Báo cáo Y học: The structure of the O-chain of the lipopolysaccharide of a prototypal diarrheagenic strain of Hafnia alvei that has characteristics of a new species under the genus Escherichia pot

The structure of the O-chain of the lipopolysaccharide of a prototypal diarrheagenic strain of Hafnia alvei that has characteristics of a new species under the genus Escherichia Reine Eserstam 1 , Thushari P. Rajaguru 1,2 , Per-Erik Jansson 1 , Andrej Weintraub 3 and M. John Albert 4 1 Clinical Research Center, Analytical unit, Karolinska Institute, Huddinge Hospital, Huddinge, Sweden; 2 Department of Chemistry, University of Peradeniya, Peradeniya, Sri Lanka; 3 Karolinska Institute, Department of Microbiology, Pathology and Immunology, Division of Clinical Bacteriology, Huddinge University Hospital, Sweden; 4 Department of Microbiology, Faculty of Medicine, Kuwait University, Safat, Kuwait The structure of the O-polysaccharide of the lipopolysac- charide from a diarrheal strain isolated in Bangladesh was studied with sugar, and methylation analysis, NMR spectroscopy, mass spectrometry and partial acid hydrolysis. The strain was first designated as Hafnia alvei, but later found to be a possible new species in the genus Escherichia. Two different polysaccharides were detected, a major and a minor one. The structure of the major polysaccharide is gi- ven below, while the structure of the minor one was not investigated. The structure of the repeating unit was estab- lished as →6)-β- D -Galf-(1→3)-β- D -GalpNAc-(1→3)-β- D -Galp-(1→ α-NeuAc ↑ 6 2 The structure does not resemble any of the previously investigated lipopolysaccharide O-chains from Escherichia coli or H. alvei, but could fit in either group based on types of sugar residues and acidity. Phenotypic microbiological studies cannot definitely assign it to either species of the two genera. Genetic hybridization studies indicate that the Bangladeshi isolates may require a new species designation under the genus Escherichia. Keywords: lipopolysaccharide; Escherichia; Hafnia alvei; diarrhea; neuraminic acid. Hafnia alvei is a Gram negative bacterium and a member of the family Enterobacteriaceae. There are reports of associ- ation of H. alvei with diarrhoea in Canada [1] and Finland [2], but the mechanism of diarrhoea caused by this organism in these locations remains unknown [3]. However, some isolates of a bacterium typed as H. alvei from patients with diarrhoea in Bangladesh produced diarrhoea in rabbits by attaching and effacing (AE) lesions in the intestinal mucosa that are characteristic of the lesions produced by entero- pathogenic Escherichia coli [4]. Like enteropathogenic E. coli,theseH. alvei isolates possess a homologous patho- genicity island in the chromosome locus for enterocyte effacement (LEE), which is responsible for producing attaching and effacing lesions [5]. LEE encodes a type III secretory system [6]. Secretion of the virulence factors leads to effacement of the microvillus structure and reorganiza- tion of the actin cytoskeleton to form a pedestal-like structure, the attaching and effacing lesion [7]. AE lesion formation is critical in mediating diarrhoea production in the host, but its exact role in disease is not known. Recent results from conventional biochemical analyses, testing of susceptibility to cephalothin, lysis by a Hafnia-specific phage, and amplification of the outer membrane protein gene phoE with species-specific primers support the identi- fication of these isolates as members of the genus Escheri- chia rather than Hafnia alvei [8]. We studied the structure of the O-chain of the lipopolysaccharide of one them. MATERIALS AND METHODS