ABSTRACT
Bacillus pumilus strain Sh18 cell wall polysaccharide (CWP), cross-reactive with the capsular polysaccharide of Haemophilus influenzae type b, was purified and its chemical structure was elucidated using fast atom bombardment mass spectrometry, nuclear magnetic resonance techniques, and sugar-specific degradation procedures. Two major structures, 1,5-poly(ribitol phosphate) and 1,3-poly(glycerol phosphate), with the latter partially substituted by 2-acetamido-2-deoxy-α-galactopyranose (13%) and 2-acetamido-2-deoxy-α-glucopyranose (6%) on position O-2, were found. A minor component was established to be a polymer of →3-O-(2-acetamido-2-deoxy-β-glucopyranosyl)-1→4-ribitol-1-OPO3→. The ratios of the three components were 56, 34, and 10 mol%, respectively. The Sh18 CWP was covalently bound to carrier proteins, and the immunogenicity of the resulting conjugates was evaluated in mice. Two methods of conjugation were compared: (i) binding of 1-cyano-4-dimethylaminopyridinium tetrafluoroborate-activated hydroxyl groups of the CWP to adipic acid dihydrazide (ADH)-derivatized protein, and (ii) binding of the carbodiimide-activated terminal phosphate group of the CWP to ADH-derivatized protein. The conjugate-induced antibodies reacted in an enzyme-linked immunosorbent assay with the homologous polysaccharide and with a number of other bacterial polysaccharides containing ribitol and glycerol phosphates, including H. influenzae types a and b and strains of Staphylococcus aureus and Staphylococcus epidermidis.
Cell wall polysaccharides (CWP) of gram-positive bacteria, also referred to as teichoic acids, are structurally diverse linear polymers of polyols or carbohydrates linked through phosphodiester bonds, often substituted with different glycosyl or amino acid residues and terminally linked to the muramic acid of peptidoglycan (5). The functions and structures of different types of teichoic acids have been reviewed (23). These compounds comprise 20 to 50% of the weight of the cell wall and are major surface antigens in noncapsulated gram-positive bacteria. In capsulated bacteria, the capsule is the surface antigen covering other structures such as CWP in gram-positive bacteria and lipopolysaccharides in gram-negative bacteria. Antibodies to the capsular polysaccharides (CP) of pathogenic bacteria are protective, and the CP are used as vaccines, either alone or as protein conjugates (1).
Bacillus pumilus strain Sh18, a nonpathogenic, enteric gram-positive bacterium, produces a CWP, hitherto of unknown structure, reported to cross-react with the H. influenzae type b (Hib) CP. This cross-reactivity has been attributed to poly(ribitol phosphate), a known component of teichoic acids of bacilli. Ribitol phosphate, but not ribitol alone, inhibited precipitation of anti-Hib serum with B. pumilus Sh18 CWP (3, 9). No cross-reactivity, however, was observed with H. influenzae type a (Hia) CP, also containing ribitol phosphate in its subunit (Fig. 1). The induction of antibodies against Hia would be of interest, since this type has been reported in several countries to cause up to 10% of systemic infections due to H. influenzae (26, 35, 36). Here, we describe the purification of B. pumilus Sh18 CWP, the determination of its chemical structure, and the preparation and characterization of CWP-protein conjugates.
Structures of CP and CWP cross-reacting with B. pumilus Sh18 CWP.
MATERIALS AND METHODS
Bacteria and cultivation. B. pumilus Sh18 and Sh17, Hib strains Eagan and Rab, Hia strain Harding, Staphylococcus aureus type 5 strain Lowenstein, and Staphylococcus epidermidis RP-62A (ATCC 35981) (22, 32) were used. Sh18 and Sh17 were cultured in ultrafiltered tryptic soy broth (Difco), and Hib, Hia, and S. aureus type 5 were cultured as reported previously (13, 25). S. epidermidis was grown in a chemically defined medium (14). The structures of the polysaccharides are shown in Fig. 1.
Polysaccharides.CP from Hib and Hia and CWP from S. aureus type 5, Sh18, and Sh17 were prepared as described elsewhere (17, 25, 34) with additional passage through a Sepharose CL-6B column (1 by 100 cm; 0.2 M NaCl as eluent). The identity of Hib and Hia CP was confirmed by precipitation in double immunodiffusion with type-specific sera (22) and by nuclear magnetic resonance (NMR) spectroscopy by comparison to the published spectra (19, 38). The CWP of S. aureus type 5 was further separated from its CP by DEAE Sephadex (5 by 15 cm) chromatography. Fractions showing a positive reaction with rabbit anti-S. aureus teichoic acid serum and a negative reaction with rabbit anti-S. aureus type 5 CP were collected (17). S. epidermidis CWP was precipitated with 80% ethanol from the culture supernatant, purified as described elsewhere (34), and chromatographed on a BioGel P100 (1- by 100-cm) column equilibrated in phosphate-buffered saline (PBS). Anti-S. epidermidis sera were prepared by intravenous immunization of rabbits with acetone-dried bacterial cells as described elsewhere (2). Sh18 CWP was further purified by DEAE-Sephadex chromatography, and fractions reacting with Hib antiserum were collected.
