Structural studies of the capsular polysaccharide and lipopolysaccharide O-antigen of Aeromonas salmonicida strain 80204-1 produced under in vitro and in vivo growth conditions Zhan Wang 1 , Suzon Larocque 1 , Evgeny Vinogradov 1 , Jean-Robert Brisson 1 , Andrew Dacanay 2 , Marshall Greenwell 2 , Laura L. Brown 2 , Jianjun Li 1 and Eleonora Altman 1 1 Institute for Biological Sciences, National Research Council of Canada, Ottawa, Canada; 2 Institute for Marine Biosciences, National Research Council of Canada, Halifax, Canada Aeromonas salmonicida is a pathogenic aquatic bacterium and t he causal agent of furunculosis in salmon. In the course of this study, it was found that when grown in vitro on tryptic soy agar, A. salmonicida strain 80204-1 produced a capsular polysaccharide with the identical structure to that of the lipopolysaccharide O-chain polysaccharide. A com- bination of 1D and 2 D NMR methods, including a series of 1D analogues of 3D experiments, together with capillary electrophoresis-electrospray MS (CE-ES-MS), composi- tional and methylation analyses and specific modifications was used to determine the structure of these polysaccharides. Both polymers were shown to be composed of linear trisaccharide repeating units consisting of 2-acetamido- 2-deoxy- D -galacturonic acid (GalNAcA), 3-[(N-acetyl-L- alanyl)amido]-3,6-dideoxy- D -glucose{3-[(N-acetyl- L -alanyl) amido]-3-deoxy- D -quinovose, Qui3NAl aNAc} and 2-ace- tamido-2,6-dideoxy- D -glucose (2-acetamido-2-deoxy- D - quinovose, QuiNAc) and having the following structure: [fi3)-a - D -GalpNAcA-(1fi3)-b- D -QuipNAc-(1fi4)-b- D - Quip3NAlaNAc-(1-] n , where GalNAcA i s partly presented as an amide and AlaNAc represents N-acetyl- L -alanyl group. CE-ES-MS analysis of CPS and O-chain polysac- charide confirmed that 40% of GalNAcA was presen t in the amide form. Direct CE-ES-MS/MS analysis of in vivo cul- tured cells confirmed the formation of a novel polysaccha- ride, a structure also formed in vitro, which was previously undetectable in bacterial cells grown w ithin implants in fish, and in which GalNAcA was fully amidated. Keywords: Aeromonas salmonicida; capsular polysaccharide; lipopolysaccharide; NMR. Aeromonas salmonicida is the aetiological agent of fu runcu- losis in s almonid fish, a disease which causes high mort alities in aquaculture. Considerable effort has been devoted to the development of e ffective vaccines a gainst furunculosis. Known extracellular virulence factors of the in vitro-grown A. salmonicida include surface layer (A-layer) [1], proteases [2], haemolysins [3], type IV pili [4] and LPS [5]. Very little is known about the role of these factors in vivo and their role in furunculosis. The A-layer is believed t o play a crucial role in the bacterial protection against c omplement-mediated killing in vitro and contributes to bacterial survival in phagocytes or host tissues [6]. In addition to the A-layer, lipopolysaccharide (LPS) is another at least partially exposed cell surface antigen that a ppears t o m ediate invasion [7]. Mo noclonal and polyclonal antibody analysis of a variety of A. salmonicida strain s h as shown that like the A-layer, the surface-exposed regions of LPS are antigeni- cally cross-reactive [8]. Formation o f capsular polysaccha- ride (CPS) covering the A-layer has been reported to be produced during the in viv o culture of A. salmonicida in surgically implanted intraperitoneal culture chambers [9]. Moreover, Merino et al . [10] have reported that when grown under c onditions promoting c apsule formation, strains of A. salmonicida exhibited s ignificantly higher ability to invade fish cell lines. It suggests that, as with the A-layer and LPS, CPS is an important virulence factor, essential for host cell invasion and bacterial survival. Previous studies have determined the structure of the O-chain polysaccharide of A. salmonicida strain SJ-15 [11]. Partial structure of the core oligosaccharide from the same strain of A. salmonicida was also determined [12]. In both instances A. salmonic ida strain SJ-15 was cultured in tryptic soy broth (TSB) at 25 °C. In addition, other reports des- cribe capsular material isolated from cells grown on yeast extract-peptone-glucose-mineral salts [13]. The relevance of these structures to in vivo-cultured bacteria and their role in pathogenesis has not been established. In the present study we have isolated and c haracterized the cell-surface carbohydrate antigens of A. salmonicida Correspondence to E. Altman, Institute for Biological Sciences, National Research Council of Canada, Ottawa, Ontario, K1A OR6, Canada. Fax: +1 613 941 1327, Tel.: +1 613 990 0904, E-mail: eleonora.altman@nrc-cnrc.gc.ca Abbreviations: A-layer, Ae r omonas surfa ce layer; CE- E S-MS, capillary electrophoresis-electrospray MS; CPS, capsular polysaccharide; GalNAcA, 2-acetamido-2-deoxy- D -galacturonic acid; GalNAcAN, 2-acetamido-2-deoxy- D -galacturonamide; LPS, lipopolysaccharide; Qq-TOF, hybrid quadrupole time-of-flight; TSA, tryptic s oy agar; TSB, tryptic soy broth. (Received 18 June 2004, revised 7 September 2004, accepted 4 October 2004) Eur. J. Biochem. 