Structural Characterization of Core Region in Erwinia amylovora Lipopolysaccharide

Erwinia amylovora (E. amylovora) is the first bacterial plant pathogen described and demonstrated to cause fire blight, a devastating plant disease affecting a wide range of species including a wide variety of Rosaceae. In this study, we reported the lipopolysaccharide (LPS) core structure from E. amylovora strain CFBP1430, the first one for an E. amylovora highly pathogenic strain. The chemical characterization was performed on the mutants waaL (lacking only the O-antigen LPS with a complete LPS-core), wabH and wabG (outer-LPS core mutants). The LPSs were isolated from dry cells and analyzed by means of chemical and spectroscopic methods. In particular, they were subjected to a mild acid hydrolysis and/or a hydrazinolysis and investigated in detail by one and two dimensional Nuclear Magnetic Resonance (NMR) spectroscopy and ElectroSpray Ionization Fourier Transform-Ion Cyclotron Resonance (ESI FT-ICR) mass spectrometry.


Introduction
Erwinia amylovora is the causal agent of fire blight, a disease of nutritionally important members of the family Rosaceae, such as apple and pear trees. The symptoms in apple plants are present on rootstocks, blossoms, shoots, and fruits [1]. On fruits, the disease provokes the developent of ooze, which is composed of bacteria, polysaccharides, and plant sap [1]. Bacteria are transported by insects, rain, birds, wind-wipping, and hail from the ooze to flowers. From flower infection, it can also be transferred to lateral parts of the plant through an endophytic mechanism [2].
Two major virulence determinants are known for the pathogenesis of Erwinia: (i) one involves the hrp/dsp gene cluster, the role of which is to secrete and deliver proteins from bacteria to plant apoplasts and cytoplasm [1]; (ii) the second is the production of two types of exopolysaccharides (EPS), amylovoran and levan [3,4].
The outer membrane (OM) of almost all Gram-negative bacteria and of some cyanobacteria [5][6][7][8] contains lipopolysaccharides (LPSs), where they constitute approximately 75% of the outer surface. They are amphiphilic endotoxic molecules necessary for the viability and survival of Gram-negative bacteria, as they seriously contribute to the structural integrity of the OM and to the protection of the bacterial cell envelope [9].

Preparation and Structural Characterization of Oligosaccharides from Ea∆waaL LPS
The LPSs from the waaL, wabH, and wabG mutants (Ea∆waaL, EawabH, and Ea∆wabG, respectively) of the Erwinia amylovora strain CFBP1430 were extracted by the phenol-chloroform-light petroleum (PCP) method. WabG is responsible for the transfer of D-GalA to the O-3 position of L,D-Hep II, and WabH transfers a D-GlcNAc residue from UDP-GlcNAc to the D-GalA [15]. The monosaccharides composition was performed as already reported [16]. In particular, the Gas Chromatography-Mass Spectrometry (GC-MS) analysis of methyl and octyl glycosides for Ea∆waaL and Ea∆wabH mutant LPSs revealed the following sugars; D-Glc, D-GalA, D-GlcN, L,D-Hep, D,D-Hep, and Kdo. When this analysis was performed on Ea∆wabG LPS, the lack of D-GalA and D,D-Hep was observed.
The removal of fatty acids from Ea∆waaL LPS was performed both by strong alkaline and acetic acid hydrolyses. The obtained products were analyzed by ESI FT-ICR mass spectrometry. In addition, mono-and two-dimensional NMR spectroscopy ( 1 H, 1 H DQF-COSY, 1 H, 1 H TOCSY, 1 H, 1 H ROESY, 1 H, 13 C HSQC-DEPT, 1 H, 13 C HSQC-TOCSY, and 1 H, 13 C HMBC) was allowed to assign all the proton and carbon chemical shifts of each mutant. The anomeric configurations were identified on the basis of both 13 C chemical shift values and 3 J H1,H2 coupling constants. The sequence of the residues in all the oligosaccharides was obtained both by long-range scalar couplings and by nuclear Overhauser enhancement (NOE) data.
Starting from the totally deacylated product (L-OS KOH , Scheme 1, structures 1a-c), the negative ion mode ESI FT-ICR mass spectrum of the sample identified eight main species (Figure 1 and Table 1). Species M1, occurring at 2220.595 Da (calculated molecular mass 2220.586 Da), represented the higher molecular mass oligosaccharide chain containing the phosphorylated lipid A backbone and the core structure.   Figure 1. Charge deconvoluted ElectroSpray Ionization Fourier Transform-Ion Cyclotron Resonance (ESI FT-ICR) mass spectrum of the totally deacylated lipopolysaccharide (LPS) fraction (L-OSKOH) isolated from the E. amylowora EaΔwaaL mutant. The spectrum was acquired in negative ion mode. The mass numbers given refer to the monoisotopic peak of the neutral molecular species.

