohol, the precipitate was washed with solvents (ethanol, acetone and diethyl ether), and then extracted with 0.1 M sodium acetate (pH 7.0) five times, and crude polysaccharide was precipitated from the combined extract supernatants with CaCl2 and absolute alcohol. Subsequently, this crude polysaccharide was subjected to enzymatic treatments with RNase A, DNase II and proteinase K (all from Sigma) and freeze-dried to yield purified CPS as determined previously (Corcoran et al., 2006).
Colorimetric and chromatographic assays
All collected fractions were lyophilized and resuspended in 1 mL H2O for subsequent analysis. The carbohydrate content of individual fractions was determined using the phenol–sulphuric acid assay of Dubois (1956). The total hexosamine content of each fraction (i.e. GlcpNAc from polysaccharide B) was determined after acid hydrolysis (4 M HCl at 100°C for 15 h) by the Morgan–Elson reaction as modified by Strominger (1959). The presence and quantification of Kdo was performed after 0.1 M HCl hydrolysis (100°C for 2 h) and determination of thiobarbituric acid-reactive material colorimetrically as Kdo equivalents (Warawdekar & Saslaw, 1959). After liberation with 2 M fluoroacetic acid (120°C, 2 h), gas chromatographic analysis of sugars as alditol acetate derivatives was performed as described previously (Muldoon et al., 2002).
Detection of endotoxin (LPS/LOS) was performed by a turbidimetric Limulus amoebocyte lyaste (LAL) assay (Pyrotell-T Associates of Cape Cod, East Falmouth, MA) with a sensitivity of 0.001 EU mL−1, according to the manufacturer's instructions.
Pooled fractions for NMR analysis were desalted on a BioGel P-2 (Bio-Rad) column (23 × 2 cm) using water as an eluant and fractions of 1 mL were collected. Carbohydrate-containing fractions were detected by spotting aliquots of each fraction onto silica gel-coated thin-layer chromatography plates (Sigma), staining with 10% H2SO4-ethanolic solution and subsequent charring at 120°C for 2 min. Delipidation was performed under standard mild acid conditions (Muldoon et al., 2002). De-O-acetylation was performed with 12.5% aqueous ammonia at 37°C for 16 h.
Samples were deuterium-exchanged by freeze-drying three times from D2O and then examined in solutions of 99.97%2H2O, using internal acetone as a reference (δH, 2.225; δC, 31.45). All spectra were recorded at a temperature of 60°C on a JEOL Lambda 400 MHz spectrometer equipped with a DEC AXP 300 computer workstation or a Bruker DRX-500 MHz spectrometer and processed using standard Bruker software (XWINNMR 1.2). The experimental parameters used for 2D acquisitions were essentially the same as described previously (Muldoon et al., 2002).
Results and discussion
Electrophoretic analysis and immunoblotting
Polysaccharide material produced by C. jejuni 81116 was extracted from freeze-dried biomass by the hot phenol–water method (Westphal & Jann, 1965) into the water phase with a yield of 18% of bacterial dry weight. Nucleic acid and protein contamination was absent from the extract as determined spectrophotometrically (Moran et al., 1992). Lack of protein contamination was confirmed by the absence of Coomassie blue staining of SDS-PAGE gels (data not shown).