271, 4507–4516 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04410.x strain 80204-1 produced under in vitro growth conditions on tryptic soy agar (TSA) and in vivo and have demonstrated that th eir structures are chemically and a ntigenically distinct from the previously described O-chain polysaccharide [11] and capsule [13]. Experimental procedures Bacterial culture conditions A-layer – A. salmonicida avirulent strain 80204-1, a laborat- ory-derived mutant of s train 80204 obtained b y subculture [14], was grown on TSA plates. The inoculum was cultured in TSB (Difco Laboratories, Detroit, MI, USA) at 15 °C until it reached D 600 of 0.17 and spread a cross t he su rface of TSA plates. The plates were incubated at 15 °Cfor5days, washed with 0.01 M NaCl/P i pH 7.4 and subjected to a low speed centrifugation (3000 g,4°C, 10 min). The cells were killed with in 1% (w/v) phenol solution (4 h, 22 °C) and harvested by centrifugation(yield25g,wetweight). Isolation and purification of CPS and LPS The cells were washed with 2.5% saline (w/v) and extracted by the method of Westphal et al. [15]. Phenol and water layers were collected separately, dialysed against tap water and lyophilized. The lyophilizates were redissolved in 1% saline (w/v) and the clear solution w as subjected to ultracentrifugation (105 000 g,4°C, 10 h); the LPS pellets were redissolved in water a nd lyophilized. Clear supernatant was dialysed until salt-free, lyophilized and used f or isolation of CPS. Pure CPS was obtained by gel filtration on Sephadex G-100 column (yield 12 mg). Analytical methods CPS and LPS samples (0.5 mg) were hydrolysed with 3% (w/v) methanolic hydrogen chloride at 85 °Cfor5hand the reaction mixture was neutralized with silver carbonate (Aldrich, Oakville, O N, Canada). Resultant methyl g ly- cosides were converted to acetates an d analysed by GLC- MS using a Hewlett–Packard chromatograph equipped with a 30 m DB-17 capillary column [180 °C(30min)to 260 °Cat2°CÆmin )1 ] and mass spectra in the electron impact mode (EI) were recorded using a Varian Saturn 2000 mass spectrometer. To confirm the ring configuration of constituent sugars, the LPS sample (20 mg) was dissolved in dry methanol (5 mL), cooled in dry ice/acetone bath and acetyl chloride (5 mL) added. The reaction mixture was kept a t 70 °C for 4 h and dried under a stream of N 2 . M ethanolysis products were separated by HPLC using a C18 column (10 · 250 mm, Aqua, Phenomenex Torrance, CA, USA) in 3% MeCN (20 min, isocratic) to 90% MeCN gradient at 3mLÆmin )1 with the UV detection at 220 nm. Fractions were dried and analysed by NMR. Fatty acids were determined by GLC-MS analysis of their methyl esters derived by sealed-tube hydrolysis of LPS with 3% (w/v) methanolic hydrogen chloride at 100 °Cfor 16 h and then neutralized with silver carbonate (Aldrich). For GLC analysis a 30 m DB-5 capillary column [160 °C (2 min) to 260 °Cat1°CÆmin )1 ] was used and the identity of each fatty acid was established by comparison of its MS with that of the reference compound. The absolute configuration of glycoses was established by capillary GLC o f their ace tylated (-)-2-butyl glycosides, according to the method of Leontein et al.[16].Theidentity of each glycose derivative was established by comparison of its GLC retention time and MS with that of an authentic reference sample. The synthetic 2-acetamido-2-deoxy- L -quinovose was a gift from M. B . Perry (National Research Council, Ottawa, ON, Canada). The absolute configuration of the 2-acetamido-2-deoxy-galacturonic acid was deter- mined following the hydrolysis of the carboxyl-reduced CPS. Presence of L -alanine was confirmed by amino acid analysis. F or this, 0.6 mg of CPS was hydrolysed in 5.7 M hydrochloric acid containing 0.1% phenol (w/v) for 1 h at 160 °C under vacuum. The acid was removed by vacuu m centrifugation (speed-vac) with NaOH trap and the con- centrated sample was injected into the amino acid analyser based on the cation exchange chromatography with amino acid detection by ninhydrin colour at 570 nm except for proline at 440 nm. Methylation analysis The CPS and L PS samples w ere methylated a ccording to the method of Ciucanu & Kerek [17]. Methylated polysaccharide was subjected to hydrolysis as described by Stellner et al. [ 18] a nd methylation analyses were made according to previously reported conditions for alditol acetates. Carboxyl group reduction Carboxyl group reduction of the CPS and LPS samples was performed as previously described [19]. Briefly, LPS (10 mg) was d issolved in distilled water (10 mL) and following the addition of 1-cyclohexyl-3-(2-morpholino- ethyl) carbodiimide metho-p-toluenesulfonate (113 mg), the stirred mixture was maintained at pH 4.7 by titration with 0.1 M HCl for 3 h. Following completion of the reaction a 2 M solution of sodium borohydride (12.5 mL) was added slowly a nd the r eaction mixture was main- tainedatpH7bytitrationwith4 M HCl. The reaction wasallowedtoproceedfor2hat22°C, and the solution was dialysed and lyophilized. The product was purified by gel permeation chromatography on Sephadex G-100 and lyophilized (yield 6 mg). NMR spectroscopy NMR spectra were performed on Varian INOVA 500 and 600 M Hz spectrometers using standard software. NMR measurements were made at 25 °Cand60°ConCPSor LPS samples dissolved in D 2 O at pD 6.5 at concentration s of 6 mgÆmL )1 for CPS or 20 mgÆmL )1 for LPS. For the detection of NH protons spectra were recorded at 25 °Cin 90% H 2 O/10% D 2 O(v/v). All NMR experiments were performed using a 5-mm or 3-mm indirect detection probe with the 1 H coil nearest to the sample. The observed 1 H chemical shifts are reported relative to external acetone (2.225 p.p.m.), and t he 13 C 4508 Z. Wang et al.(Eur. J. Biochem. 271) Ó FEBS 2004 chemical shifts are quoted relative to the methyl group of external acetone (31.07 p.p.m.). 