Bacterium, cultivation and isolation of lipopolysaccharide and O-specific polysaccharide The Hafnia alvei, strain number 10457, was from the culture collection of the International Centre for Diarrhoeal Disease Research, Bangladesh (ICDDR, B), Dhaka. This strain was isolated from a patient with diarrhoea and was positive for the AE property [15]. The bacterium was grown in TY-medium, and the lipopolysaccharide isolated by centrifugation and extraction of bacterial cells with hot aqueous phenol [16]. The polysaccharide was analysed as Correspondence to P E. Jansson, Karolinska Institute, Clinical Research Center, Novum, Huddinge University Hospital, S-141 86 Huddinge, Sweden. Fax: + 46 8 58583820, Tel.: + 46 8 58583821, E-mail: pererik.jansson@kfc.hs.sll.se Abbreviations: AE, attaching and effacing; Hex, hexose; DEPT, dis- torsionless enhanced polarization transfer; HMBC, heteronuclear multiple-bond correlation; HSQC, heteronuclear single-quantum coherence; LEE, locus for enterocyte effacement; TMS, trimethylsilyl. (Received 19 March 2002, revised 6 May 2002, accepted 21 May 2002) Eur. J. Biochem. 269, 3289–3295 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03009.x the lipopolysaccharide, or degraded with 0.1 M acetic acid/ sodium acetate, pH 4.2, for 4 h at 100 °Ctogivethe O-polysaccharide. The O-polysaccharide was isolated by gel-permeation chromatography on a column (70 · 2.6 cm) of Sephadex G-50 using 0.05 M pyridinium acetate, pH 4.5, as eluent and monitoring with a differential refractometer. Sugar analysis Hydrolysis was performed with 2 M trifluoroacetic acid (120 °C, 2 h), and monosaccharides identified by GLC as their alditol acetates. Sugars were analysed on a Hewlett– Packard 5880 GLC instrument on a DB-5 fused-silica capillary column and a temperature gradient of 160 °C (1 min) to 250 °Cat3°CÆmin )1 . The absolute configura- tions were determined by GLC of acetylated glycosides with (+)-2-butanol, as described previously, but with the modification that acetates were used [9–11]. Neuraminic acid methyl glycoside methyl ester was analyzed as the trimethylsilyl (TMS)-derivative and authentic reference. A colorimetric test for Kdo using thiobarbituric acid was also made [17]. Methylation analysis Methylation was carried out with methyl iodide in dimethyl sulfoxide in the presence of sodium methylsulfinylmethanide [18]. The methylated polysaccharide was purified using Sep– Pak C18 cartridges. Hydrolysis was performed as described for sugar analysis; partially methylated monosaccharides were converted into alditol acetates and analyzed by GLC and GLC-MS on a Hewlett Packard 5890 chromatograph equipped with a NERMAG R10–10 L mass spectrometer, using the above conditions. Identification was made using reference data. NMR spectroscopy 1 H- and 13 C-NMR spectra were recorded with a JEOL GSX-270 or JEOL JNM ECP500 instruments or solutions in 2 H 2 Oat70or85°C. Chemical shifts are reported with internal acetone (d H 2.25, d C 31.00) as reference. Mixing times of 30–160 ms were used in TOCSY experiments, and for NOESY 100 and 300 ms. MALDI mass spectrometry MALDI mass spectrometry in the positive mode was run on a Finnigan Lasermat instrument using dihydroxybenzoic acid acid as matrix. Between 10 and 20 scans were accumulated and added. The neuraminic-acid-free polysac- charide was treated with 0.01 M acetic acid for 1 h at 65 °C and then neutralized with dilute sodium hydroxide solution. RESULTS AND DISCUSSION Hafnia alvei strain number 10457 was grown in tryptone- yeast (TY) medium and harvested by centrifugation. Extraction with hot phenol/water (1 : 1, v/v) gave a lipo- polysaccharide in the aqueous phase, which was recovered