Structure of the sialic acid-containing O-specific polysaccharide from Salmonella enterica serovar Toucra O48 lipopolysaccharide

Authors


A. Gamian, Ludwik Hirszfeld Institute of Immunology and Experimental Therapy, Polish Academy of Sciences, Weigla 12, 53–114 Wrocław, Poland. Fax: + 48 71 3732587, Tel.: + 48 71 3732316, E-mail: gamian@immuno.iitd.pan.wroc.pl

Abstract

Lipopolysaccharide was extracted from cells of Salmonella enterica serovar Toucra O48 and, after mild acid hydrolysis (1% AcOH, 1 h, 100 °C or 0.1 m NaOH-AcOH, pH 4.5, 5 h, 100 °C), the O-specific polysaccharide was isolated and characterized. The core and an oligosaccharide containing a fragment of the repeating unit linked to the core region were also obtained, depending on hydrolysis conditions. On the basis of sugar and methylation analyses and NMR spectroscopy of the hydrolysis products, the biological repeating unit of the O-specific polysaccharide was shown to be the following trisaccharide:

→4)-α-Neup5Ac(2→3)-l-α-FucpNAc(1→3)-d-β-GlcpNAc(1→

The polysaccharide O-chain was substituted with a single molar equivalent of O-acetyl group, distributed between the Neu5Ac O-9 and O-7 positions, in an approximate ratio of 7 : 3.

Abbreviations
CPS

capsular polysaccharide

FucN

fucosamine

GlcN

glucosamine

LPS

lipopolysaccharide

TSP-d4

2,2,3,3-tetradeutero-3-trimethylsilylpropionic acid

WHSQC

1H-detected heteronuclear single quantum coherence spectroscopy

PFG

pulsed field gradient

Bacterial lipopolysaccharide (LPS) generally consists of an O-specific polysaccharide chain attached through a core oligosaccharide to lipid A. Sialic acid may be present in LPS, and has been reported to be a component of the Salmonella enterica serovar Toucra O48 LPS [1]. When present in bacterial capsular polysaccharides (CPSs) or LPSs, sialic acid may contribute to the pathogenicity of the microorganism by mimicking a host tissue component [2,3]. For example, the core oligosaccharides of LPS from Campylobacter jejuni strains, which are associated with the development of the neurological disorder Guillain–Barré syndrome, mimic gangliosides [4], while the low-molecular-mass LPS from pathogenic Neisseria spp. mimic sialylated lactoneo glycosphingolipids, which may serve to camouflage the bacterium from the host [4,5]. The LPS of S. Toucra O48 has previously been shown to contain a long-chain sialic acid-containing O-specific polysaccharide [1,6], but the role of sialic acid in the O-chain has not been defined. Elucidation of the structures of these saccharides may allow a better understanding of the functions of the Neu5Ac residue in prokaryotes, and its role in pathogen–host interactions. Early reports indicated that S. Toucra O48 cross reacted with Citrobacter freundii O37, which also expresses a sialic acid-containing O-specific polysaccharide [7], but recent work has failed to confirm this [6]. While both LPS preparations contain terminal nonreducing Neu5Ac, the internal sialic acid in S. Toucra O48 LPS is 4-substituted, whereas in C. freundii O37 it has been shown to be 7-substituted [8]. Differences in the pattern of O-acetylation may also be significant. We report here the structure of the S. Toucra O48 O-specific polysaccharide.