Analyses.Sugars were analyzed according to the method of Sawardeker et al. (27). A 0.5-mg portion of each polysaccharide was hydrolyzed in 48% HF for 1 h at 60°C (15) and/or in 10 M HCl for 30 min at 80°C and, after reduction and peracetylation, analyzed by gas-liquid chromatography-mass spectrometry (GLC-MS) using a Hewlett-Packard model HP 6890 apparatus with a type HP-5 glass capillary column (0.32 mm by 30 m) and temperature programming at 8°C/min, from 125 to 250°C in the electron ionization (106 eV) mode. Ribitol was distinguished from ribose by performing the reduction step with sodium borodeuteride. Amino acids were analyzed after hydrolysis with 6 M HCl at 150°C for 1 h and derivatization to volatile N-heptafluorobutyryl isobutyl esters of amino acids (21) and muramic acid after trimethylsilyl derivatization (7), using the same GLC-MS apparatus and temperature program as above. Protein concentration was assayed by the Lowry method (20), phosphate content was determined according to the method of Chen et al. (10), and the amount of hydrazide groups was determined in a 2,4,6-trinitrobenzenesulfonic acid (TNBS) assay (29).
Hydrofluoric acid treatment and methylation analysis.Sh18 CWP (15 mg) was treated with 1 ml of aqueous 48% HF at 4°C for 48 h. HF was evaporated under nitrogen, and the hydrolyzed components were separated on a BioGel P2 column equilibrated with water. Separated materials were lyophilized and analyzed. Methylation was performed as described previously (11). Methylated compounds were hydrolyzed, converted to alditol acetates, and analyzed by GLC-MS as above.
Periodate oxidation.Sh18 CWP (15 mg) was treated with 1 ml of 0.1 M NaIO4 at 4°C for 48 h in the dark. Excess periodate was then destroyed by the addition of 0.2 ml of ethylene glycol, followed by the addition of 20 mg of sodium borohydride. After 15 h at 4°C, the solution was desalted on a prepacked PD10 column, applied to a BioGel P60 column equilibrated in 20 mM ammonium bicarbonate, and connected to the Uvicord SII detector (A206 and A280).
Mild acid hydrolysis.Sh18 CWP was treated with 10 mM HCl for 10 min at 100°C as described elsewhere (18). Acid was evaporated under nitrogen, and the material was applied to a BioGel P60 column in 20 mM ammonium bicarbonate and connected to the Uvicord SII detector (A206 and A280).
Fast atom bombardment-MS.Mass spectra were recorded by using a JEOL SX102a magnetic sector instrument with xenon and 6 keV atoms to ionize samples from a 3-nitrobenzyl alcohol or glycerol matrix.
NMR spectroscopy.NMR spectra were acquired at 300 K by use of a Bruker DRX-500 spectrometer. Solutions of 5 to 13 mg of compound in D2O (99.96 atom% D) were used, with acetone as an internal reference at 2.225 and 31.0 ppm, respectively, for 1H and 13C NMR. In most cases, 32,768-point data sets were used for 1D spectra, in some instances with zero-filling or complex, forward linear prediction to 32,768, 65,536, or 131,072 points. 1D 1H NMR spectra were recorded at 500 MHz with a spectral width of 3.21 or 4.25 kHz, a 30° pulse, and a recycle time of 6 s. 1D 13C NMR spectra were acquired at 126 MHz, using a spectral width of 25.1 kHz. Methylene 13C resonances were identified by the DEPT (distortionless enhancement by polarization transfer) method. 1H-coupled 13C-NMR spectra were acquired with the nuclear Overhauser effect (NOE) by use of gated, WALTZ-16 irradiation at the 1H frequency. 1H-decoupled, 1D 31P NMR spectra were recorded at 202 MHz by using 16,384-point data sets, a spectral width of 6.07 kHz, a 90° pulse, a recycle time of 4.85 s, and 85% H3PO4 containing 10% D2O as an external reference at −0.73 ppm. 1H-coupled 1D 31P NMR spectra were acquired without the NOE.