and freeze dried. Ultracentrifugation of the lipopolysaccha- ride gave a pellet and an upper phase, the latter containing most of the material. SDS/PAGE of the two materials (Fig. 1) in the upper phase and the pellet showed identical patterns and it was therefore concluded that the same polysaccharide was present. A hydrolysate of the upper phase, analyzed as alditol acetates, revealed as D -glucose, D -galactose, D -galactosamine, L -glycero- D -manno-heptose, and D -glucosamine in the proportions 6 : 65 : 19 : 6 : 3. The relatively high proportion of heptose may be the result of short chains. It is not a component in the polysaccharide as demonstrated in the MS analysis (see below). The absolute configurations of the sugars were established by GLC analysis of the acetylated (+)-2-butyl glycosides [9–11]. Methanolysis of the sample and analysis by GLC- MS gave, in addition to the sugars mentioned, neuraminic acid. The pellet showed essentially the same compounds. To verify that the material was an O-polysaccharide or exclude the possibility, the content of Kdo was checked with the thiobarbiturate method. It showed that the content of Kdo in the crude material, the pellet, and the supernatant was 11, 15, and 12 lgÆmg )1 , respectively, corresponding to one Kdo per 70 sugars or two per 35, i.e a significant amount of Kdo. Treatment of the upper phase with acetic acid buffer of pH 4.2 followed by gel chromatography on a Sephadex G-50 column gave a major O-polysaccharide peak at the void volume (O-polysaccharide) and a minor peak just after. The material in the major peak was devoid of neuraminic acid. The second minor peak, which was included in the column and had four signals in the anomeric region of the 1 H-NMR spectrum at d 5.07, 4.92, 4.80, and 4.49, clearly different from the major compound. The presence of two different polysaccharides in Hafnia has not been observed before. It was not clear whether it was an lipopolysaccharide or a capsular polysaccharide and the fraction was not further investigated. The proportion of the minor polysac- charide was indicated by the size of the minor peaks in the 1 H-NMR spectrum (Fig. 2), especially as the peak near d 5.0 was separate enough to be able to make a quantitative estimation, approximately 5%. In the methylation analysis of the O-polysaccharide, derivatives corresponding to 6-substituted galactofuranose, 3-substituted galactopyranose and 3-substituted galactosa- Fig. 1. SDS/PAGE of the upper phase (1) and the pellet (2) obtained on ultracentrifugation of the Hafnia alvei lipopolysaccharide. For com- parison lipopolysaccharide from a smooth (Shigella flexneri,3)anda rough bacterium (Salmonella typhimurium Ra, 4) was run simulta- neously. 3290 R. Eserstam et al. (Eur. J. Biochem. 269) Ó FEBS 2002 mine were detected, thus indicating a linear polysaccharide consisting of repeats with three sugar residues. This was corroborated by the 1 H-NMR spectrum, which showed signals for anomeric protons at d 5.11 (J small), 4.76 (J 7.7 Hz) and 4.46 (J 7.7 Hz), for ring protons, and for an N-acetyl methyl group at d 2.05. The first chemical shift should belong to a furanoside as these normally have small J-values, the second and third signal have typical J-values for b-linked sugars with galacto-pyranose configuration. The absence of NeuAc was evident as no methylene- deoxyresonances could be detected. The 13 C-NMR spec- trum showed signals inter alia for anomeric carbons at d 109.9, 104.1 and 103.6. The first value is very high and characteristic of a b-furanosidic sugar. A distorsionless enhanced polarization transfer (DEPT) spectrum revealed that the substituted hydroxymethyl group, C-6 