NMR and MS evidences for a random assembled O-specific chain structure in the LPS of the bacterium Xanthomonas campestris pv. Vitians A case of unsystematic biosynthetic polymerization Antonio Molinaro 1 , Cristina De Castro 1 , Rosa Lanzetta 1 , Michelangelo Parrilli 1 , Bent O. Petersen 2 , Anders Broberg 2, * and Jens Ø Duus 2 1 Dipartimento di Chimica Organica e Biochimica, Universita ` di Napoli Federico II, Napoli, Italy; 2 Department of Chemistry, Carlsberg Laboratory, Copenhagen, Denmark Xanthomonas campestris pv. vitians is a Gram-negative plant-associated bacterium that acts as causative agent of bacterial leaf spot and headrot in lettuce. The lipopolysac- charide of this bacterium is suspected to be an important molecule for adhesion to and infection of the plants. The lipopolysaccharide has been isolated from the phenol phase and the O-specific chain characterized by compositional analysis, high field NMR and MALDI-TOF MS. It consists of a nonrepetitive branched polysaccharide with a rhamnan backbone to which Fuc3NAc is linked. The NMR and MS approach led to the characterization of the fine structure of the polymer, which is randomly assembled. The rhamnan backbone is built up of b-Rhap and a-Rhap,thislastis present in one, two or three adjacent units and branched by an a-Fucp3NAc unit. This is a real case of a random con- stituted O-specific chain, therefore biosynthetic studies towards the comprehension of this irregular biosynthesis are needed. Keywords: Xanthomonas campestris; lipopolysaccharide; phytopathogen; O-specific chain random polymerization. Until recently, structural data on lipopolysaccharides (LPSs) of phytopathogenic bacteria have been rather limited, but the interest in their structure is increasing, especially in the structural analysis of O-specific polysaccharides (OPSs) [1]. The main purposes of these investigations are to establish a chemotaxonomic relationship among strains and species of the same genus, to clarify the biosynthesis of the LPS and finally to evaluate their role in the interaction between the bacterium and the vegetal host organisms. Xanthomonas campestris pv. vitians is a Gram-negative bacterium, which acts as causative agent of bacterial leaf spot and headrot in lettuce. This disease is easily recognized by translucent and water-soaked brown lesions that get dark after a while. The molecular basis of this disease is not understood. Recently, the O-specific polysaccharide (OPS) of the LPS extracted from the aqueous phase of Xanthomonas campestris pv. vitians (X. hortorum pv. vitians) has been described [2]. The structure consists of a linear non-strictly repetitive rhamnan: [fi3)-a-L-Rhap-(1fi] n 3)-b-L-Rhap-(1fi where n is more frequently equal to two but it also assumes values equal to one and or three. This rhamnan backbone is identical to the Smith degraded product of the OPS derived from the LPS contained in the phenol phase, which differed by the additional presence of terminal Fuc3NAc units. However, this apparent minimal structural difference made both 1 Hand 13 CNMRspectra much more intricate than those of the backbone obtained by Smith degradation [2]. In this paper a detailed NMR and MALDI-TOF MS analysis is reported and based on the results a random structure for the OPS of the phenol- phase LPS is proposed. MATERIALS AND METHODS Growth of bacteria, isolation of LPS and OPS X. campestris pv. vitians strain 1839 (NCPPB), obtained from the National Collection of Plant Pathogenic Bacteria, Harpenden, UK, was grown as described [3] and then lyophilized. Lipopolysaccharide was isolated from dried bacterial cells (4.45 g) by extraction using hot phenol/water extraction [4]. The phenol phase LPS was a mixture containing a large amount of low molecular mass glucan, therefore it was further purified by GPC, on Bio-Gel A-15 m column (3 · 150 cm) with 300 m M triethylamine-EDTA buffer (pH 7) as eluent, monitored with a Waters differential refractometer. The pure LPS was recovered in the void volume and further purified by precipitation with 2-propanol Correspondence to Antonio Molinaro, Dipartimento di Chimica Organica e Biochimica, Universita ` di Napoli Federico II, Complesso Universitario Monte S. Angelo, via Cynthia 4, 80126 Napoli, Italy. Fax: +39 081674393, Tel.: +39 081674123, E-mail: molinaro@unina.it Abbreviations: Fuc3NAc, 3-acetamido-3,6-dideoxy-galactose; HPAEC, high performance anion exchange chromatography; LPS, lipopolysaccharide; OPS, O-specific polysaccharide chain; PSD, post source decay; Rha, rhamnose. *Present address: Department of Chemistry, Swedish University of Agricultural Sciences, SE-750 07 Uppsala, Sweden Dedication: dedicated to Prof Lorenzo Mangoni on the occasion of his 70th birthday. (Received 1 March 2002, revised 26 June 2002, accepted 25 July 2002) Eur. J. Biochem. 