ABSTRACT
Hafnia alvei, a gram-negative bacterium, is an opportunistic pathogen associated with mixed hospital infections, bacteremia, septicemia, and respiratory diseases. Various 3-deoxy-d-manno-oct-2-ulosonic acid (Kdo)-containing fragments different from known structures of core oligosaccharides were previously found among fractions obtained by mild acid hydrolysis of some H. alvei lipopolysaccharides (LPSs). However, the positions of these segments in the LPS structure were not known. Analysis of de-N,O-acylated LPS by nuclear magnetic resonance spectroscopy and mass spectrometry allowed the determination of the location of a Kdo-containing trisaccharide in the structure of H. alvei PCM 32 LPS. It was established that the trisaccharide {l-α-d-Hepp-(1→4)-[α-d-Galp6OAc-(1→7)]-α-Kdop-(2→} is an integral part of the outer-core oligosaccharide of H. alvei 32 LPS. The very labile ketosidic linkage between →4,7)-α-Kdop and →2)-Glcp in the core oligosaccharide was identified. Screening for this Kdo-containing trisaccharide was performed on the group of 37 O serotypes of H. alvei LPSs using monospecific antibodies recognizing the structure. It was established that this trisaccharide is a characteristic component of the outer-core oligosaccharides of H. alvei 2, 32, 600, 1192, 1206, and 1211 LPSs. The weaker cross-reactions with LPSs of strains 974, 1188, 1198, 1204, and 1214 suggest the presence of similar structures in these LPSs, as well. Thus, we have identified new examples of endotoxins among those elucidated so far. This type of core oligosaccharide deviates from the classical scheme by the presence of the structural Kdo-containing motif in the outer-core region.
Lipopolysaccharide (LPS) (endotoxin) is the main surface antigen and an important virulence factor of most of the gram-negative bacteria that are pathogenic for humans and animals (46). LPS contributes greatly to the structural integrity of bacteria and constitutes a “pathogen-associated molecular pattern” for host infection (46). As one of the most potent natural activators of the innate immune system, LPS is recognized by different classes of receptors present on macrophages, monocytes, B and T cells, neutrophils, endothelial cells, and epithelial cells (46). Endotoxins stimulate these cells to produce multiple inflammatory mediators responsible for immunotoxicity (e.g., tumor necrosis factor alpha, interleukin 1 [IL-1], IL-6, IL-8, gamma/alpha interferon, NO, platelet-activating factor, and endorphins). Interaction of LPS with the CD14/Toll-like receptor 4/MD-2 receptor complex constitutes a major mechanism responsible for the innate immune response to infection by gram-negative bacteria (1, 46). A large amount of LPS released into the bloodstream triggers the excessive inflammatory response of the innate immune system, leading to sepsis and septic shock (6). High levels of inflammatory mediators have profound effects on the cardiovascular system, kidneys, lungs, liver, central nervous system, and coagulation system. Following their action, renal failure, myocardial dysfunction, acute respiratory distress syndrome, hepatic failure, and disseminated intravascular coagulation can occur, which may result in death (6). Despite intense research on the etiology and treatment of sepsis, its severe form still carries a high mortality rate (6, 46).
Hafnia alvei has been reported to be an opportunistic human pathogen. This gram-negative bacterium and its LPS are among the identified causative agents of bacteremia and septicemia in humans and animals (19). For the years 2001 to 2003, up to 42 cases of H. alvei bacteremia were reported annually in the United Kingdom. Most of them were monomicrobic infections, and in ∼33% of the cases, H. alvei was isolated, not only from the blood, but also from hepatic abscesses, pancreatic pseudocyst fluid, sputum, feces, and central venous catheters (19). Besides bacteremia and sepsis, which seem to be the most common syndromes reported, H. alvei is also associated with respiratory diseases and mixed hospital infections in humans. Since the gastrointestinal and respiratory tracts represent very common habitats for hafniae, most cases of H. alvei bacteremia originate there. H. alvei sepsis is also a serious clinical problem in the animal production industry, as infections of H. alvei can be severe, causing septicemia in commercial laying hens, pullets, and rainbow trout (19).
Our knowledge of the pathogenicity of H. alvei is limited. LPS is the major virulence factor in cases of H. alvei septicemia and bacteremia (19). Studies of other virulence factors of H. alvei have reported only on the iron-scavenging mechanism, mannose-sensitive/mannose-resistant hemagglutinins, and plasmids encoding bacteriocins and alveicins (19).
Most of the elucidated structures of H. alvei LPS are smooth-type molecules built up of O-specific polysaccharide, core oligosaccharide (OS), and lipid A. The O antigens of H. alvei are subdivided into 40 O serotypes (2, 28, 42). The structures of the O-specific polysaccharides from 30 serologically different H. alvei strains have been elucidated (15, 24, 26, 28, 42).
So far, four types of core OS have been identified for H. alvei LPSs (9, 17, 25, 27, 30, 43). The most common core OS, isolated by mild acidic hydrolysis from LPSs of smooth H. alvei strains, is a hexasaccharide composed of two d-Glc, three l,d-Hep, and one 3-deoxy-d-manno-oct-2-ulosonic acid (Kdo) residues. Two l,d-Hep residues are substituted by phosphoethanolamine (PEtn) and phosphoryl (P) groups (9, 17, 25). In LPSs isolated from H. alvei PCM 1185 and 1204, core OSs are terminated with d-Galp instead of d-Glcp (27). LPSs of H. alvei containing nontypical core OSs, identical with those found in LPSs of Escherichia coli R4 (strains 23 and 1222) and Salmonella enterica Ra (strain 39), were also identified (43).
The chemical structures of H. alvei O-specific chains and core OSs were elucidated using fractions obtained by mild acid hydrolysis of LPS. The procedure was optimized for the delipidation of LPS, exploiting the susceptibility of a ketosidic linkage between the inner core and lipid A to acid. However, other acid-labile linkages within the LPS could also be affected, leading to partial degradation of the isolated molecules.