Electrophoretic analysis of the extract with silver staining, after periodate oxidation, showed a low-Mr band, suggesting LOS production (Fig. 3b) when compared with E. coli LPS as a standard (Fig. 3a). However, silver staining of gels in one of four experiments not only showed the presence of mid-Mr bands but also high-Mr bands with a ladder-like pattern similar to LPS in experiments with the same extract preparation (Fig. 3c). Also, to exclude the possibility of laboratory-based variation in silver staining, a standardized, commercial silver staining kit, with a periodate oxidation step included, was used but gave the same variable staining. This variable staining may have led to these bands being overlooked previously in silver-stained gels in previous studies (Preston & Penner, 1987). However, after electrophoresis, reduction of the gel fixation time from overnight incubation to 1 h resulted in repeatable staining of the high-Mr bands with the laboratory-based and commercial silver stains. Although the intensity of silver staining depends upon the properties of individual sugars and their linkages, the variable staining of these bands may reflect the sensitivity of the stain because of a low concentration of the high-Mr bands. Combined with this, elution of some molecules may occur during the fixation step, particularly over the longer and more usual fixation period, thereby reducing their level below a critical concentration for staining that can vary between experiments. Of the bands, only the mid-Mr bands stained with Alcian blue (Fig. 3d), thus corresponding to CPS molecules (Karlyshev & Wren, 2001). Moreover, a purified CPS preparation from C. jejuni 81116 (Corcoran et al., 2006) exhibited identical mobility and staining. The CPS-related bands were detectable by immunoblotting with HS:6 antiserum, consistent with the serodominance of these molecules in this serotype (Oldfield et al., 2002), whereas the high-Mr bands were unreactive with this antiserum (Fig. 3e). On the other hand, C. jejuni 81116 types as both HS:6 and HS:7 (Karlyshev et al., 2000). However, only the high-Mr bands were detectable in immunoblotting with HS:7 antiserum (Fig. 3f) and corresponded to the high-Mr banding in the silver-stained gel of purified polysaccharide B (Fig. 3g), which had been purified as described below. The absence of contaminating low-Mr bands (indicative of LOS) and mid-Mr bands (indicative of CPS) from the purified polysaccharide B preparation is noteworthy.
Isolation of polysaccharide B
A quantity of the untreated phenol–water extract (63.8 mg) was centrifuged to give a supernatant, designated sp1, and a pellet, termed pel1 (Fig. 2). The 1H-NMR spectrum of lyophilized and deuterium-exchanged sp1 showed a mixture of polysaccharides A and B. In order to support this conclusion and verify that the polysaccharides under examination were indeed the same as previously reported (Muldoon et al., 2002), a portion of the material was delipidated and de-O-acetylated, and then the 1H-spectrum (Fig. 4) and 13C-NMR spectrum were acquired and assigned by 2D homonuclear COSY and TOCSY and heteronuclear HSQC spectra (Table 1). Comparison of the assigned chemical shifts and coupling constant values with previously reported spectra of the mixed polysaccharides that had been similarly modified showed good agreement (Muldoon et al., 2002), and thus confirmed the identities of the polysaccharides under investigation. Subsequently, this sample was fractionated on a TSK-60 column to give two poorly resolved peaks; fractions representing nonoverlapping portions of the peaks were collected (Fr1 and Fr2, respectively) and analyzed by 1H-NMR spectroscopy (Fig. 2). Fr1 consisted of a mixture of polysaccharides A and B in a 5 : 1 ratio, whereas Fr2 was comprised of the same mixture but in a 2 : 1 ratio. Both spectra also contained signals for core sugars and lipid-derived fatty acid residues. Furthermore, comparison of the relative intensities of the fatty acid signals with the anomeric signals of both polysaccharides A and B indicated that the former signals were associated with polysaccharide B. Thus, it was inferred that polysaccharide B, but not polysaccharide A, was LPS-associated.
Chemical shifts for N-acetyl are δ 2.03 and 2.08.
Assignment may be interchanged.
Meanwhile, the material of pel1 was treated with mild acid (2% acetic acid at 100°C for 2 h) and further centrifuged (Fig. 2). As the amount recovered was negligible, the resultant pellet (designated pel2) was discarded. The supernatant of this acid treatment (termed sp2), which yielded about 1 mg of material, was subjected to GPC on a Sephadex G-50 column (Fig. 2), and then examined by 1H-NMR spectroscopy. The 1H-NMR spectrum contained only signals from the neutral polysaccharide B and those associated with fatty acid and core sugar residues, but none derived from polysaccharide A (Fig. 5). The spectrum had minor signals typical of Kdo and core sugar residues, namely δ~1.10, 1.15, 1.19, 1.60, 1.98, 4.74, 4.82 and 5.00 (for comparison, see Aspinall et al., 1995; Hanniffy et al., 2001; Müller-Loennies et al., 2002; Bystrova et al., 2003; Molinaro et al., 2003) and fatty acids, i.e. derived from lipid A, namely CH3, δ~0.65–0.85; CH2, δ~1.35–1.58 (for comparison, see Hauksson et al., 1995; Gamian et al., 1996; Pasciak et al., 2003). Again, from this evidence, it can be deduced that the neutral polysaccharide B is associated with LPS. Consistent with the LPS origin of polysaccharide B, this purified preparation was reactive in the endotoxin-detecting LAL assay. On the basis of solubility and fractionation properties, the polysaccharide B of sp2 was deduced to be of lower molecular weight and short-chained compared with a higher Mr, longer chain version that was present in the spectrum of sp1. SDS-PAGE analysis of sp2 with silver staining showed similar, but slightly shorter, ladder-like high-Mr bands (Fig. 3g) to those of the original extract (Fig. 3c), confirming this conclusion.