Materials and methods

Bacterial strains, preparation of LPS and O-specific polysaccharides

S. enterica subspecies enterica serovar Toucra [(48:z:1,5,(z58)] strain KOS 1386 (PCM 2359) (formerly Salmonella toucra O48) was obtained from R. Glośnicka at the National Salmonella Centre, Gdańsk. The nomenclature of Salmonella strains is according the Kaufmann–White scheme revised by Popoff and Le Minor [9]. Bacteria were cultivated in Davis liquid broth supplemented with casein hydrolysate and yeast extract (Difco), with aeration at 37 °C for 24 h, harvested and freeze dried. LPS was isolated by phenol/water extraction [10] and purified on a column of Sepharose 2B, as described [6,11]. A portion of LPS (200 mg) was treated with 1% AcOH (20 mL) for 60 min at 100 °C and the carbohydrate-containing supernatant was fractionated on a column of Bio-Gel P-4 [1.6 cm × 100 cm]. Alternatively, a second portion of LPS (200 mg) was degraded with 0.1 m NaOH-AcOH (pH 4.5) at 100 °C for 5 h. Gel-permeation chromatography in 0.05 m aqueous pyridinium acetate buffer (pH 5.6), with monitoring of the eluate with a Knauer differential refractometer yielded eight fractions. The earliest eluting fraction in both cases, containing polysaccharide material, is the subject of the present investigation.

Analytical procedures

Sugar components were analysed as described [11–13]. Oligosaccharide and polysaccharide samples were hydrolysed with 1 m HCl for 4 h at 100 °C, followed by evaporation under a stream of N2. For GLC-MS analysis, a Hewlett-Packard 5971 A system equipped with a HP-1 glass capillary column (0.2 mm × 12 m), a temperature program of 8° min−1 from 150–270 °C, and 70 eV ionization potential was used. Sialic acid was determined with resorcinol reagent [14] and O-acetyl groups by the method of Hestrin [15]. For the analysis of N-acetyl groups the polysaccharide (0.4 mg) was hydrolysed as above, reduced with NaBH4 (20 mg, 16 h, 4 °C), acetylated with trideuteroacetic acid anhydride (Sigma) and pyridine (1 : 1, v/v), and the products analysed by GLC-MS.

Determination of the absolute configuration of the monosaccharides

The O-specific polysaccharide (1 mg) was hydrolysed with 4 m HCl for 12 h at 100 °C to release hexosamines. To determine the absolute configuration of glucosamine, half the material was treated with hexokinase in the presence of ATP. Complete phosphorylation of glucosamine was achieved as monitored by paper chromatography. For the determination of the configuration of fucosamine, the second part of hydrolysate was analysed by GLC-MS of the trimethylsilylated (R)-(–)-2-butyl glycosides [16]. The LPS from Hafnia alvei PCM 1220 (a gift from E. Katzenellenbogen) was hydrolysed and analysed in the same way to provide an l-FucNAc standard.

Methylation analysis

Methylation was performed according to Hakomori [17] and the product purified using a Sep Pak C18 cartridge. The methylated product was hydrolysed with 0.6 m HCl in 80% acetic acid at 80 °C for 18 h as previously shown [13], reduced with NaBD4 and acetylated for GLC-MS analysis, as described above. Part of the methylated material was methanolysed (1 m HCl in MeOH, 80 °C, 4 h) and peracetylated for GLC-MS analysis. The original lipopolysaccharide was also permethylated with the method of Prehm [18], purified by dialysis against water and by Sep Pak C18 cartridge, then methanolysed, peracetylated and analysed by GLC-MS as above.

NMR spectroscopy

Samples (5 mg) were dissolved in 300 µL of deuterated water, lyophilized, redissolved in the same volume of deuterated water, and introduced into a 5 mm susceptibility matched Shigemi tube. Spectra were recorded on a Varian Unity 500 NMR spectrometer equipped with a 5 mm triple-resonance pulsed field gradient (PFG) probe. Spectra were collected at a nominal probe temperature of 30 °C using standard Varian pulse sequences except for: the introduction of a spin echo sequence into the TOCSY and ROESY experiments; the use of the WHSQC experiment developed by Wider and Wüthrich [19] using pulsed field gradients to suppress artifacts; and the WHSQC-TOCSY sequence, an adaptation of the HMQC-TOCSY experiment of Crouch et al. [20] to incorporate the WHSQC sequence. Chemical shifts are referenced against internal TSP-d4 at zero p.p.m. (1H) and −1.8 p.p.m. (13C) [21]. The ROESY spectrum used a 150-ms 2.5-kHz spin lock generated by continuous low power irradiation, whilst the TOCSY spectrum used an 80-ms 10-kHz spin lock generated using an MLEV17 sequence. The spin lock period in the WHSQC-TOCSY spectrum was 20 ms, using a 10-kHz spinlock field generated as before.