31 P-NMR experiments were performed on a Varian INOVA 200 MHz spectrometer, chemical shifts are given relative to the external 85% H 3 PO 4 (d P 0.0 p.p.m.). Standard homo- and heteronuclear correlated 2D tech- niques were used for general assignments of the CPS and LPS O-chain polysaccharide: COSY, TOCSY, NOESY and HSQC. Due to overlap in proton resonances, selective TOCSY and TOCSY–TOCSY experiments were used to complete the assignments [20]. MS All experiments were performed as described previously in detail [21]. Briefly, a Crystal Model 310 CE instrument (ATI Unicam, Cardiff, CA, USA) was coupled to an API 3000 or Q-Star quadrupole/TOF (Qq-TOF) mass spectrometer (Applied Biosystems/Sciex, Foster City, CA, USA) via a microIonspray interface. Sheath solution (isopropanol/ methanol, 2 : 1 ) was delivered at a flow rate of 1 lLÆmin )1 . An electrospray stainless steel needle (27 g) w as butted against the low dead volume tee and enabled the delivery of the sheath solution t o the end of the capillary column. The separations were obtained o n  90-cm long bare fused-silica capillary using 10 m M ammonium acetate in deionized water pH 9.0, containing 5% methanol. A voltage of 2 5 k V was typically applied at the injection. The outlet of the capillary was tapered to  15 lm internal diameter using a laser puller (Sutter Instruments, N ovato, CA, USA). Mass spectra were acquired with an orifice voltage of +120 V. Fragment ions formed by collision activation of selected precursor ions with nitrogen in the reference frame-only quadrupole collision cell, were registered by mass and analysed by TOF. Analysis of the in vivo cultured A. salmonicida strain 80204-1 cells In vivo culture was performed using ligated dialysis tubing bags filled with bacterial suspension of A. salmonicida strain 80204-1, surgically implanted in the peritoneal cavity of juvenile Atlantic salmon, and harvested at 72 h postsurgery as in Daca nay et al. [22]. Bacterial cells, 2.5 · 10 11 colony forming units, were w ashed with 2.5% (w/v) saline, and the pellet recovered by low-speed centrifugation (3000 g,4°C, 10 min) and lyophilized. The supernatant was dialysed against distilled water and l yophilized. It was tre ated with proteinase K (final concentration 250 lgÆmL )1 in 0.01 M NaCl/P i pH 7.2, 1 h , 60 °C) and, following the enzyme deactivation (5 min, 100 °C) and low-speed centrifugation, lyophilized sample was analysed directly by capillary electrophoresis-electrospray MS (CE-ES-MS) using Qq-TOF. The lyophilized pellet was treated with RNase and DNasetoreleaseLPS(finalconcentration10lgÆmL )1 in 0.02 M ammonium acetate, pH 7.5, 2 h, 37 °C) and lyoph- ilized following low-speed centrifugation (yield 2 7 m g, dry weight). It was treated with proteinase K as described above and bacterial cells were recovered by low-speed centrifuga- tion. Lyophilized sample was treated with 1% acetic acid (1 h, 100 °C), desalted using a centrifugal filter device (Pall Corporation, East Hills, New York, USA) and analysed directly by CE-ES-MS using a Qq-TOF mass spectrometer. In addition, lyophilized cells were subjec ted to methanolysis and methylation analyses as described above for purified CPSandLPSsamples. Results Cells of the A-layer – avirulent s train o f A. salmonicida, 80204-1, were grown on TSA, harvested, washed with 2.5% saline and subjected to the phenol/water extraction [15] followed by purification of aqueous- and phenol-phase soluble LPS by ultracentrifugation. Crude CPS was recov- ered from the initial 2.5% saline wash of bacterial cells and from 1% saline wash of the aqueous- and phenol-phase soluble LPS and purified by gel p ermeation chromatogra- phy on Sephadex G-100 column. CPS eluted as a broad peak in a void volume of a Sephadex G-100 column. Fractions were collected and analysed colorimetrically for aldose [23]. Three fractions, designated FI–FIII (not shown), were pooled and analysed by NMR. Fraction FI was contaminated by an a1,6-glucan while fractions FII and FIII were homogeneous yielding a glucan-free CPS that was used for further analysis. Methanolysis of CPS with 3% methanolic hydrogen chloride, followed by acetylation and GLC a nalysis of the resultant methyl glycosides afforded 2-amino-2,6-dideoxy- glucose, 3-amino-3,6-dideoxy-glucose and 2-amino-2- deoxy-galacturonic a cid in the approximate molar ratio of 1.0 : 1.0 : 0.9. LPS O-chain components previously identi- fied by Shaw et al. [11] and produced when A. salmonicida wasculturedinTSB,namely L -rhamnose, 2-amino-2- deoxy- D -mannose and D -glucose, and core oligosaccharide components, D -glucose, D -galactose, 2-amino-2-deoxy- galactose, 2-amino-2-deoxy-glucose and L -glycero- D -manno- heptose [12], were also detected ( 10%). In addition, phenol-phase soluble LPS was found to contain an a1,6- linked glucan, as demonstrated by both composition and methylation analyses. A significant amount of 2-amino-2- deoxy- D -galactose was identified in the hydrolysis products of both c arboxyl-reduced CPS and LPS [19], confirming the presence of GalNAcA in the native CPS and LPS. This was further corroborated by NMR and MS analyses performed on CPS and carboxyl-reduced LPS. Amino acid analysis confirmed the presence of L -Ala in both polysaccharides. Fatty acid analysis of both phenol- and aqueous-phase soluble LPS showed the presence of dodecan oic acid (C12 : 0), 3-hydroxytetradecanoic acid ( 3-OH C14 : 0), hexadecanoic acid ( C16 : 0) and 9 -hexadec enoic acid (C16 : 1n9) as major constituents with 2-hydroxydodeca- noic acid (2-OH C12 : 0), 3-hyd roxydodecanoic acid (3-OH C12 : 0) and 9-octadecenoic acid (C18 : 1n9) being minor components. Fatty acids accounted for 7% (w/w) of the aqueous-phase LPS and 12% (w/w) of the phenol-phase LPS suggesting an under-acylation pattern. The 31 P-NMR spectrum of aqueous-phase LPS in D 2 O, pH 6.5 showed two distinct groups of r esonances, indicating the presence of phosphate monoester at 0.49 p.p.m. and phosphate diester centred at )1.79 p.p.m. Methylation analysis of t he carboxyl-reduced CPS and subsequent GLC-MS analysis revealed the presence of 2,6-dideoxy-4-O-methyl-2-(N-methylacetamido)-glucose, Ó FEBS 2004 CPS and LSP O-antigen of A. salmonicida (Eur. J. Biochem. 