269, 4185–4193 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03138.x and in succession chromatographed on Biogel A 1.5-m (yield 370 mg, 8% of cell dry mass, w/w). The product released by mild acid hydrolysis (aqueous 1% HOAc, 100 °C, 2 h) of the LPS was applied to a GPC, Bio-Gel P-10 column (3 · 90 cm) with ammonium bicarbonate buffer (pH 5) as eluent. The polymeric fraction eluted in the void volume was the O-polysaccharide (340 mg, yield 91% of LPS). General and analytical methods The monosaccharides were identified by GLC and GLC-MS as acetylated O-methyl glycoside derivatives: briefly, samples were treated with 1 M HCl/MeOH at 80 °C for 20 h, dried under reduced pressure and then acetylated with acetic anhydride in pyridine at 80 °C for 30 min. The absolute configuration of Rha and Fuc3NAc residues was determined by the published method [5], using GLC of acetylated (S)-2-octyl glycosides, temperature profile: 150 °C for 8 min, then 2 °Cmin )1 to 200 °Cfor0min, then 6 °Cmin )1 to 260 °C for 5 min. The Fuc3NAc retention time was compared with an authentic sample obtained by synthesis [6]. Methylation analysis of polysaccharide was carried out by standard procedure [7] and monitored by GLC-MS of the partially methylated alditol acetates. NMR spectroscopy Chemical shifts obtained by NMR spectroscopy were assigned using 2D homo- and heteronuclear experiments at 799.96 MHz for proton and 201.12 MHz for carbon, using the methyl group of acetone as reference for proton (2.225 p.p.m) and 1,4-dioxane for carbon (67.4 p.p.m). Spectra were recorded at 310K on a Varian UNITY INOVA 800. In addition to standard 1D proton spectra a series of 2D spectra were obtained. Double quantum-filtered phase- sensitive COSY experiment was performed using the Varian standard program tndqcosy (Double-Quantum-filtered- COrrelation-SpectroscopY, with water suppression), with 0.258 s acquisition time and 4096 data points in the F2 dimension. The data matrix was zero-filled in the F1 dimension to give a matrix of 4096 · 2048 points and was resolution enhanced in both dimensions by a shifted sine- bell function before Fourier transformation. Similarly, the nuclear Overhauser experiment was performed using the Varian standard tnnoesy program (Nuclear-Overhauser- Enhancement – SpectroscopY, with water suppression), with a mixing time of 25 ms. The TOCSY experiment was performed using standard Varian program tntocsy (TOtal- Correlation-SpectroscopY, with water suppression) with a spinlock time of 80 ms. The heteronuclear experiments were performed using pulse field gradient programs as gHSQC, gHSQC-TOCSY, gHSQC-NOESY and gHMBC. The spectra were assigned using the computer program Pronto [8], which allows the simultaneous display of different two-dimensional spectra and the individual label- ling of cross peaks. Mild hydrolysis An aliquot of the OPS (10 mg) was dissolved in a trifluoroacetic acid solution (0.01 M ) and left at 120 °Cfor 6 h. After lyophylization the sample was chromatographed on Bio-Gel P2 (2 · 100 cm), using the same conditions as above, two main products (fractions 1 and 2) were recovered and analyzed by MALDI-TOF MS. High-performance anion-exchange chromatography Further purification of fraction 1 was performed by high performance anion exchange chromatography (HPAEC) with pulsed amperometric detection (GP40 pump connected to a CarboPac PA-100 column (4 · 250 mm) and an ED40 electrochemical detector operated in the pulsed ampero- metric detection mode; Dionex, Sunnyvale, CA, USA). The eluent was 36 m M NaOH and an elution gradient was formed with NaOAc (0–250 m M in 17 min) at 0.8 mLÆmin )1 . A carbohydrate membrane desalter (Dionex; 0.1 M H 2 SO 4 as regenerant at 6 mLÆmin )1 ) was employed for on-line cation-exchange (H + /Na + ) of the eluent. A volume of 20 lL was injected from a sample solution earlier analyzed with NMR and five fractions were manually collected and freeze-dried. MALDI-TOF MS Fractions 1 and 2 from partial hydrolysis of the O-specific polysaccharide were analyzed on a Bruker Reflex III MALDI-TOF mass spectrometer (Bruker, Germany) operated in the delayed extraction mode with an acceler- ating voltage of 20 kV and a reflectron voltage of 22.8 kV. A solution of each sample was mixed 1 : 1 (v/v) with the matrix 2,5-dihydroxy-benzoic acid (20 mgÆmL )1 in H 2 O/ CH 3 CN, 3 : 2) and 1 lL of the mixture was deposited on a chromeplated stainless steel target and dried under reduced pressure. After introduction of the target in the ion-source, the laser power was adjusted to a level just above the threshold for formation of ions and the results from 50 laser shots were summed. A mixture of maltotriose, maltotetra- ose, maltopentaose, maltohexaose and maltoheptaose was used as external calibrant. MALDI postsource decay (PSD) TOF MS experiments were performed to study fragmentation patterns of oligo- saccharides isolated (fraction 3) by HPAEC-PAD. The samples for MALDI-PSD TOF MS were prepared as described above. The laser power was adjusted to a level considerably higher than the threshold value required to form ions and the reflectron voltage was stepped down from 22.8 kV in seven steps (25% decrease in voltage in each step). Combination of the recorded mass segments as well as instrument calibration using fragments from the peptide adrenocorticotropic (ACTH) hormone (18–39) clip were performed using software supplied by Bruker. The ATCH was purchased from Sigma. Smith degradation An aliquot of O-polysaccharide (20 mg) was N-deacety- lated at 120 °CwithKOH4 M for 16 h with stirring [2]. After neutralization, dialysis (cut-off 3500 Da) and lyophilization, the sample (18 mg) was submitted to Smith degradation [2]: it was treated with 50 m M NaIO 4 at 4 °C for 7 days, followed by addition of ethane-1,2- diol, reduction (NaBH 4 ), acidification (2 M acetic acid), dialysis and freeze-drying. Then the oxidized polymer 4186 A. Molinaro et al.(Eur. J. Biochem. 269) Ó FEBS 2002 was hydrolyzed with 1% HOAc at 100 °C, 1.5 h, and acid was removed by freeze-drying. The product was purified by Bio-Gel P2 (2 · 100 cm), eluted in the void volume with 50 m M ammonium bicarbonate buffer (pH 5), monitored with a Waters differential refractom- eter, and dried (15 mg). Molecular modelling Molecular modelling was carried out using the consistence valence force field in the Discover program [9]. The monosaccharide residues were constructed using standard bond lengths and angles in the Insight II program (MSI, San Diego). Molecular dynamics simulations were per- formed for the fragments containing from eight to ten residues in a water box of sidelength of 50 A ˚ for 500 ps and with a step length of 1 fs. Full coordinates were saved every 2.5 ps. Among the structures modelled were A–A–B–A–A– B–A–A, A–A–B–A– A–B–A–A, A–A–B–A–AvB–A–A and A–A–B– A–A–B–A–A (B: b-Rhap,A:a-Rhap, A: a-Fucp3NAc(1fi2)a-Rhap). The phi and psi angles are defined by H1–C1–O1–CX and C1–O1–CX–HX, respect- ively, where X is the position of glycosylation. RESULTS Isolation, characterization of the LPS and isolation of the OPS The LPS fraction was extracted from dried cells using the hot phenol/water method and isolated in the phenol phase, further purified by precipitation with 2-propanol and, in succession, chromatographed on Biogel A 1.5-m. The fatty acids composition (3-hydroxydecanoic, 3-hydroxydodeca- noic) and the presence of Kdo in the compositional analysis of purified fraction confirmed the presence of a lipopoly- saccharide. By SDS/PAGE the LPS showed a pattern indicating a wide continuous distribution of molecular mass. Mild acid hydrolysis with 1% HOAc yielded the lipid A moiety as precipitate and the OPS was isolated from the supernatant and further purified by gel-permeation chro- matography. Compositional, size of ring and linkage analysis GLC-MS analysis of the acetylated O-methyl glycosides and of the acetylated (S)-2-octyl glycoside derivatives showed that OPS is composed by L -rhamnose and 3-acetamido, 3,6-deoxy- D -galactose (Fuc3NAc). This last derivative was identified by comparison with an authentic sample. GLC-MS analysis of the partially methylated alditol acetates yielded the following sugar composition: 2,3-disubstituted L-Rhap, 2,3-substituted L-Rhap, terminal-D-Fucp3NAc in the ratios 2 : 1 : 1, respectively. NMR analysis of the OPS The presence of Rhap in both anomeric configurations a and b was previously established on the Smith degraded product [2], whereas the a configuration of Fucp3NAc was inferred by the 1 J C,H values [10] of anomeric signals measured in a coupled 13 C NMR spectrum of OPS. The 1 Hand 13 C NMR spectra of the OPS were indicative of a highly complex structure (Fig. 1). In the 13 CNMR spectrum, the a-anomeric signals attributable to Rhap and Fucp3NAc residues were broadened in the region 101– 103 p.p.m., while the anomeric signals of b-Rhap units ( 1 J C,H ¼ 160–163) [11] occurred around 97 p.p.m. In addi- tion, the region of glycosylated carbon (76–82 p.p.m) appeared very crowded suggesting a nonregular structure of the polymer. Thus, despite the simplicity of composi- tional analysis data, it appeared clear that this polymer was arranged in a rather intricate assemblage. Therefore a detailed 2D NMR analysis (Table 1) was performed at 800 MHz using homo- and heteronuclear experiments. By the combination of several 2D spectra it was possible to assign four groups of spin systems, corresponding to four different carbohydrate units (Fig. 2). One type of spin system (F) showed anomeric signals in the range of 5.043–5.086 p.p.m and a J H1,H2 of  4Hzin agreement with the a configuration. The methyl group for all signals of system F resonated at 1.16 p.p.m and 16.0 p.p.m. (H6/C6). The J H,H -values for H3–H4 and