R-α-D-ΔGalA-(1→3)-α-L,D-Hep-(1→3)-α-L,D-Hep-(1→5)-Kdo-(2→6)-β-D-GlcN4P-α-D-GlcN1P
In particular, the following composition was attributed to the M1 species: GlcN2P2Kdo2Hep5Glc1ΔGalA, where ΔGalA represents an unsaturated galacturonic acid. The presence of this acid unit in the M1 species clearly indicated that the strong alkaline conditions determined the lack of one or more sugars from GalA, due to a β-elimination reaction [17]. Species M3, M5, and M8 differed from M1 in their heptoses composition, displaying four units in M3 and three and two in M5 and M8, respectively. The remaining species, M2, M4, and M7, lacked a glucose unit respect to M1, M3, and M5, respectively, thus confirming the terminal position for this residue, while the M6 species lacked ΔGalA compared to M5. Further mass peaks originate from an additional phosphate group P (Δm = +79.966 u) and/or one linked C14:0(3OH) fatty acid (Δm = +226.193 u) due to incomplete deacylation. Finally, F1-F6 are fragments induced during the ESI process leading to the cleavage of the lipid A backbone GlcN2P2 (Δm = −500.081 u).   Figure 1. Charge deconvoluted ElectroSpray Ionization Fourier Transform-Ion Cyclotron Resonance (ESI FT-ICR) mass spectrum of the totally deacylated lipopolysaccharide (LPS) fraction (L-OSKOH) isolated from the E. amylowora EaΔwaaL mutant. The spectrum was acquired in negative ion mode. The mass numbers given refer to the monoisotopic peak of the neutral molecular species.
In particular, the following composition was attributed to the M1 species: GlcN2P2Kdo2Hep5Glc1ΔGalA, where ΔGalA represents an unsaturated galacturonic acid. The presence of this acid unit in the M1 species clearly indicated that the strong alkaline conditions determined the lack of one or more sugars from GalA, due to a β-elimination reaction [17]. Species M3, M5, and M8 differed from M1 in their heptoses composition, displaying four units in M3 and three and two in M5 and M8, respectively. The remaining species, M2, M4, and M7, lacked a glucose unit respect to M1, M3, and M5, respectively, thus confirming the terminal position for this residue, while the M6 species lacked ΔGalA compared to M5. Further mass peaks originate from an additional phosphate group P (Δm = +79.966 u) and/or one linked C14:0(3OH) fatty acid (Δm = +226.193 u) due to incomplete deacylation. Finally, F1-F6 are fragments induced during the ESI process leading to the cleavage of the lipid A backbone GlcN2P2 (Δm = −500.081 u).  In particular, the following composition was attributed to the M1 species: GlcN 2 P 2 Kdo 2 Hep 5 Glc 1 ∆GalA, where ∆GalA represents an unsaturated galacturonic acid. The presence of this acid unit in the M1 species clearly indicated that the strong alkaline conditions determined the lack of one or more sugars from GalA, due to a β-elimination reaction [17]. Species M3, M5, and M8 differed from M1 in their heptoses composition, displaying four units in M3 and three and two in M5 and M8, respectively. The remaining species, M2, M4, and M7, lacked a glucose unit respect to M1, M3, and M5, respectively, thus confirming the terminal position for this residue, while the M6 species lacked ∆GalA compared to M5. Further mass peaks originate from an additional phosphate group P (∆m = +79.966 u) and/or one linked C14:0(3OH) fatty acid (∆m = +226.193 u) due to incomplete deacylation. Finally, F1-F6 are fragments induced during the ESI process leading to the cleavage of the lipid A backbone GlcN 2 P 2 (∆m = −500.081 u).