271) 4509 3,6-dideoxy-2-O-methyl-3-( N-methylacetamido)-glucose and 2-deoxy-2-(N-methyla cetamido)-4 ,6-di-O-methyl-galactose in the approximate molar ratio of 0.8:1.0: 0.8, while GLC- MS analysis of the n ative CPS showed the presence of 2,6-dideoxy-4-O-methyl-2- (N-methylacetamido)-gluco se and 3 ,6-dideoxy-2-O-methyl-3-(N-methylacetamido)-glu- cose only, in approximate molar ratio of 0.5 : 1.0, and was consistent with the presence of GalNAcA in the native CPS. An additional minor component was also detected in GLC-MS analysis of both the native and the carboxyl- reduced CPS, its mass spectrum consistent with that of 3,6-dideoxy-2-O-methyl-3-[(N-acetyl- L -alanyl) methylami- do]-glucose, suggesting that L-Ala was located on 3-amino-3,6-dideoxy-glucose. The absolute configurations of 2-acetamido-2,6-dideoxy-glucose and 3-ac etamido- 3,6-dideoxy-glucose were determined by GLC-MS of the corresponding (R)-2-bu tyl-glucoside derivatives and found to be D , while the absolute configuration of the 2-acetam- ido-2-deoxy-galacturonic acid was determined following the hydrolysis of the carboxyl-reduced CPS and was also found to be D . The results suggest that A. salmonicida CPS and O-chain LPS are composed of linear trisaccharide repeating units containing 3-linked 2-acetamido-2-deoxy- D -quinovose, 4-linked 3-[(N-acetyl- L -alanyl) amido]-3-deoxy- D -quinovose and 3-linked 2-acetamido-2-deoxy- D -galacturonic acid. The sequence of constituent glycoses and the location of L -Ala were confirmed by 2D NMR analysis performed on both CPS and aqueous-phase LPS and their methanolysis products, and CE-ES-MS methods. In order to sharpen broad resonances due to the extreme viscosity of the CPS sample in water, some 2D NMR experiments were carried out on a CPS sample dissolved in D 2 O, pD 6.5 at 60 °C. Due to the structural identity of CPS and O-chain polysaccharide based on their compositional and methylation analyses, 1D 1 H-NMR spectra (Fig. 1,B) and a good solubility of LPS probably attributable to a relatively low fatty acid content, aqueous-phase LPS was used in most of 1D analogues of 2D NMR experiments. The 1D 1 H-NMR spectrum of CPS showed three reso- nances in the low-field region ( 5.3–4.4 p.p.m.) in the relative ratio of 1 : 1 : 3, of which three were later attributed to resonances from anomeric protons by direct correlation with corresponding 13 C resonances in a 2D heteronuclear 1 H)1 3 C correlation HSQC spectrum. Initial assignments were made from 2D COSY and TOCSY e xperiments carried out on the CPS sample (Table 1). From the HSQC spectrum in Fig. 1C, the anomeric protons of three sugar residues were designated a–c, according to the decreasing order of their proton chemical shifts. For residue a only H-1, H -2 (from 2D COSY) and H-3, H-4 (from 2D TOCSY) (Fig. 2A) could be identified, due to a small J 4,5 coupling constant which prevented magnetization transfer past H-4a,suggestinga galacto configuration. The HSQC 1 H- 13 C experiment identified C-2 at 49.3 p.p.m. confirming residue a being 2-amino glycose. The H-5a resonance was detected in the 2D NOESY experiment. A high chemical shift of H-4a at 4.43 p.p .m. was typical of a u ronic acid, confirming residue a to be 2-amino-2-deoxy-a- D -galacturonic acid (Table 1) substituted at position O-3 (C-3a at 77.9 p.p.m. due to a glycosylation effect [24]). This was further confirmed by CE-ES-MS and CE-ES-MS/MS analyses performed on CPS and LPS samples. Several attempts to carry out HMBC experiments on both CPS and aqueous-phase LPS samples to confirm the presence of carboxyl group were unsuccessful, possibly due to a line broadening effect. For residues b and c it was possible to make partial assignments of H -1, H -2 ( from 2 D C OSY) and H -3 ( from 2 D TOCSY). Presence of two methyl groups corresponding to H-6 of 6-deoxy sugars was observed at 1.27 p.p.m. and 1.29 p.p .m., their H-5 resonances at 3.56 p.p.m. and 3.34 p.p .m., respectively, could be traced through 2D COSY connectivities. In order to assign unambiguously the reso- nances for residues b and c selective 1D T OCSY–TOCSY experiments were c arried out with selective excitation of both H-6b and H-6c in the first step followed by selective excitation of H-5c and H-5b, respectively, permitting complete assign- ment of resonances for residues b and c (Table 1, Fig. 2C,D). Assignments of 13 C resonances were carried out by direct correlation of 1 H resonances with 13 C resonances in a heteronuclear 1 H- 13 CHSQCexperiment(Table 1,Fig.1C). Fig. 1. 1 H-NMR spectra for CPS (A), aque- ous-phase LPS (B) and HSQC spectrum for A. salmonicida CPS (C), recorded at 600 MHz, 60 °C, pD 6.5. 4510 Z. Wang et al.(Eur. J. Biochem. 271) Ó FEBS 2004 Most NH resonances of CPS and O-chain polysaccharide were assigned on the basis of their large coupling constants withtheringprotonsH-C-N-H(9–10Hz)viaCOSYand TOCSY experiments on samples in 90% H 2 O/10% D 2 O (Table 1). The NH-2a was assigned on the basis of intra- residue NOE with H-1a.TheNH-2c was assigned based on the intraresidue NOEs with H-1c,H-2c and H-3c (Fig. 3B). The presence of the intraresidue NOEs between the N-acetyl proton 3b-NH and the C H (2-Ala), CH 3 (3-Ala), and NH protons of L -alanine (NHAc-Ala), confirmed by NOESY experiment in 90% H 2 O/10% D 2 O, demonstrated that N-acetyl- L -alanyl substituent was located on position 3 of residue b (Fig. 3B). Proton resonances for the CPS sample did not appear as resolved multiplets due to heterogeneity resulting from the presence of both the amide a nd nonamide forms of GalNAcA and broadening due to high viscosity of the polysaccharide. Hence, in order to confirm the configur- ation of the sugars and structure, methanolysis products were purified and analysed by NMR to extract coupling constants. LPS sample was subjected to methanolysis