H4–H5 were indicative of a galacto configuration (3–4 Hz and less than 1 Hz, respectively). The carbon chemical shifts indicated no substitution except for C3 having a shift at 51.9 p.p.m. in agreement with the presence of an acetamido group, and the NAc could be assigned with a proton chemical shift at 2.05 p.p.m and a carbon chemical shift at 22.8 p.p.m. Thus, all of signals in the spin system F were identified as terminal a-Fuc3NAc in different chemical/ magnetic environments. Three other spin systems (B, A and A) were all recognized as Rha residues as they possessed a small J H2-H3 value ( 3 Hz) and H6/C6 chemical shifts (1.3/17.4 p.p.m) distinctive of a methyl group. One type of the Rha spin systems (B) was characterized by the anomeric signals at  4.8 p.p.m and  98 p.p.m. (Fig. 2). The anomeric 1 J C,H -value (161 Hz) indicated the b configuration. A further indication of the b configuration was the observation of a coupling constant H1 to H2, between 0.5 and 1 Hz, and the presence of the nuclear Fig. 1. The 1 H- and 13 C-NMR spectra of OPS. Spectra were recorded in D 2 O at 310K at 799.96 MHz for proton and 201.12 MHz for car- bon, using methyl signal of acetone as reference for proton (2.225 p.p.m) and 1,4-dioxane for carbon (67.4 p.p.m). Ó FEBS 2002 O-chain structure from Xanthomonas campestris pv. vitians (Eur. J. Biochem. 269) 4187 Overhauser effect among H1, H3 and H5. The carbon chemical shifts were indicative of a substitution at C3, as found by a downfield shift of this carbon signal ( 81 p.p.m). All data identified spin system B as 3-substi- tuted-b-Rha. Thelasttwotypesofspinsystems( AandA)were both endorsed as a-Rha residues ( 1 J C,H ¼ 174 Hz and 3 J H1-H2 ¼ 1 Hz). The anomeric protons were present at  5.3 p.p.m and 5.1 p.p.m and correlated to two different carbon signals in the HSQC spectrum at  101 p.p.m and 103 p.p.m., respectively (Fig. 3). The a-Rha unit with anomeric proton occurring around 5.1 p.p.m. (A) was assigned as 3-substituted-a-Rha owing to C3 low field chemical shift ( 78 p.p.m). The a-Rha residue with an anomeric resonance occurring around 5.3 p.p.m. ( A) was identified as 2,3-di-substituted-a-Rha because of the downfield chemical shifts of C2 and C3 carbons at  76 or 79 p.p.m., respectively. Hence, this last residue was identi- fied as the nodal unit. The a-Fuc3NAc residue was linked to the 2 position of the nodal a-Rha. This was deduced by an interresidual nOe between the anomeric proton of the a-Fuc3NAc and the H1 and H2 of the a-Rha and by the cross peak in the gHSQC- NOESY of the same anomeric proton to C2 of the a-Rha (Fig. 3). Similarly, by NOESY and gHSQC-NOESY (Fig. 3) the b-Rha residue could be assigned to be substituted by an a-Rha unit at the 3 position and linked (1fi3) to a-Rha. From the NOESY spectrum (Fig. 4), it was possible to establish that the a-Rha residue could be both substituted at the 3 position by either a-Rha or b-Rha residues and linked (1fi3) to either a-Rha or b-Rha residues. Table 1. Chemical shift for the assigned residues in identified fragments in the LPS. Fragment a Residue Type 1 2 3456NAcNOE b A–B–A–A/B b-Rha 1 H 4.814 4.128 3.65 3.52 3.451 1.318 13 C 97.8 71.4 81.1 72.3 72.8 17.4 A–B–A–A/B b-Rha 1 H 4.818 4.083 3.66 3.517 3.447 1.318 4.18 13 C 97.8 71.4 80.4 72.3 72.8 17.4 A–B–A– A b-Rha 1 H 4.764 4.113 3.66 3.50 3.403 1.307 13 C 97.3 71.4 81.3 72.3 72.8 17.4 A–B–A–A b-Rha 1 H 4.770 4.060 3.66 3.50 3.390 1.307 4.07 13 C 97.3 71.4 80.5 72.3 72.8 17.4 B–A–B–A a-Rha 1 H 5.068 4.239 4.07 3.57 3.88 1.28 3.65 13 C 102.8 68.2 78.3 72.4 69.7 17.4 B–A–B–A a-Rha 1 H 5.072 4.232 4.07 3.57 3.88 1.28 3.65 13 C 102.8 68.2 78.3 72.4 69.7 17.4 B–A–A a-Rha 1 H 5.084 4.239 4.07 3.57 3.88 1.28 3.92 13 C 102.8 68.2 78.3 72.4 69.7 17.4 B–A–A– A a-Rha 1 H 5.119 4.213 3.950 3.57 3.88 1.28 3.95 13 C 102.5 68.2 77.6 72.4 69.7 17.4 B–A– A–A a-Rha 1 H 5.124 4.230 3.962 3.57 3.88 1.28 4.05 13 C 102.5 68.2 77.6 72.4 69.7 17.4 A–A–B a-Rha 1 H 5.016 4.140 3.918 3.57 3.89 1.30 3.65 13 C 103.0 70.7 79.1 72.4 69.9 17.4 A–A–B a-Rha 1 H 5.004 4.136 3.927 3.57 3.89 1.30 3.91 13 C 103.0 70.7 78.7 72.4 69.9 17.4 B–A–A–A a-Rha 1 H 5.034 4.142 3.918 3.57 3.89 1.30 3.91 13 C 103.0 70.7 78.7 72.4 69.9 17.4 B–A–B a-Rha 1 H 5.261 4.250 4.18 3.74 3.88 1.33 3.66 13 C 101.5 76.2 77.8 71.2 69.9 17.4 Fuc3NAc 1 H 5.060 3.78 4.18 3.74 4.11 1.16 2.05 4.25 13 C 100.9 67.2 51.9 71.0 67.9 16.0 22.8 B– A–A a-Rha 1 H 5.273 4.250 4.180 3.74 3.88 1.31 3.92 13 C 101.5 76.2 77.8 71.2 69.9 17.4 Fuc3NAc 1 H 5.086 3.80 4.18 3.746 4.11 1.16 2.05 4.25 13 C 100.9 67.2 51.9 71.0 67.9 16.0 22.8 B– A–A a-Rha 1 H 5.327 4.231 4.07 3.75 3.88 1.35 4.06 13 C 101.4 75.8 77.4 71.2 70.0 17.4 Fuc3NAc 1 H 5.043 3.78 4.18 3.74 4.11 1.16 2.05 4.23 13 C 100.9 67.2 51.9 71 67.9 16.0 22.8 A–A–B a-Rha 1 H 5.25 4.09 4.08 3.75 3.89 1.32 3.66 13 C 101.5 79.8 77.4 72.9 69.9 17.4 Fuc3NAc 1 H 5.066 3.78 4.18 3.74 4.11 1.16 2.05 4.09 13 C 101.7 67.2 51.9 71 67.9 16.0 22.8 a The following abbreviations are used: B; b-Rha, A; a-Rha, A; a-Fuc3NAc (1–2) a-Rha). b In this column the inter residue NOEs from the anomeric 1 H used to assign the linkages are given. 