The presence of Kdo-containing OSs among fractions obtained by mild acid hydrolysis of LPSs, other than previously identified core OSs, makes the structural analysis of entire H. alvei LPSs difficult. Two types of trisaccharides were previously identified: (i) l-α-d-Hepp-(1→4)-[α-d-Galp-(1→7)]-α-Kdop (l-α-d-Hep is α-l-glycero-d-manno-heptose) for strains 2, 1211, 32, and 1192 (16, 23) and (ii) α-d-Galp-(1→2)-l-α-d-Hepp-(1→4)-α-Kdop for strains 1188 and 1196 (22). These Kdo-containing motifs were never located in any of the LPS segments. Thus, it is of interest to complete the structure of H. alvei LPSs with the locations of such acid-labile motifs in the structures of LPSs isolated from these bacteria.
We now report on structural studies of de-N,O-acylated LPS of H. alvei 32 containing a carbohydrate backbone of lipid A, core OS, an additional trisaccharide in the outer region of the core OS, and all of the acid-labile substituents. Additionally, data obtained previously for a trisaccharide isolated from H. alvei 32 LPS (16) was complemented with detailed 1H and 13C nuclear magnetic resonance (NMR) analyses and the assignment of all proton and carbon signals. Screening for the presence of these acid-labile trisaccharides, identical with those found in the strain 32 LPS, was performed on 37 different O serotypes of H. alvei LPSs with antibodies against the conjugate of the de-N,O-acylated H. alvei 32 endotoxin fragment with bovine serum albumin (BSA), specific for the isolated trisaccharide.
(Part of this work was presented at the 21st International Carbohydrate Symposium, Cairns, Australia, 7 to 12 July 2002, and the 3rd German-Polish-Russian Meeting on Bacterial Carbohydrates, Wroclaw, Poland, 6 to 9 October 2004.)
MATERIALS AND METHODS
Bacteria. H. alvei strains PCM 1, 1 M, 2, 17, 23, 31, 32, 37, 38, 39, 481, 600, 744, 974, 981, 1188, 1190, 1191, 1192, 1195, 1198, 1200, 1203, 1204, 1205, 1206, 1207, 1209, 1210, 1211, 1212, 1213, 1214, 1215, 1221, 1224, and 4221 were obtained from the Polish Collection of Microorganisms (PCM) at the Institute of Immunology and Experimental Therapy (Wroclaw, Poland). The bacteria were grown in Davis medium, killed with 0.5% phenol, and centrifuged using a CEPA flow laboratory centrifuge (39).
Preparation of LPSs and core OSs.LPSs were extracted from bacterial cells by the hot-phenol/water method (53) and purified by ultracentrifugation as previously described (39). The yields of LPS preparations were 2 to 3.5%. Polysaccharides and OSs were isolated by mild acidic hydrolysis (1.5% acetic acid) at 100°C for 2 h, fractionated, and purified as previously described (16, 37).
De-N,O-acylation of LPS.De-O-acylation of LPS 32 was achieved by mild hydrazinolysis (14) modified as previously described (36). Briefly, LPS (200 mg) was dissolved in anhydrous hydrazine (5 ml), and the reaction was carried out at 37°C for 30 min. The mixture was cooled and added to cold acetone (−20°C) to terminate the reaction by conversion of hydrazine to acetone hydrazone. The precipitate of the de-O-acylated LPS was collected by centrifugation (4,000 × g; −20°C; 30 min), dissolved in water, and freeze-dried. The de-O-acylated LPS was dissolved in aqueous 4 M KOH (10 ml) and hydrolyzed in a sealed tube under nitrogen (120°C; 16 h). The reaction mixture was neutralized with HClO4 at 0°C. Most of the insoluble KClO4 sediment was removed by centrifugation and further desalted using a Bio-Gel P-2 column equilibrated with 0.05 M pyridine/acetic acid buffer at pH 7.0. The fraction with the highest molecular weight was further fractionated by gel permeation chromatography performed on a column (1.6 cm by 100 cm) of Bio-Gel P-10 equilibrated with 0.05 M pyridine/acetic acid buffer at pH 7.0. Fractions were collected, freeze-dried, and checked by one-dimensional (1D) and 2D NMR spectroscopy. The fraction showing NMR resonances of the structure reporter groups typical of the intact Kdo-containing core regions (OS32) was rechromatographed and chosen for further structural analysis.
Analytical procedures.Methylation was performed on OS according to the method of Hakomori (13). Alditol acetates and partially methylated alditol acetates were analyzed by gas chromatography-mass spectrometry (GC-MS) with a Hewlett-Packard 5972 system using the HP-1 fused-silica capillary column (0.2 mm by 12.5 m) and a temperature program of 150 to 270°C at 8°C min−1. The absolute configurations of monosaccharides were determined as described by Gerwig et al. using (R)-2-butanol for the formation of 2-butyl glycosides (10, 11). Prior to the absolute-configuration analysis, dephosphorylation of residues substituted by P or PPEtn was performed as previously described (34). Briefly, OSs (2 mg) were treated with aqueous 48% HF (1 ml; 72 h; 4°C) and then concentrated to dryness by evaporation. The trimethylsilylated butyl glycosides were identified by comparison with the authentic samples produced from carbohydrate standards (Sigma, St. Louis, MO) and (R/S)-2-butanol (Fluka, Buchs, Switzerland) on GC-MS. This analysis was carried out with a Hewlett-Packard 5971A system using an HP-1 fused-silica capillary column (0.2 mm by 12 m) and a temperature program of 100 to 270°C at 8°C/min−1.