2D correlation spectroscopy (COSY) 1H NMR spectra were collected in 2,048- by 512-point data sets zero-filled to 2,048 by 2,048 points. 2D total COSY (TOCSY) 1H NMR spectra were acquired using 16,384- by 256-point data sets, zero-filled to 16,384 by 2,048 points, by use of the z-gradient-selected (GS), phase-sensitive, echo-anti-echo protocol. 1D 1H NMR subspectra of individual residues were generated by extraction of F2 slices from the 2D TOCSY spectra. For further confirmation of assignments, some 2D TOCSY experiments were conducted with either selective, digital, homonuclear 1H decoupling or continuous, WALTZ-16 31P decoupling. 2D heteronuclear single-quantum correlation (HSQC) and heteronuclear multiple-bond correlation (HMBC) 1H/13C NMR spectra were recorded as 2,048- by 512-point or 800-point data sets, zero-filled to 2,048 by 2,048 points, using the GS, sensitivity-enhanced, phase-sensitive echo-anti-echo mode for HSQC and a GS, low-pass-filtered, long-range, nondecoupled pulse sequence for HMBC, with an evolution delay of 83 ms, i.e., optimized for 2,3JCH 6.0 Hz. 2D 1H/31P HMBC was conducted at 500/202 MHz by using 8,192 (F2) by 512 (F1) point data sets, optimized for 2,3JPH 6.0 Hz.
2D 1H/13C and 1H/31P HSQC-TOCSY experiments were performed at 500/126 MHz and 500/202 MHz, respectively, by the GS sensitivity-enhanced method with the echo-anti-echo protocol, together with 2,048 or 4,096 points (F2) × 800 points (F1), 13C or 31P decoupling during acquisition, respectively, and a TOCSY mixing time of 140 ms. The phase-sensitive, 1H/13C experiment was optimized for 1JCH 145 Hz, whereas the 1H/31P method was optimal for 2,3JPH 6 Hz. Resolution enhancement of 1D and 2D NMR spectra was performed either by Gaussian multiplication using a line broadening of −0.75 to −5 Hz and a Gaussian broadening fraction of 0.3, by a minimally shifted, sine-bell squared window function, or by complex, forward linear prediction to larger data sizes.
Conjugation of Sh18 CWP. (i) Method 1.Step I: bovine serum albumin (BSA; Sigma) was derivatized with adipic acid dihydrazide (ADH) as described previously (29). Step II: BSA-AH was mixed with Sh18 CWP at a concentration of 10 mg/ml (each). The pH was adjusted to 5.8 with 0.1 M HCl, and EDAC was added to a concentration of 0.1 M. The reaction was continued at room temperature for 4 h at pH 5.8. The solution was dialyzed overnight against saline at 4°C and applied to a Sepharose CL-6B column (1 by 100 cm) equilibrated in 0.2 M NaCl. Fractions showing an identity line with anti-BSA and anti-Hib by double immunodiffusion were collected, and protein and phosphate contents were measured.
(ii) Method 2.Step I: recombinant Pseudomonas aeruginosa exotoxin A (rEPA) (8) was derivatized with ADH as described above. Step II: the Sh18 CWP was reacted with 1-cyano-4-dimethylaminopyridinium tetrafluoroborate (CDAP) (31) to form a cyanate ester derivative (CWP—O—C≡N); CWP (10 mg) was dissolved in water at 5 mg/ml, the pH was adjusted to 9.0, and 10 mg of CDAP in 0.1 ml of acetonitrile was added. After 30 s of mixing, 0.1 ml of 0.2 M triethylamine was added, followed by 10 mg of rEPA-AH in 0.6 ml of PBS over the next 2.5 min. The pH was maintained at 8.0 to 8.5 for 3 h, and the solution was left at 4°C overnight. The solution was dialyzed overnight against saline, at 4°C, and applied to a Sepharose CL-6B column as described above. Fractions showing an identity line with anti-P. aeruginosa exotoxin A (List Biological Lab., Inc.) and anti-Hib were collected, and protein and phosphate contents were assayed.
Immunization and immunological assays.Groups of 10 5- to 6-week-old female NIH general-purpose mice were injected subcutaneously three times, 2 weeks apart, with 2.5 μg of Sh18 CWP as a conjugate. Mice were exsanguinated 1 week after the last injection, and sera were stored at −20°C.
Antibody levels were measured by enzyme-linked immunosorbent assay (ELISA) with Nunc CovaLink plates as described elsewhere for DNA (24, 37). In this assay, the terminal phosphate group of polysaccharide, in the presence of carbodiimide, forms a phosphoramide bond with secondary amino group exposed on the surface of the wells. Plates were coated with the Hib, Hia, and E. coli K1 (negative control) CP and with the Sh18, Sh17, S. aureus type 5, and S. epidermidis RP-62A CWP. Polysaccharides (5 μg/ml) were dissolved in 10 mM 1-methylimidazole buffer (pH 7.0), and EDAC was added to a final concentration of 50 mM. The antigens were applied at 100 μl per well and incubated at 37°C overnight. Plates were washed six times with 0.1% Brij 35-saline and blocked with 1% human serum albumin (HSA) in PBS for 1 h at room temperature. Twofold dilutions of the sera were made in 1% HSA-0.1% Brij 35-saline and incubated at 37°C for 4 h. Plates were washed, and goat anti-mouse immunoglobulin G (IgG) conjugated to alkaline phosphatase was added and incubated at 37°C for 3 h. 4-Nitrophenylphosphate (1 mg/ml in 1 M Tris-HCl buffer [pH 9.8] and containing 0.3 mM MgSO4) was added, and the A405 was read after 30 min in an MR600 microplate reader (Dynatech). Some ELISAs were run using the avidin-biotin system (33). The murine monoclonal anti-Hib antibodies (0.52 mg/ml), kindly provided by D. V. Madore, were used as a standard for the ELISA. An inhibition ELISA was done by incubating mouse sera induced by Sh18 CWP conjugate I, diluted to the concentration that gave an A405 absorption of 1.0, with 5 or 20 μg of Hib, Hia, E. coli K1, and E. coli K93 CP or Sh18, S. aureus, and S. epidermidis CWP/ml for 1 h at 37°C and overnight at 4°C. The assay was then continued as above. Sera with and without inhibitor, at the same dilution, were compared. Percent inhibition was defined as follows: [1 − (A405 of adsorbed serum)/(A405 of nonadsorbed serum)] × 100%. Bactericidal activity of the sera was assayed using precolostral calf serum as a source of complement and Hib strain Eagan and Hia strain Harding (30).