4188 A. Molinaro et al.(Eur. J. Biochem. 269) Ó FEBS 2002 Therefore, the above spectroscopic assumptions were in full agreement with a rhamnan backbone in which one b-Rha is alternating between one, two or up to three a-Rha2: fi3)bRha(1fi3)aRha(1fi fi3)bRha(1fi3)aRha(1fi3)aRha(1fi fi3)bRha(1fi3)aRha(1fi3)aRha(1fi3)aRha(1fi The degree of fucosylation (i.e. a-Fuc3NAc(1fi2) a-Rha: a- Rha) could be estimated to be in the order of 2 : 3, by measurement of anomeric protons area. This led to 14 possible combinations: (B ¼ b-Rha, A ¼ a-Rha, A ¼ a-Fuc3NAc (1fi2) a-Rha) B–A, B– A B–A–A, B–A– A, B–A–A, B–A–A B–A–A–A, B–A–A– A, B–A–A–A, B–A–A–A, B–A–A–A, B– A–A–A, B–A–A–A, B–A–A–A Several of these combinations were found by means of an in-depth NMR analysis. The a-Fuc3NAc influenced the chemical shifts of the b-Rha residue when this latter was directly linked at nodal unit, mainly on its H2 (4.128 p.p.m. without Fuc3NAc and 4.083 p.p.m. with Fuc3NAc). In this way, it was possible to discriminate between b-Rha unit linked to a nodal or to an unsubstituted a-Rha residue. On the other hand it was also possible to assign b-Rha when it was substituted by a nodal Rha or 3-a-Rha. This was visible on the C3 chemical shift of the b-Rha unit (80.4 p.p.m. with nodal and 81.2 p.p.m. without). The a-Fuc3NAc also influenced the b-Rha chemical shifts when this was attached to Rha two units away from the nodal residue, but on different atoms. In partic- ular, changes were clearly visible on H1 and C1 and furthermore on H4, H5 and H6 (for all see Table 1). It was not possible to see any effect on the b-Rha whether it was substituted at C3 by a-Rha or by the a-Fuc3NAc(1fi2) a-Rha disaccharide. The conclusion is that b-Rha could be assigned in the following combinations (the assigned b-Rha are in bold): A-B–A–A/B, A-B–A–A/B, A-B–A–AandA–B–A–A. (Table 1). The assignment of the a-Rha has been more difficult due to many possible combinations. It was possible to identify the C3 substituent, that is to say b-Rha, a-Rha or nodal Rha, by a nOe from H3 of a-Rha to the anomeric proton of the substituting residue. The residue to which the a-Rha was linked was recog- nized examining the inter residue nOe of its anomeric proton to H3 of the substituted residue, again b-Rha, a-Rha or nodal Rha. Further information was given by the Fig. 4. The NOESY spectrum showing the significant NOEs for the interresidual assignment, e.g. A-B for the connectivity a-Rha(1–3)b-Rha. Fig. 2. Section of the 13 C- 1 H HSQC spectrum in which resonances of the four different spin systems (A, A, B, F) are shown. Fig. 3. Sections of the HSQC and HSQC/NOESY spectra in which the inter residue connectivities are shown with arrows, e.g. C1 (A) to C3 (B) for the linkage a-Rha (1fi3)b-Rha. Ó FEBS 2002 O-chain structure from Xanthomonas campestris pv. vitians (Eur. J. Biochem. 269) 4189 observation that the a-Fuc3NAc also influenced the chem- ical shift of a-Rha if linked one or two units away towards the terminal end. The following fragments were assignable (the assigned a-Rha being bold): B–A–B–A, B–A–B– A, B–A–A–A, B–A–A–A, B–A–A–A/ B, A–A–A, A–A–B, A–A–B, B–A–B, B–A–A, B–A–A, A– A–B (Table 1). In summary, eight of the 14 possible combinations could be assigned by NMR, as illustrated below. The remaining possible combinations might be present in low amounts or the chemical shifts are simply too close to the fragments identified. B–A, B– A B–A–A, B– A–A, B–A–A B–A–A–A, B– A–A–A, B–A–A–A Molecular modelling In order to explain some unusual variations in chemical shifts of the assigned fragments, a series of molecular dynamics (MD) simulations in water have been performed. The observed differences in chemical shift can largely be explained by direct glycosylation shifts and by evaluation of the executed MD simulations. The chemical shift of C2 in the a-Rha without Fuc3NAc is dependent on the substituting residue at C3, i.e. if it is substituted by a-Rha the C2 chemical shift is  70 p.p.m and if substituted by b-Rha it is  68 p.p.m. The difference is as expected from previous studies of glycosylation effect [12]. For the C-2 of a-Rha in the fragment A- A-B, a rather downfield chemical shift is observed, not only explicable by the normal glycosylation shifts. This can be partially explained by a fairly restricted conformation of the Fuc3NAc linkage in this fragment. The change in the conformational preference is also reflected in the chemical shifts of C1 of Fuc3NAc, which is downfield in comparison to the other fragments. A change in the / and w angles has been shown to give rise to a difference in the chemical