Despite the complexity of the 1 H-13 C Heteronuclear Single Quantum Coherence-Distortionless Enhancement by Polarization Transfer (HSQC-DEPT) experiment (Figure 2), it was possible to assign the signals of the main species, M1, M3, and M5. All the NMR data ( Table 2, structures 1a-c) strongly suggested that the fractions L-OS KOH and the structure of the oligosaccharide OS1 obtained from the previously published K. pneumoniae 52145 waaL mutant [17] possessed a common structural element constituted by the residues A, B, C, D, E, F, G, H, and I (monosaccharide units are as shown in Table 2), in agreement with the strain genomics. Despite the complexity of the 1 H-13 C Heteronuclear Single Quantum Coherence-Distortionless Enhancement by Polarization Transfer (HSQC-DEPT) experiment ( Figure 2), it was possible to assign the signals of the main species, M1, M3, and M5. All the NMR data ( Table 2, structures 1a-c) strongly suggested that the fractions L-OSKOH and the structure of the oligosaccharide OS1 obtained from the previously published K. pneumoniae 52145 waaL mutant [17] possessed a common structural element constituted by the residues A, B, C, D, E, F, G, H, and I (monosaccharide units are as shown in Table  2), in agreement with the strain genomics.    All the species possessed α-GlcN1P residue at the reducing end (A, H1, δ 5.65 ppm; C1, δ 92.0 ppm), originating from the lipid A backbone. The second residue of glucosamine of the lipid A (B, H1, δ 4.99 ppm, C1, δ 100.9 ppm) was O-4 phosphorylated, as suggested by the downfield shift of both H4 and C4 chemical shifts [18]. In all the oligosaccharides corresponding to the M1-M7 species, the core region was linked to the lipid A backbone through an α-Kdo residue C, which was substituted at O-4 position by a second residue of Kdo (D).
Residues E, G, H, L, and M were identified as five manno-configured α-heptopyranoses due to their small coupling constants 3 J H1,H2 and 3 J H2,H3 values. Downfield shifted carbon signals with respect to reference values [19] indicated substitutions at O-3 and O-4 of residue E (C3 at 77.1 and C4 at 75.1 ppm, respectively) and at O-3 and O-7 of residue G (C3 at 81.1 ppm and C7 at 71.2 ppm). Both residues H and M were identified as terminal units as none of their carbon signals were downfield shifted. All the heptose residues were found to be L,D-configured except for L; in fact, for this residue, the chemical shift of its C6 at 72.1 ppm suggested a D,D-configuration [20]. In addition, the downfield shift of its C2 resonance at 80.0 ppm clearly indicated that it was substituted at this position. Residue I, with H1/C1 at 5.58/99.6 ppm and H4/C4 at 5.82/108.9 ppm, was identified as a α-threo-hex-4-enuronopyranosyl unit (∆GalA).
Finally, residue F was assigned to a β-glucose residue, on the basis of the proton multiplicities obtained by the DQF-COSY and TOCSY experiments. No downfield carbon chemical shifts were observed for this residue, thus indicating its terminal non reducing end position.
The sequence of the residues was deduced from the HMBC experiment, which showed correlations between C1 of L and H2 of I, H1 of I and C3 of G, H1 of G and C3 of E, C1 of H and H7 of G, and C1 of F and H4 of E.
In addition, residue M was found to be linked at the O-2 position of residue L, as shown by NOE contacts among H1 of M and both H1 and H2 of L, observed in the ROESY experiment. This last completed and confirmed the above sequence by inter-residual dipolar couplings. In the same spectrum, the intra-residue NOE contacts were in agreement with the assigned relative configuration of the monosaccharides.
All these data indicated for M1 species of L-OS KOH the oligosaccharide structure 1a. Structures 1b and 1c, corresponding to the species M3 and M5, were identified on the basis of their terminal residue D,D-heptose and ∆GalA, respectively.
The LPS of Ea∆waaL mutant was then treated with 5% acetic acid at 100 • C, and the supernatant was analyzed (L-OS AcOH , Scheme 2, structures 2a,b). Gas Chromatography-Mass Spectrometry (GC-MS) analysis of the partially methylated alditol acetates of this sample indicated the presence of terminal L,D-heptose, terminal glucose, terminal D,D-heptose, 2-substituted D,D-heptose, 2-substituted glucose, 6-substituted glucose, 4-substituted glucosamine 3,4-substituted L,D-heptose, and 3,7-substituted L,D-heptose. Each sugar was recognized from both its Electronic Ionization (EI) mass spectrum and the GC column retention time, by comparison with standards. The Kdo was not revealed in this analysis, due to the presence of its anhydro form.