of the galactofuranose, is located at d 71.8, thus among those of secondary carbons. The 1 H- and 13 C-NMR spectra were assigned with 2D NMR spectra including COSY, TOCSY, NOESY, HSQC, and HMBC. Overlap in the spin systems were in some cases a problem, but could be overcome with a combination of the spectra. Residues and spin systems are denoted A–C in Table 1. Indeed, the A residue was the furanoside as evident from the correlation in the HSQC- spectrum between d 5.11/109.9. The possibility to trace signals beyond that H-2 was limited, but two of the signals in the 13 C-NMR spectrum at 82–84 p.p.m. could be shown to derive from C-2 and C-4 in A, and to correlate to proton signals at approximately d 4.07. B could be assigned to the GalNAc residue as evident from the chemical shift of C-2 signal, which appeared at d 52.2, typical for C-N signals. A downfield shift of the signal for C-3 signal to d 78.7 corroborated the linkage position. Residue C gave in the TOCSY spectrum three correlations, up to H-4, which had aresonanceatd 4.16, from the high value it was confirmed that it was galactose. In addition, the couplings of H-4 (1D/ 2D) were small, indicative of the galacto configuration. In the 13 C-NMR spectrum, it was evident that two signals were present at d 75.6, the second being assigned to C-5 in residue C, close to the value in the monomer. The sequence of Fig. 2. The 1 H-NMR spectrum of the Hafnia alvei lipopolysaccharide in D 2 O. Table 1. 1 H- and 13 C-NMR chemical shifts (d, p.p.m.) for different H. alvei polysaccharides. The N-acetyl group in the GalNAC residue appears at d 2.05/22.7/176.1 in the O-polysaccharide and in the GalNAc and NeuAc residues at d 2.05/22.8–22.9/175.8 in the lipopolysaccharide. Sugar residue 1 2 3 4 5 6 7 8 9 H. alvei O-polysaccharide fi 6)-b- D -Galf-(1 fi (A) 5.11 4.09 4.06 4.06 4.02 3.75, 4.04 109.9 82.1 77.6 83.7 70.2 71.8 fi 3)-b- D -GalpNAc-(1 fi (B) 4.76 4.05 3.82 4.02 3.71  3.77 103.6 52.2 78.7 70.2 75.6 61.6 fi 3)-b- D -Galp-(1 fi (C) 4.46 3.62 3.76 4.16 3.71  3.77 104.1 70.7 82.6 69.3 75.6 61.6 Native H. alvei lipopolysaccharide fi 6)-b- D -Galf-(1 fi (A) 5.11 4.09 4.05 4.04 4.00 3.89, 3.89 109.9 82.8 77.8 83.8 70.6 72.1 fi 3)-b- D -GalpNAc-(1 fi (B) 4.73 4.04 3.83 4.07 3.70  3.75,  3.75 103.7 52.4 78.8 68.5 75.6 61.7 fi 3,6)-b- D -Galp-(1 fi (C) 4.43 3.65 3.72 4.19 3.73 3.62, 3.92 104.2 70.6 82.8 69.0 73.4 64.1 a-NeuAc-(1 fi (D) – – 2.74, 1.68 3.74 3.84 3.70 3.78 4.03 174.2 101.2 41.0 69.0 52.6 73.8 69.0 72.4 63.4 Ó FEBS 2002 New species under the genus Escherichia (Eur. J. Biochem. 269) 3291 sugars was indicated by the following H/C correlations in the HMBC spectrum, d 5.11/78.7 (A H-1/B C-3), d 4.73/82.6 (B H-1/C C-3), and d 4.44/71.8 (C H-1/A C-6). The disaccharide elements A–B, B–C, and C-A were thus present to make up the chain as →6)-β- D -Galf-(1→3)-β- D -GalpNAc-(1→3)-β- D -Galp-(1→ A B C The next step was to analyze the native lipopolysaccha- ride. Methylation analysis of the native lipopolysaccharide gave major GLC peaks corresponding to 6-substituted galactofuranose, 3,6-disubstituted galactose, and 3-substi- tuted 2-acetamido-2-deoxy- D -galactose. In addition, smaller peaks corresponding to the minor component polysaccha- ride were observed. The repeating unit of the polysaccharide thus contains a terminal NeuAc, and the above mentioned residues. A comparison to the methylation analysis data on the O-polysaccharide, indicates that the terminal NeuAc should be substituting the galactose residue in the 6-position. For the full characterization of the lipopolysaccharide, with NeuAc still present, an NMR sample was prepared from the native lipopolysaccharide. The