Results

Lipopolysaccharide was obtained from the S. Toucra O48 strain in 3.8% yield by phenol/water extraction of dried cells and purification on a Sepharose 2B column [6,11]. The LPS exhibited a high-molecular-mass ladder-like pattern typical of smooth type LPS when analysed by SDS/PAGE and immunoblotting with specific rabbit antiserum [6]. Mild aqueous acetic acid hydrolysis released a lipid residue and a water-soluble carbohydrate, which was separated into eight fractions by gel filtration on a column of Bio-Gel P-4 (Fig. 1A). Oligosaccharide fraction 6, which contained sialic acid, was subsequently divided into two subfractions by affinity chromatography on Sepharose 4B-serotonin [11]. Fraction 6a, which was not retained, was eluted with water, and fraction 6b containing core material was desorbed with ammonium hydrogen carbonate buffer. The structures of the components of fractions 1–6 were deduced by sugar analysis, methylation analysis and NMR spectroscopy. Fractions 7 and 8 contained free sialic acid and Kdo, respectively, released from the LPS and were not considered further. When mild aqueous hydrolysis of LPS was performed with 0.1 m sodium acetate buffer, pH 4.5, a polysaccharide fraction (called fraction 1a) was obtained in high yield (Fig. 1B).

Figure 1.

Fractionation of the carbohydrate material obtained from acetic-acid-hydrolysed (A) and sodium-acetate pH 4.5 hydrolysed (B) S. Toucra O48 LPS. Fractionation was carried out on a Bio-Gel P-4 column (1.6 cm × 100 cm) equilibrated with pyridine/acetic acid, pH 5.6. The flow rate was 4 mL·h−1. The absorbance was measured at 490 nm for the phenol/sulphuric acid reaction and at 580 nm for the resorcinol reaction characteristic for sialic acid. Polysaccharide fractions (horizontal hatching), monomeric oligosaccharide unit (angled hatching), and released sialic acid (vertical hatching) are indicated.

Sugar analysis of fractions 1–6 (Fig. 1A) and 1a (Fig. 1B) after hydrochloric acid hydrolysis, GLC-MS of their alditol acetate derivatives, and colorimetric determination of sialic acid using the resorcinol method, showed that fractions 1, 1a and 6 from S. Toucra O48 had approximately the same composition, containing 6-deoxyhexosamine, neuraminic acid (Neu) and glucosamine (GlcN) (Table 1). The high proportion of glucosamine in the sugar analysis is probably due to partial degradation of 6-deoxyhexosamine and the interference with core material and the equimolar ratio of sugar residues was demonstrated by NMR, as reported below. The retention time in GLC-MS analysis of the 6-deoxyhexosamine (tRGlcN= 0.78) was the same as that of a fucosamine (FucN) standard. That the aminosugars are N-acetylated in the polysaccharide was deduced from GLC-MS spectra of the alditol perdeuterioacetates and from integration of the acetamido signals in the respective 1H NMR spectra. GlcNAc was shown to have the d configuration by quantitative reaction with hexokinase, which specifically phosphorylates the d-isomer. The l configuration was assigned to FucN on the basis of the GLC-MS analysis of the trimethylsilylated chiral butyl glycoside derivatives. That the oligosaccharide in fraction 6a originates from the O-specific polysaccharide was further corroborated by methylation analysis of polysaccharide and LPS.

Table 1. Sugar analysis (as molar ratios) of the poly and oligo-saccharides released from S. Toucra O48 lipopolysaccharide.
SugarMolar ratio for fractions
component11a234566a6b7
  1. a  Component not detected. b  Sialic acid was determined with resorcinol reagent [14].