shift of the carbons at the glycosidic linkage [13]. Likewise, C3 of b-Rha resonates at 81.2 p.p.m. if it is substituted by an a-Rha and at 80.4 p.p.m. if it is substituted by a nodal a-Rha, in fragments as in A–B–A–A/B and A–B–A–A. This small consistent difference is most likely correlated with a more restricted conformation of the a-linkage when a-Rha is substituted by Fuc3NAc, as, e.g. demonstrated by a change in the average //w angles from  27/)15 to 50/20. This should also affect the chemical shift of C1 of the a-Rha, but because this carbon also experiences a direct glycosy- lation effect it cannot be directly observed. The chemical shift of H5 in the b-Rha certainly depends on the nature of the two Rha residues towards the reducing end. If one of these two is a Rha bearing a Fuc3NAc unit, the H5 chemical shift is upfield shifted by 0.05 p.p.m., which is in agreement with a short distance in the MD simulations between the H5 of b-Rha and the methyl group of the Fuc3NAc that is folding towards the terminal nonreducing end of the polymer (Fig. 5). This short distance apparently does not give rise to any detectable nOe. The chemical shifts of the Rha residues are hardly affected by the Fuc3NAc substitution. This is in accordance with the observation that the Rha one residue towards the nonreducing end has no close contacts to the Fuc3NAc in the MD simulations (Fig. 5). MALDI-TOF-MS analysis In order to support the high structural heterogeneity of the OPS, a detailed MALDI-TOF analysis was performed on the oligosaccharide fractions (1 and 2) obtained by gel permeation chromatography after partial acid hydrolysis of OPS. The mass spectrum from MALDI-TOF analysis of fraction 1 is shown in Fig. 6A. The sample was mainly constituted of tetra- to heptasaccharides composed of Rha residues. Each oligosaccharide was carrying 0–2 Fuc3NAc substituents in accordance with the compositional and methylation analyses proving that the polysaccharide is composed of a linear rhamnan backbone substituted with Fuc3NAc residues. The mass spectrum in Fig. 6A provides some information about the distribution of the Fuc3NAc residues on the Rha backbone. The cluster of ions originating from pentasaccharides contains sodium adduct ions corresponding to oligosaccharides with the composi- tions Rha 5 ,Rha 4 Fuc3NAc and Rha 3 (Fuc3NAc) 2 .The cluster of ions from tetrasaccharides has sodium adduct ions corresponding to oligosaccharides with the compositions Rha 4 ,Rha 3 Fuc3NAc and Rha 2 (Fuc3NAc) 2 , but the latter ion is of very low intensity indicating that Fuc3NAc substituents are rarely positioned on neighbouring Rha units. Fraction 2 was found to contain larger oligosaccha- rides than fraction 1 (Fig. 6B). The Fuc3NAc distribution on the Rha backbone of these oligosaccharides was similar as in fraction 1, the cluster of ions originating from octasaccharides contained ions from sodium adducts of Rha 8 ,Rha 7 Fuc3NAc, Rha 6 (Fuc3NAc) 2 and Rha 5 (Fuc3- NAc) 3 , whereas the sodium adduct ions of Rha 10 Fuc3NAc, Rha 9 (Fuc3NAc) 2 ,Rha 8 (Fuc3NAc) 3 and Rha 7 (Fuc3NAc) 4 were prominent among ions formed from undecasaccha- rides. This indicates statistically that few neighbouring Rha units are substituted with Fuc3NAc, just as was found for fraction 1. Fig. 5. Stick and ball representation of a minimum energy conformation of A-A-B-A- A-B-A-A. The Fuc3NAc residue points back towards the Rha two residues towards the nonreducing end. (B; b-Rha, A; a-Rha, A; nodal a-Rha) 4190 A. Molinaro et al.(Eur. J. Biochem. 269) Ó FEBS 2002 In order to gain more informations on the Fuc3NAc substitution the fraction 1 was further fractionated on HPAEC. The chromatogram and a MALDI-TOF mass spectrum of the isolated fraction 3 are displayed in Fig. 7A and B. Then the sample was submitted to MALDI-PSD TOF MS analysis and the results from analysis of the ions of m/z 853.4 and 999.4 are shown in Fig. 8A and B). Figure 8A shows the PSD spectrum from analysis of the ion of m/z 853.4. The m/z of the precursor ion corresponds to the sodium adduct ion of Rha 3 (Fuc3NAc) 2 . The spectrum is dominated by B- and Y-series types of ions [14] and provides some information concerning the Fuc3NAc-substitution pattern of the precursor ion. No fragment ions corresponding to the loss of a reducing end Rha (B-type, m/z 689.4) or a nonreducing end Rha (Y-type, m/z 707.4) are visible. This suggests that Fuc3NAc is linked to both the reducing end as well as to the nonreducing end of the Rha-backbone of the oligosaccharide, i.e. the structure is Fuc3NAc–Rha–Rha–(Fuc3NAc)–RhaOH. Figure 8B shows the PSD spectrum from analysis of the ion of m/z 999.4 (Rha 4 (Fuc3NAc) 2 , sodium adduct). This spectrum also lacks prominent fragment ions correspond- ing to loss of a reducing end Rha (B-type, m/z 835.4) or a nonreducing end Rha (Y-type, m/z 853.4), indicating that