MS.Matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) MS was carried out on a Kratos Kompact-SEQ instrument as described previously (34). 2,4,6-Trihydroxyacetophenone (25 mg/ml acetonitrile-water; 1:1 [vol/vol]) was used as a matrix for the analysis of OSs. Electrospray ionization (ESI)-MS analysis of the de-N,O-acylated H. alvei 32 LPS (OS32) was performed on a micrOTOF-Q spectrometer (Bruker Daltonics, Bremen, Germany) in the negative-ion mode. The sample was dissolved in acetonitrile-water solution (1:1 [vol/vol]; 0.2 mg/ml) and analyzed by direct infusion at a rate of 3 μl/min. Spectra were scanned in the m/z range of 120 to 2,000. The ion source temperature was 180°C, the flow rate was 4 liters/min, and the pressure of nitrogen was 4 × 104 Pa. External calibration in the negative-ion mode was applied using the Tune Mix mixture (Bruker Daltonics, Germany) in quadratic-regression mode in the m/z range of 113 to 2,234.
NMR spectroscopy.All NMR spectra were obtained on Bruker DRX 400 and DRX 600 spectrometers. NMR spectra of the isolated polysaccharides and OSs were obtained for 2H2O solutions at 30°C using acetone (δH 2.225; δC 31.05) as an internal reference. In 31P NMR spectroscopy experiments, phosphorus resonances were referenced to external 85% phosphoric acid (δ 0 ppm) in a separate experiment. OSs were repeatedly exchanged with 2H2O (99.95%) with intermediate lyophilization. The data were acquired and processed using standard Bruker software. The processed spectra were assigned with the help of SPARKY (12). The signals were assigned by 2D experiments (correlated spectroscopy, clean total-correlation spectroscopy [TOCSY], nuclear Overhauser effect spectroscopy [NOESY], rotating-frame NOESY, heteronuclear multiple-bond correlation [HMBC], heteronuclear single-quantum coherence-distortionless enhancement by polarization transfer [HSQC-DEPT], and HSQC with and without carbon decoupling). In the clean-TOCSY experiments, the mixing times used were 30, 60, and 100 ms. The delay time in HMBC was 60 ms, and the mixing time for NOESY was 200 ms.
Conjugation of OS32 with BSA.BSA (10 mg) was activated with glutardialdehyde under nitrogen (1%; carbonate/bicarbonate buffer, pH 9.0; 4 h; 21°C). Excess glutardialdehyde was removed by dialysis of the mixture against carbonate/bicarbonate buffer (50 mM; pH 9.0). The activated BSA was mixed with the dodecasaccharide OS32 (5 mg), and the conjugation was carried out for 16 h at 21°C, followed by treatment with NaBH4 (10 mg/ml; 2 h), and dialyzed extensively against phosphate-buffered saline (PBS) (pH 7.4) (5, 21).
Immunization procedures.Rabbits were housed at the animal facility of the Institute of Immunology and Experimental Therapy (Wroclaw, Poland). The rabbits were immunized with 50 μg of OS32-BSA conjugate suspended in complete Freund's adjuvant, and polyclonal antibodies against the conjugate were obtained by procedures described previously (21, 32). All experiments were carried out according to the procedures approved by the local ethics commission.
Isolation of trisaccharide-specific antibodies from polyclonal serum.Trisaccharide-specific antibodies were prepared by absorbing crude antiserum obtained by immunization of rabbits with OS32-BSA conjugate with H. alvei 1207 bacterial cells, whose LPS is devoid of trisaccharide (C. Lugowski, T. Niedziela, W. Jachymek, and J. Lukasiewicz, unpublished data). Killed, freeze-dried bacteria (2 g) were suspended in PBS for 24 h. The bacteria were centrifuged (1,000 × g at 4°C for 30 min) and rinsed three times with 30 ml of PBS. The serum (10 ml; diluted 1:3 in PBS) was added to the bacteria suspended in PBS. The mixture was incubated on a rocking platform for 16 h at 22°C and centrifuged. The supernatant was added to a fresh suspension of H. alvei 1207 bacteria, incubated (16 h at 22°C), and centrifuged. The supernatant was filtered (0.2-μm pores), collected in sterile vials, and stored at −70°C.
ELISA and ELISA inhibition test.Enzyme-linked immunosorbent assay (ELISA), using LPS as the solid-phase antigen, was performed by a modification (35) of the method described by Voller et al. (52). In the inhibition studies, the trisaccharide-specific serum (100 μl) at a concentration twice as high as that giving an A 405 in the range 0.5 to 0.8 was mixed with serial dilutions of the trisaccharide isolated from H. alvei 32 (100 μl) and incubated for 1 h at 37°C. The mixture (100 μl) was then transferred into the wells of a microtiter plate coated with H. alvei 32 LPS, and the reaction was carried with shaking (15 min; 22°C; pH 7.3). Washing of the wells, reaction with a second antibody conjugated with alkaline phosphatase, and color development were performed as described for ELISA.
SDS-PAGE.The LPS was analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) according to the method of Laemmli (29) with previously described modifications (15). LPS bands were visualized by the silver-staining method (45).
Immunoblotting.Immunoblotting was performed on the SDS-PAGE-separated LPS fractions as previously described (31). Goat anti-rabbit immunoglobulin G conjugated with alkaline phosphatase (Bio-Rad) was used as the second antibody, and 5-bromo-4-chloro-3-indolyl phosphate/nitroblue tetrazolium was applied as a detection system.
RESULTS
Isolation and initial analysis of the de-N,O-acylated LPS and the Kdo-containing trisaccharide.The mixture of the de-N,O-acylated LPSs of H. alvei 32 was fractionated by gel filtration (see Fig. S1A and B in the supplemental material). All isolated fractions were checked by 1D and 2D 1H and 13C NMR spectroscopy. The fraction showing characteristic NMR resonances of three pairs of signals originating from deoxy protons of Kdo and two signals of nitrogen-bearing carbons from aminosugars of lipid A (Fig. 1 and 2) was selected for structural analysis of the intact Kdo-containing core regions (OS32; 11 mg).