Gal0.20.10.91.01.62.00.2a2.0
Glc0.30.11.01.52.02.00.21.8
FucNAc0.40.50.51.00.70.31.00.60.2
Hep0.20.10.71.01.41.60.21.7
GlcNAc1.01.01.92.41.81.01.71.01.3
Neu5Ac0.4b0.61.00.91.30.80.60.1+

Methylation analysis of fraction 1, 1a and original LPS showed that the O-specific polysaccharide was composed of the repeat unit present in fraction 6a, containing 3-substituted l-fucosamine (giving 1,3,5-tri-O-acetyl-2,6-dideoxy-4-O-methyl-2-methylacetamidohexitol) and 3-substituted 2-acetamido-2-deoxy-d-glucose (giving 1,3,5-tri-O-acetyl-4,6-di-O-methyl-2-methylacetamidohexitol). In fractions 6a and 6, the l-FucNAc is terminal, giving rise to a 1,5-di-O-acetyl-2,6-dideoxy-3,4-di-O-methyl-2-methylacetamidohexitol in the methylation analysis. These results indicate that in the polysaccharide the N-acetylneuraminic acid component is linked to O-3 of the fucosamine residue. In the S. Toucra O48 LPS, the sialic acid residue was shown to be 4-substituted, by identification of methyl 4-O-acetyl-5-N-acetyl-N-methyl-7,8,9-tri-O-methylneuraminate methyl glycoside obtained after methanolysis and acetylation of permethylated fractions 1 and 6, and intact LPS [8](Fig. 2A, peak 2). When the original LPS was subjected to methylation, methanolysis and perdeuterioacetylation, the gas chromatogram obtained also contained a peak from Neu5Ac4,5,7,8,9Me5 methyl ester methyl glycoside [8] (Fig. 2A, peak 1) indicating that the original LPS contained sialic acid in a terminal nonreducing position.

Figure 2.

Partial GLC-MS chromatogram of peracetylated methyl ethers of NeuAc methyl ester methyl glycosides from LPS of S. enterica ser. Toucra O48. Methylation was performed using the methods of Hakomori (A) and Prehm (B). Peaks from sialic acid methyl ester methyl glycoside derivatives are labelled as indicated: 1, Neu5Ac4,5,7,8,9Me5; 2, Neu5Ac5,7,8,9Me44Ac; 3, Neu5Ac5,7,8Me34,9Ac2; 4, Neu5Ac5,8Me24,7,9Ac3.

The 500 MHz 1D 1H NMR spectrum (Fig. 3) of fraction 1a at 30 °C showed high field resonances from a single C-methyl group (FucNAc H-6) at 1.20 p.p.m., four singlets from N-acetyl and O-acetyl groups at 1.95, 2.04, 2.13 and 2.30 p.p.m. (and several minor resonances with an intensity approximately 15% of that of the major residues), characteristic resonances from the Neu5Ac H-3a and H-3e at 1.80 and 2.89 p.p.m. Two major and one minor anomeric proton resonances were observed at 4.55, 5.15 and 5.18 p.p.m., respectively, and a small number of other resolved resonances. The 1H NMR spectrum of the major repeat unit was completely assigned from TOCSY and DQF-COSY spectra, and these assignments are reported in Table 2. The GlcNAc and FucNAc residues were distinguished by the magnitude of the H3/H4 coupling constant, estimated from the TOCSY spectrum, and the Neu5Ac residue was obvious from the H-3a and H-3e resonances. Assignment of spin systems began from well resolved resonances, usually anomeric but also the Neu5Ac H-3s, the FucNAc H-6s and the lowfield H-9s of the Neu5Ac(9OAc) residue. The 13C spectrum was assigned from a WHSQC spectrum optimized for 1JC,H of 150 Hz (Fig. 4), and this spectrum allowed many proton resonances from minor spin systems to be identified and partially assigned (Table 2). The assignments were confirmed by data from the ROESY spectrum, a WHSQC spectrum optimized for nJH,C of 20 Hz, and the WHSQC-TOCSY (obtained using a 20 ms mixing time). The individual N-acetyl and O-acetyl groups were assigned from the 150 ms ROESY spectrum (Fig. 5), taking into account that major resonances arose from repeat units with Neu5Ac 9-O-acetyl groups and minor resonances from repeat units containing Neu5Ac 7-O-acetyl groups, and from data on model systems [22–24].