both the nonreducing end and the reducing end of the Rha-backbone of the oligosaccharide are substituted with Fuc3NAc. Thus, the structure is Fuc3NAc–Rha– Rha–Rha–(Fuc3NAc)–RhaOH. Sodium adduct ions corresponding to Rha 3 Fuc3NAc and Rha 4 Fuc3NAc were also analyzed with MALDI-PSD-TOF MS (data not shown). As these ions only contained one Fuc3NAc residue each, these experiments could not provide any information concerning the Fuc3NAc distri- bution on the Rha backbone. These experiments showed, however, that when a Rha, not substituted with a Fuc3NAc, was present at the reducing end and/or the nonreducing end, the loss of Rha always resulted in ions Fig. 7. The HPAEC chromatogram (A) of the fraction 1 and the MALDI-TOF MS spectrum (B) of the isolated fraction 3 are displayed. Fig. 6. MALDI-TOF MS spectra of fractions 1 (A) and 2 (B) obtained by gel permeation chromatography after a partial acid hydrolysis of OPS. The main ions present in the spectra are explicated. Ó FEBS 2002 O-chain structure from Xanthomonas campestris pv. vitians (Eur. J. Biochem. 269) 4191 of considerable intensity. Thus, the absence of fragment ions corresponding to losses of reducing end or nonre- ducing end Rha in the PSD spectra shown in Fig. 8A and B, probably reflects structural features of the studied ions and not inherent low abundance of such fragment ions. It should be noted, however, that the samples analyzed by MALDI-TOF MS were produced by partial hydrolysis of the polysaccharide thus the distribution of the proposed oligosaccharide fragments in the OPS is random. On the other hand the distribution of Fuc3NAc residues should reflect the distribution in the native polysaccharide, actually no free reducing Fuc3NAc was found in the analysis of the hydrolysis products. DISCUSSION In conclusion, all the data suggest that the structure of this polymer has neither a real repeating unit nor a masked one. As it is shown below, Fuc3NAc (in italic) is a nonstoichio- metric substituents, and when present has no fixed a-Rhap to substitute. The rhamnan backbone is more frequently a trisaccharide, n is more frequently equal to two but it can also assume values equal to one and to three. The assembly of O-polysaccharide structures has been extensively studied in animal associated bacteria, in which they are for the most of cases rigorously repetitive. A structural heterogeneity has been found in few OPS of Xanthomonas campestris pvs. campestris [15], begoniae [16], vignicola [17], Xanthomonas fragrariae [18] and Pseudo- monas fluorescens [19]. In all of these polymers the lack of exacting regularity is due to the presence of a monosac- charide in nonstoichiometric amounts on the side chain. Nevertheless, despite these examples of not strictly repetitive polymers, the OPS from X. campestris seems to be the first true case of an unsystematic biosynthetic polymerization. The nonstoichiometric side branch glycosylation by Fuc3- NAc can be considered as a post polymerization decoration, this is well demonstrated by the finding of the homologous polymer without any side branch in the aqueous phase. What remains obscure still, is the assembly of the polymer backbone, which is definitely irregular in respect to the other examples already described in the literature [1]. In animal pathogen bacteria, the polymerization reaction can be conducted according to the ABC transporter pathway in which subsequent residues are added by the glycosyl transferases to the nonreducing end of the acceptor chain at the cytoplasmic face [20]. In some cases a single enzyme catalyses the formation of more linkages and this, of course, poses difficulties for the maintenance of the fidelity of the repetitive structure. Therefore biosynthetic studies towards the comprehension of this irregular biosynthesis are needed. ACKNOWLEDGEMENTS This paper is dedicated to Prof Lorenzo Mangoni on the occasion of his 70th birthday. A. M. is grateful to Prof M. Adinolfi for the kind gift of D-Fuc3NAc. The 800 MHz spectra were obtained using the Varian Unity Inova spectrometer of the Danish Instrument Center for NMR Spectroscopy of Biological Macromolecules. The authors wish to thank Dr Zoina for supplying cells of X. campestris. REFERENCES 1. Corsaro, M. M., De Castro, C., Molinaro, A. & Parrilli M. (2001) Structure of lipopolysaccharides from phytopathogenic bacteria. In Recent Research Developments in Phytochemistry (Pandalai, G., ed.). Research Signpost, 5, 119–138. Fig. 8. MALDI-PSD TOF MS spectra and fragment elucidations of the oligosaccharide fraction 3 are shown. The fragment ions are ori- ginated from the ions at m/z 853.4 (A) and 999.4 (B). The dotted lines are showing the lacking of fragment ions corresponding to the presence of a reducing end Rha or a nonreducing end Rha. 4192 A. Molinaro et al.(Eur. J. Biochem. 269) Ó FEBS 2002 2. Molinaro,A.,Evidente,A.,Lanzetta,R.,Parrilli,M.