Double immunodiffusion, immunoelectrophoresis, rocket and intermediate gel immunoelectrophoresis, and quantitative precipitation assays were done as described elsewhere (4, 16). Adsorption of burro anti-Hib serum with Sh18 CWP was done by adding 1 mg of CWP to 20 ml of serum (the ratio was based on the maximal precipitation in the quantitative precipitation assay). The solution was incubated at 37°C for 1 h and at 4°C for 2 days. The precipitate was removed by centrifugation at 37,000 × g for 10 min and washed, and the composition was analyzed by GLC-MS.
RESULTS
Isolation and characterization of B. pumilus Sh18 CWP.Sh18 CWP was eluted from a Sepharose CL-6B column with a Kd of 0.5 and from a DEAE-Sephadex column by 0.48 M NaCl. In both cases, the CWP showed size and charge heterogeneity. Compositional analysis of the eluted fractions showed that each contained ribitol, glycerol, and amino sugars, but in different molar ratios. The peak from the DEAE-Sephadex column that reacted with anti-Hib serum (Fig. 2), designated as Sh18 CWP, was further characterized.
Chromatography of B. pumilus Sh18 CWP on a Sepharose-DEAE ion-exchange column and the rocket immunoelectrophoresis of the eluted fractions with anti-Hib serum.
GLC-MS analysis of Sh18 CWP detected peracetylated derivatives of glycerol, ribitol, glucosamine, and galactosamine (Table 1), and the CWP contained 10.7 wt% of phosphorus. HF hydrolysis (48% HF; 1 h at 60°C) released only ribitol and glycerol, whereas additional hydrolysis with HCl released amino sugars, indicating that they are bound by glycosidic linkages.
Molar ratios of B. pumilus Sh18 CWP components and the compositions of fractions obtained after HF and sodium periodate degradation as determined by GLC-MS
Treatment of the Sh18 CWP with HF (48%; 4°C for 48 h) followed by a BioGel P-2 chromatography revealed low-molecular-mass components (Fig. 3). These components were subjected to GLC-MS (Table 1). The first fraction (Sh18-HF-1) contained glucosamine and small amounts of alanine, glutamic, diaminopimelic, and muramic acids, suggesting the presence of cell wall fragments. The second fraction (Sh18-HF-2) contained disaccharides: 2-acetamido-2-deoxy-hexosylribitol (pseudomolecular ion [M + 1]+ of 356) and 2-acetamido-2-deoxy-hexosylglycerol (pseudomolecular ion [M + 1]+ of 296) as determined by fast atom bombardment-MS analysis. GLC-MS revealed peracetylated glycerol, ribitol, 2-acetamido-2-deoxy-galactose, and 2-acetamido-2-deoxy-glucose at the molar ratios presented in Table 1. Methylation analysis performed with GLC-MS detected 4-O-acetyl-1,2,3,5-tetra-O-methyl-ribitol and 1,5-di-O-acetyl-3,4,6-tri-O-methyl-2-N-methylacetamido-2-deoxy-glucitol as major components, indicating the substitution of ribitol by GlcNAc on carbon C-4. Free ribitol and glycerol were recovered in fraction Sh18-HF-3.
Chromatography of HF-digested B. pumilus Sh 18 CWP on a BioGel P-2 column.
In the next step, oxidation of the Sh18 CWP with periodate was performed, and the products were separated on a BioGel P-60 column and analyzed (Fig. 4). No ribitol was detected, indicating that it had been completely oxidized. The composition of the two major fractions eluted from the column (Sh18-oxid-1 and Sh18-oxid-2) is shown in Table 1. GLC-MS and NMR analysis of Sh18-oxid-1 demonstrated the presence of 1,3-poly(glycerol phosphate). The low-molecular-mass fraction Sh18-oxid-2 was composed of glycerol and β-GlcNAc phosphate, representing the degradation products of the minor polymer: →3-O-β-GlcNAc-(1→4)-ribitol-1-OPO3→ (see below).