1H-NMR spectrum of the dodecasaccharide OS32 (structure shown at top) isolated from de-N,O-acylated H. alvei 32 LPS. The 1H-NMR spectrum was obtained for 2H2O solutions (pH ∼8.5) at 600 MHz and 30°C. Residual water was removed by processing. The capital letters in the anomeric regions of the 1H-NMR spectra refer to carbohydrate residues, as shown in the structure. The Arabic numerals refer to protons in the respective residues. Residues h and j are constituents of OS devoid of the trisaccharide (framed with a dotted line) and represent variants of residues H and J (Table 2). M1 indicates substitution of the terminal α-d-Galp residue (M) at O-6 by an O-acetyl group identified for the trisaccharide isolated by mild acid hydrolysis of H. alvei 32 LPS.
Selected parts of HSQC-DEPT and HMBC spectra of OS32 isolated from de-N,O-acylated H. alvei 32 LPS. The spectra were obtained for 2H2O solutions (pH ∼8.5) at 600 MHz and 30°C. The capital letters refer to carbohydrate residues, as shown on the structure in Fig. 1, and the numbers refer to protons and carbons in the respective residues. Residues h and j are constituents of OS devoid of the trisaccharide and represent form of residues H and J.
The trisaccharide component of the core OS of H. alvei 32 LPS was isolated as previously described (16). Briefly, polysaccharides and OSs were released by mild acidic hydrolysis of H. alvei 32 LPS and fractionated by gel filtration on Bio-Gel P-10 (see Fig. S1C in the supplemental material), followed by Bio-Gel P-2 chromatography (see Fig. S1D in the supplemental material). The isolated polysaccharides and OSs were examined by MALDI-TOF MS (data not shown) and compared with results published previously for the O-specific chain of H. alvei 32 and the core OSs of H. alvei 32 and 1192 (16, 17, 33). The trisaccharide-containing fraction was identified by 1H and 13C NMR and selected for further investigation.
Structural analysis of the purified de-N,O-acylated LPS isolated from H. alvei 32.Monosaccharide and absolute-configuration analyses of the dephosphorylated OS32 revealed the presence of d-GlcN, l-glycero-d-manno-Hep, d-Glc, and d-Gal. Methylation analysis of native OS32 showed the presence of terminal d-Glcp, 3-substituted d-Glcp, and terminal l,d-Hepp, residues identified previously for core OSs obtained by mild acid hydrolysis of H. alvei 32 LPS (17). The analysis also revealed additional terminal d-Galp and 2-substituted d-Glcp. The terminal d-Galp and 2-substituted d-Glcp had not been identified before in core OS of H. alvei 32 obtained by mild acid hydrolysis of the LPS.
OS32 was investigated by NMR spectroscopy. 2D NMR experiments (correlated spectroscopy, TOCSY, HSQC-TOCSY, NOESY, and HSQC-DEPT) were recorded and analyzed. All NMR analyses were performed at pH ∼8.5, using 1.9-mm capillary tubes.
Nine major anomeric proton and carbon resonances, two signals of nitrogen-bearing carbons, and, in addition, three Kdo spin systems were identified in spectra of OS32 (Fig. 1 and 3A and Tables 1 and 2). (Note that the capital letters denoting residues, as shown in the structure of the isolated dodecasaccharide [Fig. 1, top], refer to the corresponding residues throughout the text, tables, and figures.)
Parts of NOESY spectra of the OS32 isolated from de-N,O-acylated H. alvei 32 LPS. (A) Region of deoxy protons of α-Kdop residues. (B) Region of anomeric protons. The cross-peaks are labeled as explained in the legend to Fig. 1.
1H and 13C NMR chemical shifts of the dodecasaccharide isolated from de-N,O-acylated H. alvei 32 (OS32; pH ∼8.5) a
Selected interresidue NOE and 3 J H,C connectivities from the anomeric atoms of the dodecasaccharide OS32 (pH ∼8.5) isolated from de-N,O-acylated H. alvei 32 LPS
Residue A, with the H-1/C-1 signals at δ 5.48/94.0 ppm and a J H-1,H-2 of ∼3.4 Hz, was recognized as the 6-substituted α-d-GlcpN1P. The low chemical shift of the C-2 signal (δ 56.4), the high chemical shift of the C-6 signal (δ 70.3), and the large vicinal coupling constants between H-2 and H-3, H-3 and H-4, and H-4 and H-5 (J H-2,H-3, J H-3,H-4, and J H-4,H-5, ∼10 Hz) were observed. A 1H and 31P correlation experiment showed the connectivity between the phosphate monoester peak at δP 1.8 ppm and H-1 (δ 5.48) and H-2 (δ 3.00) of residue A (3 J H-1,P, ∼8 Hz).
Residue B, with the H-1/C-1 signals at δ 4.68/103.1 ppm and a J H-1,H-2 of 8.6 Hz, was assigned as the 6-substituted β-d-GlcpN on the basis of the typical chemical shifts of the H-1, C-1, and C-2 (δ 57.5) signals and the large J values for the vicinal couplings between all ring protons. The small downfield shift of the C-6 signal (δ 62.7) indicated substitution by a Kdo residue, which in general induces only a weak α effect on C-6 of residue B (4, 38). The residues A and B constitute the disaccharide backbone →6)-β-d-GlcpN-(1→6)-α-d-GlcpN1P of lipid A, as was further supported by HMBC data.
Residues C, D, and K were identified as different Kdo molecules. Chemical-shift values of H-3ax and H-3eq protons in the 1.7-to-2.38 ppm range, together with the chemical-shift values of H-5 and C-2, indicated the α-pyranosidic configuration of all Kdo residues (3, 47).