Figure 3.

The 500 MHz 1H NMR spectrum ofthe isolated O-chain polysaccharide released from the LPS of S. Toucra O48 by hydrolysis with 0.1 M NaOAc buffer, pH 4.5.

Table 2. Assignments for proton and carbon NMR spectra of the O-chain polysaccharide from the S. Toucra O48 LPS. Proton chemical shifts are referenced to internal TSP-d4 at 0 p.p.m. and 13C chemical shifts to internal TSP-d4 at −1.8 p.p.m.; Numbers in square brackets represent tentative assignment.


Residue

H-1
C-1
H-2
C-2
NAc
NAc

H-3/H-3′
C-3

H-4
C-4
H-5
C-5
Nac
Nac

H-6/H-6′
C-6
H-7
C-7
OAc
OAc

H-8
C-8
H-9/H-9′
C-9
OAc
OAc
→4Neu5Ac(9OAc)α2→  1.80/2.893.733.913.613.583.894.44/4.13
major  40.278.151.1
 2.04
23.3
73.969.071.066.7
2.13
21.3
→4Neu5Ac(7OAc)α2→  1.82/2.903.733.88 5.05[3.79]3.85/3.64
minor    53.9
 1.92
24.4
 70.3
 2.16
 63.6
→3FucNAcα1→5.153.974.214.014.421.20   
major97.152.174.572.267.516.3   
2.30       
20.4       
→3FucNAcα1→5.183.87  4.451.21   
minor (1)97.050.4       
Terminal FucNAcα1→5.113.92 4.004.391.20   
minor (2)98.3[53.1]       
→3GlcNAcβ1→4.553.903.733.593.493.92/3.78   
major102.756.479.369.176.761.8   
1.95       
23.3       
Figure 4.

Partial 500 MHz HSQC NMR spectrum of the O-chain polysaccharide released from the LPS of S. Toucra O48, obtained at 30 °C. A number of key resonances are labelled as indicated: 1, GlcNAc H-1; 2, major FucNAc H-1; 3, Neu5,7Ac2 H-7; 4, Neu5,9Ac2 H-9; 5, GlcNAc H-2; 6, Neu5,9Ac2 H-5; 7, FucNAc H-2; 8, Neu5,9Ac2 H-3e; 9, Neu5,9Ac2 H-3a; 10, Neu5,9Ac2 H-4; and 11, GlcNAc H-3.

Figure 5.

Partial 500 MHz ROESY NMR spectrum of the isolated O-chain released from the LPS of S. Toucra O48. This region of the spectrum shows the interactions between the N-acetyl and O-acetyl resonances and the resonances from ring protons, and provides the data used to assign the individual N-acetyl resonances. Proton resonances are indicated as follows: 1, Neu5,9Ac2 H-3e; 2, Neu5,9Ac2 H-5; 3, FucNAc H-2; 4, FucNAc H-3; 5, Neu5,9Ac2 H-9; 6, Neu5,9Ac2 H-9′; 7, Neu5,9Ac2 H-8; 8, Neu5,9Ac2 H-7; 9, Neu5,9Ac2 H-6; 10, GlcNAc H-3 or Neu5Ac H-4; 11, GlcNAc H-1; 12, FucNAc H-1.