spectrum had broadened lines but were surprisingly good with resolved couplings (Fig. 2). The 1 H-NMR spectrum of the lipopoly- saccharide showed signals for three anomeric protons at d 5.11 (J small), 4.73 (J 7.7 Hz), 4.43 (J 7.7 Hz), thus close to those observed for the O-polysaccharide. In the high field region signals for a methylene group, assigned to CH 2 in NeuAc were observed at d 2.75 and 1.68, the large difference establishing the presence of an axial carboxyl group and an a-linkage in the NeuAc residue. Signals for N-acetyl groups deriving from NeuAc and GalNAc were present at d 2.05. In the 13 C-NMR spectrum, the corresponding signals were present inter alia at d 109.9, 104.2, 103.7, and 101.2 for anomeric carbons and at d 41.0 and 22.8–22.9 for methylene and methyl groups, respectively. The spectrum resembled that of the O-polysaccharide, but some changes were evident. The signals at d73.8, 69.0, 63.4, and 52.6 were higher than the others and were subsequently assigned to the NeuAc residue. Analysis of both of the 1 H- and the 13 C-NMR spectra using 1D and 2D techniques gave the data shown in Table 1, where the residues are referred to as A–D, D being the additional NeuAc residue. Most of the signals could be unambiguously assigned. The 13 C-NMR spectrum showed 28 signals of the possible 30. The assignment was made essentially as described for the O-polysaccharide and by comparison with the spectra of the O-polysaccharide. Figures 3–6 show the COSY, TOCSY, HSQC and HMBC spectra, respectively. From the large glycosylation shifts of signals from C-6 in A, C-3 and C-6 in B,andC-3inC, the linkage positions were verified. The absence of any glycosylation shift in D further indicated that it was terminal. The chemical shift displacement of the C-6 signal in C to d 64.1, is small, typical for substitution with ketosides. An HMBC experi- ment showed the following inter-residue correlations from anomeric protons to linkage carbons: A H-1/B C-3 (5.11/ 78.8), B H-1/C C-3 (4.73/82.8) corroborating elements A–B and B–C. From the NOE spectrum the following inter- residue correlations between H-1 and protons on linkage carbons were observed: A H-1/B H-3 (5.11/3.82), demon- strating the element A–B. Correlations 4.73/3.72 and 4.43/ 3.90 are in accord with elements B–C and C–A but ambiguous due to signal overlap. From the combined data, however, the following structure can be postulated for the repeating unit →6)-β- D -Galf-(1→3)-β- D -GalpNAc-(1→3)-β- D -Galp-(1→ α-NeuAc ↑ 6 2 A B C D Fig. 3. The COSY spectrum the of the Hafnia alvei lipopolysaccharide showing the anomeric and the ring proton region. 3292 R. Eserstam et al. (Eur. J. Biochem. 269) Ó FEBS 2002 MALDI-MS of the O-polysaccharide The O-polysaccharide, i.e. the desialylated polysaccharide chain, was also characterized by MALDI-MS of the fragmented chain, anticipated to be facile to cleave with acid, as furanosides were present. Thus, the O-polysaccha- ride was treated with aqueous 0.01 M acetic acid and the samples were withdrawn at different times. Good spectra were obtained after approximately 1 h. The spectra showed three series of ions, all sodiated, the first at m/z 1626, 2149, 2677, 3204, 3730, and 4254, the second at m/z 1825, 2352, 2881, and 3401, and the third at m/z 1946, 2475, 3001, 3528, and 4054. All series are spaced with approximately 527 atomic mass units (amu), corresponding to the molecu- lar mass of the repeating unit containing two hexoses and one acetamidohexose. The first series corresponds to a multiple of the repeating unit, thought to be derived from the expected hydrolysis of the furanosidic linkage, thus (HexNAc-Hex-Hexf) n . The second series