and resultant methyl glycosides separated on a C18 column. Fractions were analysed by 1 H-NMR and methyl 3-[(N-acetyl- L -alanyl) amido]-3,6-dideoxy-a- D -glucopyrano- side, methyl 2-acetamido-2,6-dideoxy-a- D -glucopyranoside, and two disaccharides, in both methyl a-andb-glycoside form, c ould be readily identified. A number o f other products were separated as a mixture and these were not analysed further. Two products, namely methyl 3-O-(methyl-2-acetamido-2-deoxy-a- D -galactopyra- nosyluronate)-2-acetamido-2,6-dideoxy-a- D -glucopyr- anoside (Product 1) and methyl 3-[(N-acetyl- L -alanyl) amido]-3,6-dideoxy-a- D -glucopyranoside (Product 2), were fully characterized by 1D NMR a nd 2D COSY and HSQC experiments. Based on their 1 Hand 13 C chemical shifts (Table 1), coupling constants and comparison with the literature values [24], residues b and c were assigned a gluco configuration. The sequence of monosaccharides in the repeating unit of both CPS and O-chain polysaccharide was established from 2D NOESY spectrum f or anomeric resonances. Interresidue NOEs were observed between H-1a and H-3c and between H-1b and H-3a, suggesting a linear struc ture a-c-b (Fig. 2E,F). In addition, interresidue NOEs between H-1c and H-4b (F ig. 2G) indicated that residue c was linked to residue b. This was also supported by results of Table 1. 1 H- and 13 C-NMR chemical shifts (p.p.m.) for the CPS a nd O-polysaccharide of A. salmonicida strain 80204-1 and its methanolysis products [1,2]. Spectra were recorded at 6 0 °Cor25°CinD 2 O. For the detection of NH protons spectra were recorded at 25 °C in 90% H 2 O/10% D 2 O (v/v). The observed 1 Hand 13 C chemical shifts are reported relative to external acetone (d H 2.225 p.p.m., d C 31.07 p.p.m.). Error for d H ¼ 0.02 p.p.m. and error for d C ¼ 0.2 p.p.m. Sugar residue 1 23456CH 3 CO NH 1-OMe 6-OMe CPS and O-polysaccharide -3)-a-GalpNAcA(1- a 5.22 4.32 3.95 4.43 4.12 1.99 8.30 99.8 49.3 77.9 71.0 72.9 23.2 -3)-b-QuipNAc(1- c 4.43 3.65 3.59 3.29 3.34 1.29 1.99 8.69 102.0 55.7 81.4 77.4 72.6 17.9 23.2 -4)-b-Quip3NAlaNAc(1- b 4.55 3.33 3.82 3.35 3.56 1.27 8.04 105.2 72.5 56.4 80.5 73.3 17.9 AlaNAc 4.42 1.38 2.07 8.06 176.0 50.8 19.2 22.7 Product 1 a-GalpNAcA6OMe(1- a 5.31 4.20 3.93 4.32 4.52 1.97 3.86 100.3 50.9 68.2 70.6 72.6 23.2 54.3 -3)-a-QuipNAc(1-OMe c 4.66 4.03 3.75 3.42 3.78 1.31 2.05 3.38 99.5 53.8 80.5 77.4 68.8 17.9 23.4 56.4 Product 2 a-Quip3NAlaNAc(1-OMe b 4.76 3.65 4.00 3.23 3.77 1.28 3.45 AlaNAc 4.34 1.39 2.04 Fig. 2. NMR experiments for assignment and sequence determination of sugar residues in A. salmonicida CPS and LPS. (A) Proton spectrum at 25 °C. (B) Slice from a 90-ms TOCSY for 1a. (C) 1D TOCSY–TOCSY (6bc,20ms;5b, 60 ms). (D) 1D TOCSY–TOCSY (6bc,20ms;5c, 90 m s). (E–G) Slice from a 50 ms NOESY for 1 a,1b and 1c reso- nances. For each se lective experiment the irradiated resonance and the mixing time are indicated. Ó FEBS 2004 CPS and LSP O-antigen of A. salmonicida (Eur. J. Biochem. 271) 4511 methylation and CE-ES-MS analyses obtained on both CPS and LPS. CPS and the carboxyl-reduced LPS samples of A. salmonicida were subjected to CE-ES-MS analysis using an API 3000 as a detector in a positive mode with high orifice voltage (200 V) that allowed fragmentation of both polymers (Table 2). Th e CE-ES-MS spectrum of C PS was consistent with the presence of two major species at m/z 663 and m/z 1326, corresponding to one and two trisaccharide repeating unit-containing species, respectively, with the molecular masses for constituent sugar residues being m/z 258 for Qui3NAlaNAc, m/z 217 for GalNAcA and m/z 187 for QuiNAc. In addition, fragments at m/z 405 and m/z 1067 consistent with the loss of Qui3NAlaNAc were also observed. The CE-ES-MS spectrum of the carboxyl- reduced LPS showed the presence of an additional fragment at m/ z 649, a product expected following the carboxyl group reduction of GalNAcA and corresponding to the CPS trisaccharide repeating unit containing 2-acetamido- 2-deoxy-galactose residue. Presence of a minor component at m/z 662 was unexpected and suggested that some of carboxyl groups in GalNAcA were substituted by a primary amine. In order to confirm the extent of amidation of GalNAcA residue, the CPS sample was analysed by CE-ES- MS using a Qq-TOF mass s pectrometer. The extent of amidation could be determined by the intensity ratio of fragments at m/z 662.2 D a (GalNAcAN-containing trisac- charide repeating unit) and m/z 663.2 Da (GalNAcA- containing trisaccharide repeating unit) (Fig. 4A, inset), based on this ratio about 40% of G alNAcA in both CPS and O-chain p olysaccharides was present in the form of a primary amine. When the carboxyl-reduced A. salmonicida LPS was subjected to the same analysis, as expected, the ion at m/z 663.2 disappeared giving rise to a new abundant ion at m/z 649.2, corresponding to the product following the carboxyl-reduction of GalNAcA (Fig. 4B, inset) which was also consistent with the signal intensities obtained for anomeric proton resonances of GalNAc and GalNAcAN, respectively, in the 1 H-NMR spectrum of t he carboxyl- reduced LPS sample of A. salmonicida (data not shown). The proposed structure of CPS and O-chain polysaccharide of A. salmonicida strain 80204-1 is depicted in Fig. 5. The biological significance of these findings was estab- lished through the chemical and CE-ES-MS and CE-ES- MS/MS analyses of t he in vivo cultured A. salmonicida strain 80204-1 cells [20]. Methanolysis of the bacterial cells followed by GLC-MS analysis s howed the p resence of sugars consistent with the composition of CPS and LPS O-chain polysaccharide, namely 2-amino-2-deoxy- D - quinovose, 3-amino-3-deoxy- D -quinovose and 2-amino- 2-deoxy- D -galacturonic acid, in the approximate molar ratio