Residue C was identified as the 4,5-disubstituted pyranose form of 3-deoxy-α-d-manno-oct-2-ulosonic acid (α-Kdop) on the basis of characteristic deoxy proton signals at δ 1.90 ppm (H-3ax) and δ 2.21 ppm (H-3eq) and high chemical shifts of the C-4 (δ 72.3) and C-5 (δ 69.6) signals.
Residue D was identified as terminal α-Kdop. Characteristic deoxy proton signals were found at δ 1.76 ppm (H-3ax) and δ 2.14 ppm (H-3eq). The chemical-shift values of the assigned resonances were in agreement with previously published data for terminal α-Kdop (44, 47, 49).
Residue K was identified as the 4,7-disubstituted α-Kdop on the basis of characteristic deoxy proton signals at δ 1.88 ppm (H-3ax) and δ 2.30 ppm (H-3eq) and high chemical shifts of the C-4 (δ 72.3) and C-7 (δ 78.9) signals.
Residue E, with the H-1/C-1 signals at δ 5.22/100.0 ppm and a J H-1,C-1 of ∼172 Hz, was recognized as 3-substituted l-glycero-α-d-manno-Hepp4P from the 1H and 13C chemical shifts, small vicinal couplings between H-1 and H-2 and between H-2 and H-3, and the relatively large chemical shift of the C-3 signal (δ 78.2). A 1H,31P-HMQC experiment revealed the connectivity between the phosphate monoester peak at δP 2.8 ppm and H-4 (δ 4.32) of residue E, indicating that the →3)-l-α-d-Hepp residue was substituted at O-4 with a phosphate group.
Residue F, with the H-1/C-1 signals at δ 5.16/103.8 ppm, a J H-1,H-2 of <2 Hz, and a J H-1,C-1 of ∼176 Hz, was recognized as the 3,7-disubstituted l-glycero-α-d-manno-Hepp4P from the 1H and 13C chemical shifts, small J values for the vicinal couplings between H-1 and H2 and between H-2 and H-3, and the high chemical shifts of the C-3 (δ 79.4) and C-7 (δ 70.1) signals. A 1H and 31P-HMQC experiment revealed the connectivity between the phosphate monoester signal at δP 2.8 ppm and H-4 (δ 4.34) of residue F, indicating that the →3,7)-l-α-d-Hepp residue was substituted at O-4 with a phosphate group.
Residue G, with the H-1/C-1 signals at δ 4.89/101.7 ppm, a J H-1,H-2 of <2 Hz, and a J H-1,C-1 of ∼170 Hz, and residue L, with the H-1/C-1 signals at δ 5.08/98.7 ppm, a J H-1,H-2 of <2 Hz, and a J H-1,C-1 of ∼173 Hz, were recognized as terminal l-glycero-α-d-manno-Hepp due to the small vicinal couplings between H-1, H-2, and H-3.
Residue H, with the H-1/C-1 signals at δ 5.23/102.4 ppm and a J H-1,H-2 of 2.7 Hz, was assigned as 3-substituted α-d-Glcp based on the large chemical shift of the C-3 signal (δ 80.3) and the large vicinal couplings among H-2, H-3, H-4, and H-5.
Residue J, with the H-1/C-1 signals at δ 5.46/96.8 ppm and a J H-1,H-2 of 2.6 Hz, was assigned as the 2-substituted α-d-Glcp residue based on the relatively high chemical shift of the C-2 signal (δ 74.0) and the large vicinal couplings between H-2 and H-3, H-3 and H-4, and H-4 and H-5.
Residue Μ, with the H-1/C-1 signals at δ 5.20/101.8 ppm and a J H-1,H-2 of 3.8 Hz, was assigned as the terminal α-d-Galp residue due to the large coupling constant between H-2 and H-3 and the small vicinal couplings between H-4 and H-5.
All assigned resonances were in agreement with previously published data (16, 17, 20, 44, 47, 49).
Each disaccharide element in OS32 was identified by HMBC (Fig. 2) and NOESY (Fig. 3) experiments showing interresidue connectivities between adjacent sugar residues and providing the sequence of monomers in the dodecasaccharide (Fig. 1, top).
Interresidual nuclear Overhauser effects (NOEs) were identified between H-1 of L and H-4 of K, H-1 of M and H-7 of K, H-1 of J and H-3 of H, H-1 of H and H-3 of F, H-1 of G and H-7a and -b of F, H-1 of F and H-3 of E, H-1 of E and H-5 of C, and H-1of B and H-6a of A. Additionally, the interresidue NOE signals between H-3ax (δ 1.90) and H-3eq (δ 2.21) of residue C and H-6 of residue D (δ 3.65) (Fig. 3) supported the presence of a disaccharide element in the inner-core region: α-Kdop-(2→4)-α-Kdop-(2→ (3). The HMBC spectra showed cross-peaks between the anomeric proton and the carbon at the linkage position and between the anomeric carbon and the proton at the linkage position (Table 2), which confirmed the sequence of sugar residues in the dodecasaccharide.
Residues K, L, and M constitute the trisaccharide previously found among fractions released by the mild acid hydrolysis of H. alvei LPS 32 (16). HMBC correlation between C-2 of residue K and H-2 of residue J confirmed the presence of an acid-labile α-(2→2) ketosidic linkage between the α-Kdop (K) of the trisaccharide and the α-d-Glcp (J).
Two additional signals were present in all NMR spectra (Fig. 1, 2, and 3 and Tables 1 and 2). Residue j (H-1/C-1 signals at δ 5.40/99.9 ppm, a J H-1,H-2 of ∼2.7 Hz, and a J H-1,C-1 of ∼171 Hz) was recognized as terminal α-d-Glcp. Residue j represented a nonsubstituted variant of residue J [→2)α-d-Glcp] and formed a nonreducing end in OS devoid of the trisaccharide. Residue h (H-1/C-1 signals at δ 5.28/101.8 ppm and a J H-1,H-2 of <2 Hz) was identified as a variant of residue H (3-substituted α-d-Glcp) created by the lack of the trisaccharide.