contains a mul- tiple of the repeat plus an additional acetamidohexose (203 amu) and the third contains a multiple of the repeat plus two additional hexose residues (324 amu). This implies that not only the furanosidic linkage is acid-labile, but also that of the b-D-GalNAc residue. Thus, assuming that the linkage second most easily hydrolyzed is that of the acetamidohexose, the ions in the second and the third series correspond to the formulas (HexNAc-Hex-Hex) n -HexNAc and Hex-Hex-(HexNAc-Hex-Hex) n and the sequence is further established. The initial phenotypic characterization of the strain 10457 with a commercial identification system, API-20 E identified the strain as Hafnia alvei [4]. Additional phenotypic characterization and partial 16S rRNA sequencing of a set of isolates identified them not as typical Hafnia alvei, but Fig. 5. HSQC spectrum of the Hafnia alvei lipopolysaccharide showing the anomeric and the ring proton/carbon region and including high resolution 1 H- and 13 C-NMR spectra. Fig. 4. TOCSY spectra of the Hafnia alvei lipopolysaccharide showing the correlations deriving from the anomeric protons. Mixing times were from 30 to 160 ms. Ó FEBS 2002 New species under the genus Escherichia (Eur. J. Biochem. 269) 3293 unique isolates [12]. Further phenotypic characterization suggested that these isolates are neither Hafnia alvei nor Escherichia coli, but closely related to the genus Escherichia [8]. DNA hybridization studies suggested that these isolates deserve a new species name under the genus Escherichia (J. Albert, Kuwait University, Safat, Kuwait, personal communication). The unique structure of the O-chain of one of these isolates further confirms this conclusion. The lipopolysaccharide structures of both H. alvei and E. coli are known [13,14]. More than 20 strains from H. alvei have been investigated and both amino sugars and acidic sugars are frequent. Furanosidic sugars are also observed. Neuraminic acid has been found once but only as an internal residue. Of the more than 60 E. coli strains that have been investigated, many contain hexoses and hexosa- mines, as well as acid functions. Also here, neuraminic acid has been found only internally. Furanosidic galactose is present, but not common. As a whole, the structural features of the investigated strain cannot be related to any particular structure among those studied of Hafnia and Escherichia. The terminal neuraminic acid is, however, an interesting feature, normally associated with glycoproteins and glycolipids. The biological properties of the novel lipopolysaccharide of this strain remain to be elucidated. ACKNOWLEDGEMENTS The authors thank Mrs G. Alvelius for help in mass spectrometry and Mrs M. So ¨ rensson for microbiology work. This work was supported by grants from the Swedish Natural Science Research Council, the Swedish Medical Research Council (No. B95-16X-11227-01 A), and the International Science Programs, Uppsala University, Sweden. Mrs Farrah Vesali is thanked for some preliminary experiments. REFERENCES 1. Ratnam, S. (1991) Etiologic role of Hafnia alvei in human diarrheal illness. J. Clin. Microbiol. 59, 4744–4745. 2. Ridell, J., Siitonen, A., Paulin, L., Mattila, L., Korkeala, H. & Albert, M.J. (1994) Hafnia alvei in stool specimens from patients with diarrhoea and healthy controls. J. Clin. Microbiol. 32, 2335– 2337. 3. Ismaili, A., Bourke, B., de-Azavedo, J.C., Ratnam, S., Karmali, M.A. & Sherman, P.M. (1996) Heterogeneity in phenotypic and genotypic characteristics among strains of Hafnia alvei. J. Clin. Microbiol. 34, 2973–2979. 4. Albert,M.J.,Alam,K.,Islam,M.,Montanaro,J.,Rahman,H., Haider,K.,Hossain,M.A.,Kibriya,A.K.M.G.