Enriched fractions of polysaccharide A
Subsequently, attempts were focused on isolating the second polysaccharide, polysaccharide A, corresponding to the Alcian blue-staining, mid-Mr bands in SDS-PAGE (Fig. 3d). Initial attempts using several GPC columns of various dimensions with differing stationary phase (e.g. Sephacryl S-200 HR, Sephadex G-25, etc.) to isolate polysaccharide A from a phenol–water extract (50 mg) proved unsuccessful (data not shown). Nevertheless, when a Sephacryl S-300 HR high-resolution column and a detergent-containing buffer (Peterson & McGroarty, 1985; Rivera et al., 1988; Isshiki et al., 2001) were used, greater success was obtained. Applying colorimetric assays to the fractions eluted from the column, it was possible to distinguish two overlapping peaks based on the hexosamine determination (Strominger et al., 1959), indicating the occurrence of polysaccharide B (i.e. the presence of N-acetylglucosamine), the Kdo determination (Warawdekar & Saslaw, 1959) being indicative of core sugars and hence LPS occurrence, and the phenol–sulphuric acid assay (Dubois et al., 1956) indicating the presence of carbohydrates. Both peaks contained hexosamine and Kdo, but Peak 1 (90–104 mL) contained smaller proportions, and was collected for further analysis. After desalting on a BioGel P-2 column, the sample was subjected to 1H-NMR spectroscopy and showed signals for both polysaccharides A and B in a 5 : 1 ratio, demonstrating enrichment of polysaccharide A. Signals of core residues and lipid-related fatty acids were present, and again, comparison of the relative intensities with those of anomeric signals indicated that polysaccharide B, rather than polysaccharide A, was LPS-associated. Importantly, the ratios of these relative intensities were identical to those observed with purified polysaccharide B. Agreeing with this interpretation, the relative intensities of core residue and fatty acid signals of the 1H-NMR spectrum of Peak 2 (108–114 mL) indicated that polysaccharide B was LPS-related and showed a 3 : 1 ratio of polysaccharides A and B, respectively. Moreover, comparison of the absorbance values for both the hexosamine assay (indicative of polysaccharide B) and Kdo assay (indicative of lipopolysaccharide core) in Peaks 1 and 2 revealed the coincidence of these assay plots (data not shown), further indicating that polysaccharide B is LPS-related. Should Kdo have been present because of LOS contamination, it would not have associated with these particular fractions because of size differences between LOS and LPS. Nevertheless, despite the enrichment of polysaccharide A in Peak 1, attempts at further separation of the components of Peak 1 in order to isolate this polysaccharide did not prove successful.
Chemical compositional analysis of purified CPS from C. jejuni 81116 (Corcoran et al., 2006), isolated by another extraction method (Liu et al., 1971), showed the occurrence of the sugars d-mannose, d-glucose, d-glucuronic acid in the approximate molar ratios 1 : 1.9 : 0.8, consistent with the composition of polysaccharide A (Muldoon et al., 2002). Thus, the identical electrophoretic properties of the Alcian blue-staining, mid-Mr polysaccharide A (see above, Fig. 3c and d) and the isolated CPS (see above), as well their similar composition, indicate that this polysaccharide is CPS-derived.
Campylobacter jejuni 81116 produces two polysaccharides, A and B, with differing structural properties (Muldoon et al., 2002). In the present study, electrophoretic analysis of phenol–water-extracted material indicated the presence of two entities: one with variable silver staining and composed of high-Mr bands with a ladder-like pattern, and the other of Alcian-blue staining mid-Mr bands which corresponded to CPS (Karlyshev & Wren, 2001; Oldfield et al., 2002). Using a fractionation procedure, it was possible to isolate polysaccharide B and upon NMR analysis show the occurrence of LPS core sugar and fatty acid signals in this purified preparation. Consistent with this, analysis of GPC fractions in colorimetric assays showed the coincidence of the LPS core sugar Kdo and of hexosamine, which is indicative of polysaccharide B presence, in the same fractions. Also, isolated polysaccharide B in SDS-PAGE showed high-Mr bands with a ladder-like pattern similar to that of LPS and corresponded to the high-Mr bands observed in the unfractionated sample. The LPS origin of polysaccharide B was further supported by its reaction in the LAL assay and whose reactivity was not attributable to LOS contamination as low-Mr bands were absent from the preparation in SDS-PAGE.