This minor structural heterogeneity of OS32 was also examined by ESI-MS analysis (Fig. 4 and Table 3). The ESI-MS spectra of OS32 showed several multiply charged deprotonated ions, [M-2H]2− and [M-3H]3−. The main ions corresponded to the structures of dodecasaccharide (OS32) and differed only by phosphate group substitution and the presence of a terminal Kdo residue in the inner-core OS (Table 3). The ions at m/z 999.18, m/z 959.18, m/z 665.8, and m/z 639.11 corresponded to the OS32 devoid of the trisaccharide (Table 3). Additionally the [M-H2O-H]1− ion at m/z 573.14 corresponding to the trisaccharide was also identified.
Negative-ion ESI-Q-TOF mass spectrum of the dodecasaccharide OS32. The spectra were obtained for acetonitrile-water solutions (0.2 mg/ml). The numbers of charges are indicated above the m/z values. The structures (top) show the structural heterogeneity of fraction OS32. Detailed interpretation of the ions is provided in Table 1. The outer-core trisaccharide is framed with a dashed box.
Interpretation of ESI mass spectra of the de-N,O-acylated LPSs isolated from H. alvei 32 (OS32)
Intact structure of the Kdo-containing trisaccharide.The procedure used for the isolation of the intact Kdo-containing core region of H. alvei 32 LPS-OS32 entails anhydrous hydrazine and aqueous 4 M KOH treatment and always leads to the loss of the alkali-labile substituents (i.e., O-acetyl/acyl, N-acetyl/acyl, and PEtn). The structure of the trisaccharide isolated from H. alvei 32 was elucidated previously only by GC-MS, fast atom bombardment-MS, and 1D 1H and 13C NMR spectroscopy (16), but the presence of O and N substituents susceptible to hydrazine and KOH treatment was not reported. Moreover, both papers (16, 23) presented incomplete NMR assignments of spin systems of the Kdo-containing trisaccharides. Since none of the alkali-labile components were retained in OS32 we also analyzed the trisaccharide isolated by mild acid hydrolysis of LPS 32 and completed 1H and 13C NMR spectroscopy analyses (Fig. 5 and Tables 4 and 5) with emphasis on the presence of the labile constituents.
Parts of the 600-MHz HSQC-DEPT spectrum of the trisaccharide isolated by mild acid hydrolysis of H. alvei 32 LPS. The inset spectrum contains the deoxy resonances of the Kdo residue. The capital letters refer to carbohydrate residues as shown on the structure, and the numbers refer to protons and carbons in the respective residues. Spin systems assigned lowercase letters belong to constituents of the trisaccharide l-α-d-Hepp-(1→4)-[α-d-Galp6OAc-(1→7)]-α-Kdof.
1H and 13C NMR chemical shifts of the trisaccharide isolated from H. alvei 32 LPS a
Selected interresidue 3 J H,C connectivities from the anomeric atoms of the trisaccharide isolated from H. alvei 32 LPS
Residue K* was identified as a 4,7-disubstituted α-Kdop on the basis of characteristic deoxy proton signals at δ 1.91 (H-3ax) and δ 2.11 (H-3eq) ppm and the large chemical-shift values of the C-4 (δ 71.9) and C-7 signals (δ 78.6).
Residue L*, with the H-1/C-1 signals at δ 5.08/97.9 ppm, a J H-1,H-2 of <2 Hz, and a J H-1,C-1 of ∼172 Hz, was recognized as terminal l-glycero-α-d-manno-Hepp due to the small vicinal couplings between H-1, H-2, and H-3 and the similarity of the chemical shifts with those of the terminal l-α-d-Hepp.
Residue M*, with the H-1/C-1 signals at δ 5.19/101.0 ppm, a J H-1,H-2 of 2.6 Hz, and a J H-1,C-1 of ∼172 Hz, was assigned as terminal α-d-Galp6OAc due to the large couplings between H-2 and H-3 and the small vicinal coupling constants between H-3, H-4, and H-5. The H-6a and H-6b of this residue resonated at δ 4.25 and δ 4.31 ppm. This observed downfield shift is consistent with the substitution of the O-6 with an acetyl group. It was further supported by the HMBC connectivities observed between H-6a and -b of residue M* (δH 4.25 and 4.31 ppm), the acetyl carbonyl carbon (δCO 174.5 ppm), and the acetyl methyl protons (δH 2.13 ppm).
An HMBC experiment exhibited interresidue connectivities between adjacent monosaccharides and thus provided the sequence l-α-d-Hepp-(1→4)-[α-d-Galp6OAc-(1→7)]-α-Kdop-(2→ (Table 5).
Trisaccharide-specific antibodies.In immunoblotting tests, antisera obtained against the OS32-BSA conjugate reacted strongly with fast-migrating fractions of homologous H. alvei 32 LPS, as well as LPSs of H. alvei PCM 1192 and 1207, containing core OS-lipid A molecules not substituted with O-specific chain (see Fig. S2A in the supplemental material). This observation was in agreement with structural data obtained previously for the examined LPSs (15-18).
A fraction of immunoglobulins specific for the trisaccharide l-α-d-Hepp-(1→4)[α-d-Galp-(1→7)]-α-Kdop-(2→ was obtained by absorption of anti-OS32-BSA serum with H. alvei 1207 bacterial cells. H. alvei strain 1207 was chosen for absorption of antibodies specific for the core OS identical with that found for strain 32 but devoid of the trisaccharide (Lugowski et al., unpublished).