Chromatography of sodium periodate-treated B. pumilus Sh18 CWP on a BioGel P-60 column.
In order to examine whether the polymers were bound to each other by the cell wall fragments, the Sh18 CWP was treated with 10 mM HCl, as mild acid hydrolysis has been reported to cleave the linkage between the CWP and the phosphate on the muramic acid in the peptidoglycan (3). The product of this hydrolysis was eluted from a BioGel P-60 column as a single peak, in the same position as the nontreated material, indicating that it was not affected by this treatment. GLC-MS analysis revealed ribitol, glycerol, and amino sugars in molar ratios similar to those of the original material; also, NMR analysis revealed a similar spectrum.
Immunological properties.Sh18 CWP reacted by double immunodiffusion with anti-Hib and anti-S. epidermidis antibodies with a strong precipitation line and showed a weaker line with anti-S. aureus teichoic acid serum. Immunoelectrophoresis with the intermediate gel containing anti-Hib and the upper gel containing anti-S. epidermidis serum showed that part of the CWP was captured by the anti-Hib serum, whereas some migrated further and precipitated with the anti-S. epidermidis serum. Similar results were obtained when the order of the antisera in the gels was reversed. These data suggested that at least some of the poly(glycerol phosphate) and poly(ribitol phosphate) chains of Sh18 CWP were not bound. To further address the question of possible linkage between the polymers, Sh18 CWP was precipitated with anti-Hib serum, and the resulting precipitate was washed and analyzed by GLC-MS. The analysis revealed glycerol, ribitol, and glucosamine in the ratio 0.2:1.0:0.1. The lesser amount of glycerol than in the original preparation indicated that some of the poly(glycerol phosphate) may have been free and some may have been connected to the poly(ribitol phosphate).
NMR spectroscopy. 1H NMR assignments were confirmed by 2D COSY NMR and by use of 1D slices from high-resolution, 2D 1H/1H TOCSY NMR spectra. The 1H NMR assignments were used to generate most of the 13C assignments, by use of 2D HSQC correlations. H-5, H-6, and H-6′ assignments of GalNAc residues were generated by consideration of 13C assignments. 13C DEPT spectra were used to differentiate CH, CH2, and CH3 groups. Additional evidence for 13C assignments was gained from 2D 1H/13C HMBC spectra, which were also used to detect key, interresidue 1H/13C connectivities. Proton-phosphorus connectivities over two and three bonds were determined from 2D 1H/31P HMBC experiments. Confirmatory data for 1H, 13C, and 31P assignments were obtained from 2D 1H/13C and 1H/31P HSQC-TOCSY spectra. Quantitation of components of the polysaccharides or fractions was performed by integration of the CH and CH2 signals in 13C NMR spectra.
1H chemical shifts are reported in Table 2, 13C and 31P shifts are shown in Table 3, and 1H-1H, 1H-13C, and 13C-31P coupling constants are in Table 4. Anomeric configurations of hexose components were determined by measurement of the values 3J1,2 3.3 to 3.8 Hz and 1JC-1,H-1 170.2 to 172.4 Hz, which indicated the α anomeric configuration, and the values 3J1,2 8.3 to 8.4 Hz and 1JC-1,H-1 161.1 to 162.5 Hz (Table 4), which defined the β configuration. In all samples, the magnitudes of the vicinal 1H-1H coupling constants (Table 4) indicated that the sugars were present as pyranoid forms in the 4C1 conformation.
1H NMR chemical shifts of polysaccharides and their fragments from B. pumilus Sh18, S. epidermidis, and S. aureus
13C and 31P NMR chemical shifts of polysaccharides and their fragments from B. pumilus Sh18, S. epidermidis, and S. aureus
1H-1H, 1H-13C, and 13C-31P NMR coupling constants of polysaccharides and their fragments from B. pumilus Sh18, S. epidermidis, and S. aureus
B. pumilus Sh18-HF-2.NMR analysis of the fraction F2 obtained after HF treatment of Sh18 CWP was consistent with a mixture of 4-O-(2-acetamido-2-deoxy-β-glucopyranosyl)ribitol (β-GlcNAc-ribitol), 2-O-(2-acetamido-2-deoxy-α-galactopyranosyl)glycerol (α-GalNAc-Gro), and 2-O-(2-acetamido-2-deoxy-α-glucopyranosyl)glycerol (α-GlcNAc-Gro) in proportions of 57, 29, and 14 mol%, respectively, as measured by 13C NMR, and 59, 27, and 13 mol% from integration of the anomeric 1H signals of these species. 2D 1H/13C HMBC spectra showed interresidue cross-peaks for β-GlcNAc H-1 and ribitol C-4, ribitol H-4 and β-GlcNAc C-1, α-GalNAc H-1 and Gro C-2, Gro H-2 and α-GalNAc C-1, and α-GlcNAc H-1 and Gro C-2. Additionally, intraresidue HMBC cross-peaks were observed for the NAc 13C═O nuclei and the corresponding H-2 nuclei of the β-GlcNAc, α-GalNAc, and α-GlcNAc residues.