of 1.5 : 1.0 : 0.4. Composition and methylation ana- lyses o f the inoculum TSB c ulture used to prepare the growth chambers showed no sugars corresponding to the proposed novel CPS structure and were c onsistent with Fig. 3. Partial s tructure of the repeating unit of A. salmonicida CPS and O-chain polysaccha- ride with intraresidue NOE connectivities (A) and 2D NOESY spectrum for A. salmonicida LPS (B) acquired with a m ixing time of 1 00 ms in 90% H 2 O/10% D 2 O. The proton spectra are also shown along both axes with assign- ments. Table 2. Positive ion CE-ES-MS data and proposed composition of the native CPS and the carboxyl-reduced O-chain polysaccharide of A. salmonicida strain 80204-1. Monoisotope mass units were used for calculation of molec ular mass values based on propo sed compositions as follows: GalNAcA, 217.06; G alNAc, 203.08; GalNAcAN, 216.08; QuiNAc, 1 87.09; Qui3NAc, 187.09; Ala, 71.04. For a, b and c desig- nation see Table 1. a¢ ¼ GalNAc; a¢¢ ¼ GalNAc AN. Observed ion (m/z) Caculated mass (Da) Proposed composition CPS Carboxyl-reduced O-chain polysaccharide 204.04 203.08 – a¢ 217.03 216.08 a¢¢ a¢¢ 218.01 217.06 a – 391.11 390.17 – a¢-c 404.11 403.17 a¢¢-c a¢¢-c 405.08 404.15 a-c – 649.17 648.29 – a¢-c-b 662.11 661.29 a¢¢-c-b a¢¢-c-b 663.16 662.27 a-c-b – 1297.37 1296.58 – (a¢-c-b) 2 1310.34 1309.58 – a¢¢-c-b-a¢-c-b 1324.23 1323.56 a¢¢-c-b-a-c-b – 1325.18 1324.54 (a-c-b) 2 – 4512 Z. Wang et al.(Eur. J. Biochem. 271) Ó FEBS 2004 the previously identified O-chain containing L -rhamnose, 2-amino-2-deoxy- D -mannose and D -glucose [11], confirming that the novel CPS and O-chain polysaccharide were produced during the 72-h in vivo culture p eriod. In addition, sugars detected in the inoculum TSB culture were also present. Based on sugar ratios of 3-amino-3-deoxy- D -qui- novose and L -rhamnose,  10% of the newly formed CPS and LPS O-chain polysaccharide was present i n the in vivo cultured cells. Methylation analysis performed on in vivo cultured A. salmonicida strain 80204-1 cells confirmed these findings and showed the presence of 2,6-dideoxy-4- O-methyl-2-(N-methylacetamido)-glucose and 3,6-dideoxy- 2-O-methyl-3-( N-methylacetamido)-glucose, in accordance with the proposed structure of C PS. T he linkage of GalNAcA could not be verified as the methylation analysis was carried out without the carboxyl group reduction step. To confirm the sequence of this newly formed polysac- charide detected in the in vivo cultured cells, CE-ES-MS analysis was carried out on both the concentrated bacterial cell saline wash containing CPS and on the bacterial pellet following mild acid hydrolysis to release delipidated LPS. Initially, the CPS and LPS samples were subjected to CE-ES-MS analysis with the typically used orifice of 45 V. The mass spectra indicated no individually separated ion peaks and showed only a b road peak (data not shown) corresponding to a range of molecular masses which could be attributed to CPS or L PS O-chain polysaccharide repeating unit fragments of different length. Previously, we have successfully applied this approach to the partial degradation of the polysaccharide into shorter oligosaccha- ride units due to front-end collision induced dissociation [25]. It is noteworthy to mention that this technique might not be suitable for determination of molecular masses of intact CPS and LPS O -chain polysaccharide but the information provided by this pseudo pyrolysis MS could be is very useful for characterization o f t heir repeating units. Fig. 4. CE-ES-MS analysis (+ ion mode) of LPS (A) and carboxyl-reduced LPS (B) of A. salmonicida. The inset sho ws the ratio of the native and amidated forms of LPS and CPS. Separation conditions: bare fused-silica (90 cm · 50 lm i.d., 185 lm o.d), 5% meth- anol in 15 m M ammonium acetate, pH 9.0, +15 k V. The high orifice voltage of Q-Star (120 V) pro vided the environm ent to break a polysaccharide into repeating units, giving diagnostic fragments arising f rom the cleavage of glycosidic bonds. Fig. 5. The pro posed structure of th e C PS and O-chain polysaccharide of A. salmonicida strain 80204-1. Ó FEBS 2004 CPS and LSP O-antigen of A. salmonicida (Eur. J. Biochem. 271) 4513 In order to evaluate the sensitivity of the CE-ES-MS approach, a series of purified A. salmonicida LPS standards containing between 100 lgand5lg, were prepared, treated with 1% acetic acid and subjected to CE-ES-MS analysis with a high orifice voltage of 120 V (not shown). On the basis of these experiments it w as established that e ven at t he lowest dilution corresponding to 5 lg of purified LPS, the observed CE-ES-MS spectrum was consistent with the previously established fragmentation pattern, where the fragment ion at m/z 663.3 corresponded to the mass of one trisaccharide repeating unit-containing species with GalNAcA, fragment ion at m/ z 1325.5 corresponded to the mass of two t risaccharide repeating units, and fragmen t ion at m/z 1067.4 was consistent with the loss of Qui3NAlaNAc. Due to a relatively high background at lower dilutions of the LPS standard and i n o rder to con firm the CE-ES-MS assignments, tandem MS was used for the precursor ions at m/z 663.3 and m/z 662.3, corresponding to the mass of one trisaccharide repeating unit-containing species with GalN- AcA a nd GalNAcAN, r espectively. As expected, two fragment ions, at m/z 404.2 and m/z 259.2, respectively, were observed, where a fragment ion at m/z 404.2 corres- ponded to the loss of Qui3NAlaNAc ( 258 Da), and a fragment ion at m/z 259.2, corresponded to a protonated Qui3NAlaNAc species. When direct CE-ES-MS analysis was carried out on the in vivo cultured cells of A. salmonicida strain 80204-1, as shown in Fig. 6, the presence of m/z 662.3 could not be unambiguously