The reactivity of the absorbed serum was assessed in ELISA and immunoblotting with H. alvei 32, 1192, and 1207 LPSs (see Fig. S2B and C in the supplemental material). Unlike nonabsorbed serum, strong reactions were observed only for LPSs of strains 32 and 1192. The antibodies did not react with the LPS of strain 1207. The specificity of the absorbed serum was examined by an ELISA inhibition test with the trisaccharide of H. alvei 32 as the inhibitor. The trisaccharide showed 50% inhibition of the reaction of the absorbed serum with H. alvei 32 LPS at a concentration of 8 μM and 90% at a concentration of 98 μM. Similar inhibitory activity of the trisaccharide was also observed for reaction with H. alvei PCM 1192 LPS (9.8 μM and 120 μM—50% and 90% inhibitory concentrations, respectively). This suggested that the absorption yielded antibodies recognizing only epitopes of the trisaccharide of H. alvei 32 LPS.
Screening of H. alvei LPSs for the presence of Kdo-containing trisaccharide.Specific anti-trisaccharide antibodies were used to scan all available LPSs of H. alvei, comprising 37 different O serotypes, for the presence of epitopes similar to those found in H. alvei 32. LPSs separated by SDS-PAGE (Fig. 6A) were transferred from the gel onto the nitrocellulose membrane and subjected to an immunoblotting test (Fig. 6B). Most of the LPSs represented S forms and showed a high-molecular-mass pattern of bands characteristic of smooth strains (Fig. 6A). Strong reactions were observed with fast-migrating fractions of six H. alvei strains, i.e., 2, 32, 600, 1192, 1206, and 1211, suggesting the presence of a common trisaccharide epitope in outer-core regions of these strains. Cross-reactions that were observed for H. alvei 2, 32, 1192, and 1211 LPSs are in agreement with previously published results of structural analysis of trisaccharides isolated from these LPSs (16, 23). All of the observed reactions of anti-conjugate serum were detected mainly for fast-migrating LPS fractions (lipid A substituted with core OS). Weaker reactions detected for H. alvei 974, 1188, 1198, 1204, and 1214 suggest the presence of epitopes with structures similar to that of the trisaccharide.
Reactivities of affinity-purified antibodies specific for the trisaccharide l-α-d-Hepp-(1→4)-[α-d-Galp-(1→7)]-α-Kdop-(2→ with H. alvei LPSs. (B) The affinity-purified trisaccharide-specific antibodies against OS32-BSA conjugate were used for immunoblotting. (A) LPSs were analyzed by SDS-PAGE (5 μg/lane) using a 15% polyacrylamide-bisacrylamide separating gel and were visualized by silver staining. LOS of H. alvei 1 M was isolated from the rough strain and represented the low-molecular-weight form built from lipid A substituted with core OS only (41). The spot intensities of the observed cross-reactions in immunoblotting were compared with this for reaction between H. alvei 32 LPS and absorbed serum (positive, strong reaction).
DISCUSSION
The structures of LPSs isolated from H. alvei have been widely studied over the past 20 years. A number of O-specific polysaccharides and four types of core OSs have been identified for H. alvei LPSs (9, 15, 17, 24-28, 30, 42, 43). Studies of these LPSs have focused on structural analysis of isolated and chemically modified fragments, i.e., polysaccharides and OSs obtained by mild acid treatment, a procedure typically used for the delipidation of LPS. However, this method leads to hydrolysis of all ketosidic linkages in the LPS molecules. Therefore, such a strategy has to be further supplemented by other methods of structural analysis, providing insight into the structural details of the isolated LPS molecules. The standard delipidation procedure allows the determination only of structures of discrete regions: O-specific repeats, the core OS, and Kdo-containing trisaccharides (16, 18). The key part of this study was the investigation of an additional Kdo-containing fragment and its location in H. alvei 32 LPS segments. These fragments have often been found among polysaccharides and OSs obtained by mild acid hydrolysis of H. alvei LPSs (16, 22, 23, 40), but their origin has not been explained. Thus, the results presented here and the methods used have pieced together the data on regions of H. alvei 32 LPS.
We present here the complete structure of the dodecasaccharide OS32 obtained by de-N,O-acylation of H. alvei 32 LPS, with all acid-labile ketosidic linkages preserved. The dodecasaccharide OS32 represented the complete structure of the core OS containing a →4,5)-α-Kdop residue in the inner-core region substituted by terminal α-Kdop and the trisaccharide l-α-d-Hepp-(1→4)-[α-d-Galp6OAc-(1→7)]-α-Kdop-(2→ as an integral part of the outer-core OS.
The acid-labile α-(2→2) linkage between α-Kdop (residue K) of the trisaccharide and α-d-Glcp (J) of the outer-core OS (Fig. 1, top) was identified using NMR analyses, and the trisaccharide was located in the outer-core region of H. alvei 32 LPS. The presence of →2)-Glcp among the components of OS32 was further supported by the methylation analysis and suggested that this residue was substituted by the trisaccharide. The 2-substituted α-d-Glcp (residue J) had never been identified among constituents of H. alvei core OSs obtained by mild acid hydrolysis (16, 33). Due to the heterogeneity of OS32 revealed by NMR and ESI-MS analyses and corresponding to the presence or lack of the trisaccharide, two forms of α-d-Glcp were observed: →2)-α-d-Glcp (residue J) and terminal α-d-Glcp (j). The presence of the terminal α-d-Glcp could be explained by the even higher susceptibility of the ketosidic linkage between →4,7)-α-Kdop and →2)-α-d-Glcp to acidic conditions than the bond between →4,5)-α-Kdop and lipid A. The interresidue connectivities observed in the NOESY experiment identified linkage between two other α-Kdop residues (residue C and residue D) and thus allowed an unambiguous localization of the α-Kdop-(2→4)-α-Kdop-(2→ disaccharide segment in the inner-core region. Furthermore, the assigned resonances of the remaining OS32 residues [→3)-α-Glcp, →3,7)-l-α-d-Hepp, →3)-l-α-d-Hepp, and terminal l-α-d-Hepp] were similar to these described previously for core OSs (17, 20, 33, 44, 47, 49).