The linkages between sugar residues were determined from a ROESY spectrum obtained with a mixing time of 150 ms and a WHSQC spectrum optimized for nJC,H of 20 Hz, and were in full agreement with the pattern of 13C glycosidation shifts. In the ROESY spectrum, a strong NOE was observed between the major FucNAc H-1 and a resonance at 3.73 p.p.m., and weaker NOEs to resonances at 3.99, 3.58, and 3.49 p.p.m., assigned as GlcNAc H-3, H-2, H-4 and H-5, respectively. A long-range, trans-glycosidic heteronuclear correlation was observed between the FucNAc H-1 and the lowfield GlcNAc C-3 at 79.3 p.p.m. Together these data confirm the FucNAcα1→3GlcNAc linkage. A weak interresidue heteronuclear correlation was observed between the GlcNAc H-1 and the Neu5Ac C-4 at 78.1 p.p.m., but the expected interresidue NOE is coincident with the intraresidue NOE between the GlcNAc H-1 and H-3. The data is consistent with a GlcNAcβ1→4Neu5Ac linkage. Inter-residue NOEs were observed between the Neu5Ac H-3e and the FucNAc H-1 and between the Neu5Ac H-3a and the FucNAc H-3 and H-4 resonances at 4.21 and 4.01 p.p.m., confirming the Neu5Acα2→3FucNAc linkage. These results allow the following saccharide structure to be proposed for the repeat unit of the O-specific polysaccharide: →4)-α-Neup5Ac(2→3)-l-α-FucpNAc(1→3)-d-β-GlcpNAc(1→. The 13C NMR data is consistent with the proposed substitution pattern, by comparison with data for the →3-l-FucNAcα1→3-d-GlcNAcβ1→ substructure in the O-chain from the LPS of Proteus vulgaris O19 [25] and the →4Neu5Acα2→ residues in the CPSs from Neisseria meningitidis groups Y and W-135 [26].

The isolated O-chain (in fractions 1 and 1a) was substituted with a single equivalent of O-acetyl groups, distributed between the Neu5Ac O-9 and O-7 in an approximate ratio of 10 : 1. The locations of the O-acetyl groups were determined from the proton and 13C chemical shifts at the acetylation sites, and compared with data obtained for free sialic acids [22,23] and the CPSs from N. meningitidis groups C, Y and W135 [24]. For the more intense Neu5Ac spin system, the lowfield chemical shifts of H-9 and H-9′ (4.44 and 4.13 p.p.m., 13C for C-9 at 66.7 p.p.m.) are characteristic of 9-O-acetylation, whilst in the less intense Neu5Ac spin system, the lowfield chemical shift of H-7 (5.04 p.p.m.) indicates 7-O-acetylation. During the period of approximately three weeks over which the NMR spectra were collected, minor changes in the spectra indicated that the sample was slowly degrading, although this was not sufficient to compromise the assignment of the spectra or the structure determination. In the original lipopolysaccharide, the O-acetyl groups were distributed between the Neu5Ac O-9 and O-7 in an approximate ratio of 7 : 3, as deduced from methylation analysis of LPS performed with the method of Prehm (Fig. 2B). This method of permethylation of sodium salt of LPS with methyl trifluoromethanesulphonate in trimethyl phosphate preserves O-acetyl substituents [18]. Two major peaks of Neu5Ac derivatives were identified as methyl 4,9-di-O-acetyl-5-N-acetyl-N-methyl-7,8-di-O-methylneuraminate methyl glycoside (peak 3, ions of m/z 404, 346, 298, 254, 157, 117) from 9-O-acetylation and as methyl 4,7,9-tri-O-acetyl-5-N-acetyl-N-methyl-8-O-methylneuraminate methyl glycoside (peak 4, ions of m/z 432, 298, 157, 117) derived from 7,9-di-O-acetylation.