However, attempts to purify polysaccharide A from the phenol–water-derived fractions, using differing GPC strategies, proved more difficult. Even the use of a high-resolution column (Sephacryl S-300 HR) with a detergent-containing buffer, which has previously proven successful in separating CPS and high-Mr LPS and resolving polysaccharides of differing repeat-unit sizes (Peterson & McGroarty, 1985, Rivera et al., 1988; Isshiki et al., 2001), only allowed enrichment of polysaccharide A. Comparison of the relative intensities of core sugar residue and fatty acid signals with anomeric signals in NMR analysis of this enriched mixture again indicated that polysaccharide B, but not polysaccharide A, was LPS-associated. Despite the enrichment of polysaccharide A, attempts at further separation with a variety of techniques, including ion-exchange chromatography, did not prove successful (M. Kilcoyne et al., unpublished results), because some of the LPS-associated polysaccharide B molecules are of the same molecular weight range as polysaccharide A and exhibit strong interaction properties. Thus, separation of polysaccharide A from B on the basis of charge properties alone would not prove feasible.
Only the mid-Mr bands in SDS-PAGE stained with Alcian blue, thus corresponding to CPS molecules (Karlyshev & Wren, 2001). Binding of the cationic dye Alcian blue to polysaccharides is due to the occurrence of negative charges on the macromolecules (Bonnell et al., 1999), as occurs in acidic polysaccharide A, but not in polysaccharide B. Moreover, a purified polysaccharide, extracted by the method of Liu (1971) and previously characterized as the CPS of C. jejuni 81116 (Corcoran et al., 2006), had the same electrophoretic characteristics as the mid-Mr polysaccharide A, and in particular had a similar chemical composition. Thus, it could be concluded that polysaccharide A was CPS-derived.
Importantly, the immunoblotting reactions of mid-Mr CPS-related bands with HS:6 antiserum and high-Mr LPS-related bands with HS:7 antiserum are consistent with C. jejuni 81116 typing as both HS:6 and HS:7. Further supporting our findings, Karlyshev (2000) reported that mutation of kpsM in C. jejuni 81116 led to loss of expression of the mid-Mr CPS-associated bands and loss of HS:6 typing. Also, consistent with the LPS-relatedness of polysaccharide B, mutation of waaF in C. jejuni 81116, affecting the inner core of LOS/LPS, led to loss of expression of the high-Mr HS:7-reactive bands (A.P. Moran et al., unpublished results). Hence, the dual reaction of C. jejuni 81116 with HS:6 and HS:7 typing antisera can be explained by the strain's production of two different and serologically independently recognizable polysaccharides.
Collectively, the results of this study show that C. jejuni 81116 produces a neutral polysaccharide that is LPS-associated (polysaccharide B), as well as an acidic CPS (polysaccharide A). In other strains, CPS has been described and deduced to occur with LOS alone (Aspinall et al., 1995; Hanniffy et al., 1999; Karlyshev et al., 2000; Bacon 2001; St. Michael et al., 2002; Karlyshev et al., 2005; McNally et al., 2005; Papp-Szabo et al., 2005). However, preliminary evidence suggests that additional polysaccharides may be expressed by some C. jejuni strains (Corcoran & Moran, 2005; McLennan et al., 2005). As the nature and molecular origins of the polysaccharides expressed by C. jejuni are complex, and variation may occur between strains in their production, further investigations are required on the pathogenic relevance of these differing polysaccharides produced by C. jejuni.
We thank T.U. Kosunen (University of Helsinki, Finland) for donation of HS:6 and HS:7 typing antisera. We are grateful to the Wellcome Trust (London) for a grant to A.V.S. towards the purchase of the NMR instrument. This research was supported by an Enterprise Ireland International Collaboration grant (to A.V.S.) and the National Development Plan under the Higher Education Authority (HEA) PRTL-3 programme and the Irish Health Research Board (to A.P.M.).