We isolated and analyzed the de-N,O-acylated population of H. alvei 32 LPS molecules not substituted with O-specific polysaccharide (the rough-type population). Further studies will be required to characterize the position of substitution of the core OS by O-specific polysaccharide. Mild acid hydrolysis of some LPSs isolated from other species of gram-negative bacteria usually gives fractions that contain core OSs substituted by one or a few repeating units. Structural analysis of these fractions provides information on the biological repeating unit and the linkage between the core OS and the O-specific chain. Such fractions, containing previously identified hexasaccharide core OS substituted directly with one O repeat, have never been isolated among polysaccharides and OSs obtained by mild acid hydrolysis of 32 LPS (Lugowski et al., unpublished). The lack of such a fraction could be explained by the presence of Kdo-containing fragments between the O-specific polysaccharide and core OS in H. alvei LPSs.
Ravenscroft and coworkers (40) in their studies of H. alvei 2 LPS identified an OS that was built from the fragment of the O repeat linked to C-7 of Kdo in the disaccharide →7)-[l-α-d-Hepp-(1→8)]-α-Kdop. The authors suggested that this Kdo-containing element could constitute an alternative or a predominant way that the O-specific polysaccharide was linked to the core OS, but this linkage in H. alvei 2 LPS was not proven (40). Katzenellenbogen et al. isolated fractions from H. alvei LPSs 1185, 1188, 1189, 1196, 1199, 1204, 1205, 1211, 1216, and 1546 built of sugar residues characteristic both of core OS (Hep) and of O repeats, as shown by sugar and methylation analyses (25). Analysis of constituents and their molar ratios in these fractions and fractions of unsubstituted core OSs revealed biological repeating units of 10 studied strains, but also indicated that neither the outer-core hexoses nor the terminal heptose residue was substituted with the O repeat of these LPSs (25).
In immunoblotting analysis, identical or similar elements have been found in seven H. alvei LPSs, among them three LPSs (strains 1188, 1204, and 1211) examined by Katzenellenbogen et al. (25). Another previously identified type of trisaccharide, Galp-(1→2)-l-α-d-Hepp-(1→4)-α-Kdop (22), sharing part of the epitope of the trisaccharide of LPS 32, could explain weaker cross-reactivity of antibodies specific for the trisaccharide of H. alvei 32 with the LPS of strain 1188. All these data suggest that the Kdo-containing trisaccharide is an interlinking segment for the O-specific chain in these LPSs.
A hypothesis formulated by Ravenscroft et al. for H. alvei LPS (40) holds for Klebsiella pneumoniae LPSs, for which similar Kdo-containing OS fragments were identified as an interlinking segment between the O-specific chain and the core OS (49). The presence of the third Kdo residue in the core part of the LPS was detected previously for K. pneumoniae O3, and the disaccharide Hepp-(1→4)-α-Kdo-(2→ was identified in the outer region of the core OS (50). Further studies showed that this region represents the conserved feature of K. pneumoniae LPS. Moreover, Vinogradov and coworkers identified this acid-labile element as the ligation site for the O antigens in the LPSs of K. pneumoniae (49). Position C-5 of the →4)-α-Kdo is a common attachment point of the O-specific polysaccharides to the core OSs in Klebsiella LPSs (49). In Rhizobium etli CE358, CE359, and CE166, the Kdo residue either is an attachment point for O-specific polysaccharide (7) or constitutes the outer-core region of the LPS (8).
A similar feature could also be characteristic of H. alvei LPSs. Different Kdo-containing motifs might represent the way the O-specific polysaccharide is linked to the core OS, as a core hexasaccharide substituted directly with one or a few repeating units has not been identified among fractions obtained by mild acid hydrolysis of a majority of H. alvei LPSs. In our serological studies, we have found that 6 out of 37 LPSs of H. alvei have outer-core regions {l-α-d-Hepp-(1→4)-[α-d-Galp-(1→7)]-α-Kdop-(2→} identical with the structure described here. Six other strains could have similar structures. Different types of outer-core Kdo-containing OSs identified previously (16, 22, 23), with a Kdo residue as a common feature responsible for high sensitivity to acid hydrolysis, suggest that other types of linkage between the core OS and the O-specific polysaccharide in H. alvei LPSs cannot be excluded.
LPSs of H. alvei are yet another example of enterobacterial endotoxins deviating from the classical scheme of LPS by the presence of an additional Kdo residue in the outer-core region. This finding demonstrates how important it is to use complementary instrument techniques and chemical analyses in LPS structure elucidation to avoid the loss of significant structural information. Interesting cases of acid-labile interlinking LPS segments among K. pneumoniae, R. etli, Serratia marcescens (51), and H. alvei could prompt researchers to look for similar motifs in structural analyses of other LPSs.
ACKNOWLEDGMENTS
We thank Zbigniew Szewczuk, Faculty of Chemistry, University of Wroclaw, for his help, comments, and assistance with the ESI-Q-TOF measurements.
This work was supported by grant 6 PO 04A 069 19 from the State Committee of Scientific Research (KBN), Poland, and by the Swedish Research Council. Part of this study concerning ESI-MS analyses of de-N,O-acylated LPSs was supported by grant N401 084 32/1944 from the Ministry of Sciences and Higher Education, Poland. The collaboration between Polish and Swedish groups was supported by funds from The Royal Swedish Academy of Sciences and The Polish Academy of Sciences.
FOOTNOTES
- Received 30 June 2008.
- Accepted 5 November 2008.
- Copyright © 2009 American Society for Microbiology