confirmed due to a high background which also resulted in the loss of other characteristic fragment ions at m/z 1067.4 and m/z 1325.5. However, CE-ES-MS/MS analysis of the precursor ion at m/z 662.3 was consistent with its previously determined composition of a trisaccha- ride repeating unit consisting of Qui3NAlaNAc, GalN- AcAN and QuiNAc and confirming its presence in t he in vivo cultured cells. Similarly, based on the CE-ES-MS/ MS analysis of the precursor ion at m/z 662.3, the same trisaccharide repeating unit could be a lso detected in 2.5% saline wash of the in vivo grown bacterial cells. It is noteworthy that the fragment ion at m/z 663.3 correspond- ing to a trisaccharide-repeating unit containing GalNAcA was not detected in the in vivo cultured cells, suggesting a complete amidation of GalNAcA. Discussion In this study we have established the structure of the CPS and LPS O-chain polysaccharide of A. salmonicida strain 80204-1 produced under in vitro growth conditions on TSA. Both polysaccharides were shown by c omposition, methy- lation analysis, NMR and MS methods to be composed of linear t risaccharide repeating units containing 3-linked 2-acetamido-2-deoxy- D -quinovose, 4-linked 3-[(N-acetyl- L -alanyl)amido]-3-deoxy- D -quinovose and 2-acetamido- 2-deoxy- D -galacturonic acid. To our knowledge, this is the first report describing the presence of the CPS in A. salmonicida. We have confirmed by direct CE-ES-MS analysis that both CPS and O-chain polysaccharide were also present in the in vivo -grown cells of A. salmonicida strain 80204-1 harvested at 72 h postimplant surgery. These polysaccha- rides were not detected in the in vitro-grown bacterial inoculum TSB culture used for the implants. To date a limited number of bacteria have been reported to produce capsular and O-chain polysaccharides with identical structures. It appears that this property is not uncommon for fish pathogens and we have previously reported similar findings for Listonella (formerly Vibrio) anguillarum and V. ordalii [26,27]. It should be noted that the structures of the CPS and O-chain polysaccharide of L. anguillarum and V. ordalii have recen tly been re-examined and that the galacto configuration of the Fig. 6. CE-ES-MS and CE-ES-MS/MS ana- lysis(+ionmode)ofLPSfromin vivo A. salmonicida strain 80204-1 cells. Extracted mass spectra for LPS. Extracted MS/MS spectra of precursor ion m/ z 662.3 promoted using an orifice voltage of 120 V. Other conditions were the same as in Fig. 4. 4514 Z. Wang et al.(Eur. J. Biochem. 271) Ó FEBS 2004 2,3-diacetamido-2,3-dideoxy-hexuronic acid in both struc- tures should be revised in favour of the gulo configuration (Y. A. Knirel & A. S. Shashkov, personal communication). In the p resent investigation, we have shown that t he structures of both CPS and O-chain polysaccharide were distinct from the previously reported O-chain polysacchar- ide of A. salmonicida produced in TSB and consisting of L -rhamnose, D -mannosamine and D -glucose [11] which was also detected in the bacterial inoculum TSB culture used to prepare the in vivo growth chambers. Direct CE-ES-MS analysis of the in vivo bacterial cells cultured within the semipermeable growth chambers for 72 h has confirmed the formation of a novel polysaccharide. It is possible t hat the use of dialysis tubing in the in vivo growth chambers provides an environment more clo sely resembling that seen in a systemic infection within salmon. That environment may b e critical for formation of the polysaccharide [22]. The devised microanalytical procedure is suitable for direct analysis of both CPS and LPS and could detect as little as 2 lgofLPSinthesample. The present studies suggest that caution s hould be exercised when in vitro -cultured cells are used for isolation and structural analysis of bacterial polysaccharides as the resultant structural i nformation may not be biologically relevant to in vivo conditions. Application of CE-ES-MS methods for direct analysis of cells from experimental models of infection can overcome these limitations. CE-ES- MS has been proven to b e a power ful technique to distinguish different glycoforms when applied to lipopoly- saccharides from Neisseria meningitidis [28]. However, in this particular application we a ttempted to improve the detection sensitivity by separating the C PS and LPS from salts and other matrices associated with the sample preparation. These findings suggest that additional viru lence factors such as CPS and novel LPS O-chain polysaccharide contribute to the pathogenesis of A. salmonicida in vivo and emphasize a critical role of a host in host–pathogen interactions. Acknowledgements We thank Dr Giles Olivier from the Gulf Sciences Centre, Department of Fishe ries and Oceans Canada, Moncton, NB for the kind gift of A. salmonicida strains used in this study, Dr Jessica Boyd from the Institute o f Marine Biosciences, National Research Council of Canada for providing A. salm onicida stock cultures and the Dalhousie University Aquatron for wetlab s pace. We also thank Mrs Vandana Chandan from the Institute of Biological Sciences, National Research Council of Canada for excellent training in bacterial culture techniques. References 1. Bellard, R.J. & Trust, T.J. (1985) Synthesis, export and assembly of Aeromonas salmonicida A-layer analyzed by transpo son muta- genesis. J. Bacteriol. 163, 877–881. 2. Whitby, P.W., Landon, M . & Coleman, G. (1992) The cloning and nucleotide sequence of the serine protease gene (aspA) of Aeromonas salmonicida Ssp. Salmonicida. FEMS Microbiol. Lett. 78, 65–71. 3. Hirono, I. & Aoki, T. (1993) Cloning and characterization of three hemolysin genes from Aeromonas salmonicida. Microb. Pathog. 15, 5812–5823. 