B. pumilus Sh18-HF-3.The F3 fraction obtained after HF treatment of Sh18 CWP displayed 1H, 13C, and 13C DEPT NMR spectra that were identical with those of ribitol (Sigma).
B. pumilus Sh18 CWP.NMR spectroscopy of the B. pumilus Sh18 CWP preparation indicated that it was a heterogeneous polysaccharide comprised of 1,5-poly(ribitol phosphate) (56 mol%), 1,3-poly(glycerol phosphate) (28 mol%), poly(→3-O-2-acetamido-2-deoxy-β-glucopyranosyl-1→4-ribitol-1→phosphate→) (10 mol%), 1,3-poly(2-acetamido-2-deoxy-α-galactopyranosyl-1→2-glycerol phosphate) (4 mol%), and 1,3-poly(2-acetamido-2-deoxy-α-glucopyranosyl-1→2-glycerol phosphate) (2 mol%). Recalculation of the proportions of each of the three, acetamido-deoxy sugar-containing components as percentages of the total of these components gave the ratio of β-GlcNAc-ribitol-P to α-GalNAc-Gro-P to α-GlcNAc-Gro-P of 59:28:13, respectively, which agreed well with the proportions reported above for the corresponding nonphosphorylated glycopyranosyl polyols in the HF hydrolysis product B. pumilus Sh18-HF-2. These quantitative measurements also indicated that of the total 1,3-poly(glycerol phosphate), 13% is glycosylated with α-GalNAc and 6% with α-GlcNAc.
The high proportion of 1,5-poly(ribitol phosphate) in B. pumilus Sh18 CWP was shown by the appearance in its 13C NMR spectrum of intense doublets for C-1,5 and C-2,4 of the ribitol-P and a strong singlet for C-3 (see Tables 3 and 4 for 13C chemical shifts and 13C-31P coupling constants, respectively). An intense doublet was also observed for C-1,3 of 1,3-poly(glycerol phosphate) [1,3-poly(Gro-P)] and a weaker triplet for C-2, reflecting the presence of 31P spin coupling to all of the Gro 13C nuclei. Partial glycosylation of C-2 of 1,3-poly(Gro-P) by α-GalNAc and α-GlcNAc was confirmed by the observation of 2D 1H/13C HMBC connectivities for α-GalNAc H-1 and Gro-P C-2 and of α-GlcNAc H-1 and Gro-P C-2, respectively.
The presence of a phosphate group at C-3 of the β-GlcNAc moiety was indicated by the observation of its H-3 signal as a wide quartet in the 1H/1H TOCSY NMR subspectrum (Fig. 5a), which collapsed to a wide triplet on broad-band irradiation of 31P (Fig. 5b). Confirmation of the phosphorylation of O-3 of the β-GlcNAc was obtained from the 13C NMR spectrum of the B. pumilus Sh18 CWP, which displayed the C-2, C-3, and C-4 signals of the β-GlcNAc residue as doublets containing 31P couplings (Table 4). 31P NMR spectroscopy (1D spectrum at the left side of Fig. 6) displayed four main resonances at δP of 3.64, 3.08, 2.84, and 2.59 ppm, which were assigned to 1,5-poly(ribitol-P), 1,3-poly(Gro-P), unresolved 1,3-poly(α-GalNAc-1→2-Gro-P) and 1,3-poly(α-GlcNAc-1→2-Gro-P) signals and the phosphate diester at C-3 of β-GlcNAc, respectively. Agreement was observed between the 1H/13C and 1H/31P cross-peak patterns of the CH2 protons of the 1,5-poly(ribitol-P) and 1,3-poly(Gro-P) components in both the 2D 1H/13C HSQC and 1H/31P HMBC spectra (Fig. 7). Evidence for the second attachment point of the phosphate diester at C-3 of β-GlcNAc was obtained from the 2D 1H/31P HMBC spectrum of the CWP (Fig. 7b), which showed signals at a δP of 2.59 ppm that represented 31P correlations with H-3 of β-GlcNAc and H-1 and H-1′ of a phosphorylated ribitol. Although the 13C-1 NMR signal of this ribitol was not resolved, the substitution of C-1 by phosphate was indicated by observation of the ribitol C-2 and C-3 signals as doublets, owing to 13C-31P spin coupling over three and four bonds, respectively (Table 4), in the situation where the C-4 and C-5 resonances of the ribitol did not exhibit 31P splitting. The 13C chemical shift (61.51 ppm) of this C-5 was characteristic of a nonphosphorylated CH2 group, whereas C-4 experiences a glycosylation shift to a lower field (82.42 ppm) due to attachment of the β-GlcNAc. Confirmation of the substitution of the β-GlcNAc at C-4 of the ribitol was obtained by detection of 2D 1H/13C HMBC cross-peaks for β-GlcNAc H-1/ribitol C-4, and ribitol H-4/β-GlcNAc C-1. Also, in agreement with the linkage of a phosphate diester group to C-3 of β-GlcNAc and C-1 of ribitol, the 1H-coupled 31P spectrum of the B. pumilus Sh18 CWP displayed this phosphate group as a quartet at a δP of 2.59 ppm, owing to coupling of the 31P nucleus with H-3 of β-GlcNAc and H-1 and H-1′ of ribitol. Therefore, the β-GlcNAc-containing component of the Sh18 CWP may be described as poly(→3-O-2-acetamido-2-deoxy-β-glucopyranosyl-(1→4)-ribitol-1→phosphate→).