&Tzipori,S. (1991) Hafnia alvei: a probable diarrheal pathogen of humans. Infect. Immun. 59, 1507–1513. 5. McDaniel, T.K., Jarvis, K.G., Donnenberg, M.S. & Kaper, J.B. (1995) A genetic locus of enterocyte effacement conserved among diverse enterobacterial pathogens. Proc. Natl Acad. Sci. 92, 1664– 1668. 6.Jarvis,K.G.,Giron,J.A.,Jerse,A.E.,McDaniel,T.K., Donnenberg, M.S. & Kaper, J.B. (1995) Enteropathogenic Escherichia coli contains a putative type III secretion system necessary for the export of proteins involved in attaching and effacing lesion formation. Proc.NatlAcad.Sci.USA92, 7996– 8000. 7.Rosenshine,I.,Ruschkowski,S.,Stein,M.,Reinscheid,D.J., Mills, S.D. & Finlay, B.B. (1996) A pathogenic bacterium triggers epithelial signals to form a functional bacterial receptor that mediates actin pseudopod formation. EMBO J. 15, 2613–2624. 8. Janda, J.M., Abbott, S.L. & Albert, M.J. (1999) Prototypal diar- rheagenic strains of Hafnia alvei are actually members of the genus Escherichia. J. Clin. Microbiol. 37, 2399–2401. 9. Gerwig, G.J., Kamerling, J.P. & Vliegenthart, J.F.G. (1978) Determination of the D and L configuration of neutral mono- saccharides by high resolution capillary glc. Carbohydr. Res. 62, 349–357. 10. Gerwig, G.J., Kamerling, J.P. & Vliegenthart, J.F.G. (1979) Determination of the absolute configuration of monosaccharides in complex carbohydrates by capillary glc. Carbohydr. Res. 77, 1–7. 11. Leontein, K., Lindberg, B. & Lo ¨ nngren, J. (1978) Assignment of the absolute configuration by glc of their acetylated glycosides formed from chiral alcohols. Carbohydr. Res. 62, 159–162. 12. Ridell, J., Siitonen, A., Paulin, L., Lindroos, O., Korkeala, H. & Albert, M. (1995) Characterization of Hafnia alvei with bio- chemical test, RAPD-PCR and partial sequencing of the 16S rRNA gene. J. Clin. Microbiol. 33, 2372–2239. Fig. 6. The full HMBC spectrum of the Hafnia alvei lipopolysaccharide. 3294 R. Eserstam et al. (Eur. J. Biochem. 269) Ó FEBS 2002 13. Knirel, Y.A. & Kochetkov, N.K. (1994) The structure of lipo- polysaccharides of gram-negative bacteria. III. The structure of O-antigens. Biochemistry 59, 1325–1383. 14. Jansson, P E. (1999) The chemistry of O-polysaccharide chains in bacterial lipopolysaccharides. EndotoxininHealthandDisease (Brade, H., Opal, S.M., Vogel, S.N. & Morrison, D.C., eds), pp. 155–178. Marcel Dekker, New York. 15. Albert, M.J., Faruque, S.M., Ansaruzzaman, M., Islam, M.M., Haider, K., Alam, K., Kabir, I. & Robins-Browne, R. (1992) Sharing of virulence-associated properties at the phenotypic and genetic levels between enteropathogenic Escherichia coli and Hafnia alvei. J. Med. Microbiol. 37, 310–314. 16. Westphal, O. & Jann, K. (1965) Bacterial lipopolysaccharides. Extraction with phenol-water and further applications of the procedure. Methods Carbohydr. Chem. 5, 83–91. 17. Karkhanis, Y.D., Zeltner, J.Y., Jackson, J.J. & Carlo. D.J. (1978) A new and improved microassay to determine 2-keto-3-deoxy- octonate in lipopolysaccharide of Gram-negative bacteria. Anal. Biochem. 85, 595–601. 18. Jansson, P E., Kenne, L., Liedgren, H., Lindberg, B. & Lo ¨ nng- ren, J. (1976) A practical guide to the methylation analysis of carbohydrates. In Chemical Communications. pp. 1–75. University of Stockholm, Stockholm. Ó FEBS 2002 New species under the genus Escherichia (Eur. J. Biochem. 269) 3295 . The structure of the O-chain of the lipopolysaccharide of a prototypal diarrheagenic strain of Hafnia alvei that has characteristics of a new species under. indicate that the Bangladeshi isolates may require a new species designation under the genus Escherichia. Keywords: lipopolysaccharide; Escherichia; Hafnia alvei; diarrhea;