The type of linkage between O-specific polysaccharide and the core oligosaccharide was deduced by the comparative analysis of fractions 4 and 5 (Table 1). From the sugar and the methylation analysis, fraction 5 appeared to contain a complete core oligosaccharide of Salmonella Ra type, with a terminal GlcNAc residue. Fraction 4 contained an additional disaccharide fragment linked to an Ra core oligosaccharide at O-4 of the subterminal glucopyranose unit, since it yielded the same methylated derivatives as oligosaccharide 5 but with additional stoichiometric amounts of FucNAc and GlcNAc methyl ethers corresponding to the l-α-FucpNAc(1→3)-d-β-GlcpNAc sequence. Also, the presence of 3,6-di-O-methyl-glucose in place of a 3,4,6-tri-methyl ether from fraction 5, indicated that the site of attachment of the O-specific chain to the core is at position O-4 of its subterminal d-glucose unit.

Discussion

Sialic acid was detected in serotype O48 of Salmonella in 1963 [27], and a serological relationship with one of the Citrobacter strains demonstrated [28]. As it was thought at that time that sialic acid is not a component of LPS, its presence in Salmonella O48, Arizona O29 and Citrobacter O37 was attributed to contaminants. Mikulaszek and coworkers, in 1968, established that sialic acid in Salmonella O48 is a component of LPS [29], a point later proven by Ke¸dzierska's work on S. Toucra O48 [1]. The antigen O48 of Salmonella occurs in several serovars and, historically, was classified as serogroup Y [9]. Serotype O48 belongs to clinically important bacteria causing intestinal disfunctions and diarrhoea in infants and children [30].

The lipopolysaccharide S. Toucra O48 is smooth type, with a long-chain polysaccharide [6]. The LPS contains both terminal and 4-O-substituted sialic acid [8], indicating that the biological repeating unit has a sialic acid residue at the nonreducing terminus.

In contrast to previous observations [1,7,28], no distinct serological cross-reactions were detected between S. Toucra O48 and C. freundii O37, when tested in immunoblotting and passive hemagglutination experiments [6]. The only difference found to date between these two polysaccharides is the substitution of the sialic acid residue. In the case of S. Toucra O48, O-acetyl residues present in this polysaccharide play an important role in the interaction with specific antibodies (A. Korzeniowska-Kowal, unpublished observations). In Salmonella arizonae O21, a corresponding trisaccharide repeating unit possesses a 7-substituted Neu5Ac residue, a GlcNAc residue O-acetylated at position 6 and 2-acetamidino-Fuc [31]. In Salmonella Djakarta, one of the O48 serovars, sialic acid is present exclusively as terminal residues at the 2-position of a glucose unit in the Ra core region in short S-type LPS [32]. The LPS of S. Isaszeg O48 consists of regular O-specific repeating units, containing both O-acetylated terminal and 4-substituted Neu5Ac residues [32]. The studies of Mayer and coworkers on S. Djakarta and S. Isaszeg [32] and these on S. Toucra [6,8,33] LPSs, indicate that serotype O48 is a heterogeneous group of sialic acid containing cross-reacting structures. Further structural investigation of the remaining O48 serovars may shed light on the role of sialic acid.

The linkage of the O-specific polysaccharide to the O-4 position of the subterminal glucose residue of the core oligosaccharide in S. Toucra O48 LPS is the same as in other strains of this genus [34].

The location of the O-acetyl group in S. Toucra O48 LPS on Neu5Ac O-9 and O-7 was also found in Escherichia coli O104 LPS [12]. It is not possible to deduce from these data the position on the Neu5Ac to which the O-acetyl group is delivered during LPS biosynthesis. The spontaneous O-acetyl migration is a process that would have been promoted by the conditions used to liberate the O-chain from lipid A. Methylation analysis of the original LPS is a way to determine the actual location of O-acetyl groups.

Acknowledgements

This work was supported by grants 51/99 from the Polish Academy of Sciences and 4.P05A.038.10 from the Committee for Scientific Research (KBN), Poland. Mr Xavier Lemercinier (NIBSC) is thanked for programming the WHSQC-TOCSY pulse sequence, and Dr E. Katzenellenbogen for the l-FucNAc reference material.

Footnotes

  1. *Present address: Chiron SpA., Via Fiorentina 1, 53100 Siena, Italy.

Ancillary