1D F2 slices of 2D 1H/1H TOCSY NMR spectra of the B. pumilus Sh18 CWP in D2O at 500 MHz. (a) Subspectrum of the β-GlcNAc residue without decoupling. (b) Subspectrum with 31P spin decoupling, showing the collapse of the H-3 quartet to a triplet. x, artifact.
2D 1H/31P HSQC-TOCSY NMR spectrum of B. pumilus Sh18 CWP in D2O at 500 and 202 MHz, showing the connectivity of the 31P nuclei of the phosphate groups with the proton spin systems of the polyol and amino sugar residues.
Heteronuclear NMR spectra of B. pumilus Sh18 CWP in D2O. (a) Partial 2D 1H/13C HSQC spectrum, 2,048 (F2) by 800 (F1) data points, zero-filled to 2,048 by 2,048 points. (b) 2D 1H/31P HMBC spectrum, 8,192 (F2) by 512 (F1) data points, with forward, complex linear prediction to 8,192 by 2,048 points. The spectra illustrate how the 1H assignments for the terminal protons of the ribitol phosphate (ribitol-P) and glycerol phosphate (Gro-P) residues are consistent in the 1H/13C (a) and 1H/31P (b) chemical shift correlation spectra. In panel b the connectivity of the 31P nucleus of the linking phosphate group in the ribitol-P-β-GlcNAc moiety with H-3 of the β-GlcNAc residue and H-1 and H-1′ of the ribitol-P subunit is shown.
S. epidermidis CWP.The NMR data for S. epidermidis CWP were consistent with a major proportion of 1,3-poly[2-O-(α-glucopyranosyl)glycerol phosphate], the position of the α-glucopyranosyl group at C-2 of the glycerol moiety, being established by the observation of a 2D HMBC cross-peak between H-1 of the α-Glc residue and C-2 of Gro. Phosphorylation of O-1 and O-3 of the Gro residue was indicated by the detection of 31P splittings in the C-1, C-2, and C-3 signals of Gro and by deshielding of C-1 and C-3 in this residue (Table 3).
S. aureus CWP.NMR spectroscopy of the S. aureus CWP indicated that it was 1,5-poly[4-O-(2-acetamido-2-deoxy-β-glucopyranosyl)ribitol phosphate]. The position of the β-GlcNAc substituent at C-4 of the ribitol residue was defined by the observation of a 2D HMBC cross-peak between H-1 of β-GlcNAc and C-4 of ribitol and by strong deshielding of this C-4 (Table 3). Splitting of the C-1, C-2, C-4, and C-5 resonances of ribitol by 31P coupling (Table 4) demonstrated the phosphorylation of C-1 and C-5, as did the deshielding of these nuclei (Table 3). The 1H assignments for the ribitol protons were supported by 2D 1H/13C HSQC and 1H/31P HMBC experiments.
Sh18 conjugates.Two types of Sh18 conjugates were prepared: (i) conjugate I, BSA-AH-(EDAC)-Sh18 CWP, in which the EDAC-activated terminal phosphate group of the CWP was bound to ADH-derivatized protein; and (ii) conjugate II, rEPA-AH-(CDAP)-Sh18 CWP, in which CDAP-activated hydroxyl groups of the CWP were bound to ADH-derivatized protein.
The protein/CWP weight ratios in conjugate I and conjugate II were 2.5:1.0 and 2.7:1.0, respectively. The ratios of glycerol, ribitol, GlcNAc, and GalNAc, measured by GLC-MS, were 0.45:1.0:0.25:0.11 and 0.50:1.0:0.12:0.08, respectively. 31P NMR analyses of the conjugates showed similar spectra to those of the native CWP (Fig. 8). Double immunodiffusion analysis of both showed a line of identity with anti-Hib and anti-protein sera. Antibody levels induced by the conjugates and the inhibition by cross-reactive polysaccharides are presented in Fig. 9 and 10. Conjugate I-induced antibodies reacted with ribitol phosphate-containing polysaccharides of Hib, Hia, and S. aureus and with glycerol phosphate-containing polysaccharides of Sh18, Sh17, and S. epidermidis, whereas conjugate II-induced antibodies reacted only with polysaccharides containing glycerol phosphate. No bactericidal activity against Hib and Hia was detected in sera induced by the conjugates diluted 1:4. Quantitative analysis of these sera assayed in comparison with a monoclonal anti-Hib serum showed levels of anti-Hib antibodies of about 1.0 μg/ml.