4. Masada, C.L., LaPatra, S.E., Morton, A.W. & Strom, M.S. (2002) An Aeromonas salmonicida type IV pilin is required for virulence in rainbow trout Oncorhynchus mykiss. Dis. Aquat. Organ. 51, 13–25. 5. Lee, K.K. & Ellis, A.E. (1990) Glycerophospholipid: cholesterol acyltransferase complexed with lipopolysaccharide (LPS) is a major lethal exotoxin and cytolysin of Aeromonas salmonicida: LPS stabilizes and enhances toxicity of the enzyme. J. Bacteriol. 172, 5382–5393. 6. Gardun ˜ o, R.A., Moore, A.R., Olivier, G., Lizama, A.L., Gard- un ˜ o, E. & Kay, W.W. (2000) Host cell invasion and intracellular residence by Aeromonas salmonicida: Role of the S-layer. Can. J. Microbiol. 46, 660–668. 7. Kay, W.W., Buckley, J.T., Ishiguro, E.E ., Phipps, B.M., Monette, J.P.L. & Trust, T.J. (1981) Purification and disposition of a surface protein associated with viru le nce of Ae romonas salmonicida. J. Bacteriol. 147, 1077–1084. 8. Chart, H., Shaw, D.H., Ishiguro, E.E. & Trust, T.J. (1984) Structural and imm unochemi cal homo geneity of Aeromonas sal- monicida lipopolysaccharide. J. Bacteriol. 15 8, 16–22. 9. Gardun ˜ o, R.A., Thornton, J.C. & Kay, W.W. (1993) Aeromonas salmonicida growninvivo.Infect. Immun. 61, 3854–3862. 10. Merino, S., Albertı ´ ,S.&Toma ´ s, J.M. (1994) Aeromonas salmo- nicida resistance to complement-mediated killing. Infect. Immun. 62, 5483–5490. 11. Shaw, D.H., Lee, Y Z., Squires, M.J. & Lu ¨ deritz, O. (1983) Structural studies on the O-antigen of Aeromonas salmonicida. Eur. J. Biochem. 131 , 633–638. 12. Shaw,D.H.,Hart,M.J.&Lu ¨ deritz, O . (1992) Structure of the co re oligosaccharide isolated from Aeromonas salmonicida ssp. salmonicida. Carbohydr. Res. 231, 83–91. 13.Garrote,A.,Bonet,R.,Merino,S.,Simo ´ n-Pujol, M.D. & Congregado, F. (1992) Occurrence of a capsule in Aeromonas salmonicida. FEMS Microbiol. Lett. 95, 127–132. 14. Olivier,G.,Moore,A.R.&Fildes,J.(1992)ToxicityofAeromonas salmonicida cells to Atlantic salmon Salmo salar peritoneal mac- rophages. Dev. Comp. Immunol. 16, 49–61. 15. Westphal, O. & Jann, K. (1965) Bacterial polysaccharides. Extraction with phenol-water and further applications of the procedure. Methods Carbohydr. Chem. 5, 83–91. 16. Leontein, K., Lindberg, B. & Lo ¨ nngren, J. (1978) Assignment of absolute configuration of sugars by g.1.c. of their acetylated glycosides formed from chiral alcohols. Carbohydr. Res. 62, 359– 362. 17. Ciucanu, I. & K erek, F. (1984) A simple and rapid method for the permethylation of carbohydrates. Carbohydr. Res. 131, 209–217. 18. Stellner, K., Saito, H. & Hakomori, S I. (1973) De termination of aminosugar linkages in glycolipids by methylation. Aminosugar linkages of ceramide pentasaccharides of rabbit erythrocytes and of Forssman antige n. Arch. Biochem . Biophys. 155, 464–472. 19. Taylor, R.L. & Conrad, H.E. (1972) Stoichiometric depolymeri- zation of polyuronides and glycosaminoglycouronans to mono- saccharides following reduction of their carbodiimide-activated carboxyl groups. Biochemistry 11, 1383–1388. 20. Brisson, J. -R., Su e, S.C., W u, W.G., McManus, G., Nghia, P.T. & Uhrin, D. (2002) NMR o f carbohydrates: 1D h omonuclear selective methods. I n NMR Spectroscopy of G lycoconjugates. (Jimenez-Barbero, J. & Peters, T., eds), pp. 59–93. Wiley-VCH, Weinhem. 21. Li, J., Thibault, P., Martin, A., Richards, J.C., Wakarchuk, W.W. & van der Wilp, W. (1998) Development of an on-line pre- concentration method for the analysis of pathogenic lipopoly- saccharides u sing capillary elec trophoresis-electrospray mass spectrometry: Application to small colony isolates. J. Chromatogr. A 817, 325–336. 22. Dacanay, A., Johnson, S.C., Bjornsdottir, R., Ebanks, R.O., Ross,N.W.,Reith,M.,Singh,R.K.,Hiu,J.&Brown,L.L.(2003) Ó FEBS 2004 CPS and LSP O-antigen of A. salmonicida (Eur. J. Biochem. 271) 4515 Molecular characterization and quantitative analysis of super- oxide dismutases in virulent and avirulent strains o f Aeromonas salmonicida Subsp. Salmonicida. J. Bacteriol. 185, 4336–4344. 23. Dubois, D.M., Gilles, K.A., Hamilton, J.K., Rebers, P.A. & Smith, F. (1956) Colorimetric method for determination of sugars and related substances. Anal. Chem. 28, 350–356. 24. Bock, K. & Thøgersen, H. (1982) Nuclear magnetic resonance spectroscopy in the study of m ono- and oligosaccharides. Annu. Report NMR Spectrosc. 13, 1–75. 25. Szymanski, C.M., St. Michael, F.S., Jarrell, H.C., Li, J., Gilbert, M., Larocque, S., Vinogradov, E. & Brisson, J R. (2003) Detection of conserved N-linked glycans and phase variable lipo - oligosaccharides and capsules from campylobacter cells by mass spectrometry and high resolution magic angle spinning NMR spectroscopy. J. Biol. Chem. 278, 24509–24520. 26. Sadovskaya, I., Brisson, J R., Mutharia, L.M. & Altman, E. (1996) Structural studies of th e lipopolysaccharid e O-antigen and capsular polysaccharide of Vibrio anguillarum serotype O: 2. Carbohydr. Res. 283, 111–127. 27. Sadovskaya, I., B risson, J R., Kh ieu, N.H., Mutharia, L.M. & Altman, E. (1998) Structural characterization of the lipopoly- saccharide O-antigen and capsular polysaccharide of Vibrio ordalii serotype O: 2. Eur. J. Biochem. 253, 319–327. 28. Li, J., Cox, A.D., Hood, D.M., Moxon, E.R. & Richards, J.C. (2004) Application of capillary electrophoresis-electrospray mass spectrometry to the separation and characterization of isomeric lipopolysaccharides of Neisseria meningitidis. Electrophoresis 25, 2017–2025. 4516 Z. Wang et al.(Eur. J. Biochem. 271) Ó FEBS 2004 . Structural studies of the capsular polysaccharide and lipopolysaccharide O-antigen of Aeromonas salmonicida strain 80204-1 produced under in vitro and in. grown in vitro on tryptic soy agar, A. salmonicida strain 80204-1 produced a capsular polysaccharide with the identical structure to that of the lipopolysaccharide