&Zoina,A. (2000) O-specific polysaccharide structure of the aqueous lipopo- lysaccharide fraction from Xanthomonas campestris pv. vitians strain 1839. Carbohydr. Res. 328, 435–439. 3. Molinaro, A., Lanzetta, R., Evidente, A., Parrilli, M. & Holst, O. (1999) Isolation and characterization of the lipopolysaccharide from Xanthomonas hortorum pv. vitians. FEMS Microbiol. Lett. 181, 49–53. 4. Westphal, O. & Jann, K. (1965) Bacterial lipopolysaccharides: Extraction with phenol-water and further applications of the procedure. Methods. Carbohydr. Chem. 5, 83–91. 5. Leontein,K.&Lo ¨ nngren, J. (1978) Determination of the absolute configuration of sugars by gas-liquid chromatography of their acetylated 2-octyl glycosides. Methods Carbohydr. Chem. 62, 359– 362. 6. Adinolfi, M., Barone, G., Corsaro, M., Lanzetta, R., Mangoni, L. & Monaco, P. (1995) Synthesis of methyl 3-acetamido-3, 6-dideoxy- L -galactopyranosides and of methyl 3-acetamido- 3,6-dideoxy- L -gulopyranosides by reduction of 3-ulose O-methyl- oximes. J. Carbohydr. Chem. 14, 913–928. 7. Ciucanu, I. & Kerek, F. (1984) A simple method for the per- methylation of carbohydrates. Carbohydr. Res. 131, 209–217. 8. Kjaer, M., Andersen, K.V. & Poulsen, F.M. (1994) Automated and semiautomated analysis of homo- and heteronuclear multi- dimensional nuclear magnetic resonance spectra of proteins: the program PRONTO. Methods Enzymol. 239, 288–308. 9. Hagler, A.T., Dauber, P. & Lifson, S. (1979) Consistent force field studies of intermolecular forces in hydrogen-bonded crystals. 3. The C: OÆÆÆH–O hydrogen bond and the analysis of the energetics and packing of carboxylic acids. J. Am. Chem. Soc. 101, 5131– 5141. 10. Bock, K. & Pedersen, C. (1974) Carbon-13-hydrogen coupling constants in hexopyranoses. J. Chem. Soc., Perkin Trans. 2, 293– 297. 11. Lipkind, G.M., Shashkov, A.S., Knirel, Y.A., Vinogradov, E.V. & Kochetkov, N.K. (1988) A computer-assisted structural analysis of regular polysaccharides on the basis of 13 C-NMR data. Car- bohydr. Res. 175, 59–75. 12. Shashkov, A.S., Lipkind, G.M., Knirel, Y.A. & Kochetkov, N.K. (1988) Stereochemical factors determining the effects of glycosy- lation on the 13 C chemical shifts in carbohydrates. Magn. Reson. Chem. 26, 735–747. 13. Bock, K., Brignole, A. & Sigurskjold, B.W. (1986) Conforma- tional dependence of 13 C nuclear magnetic resonance chemical shifts in oligosaccharides. J. Chem. Soc., Perkin Trans. 2, 1711– 1713. 14. Domon, B. & Costello, C.E. (1988) A systematic nomenclature for carbohydrate fragmentations in FAB-MS/MS spectra of glyco- conjugates. Glycoconjugate J. 5, 397–409. 15. Molinaro, A., Evidente, A., Fiore, S., Iacobellis, N.S., Lanzetta, R. & Parrilli, M. (2000) Structure elucidation of the O-chain from the major lipopolysaccharide of the Xanthomonas campestris strain 642. Carbohydr. Res. 325, 222–229. 16. Senchenkova, S.N., Shashkov, A.S., Laux, P., Knirel, Y.A. & Rudolph, K. (1999) The O-chain polysaccharide of the lipopoly- saccharide of Xanthomonas campestris pv. begoniae GSPB 525 is a partially 1-xylosylated 1-rhamnan. Carbohydr. Res. 319, 148–153. 17. Senchenkova, S.N., Shashkov, A.S., Kecskes, M.L., Ahohuendo, B.C., Knirel, Y.A. & Rudolph, K. (2000) Structure of the O-specific polysaccharide of the lipopolysaccharides of Xantho- monas campestris pv vignicola GSPB 2795 and GSPB. Carbohydr. Res. 329, 831–838. 18. Molinaro, A., Evidente, A., Iacobellis, N., Lanzetta, R., Manzo, E. & Parrilli, M. (2001) Structural determination of the O-specific polysaccharide from the Xanthomonas fragariae lipopolysac- charide fraction. Eur. J. Org. Chem. 5, 927–931. 19. Knirel, Y.A., Zdorovenko, G.M., Paramonov, N.A., Vere- meychenko, S.P., Toukach, F.V. & Shashkov, A.S. (1996) Somatic antigens of pseudomonads: structure of the O-specific poly- saccharide of the reference strain for Pseudomonas fluorescens (IMV 4125, ATCC 13525, biovar A). Carbohydr. Res. 291, 217– 224. 20. Keenleyside, W.J. & Whitfield, C. (1999) Genetics and biosynth- esis of lipopolysaccharide O-antigens. In EndotoxininHealthand Disease (Morrison, D.C., Brade, H., Opal, S. & Vogel, S., eds), pp. 331–358. M. Dekker, Inc., New York. Ó FEBS 2002 O-chain structure from Xanthomonas campestris pv. vitians (Eur. J. Biochem. 269) 4193 . (B ¼ b-Rha, A ¼ a- Rha, A ¼ a- Fuc3NAc (1fi2) a- Rha) B A, B– A B A A, B A A, B A A, B A A B A A A, B A A A, B A A A, B A A A, B A A A, B– A A A, B A A A, B A A A Several of these combinations were. NMR and MS evidences for a random assembled O-specific chain structure in the LPS of the bacterium Xanthomonas campestris pv. Vitians A case of unsystematic biosynthetic polymerization Antonio. 50 A ˚ for 500 ps and with a step length of 1 fs. Full coordinates were saved every 2.5 ps. Among the structures modelled were A A B A A B A A, A A B A A B A A, A A B A AvB A A and A A B– A A B A A