1H-decoupled 31P NMR spectra at 202 MHz, showing the similarity of the 31P spectra of B. pumilus Sh18 CWP (a) and its protein conjugate (b).
IgG antibody levels against Hib CP, Hia CP, S. aureus CWP, B. pumilus Sh18 CWP, B. pumilus Sh17 CWP, and S. epidermidis CWP induced by B. pumilus Sh18 CWP-protein conjugates. Groups of 10 mice were immunized three times, 2 weeks apart, subcutaneously with 2.5 μg of Sh18 CWP as conjugate and bled 1 week after the last injection. IgG levels were measured in an ELISA with Nunc CovaLink plates, and geometric means for each group are depicted.
Inhibition of anti-B. pumilus Sh18 CWP-BSA conjugate with different polysaccharides in an ELISA. Sera from eight mice of the highest responders to the conjugate were pooled, diluted to the concentration providing an A405 of 1.0, and assayed by ELISA after incubation with 20 μg of E. coli K1 CP, E. coli K93 CP, Hib CP, Hia CP, S. aureus CWP, B. pumilus Sh18 CWP, and S. epidermidis CWP. Percent inhibition is defined as [1 − (A405 of adsorbed serum)/(A405 of nonadsorbed serum)] × 100%.
Antibody levels measured for polysaccharides covalently bound between the terminal phosphate group and the secondary amino groups on the surface of the CovaLink plates or by the avidin-biotin system were similar. The advantage of using CovaLink plates was that the technique was simple and did not require previous derivatization of hydroxyl groups of the polysaccharides with ADH.
DISCUSSION
B. pumilus Sh18 CWP has been reported to cross-react with Hib CP, but its exact composition was not known (1). This report elucidated the structure of the Sh18 CWP. Two major structures were identified: unsubstituted 1,5-poly(ribitol phosphate) and 1,3-poly(glycerol phosphate) partially substituted by 2-acetamido-2-deoxy-α-galactopyranose (13%) and 2-acetamido-2-deoxy-α-glucopyranose (6%) on position O-2, together with one minor structure, →3-O-β-GlcNAc-(1→4)-ribitol-1-OPO3→ (Fig. 1). The ratios of the three components were 56, 34, and 10 mol%, respectively. The question of a possible linkage between these polymers remained unresolved. Some of our experiments suggested that the polymers were bound to each other; for instance, all three components were present in all our preparations, and their ratios in our conjugates were similar. Other experiments indicated that they might be a mixture. Intermediate gel immunoelectrophoresis suggested the existence of two separate molecules. Periodate oxidation of the B. pumilus Sh18 CWP destroyed the ribitol residues and released a single chain of poly(glycerol phosphate) of the same size as the untreated CWP, implying the presence of separate chains of similar size. Also, NMR experiments could not identify any attachment points between the chains. The isolation of teichoic acids with different structures from one bacterial strain has been described for other gram-positive bacteria (28).
Two methods of conjugation were compared. Both conjugates had similar physico-chemical characteristics: the CWP/protein ratio, molecular size estimated by Sepharose CL-6B gel filtration, and the presence of all three components in similar ratios, as shown by GLC-MS and 31P-NMR analyses. Both conjugates retained the antigenicity of their CWP and protein components, but only one conjugate prepared by activation of a terminal phosphate group with EDAC induced antibodies that were reactive with both ribitol phosphate-containing and glycerol phosphate-containing polysaccharides (Fig. 1). In contrast, conjugate prepared by activation of hydroxyl groups with CDAP induced antibodies that were reactive only with the glycerol phosphate-containing polysaccharides. The fact that CDAP reacted with hydroxyl groups of ribitol may explain the loss of the immunogenicity of the ribitol phosphate in Sh18 CWP. The level of antibodies induced by these conjugates was too low to detect bactericidal activity. Further work on improvement of the immunogenicity of the conjugates is planned.
In summary, we present a complex structure of the CWP of B. pumilus Sh18, preparation of CWP conjugates, and evaluation of their immunogenicity in mice. The method of conjugation using a terminal phosphate group of a polysaccharide for covalent binding is reported for the first time and may be applicable to other polysaccharides with an appropriate structure. The broad cross-reactivity of our conjugate may be useful for development of a single vaccine for several pathogens.
FOOTNOTES
- Received 5 March 2004.
- Accepted 15 July 2004.
- Copyright © 2004 American Society for Microbiology