The negative mode ESI FT-ICR mass spectrum of 2 ( Figure 3 and Table 3) and NMR data ( Table 2)         We compared our NMR data with the data already published for deacylated oligosaccharides from Serratia marcenscens [19], and we found very similar chemical shifts, even if we recognized in Erwinia an additional minor component (N1 species, Table 2 and Figure 3). In fact, the molecular mass of the N1 species indicated an additional hexose with respect to that of N2. This residue (Q, Table 2) was identified as gluco-configurated with an α anomeric configuration, as suggested by its 3 J H1,H2 anomeric coupling constant of 3.5 Hz. A long-range heteronuclear scalar coupling between H1 of Q and C2 of P indicated that in the N1 species the oligosaccharide substituting the O-4 of GalA had the structure: α-Glc-(1 → 2)-β-Glc-(1 → 6)-α-Glc-(1 → 4)-α-GlcN. All these data were in agreement with the methylation analysis, except for the 2,4-disubstituted galacturonic acid residue, which was absent in the GC-MS chromatogram. This fact could be due to the hindrance of both oligosaccharides substituting O-2 and O-4 positions of GalA, thus preventing its methylation.

Preparation and Structural Characterization of Oligosaccharides from Ea∆wabH and Ea∆wabG LPSs
The LPSs from Ea∆wabH and Ea∆wabG mutants were completely deacylated by hydrazine followed by the KOH reaction. After purification on a gel filtration Sephadex G10 column, the samples, named H-OS KOH and G-OS KOH respectively, were analysed by mono-and two-dimensional NMR experiments ( Table 2). The acetic acid hydrolysis was not performed on these two mutants as no β-degradation was observed for these samples, indicating that no information was lost.
The 1 H NMR spectrum of the totally deacylated Ea∆wabH LPS (H-OS KOH , Figure S1, Supporting Information) showed nine main anomeric proton signals, assigned to residues A-M, in the range 5.6-4.4 ppm. Signals in the range 2.2-1.5 ppm, attributable to the presence of the Kdo units, were also present.
The galacturonic acid residue (residue I, Scheme 3, structure 3) was recognized from the chemical shift of its C6 at δ 176.8 ppm (Table 2), which correlated with its H5 proton at δ 4.34 ppm in the Heteronuclear Multiple Bond Correlation (HMBC) experiment. We compared our NMR data with the data already published for deacylated oligosaccharides from Serratia marcenscens, [19], and we found very similar chemical shifts, even if we recognized in Erwinia an additional minor component (N1 species, Table 2 and Figure 3). In fact, the molecular mass of the N1 species indicated an additional hexose with respect to that of N2. This residue (Q, Table 2) was identified as gluco-configurated with an α anomeric configuration, as suggested by its 3 JH1,H2 anomeric coupling constant of 3.5 Hz. A long-range heteronuclear scalar coupling between H1 of Q and C2 of P indicated that in the N1 species the oligosaccharide substituting the O-4 of GalA had the structure: α-Glc-(1 → 2)-β-Glc-(1 → 6)-α-Glc-(1 → 4)-α-GlcN. All these data were in agreement with the methylation analysis, except for the 2,4-disubstituted galacturonic acid residue, which was absent in the GC-MS chromatogram. This fact could be due to the hindrance of both oligosaccharides substituting O-2 and O-4 positions of GalA, thus preventing its methylation.

Preparation and Structural Characterization of Oligosaccharides from EaΔwabH and EaΔwabG LPSs
The LPSs from EaΔwabH and EaΔwabG mutants were completely deacylated by hydrazine followed by the KOH reaction. After purification on a gel filtration Sephadex G10 column, the samples, named H-OSKOH and G-OSKOH respectively, were analysed by mono-and two-dimensional NMR experiments ( Table 2). The acetic acid hydrolysis was not performed on these two mutants as no β-degradation was observed for these samples, indicating that no information was lost.
The 1 H NMR spectrum of the totally deacylated Ea∆wabH LPS (H-OSKOH, Figure S1, Supporting Information) showed nine main anomeric proton signals, assigned to residues A-M, in the range 5.6-4.4 ppm. Signals in the range 2.2-1.5 ppm, attributable to the presence of the Kdo units, were also present.
The galacturonic acid residue (residue I, Scheme 3, structure 3) was recognized from the chemical shift of its C6 at δ 176.8 ppm (Table 2) In addition, the O-substitution at position 2 of I was inferred by the long-range between H1 of L and C2 of I. The value of the coupling constant 3 JH1,H2 of the anomeric proton was found to be 3.5 Hz, thus indicating for residue I an α configuration. The chemical shifts of heptoses E, G, H, L, and M were in agreement with that already found for structure 1, as well as that of terminal β-Glc F. Finally, both the HMBC ( Figure S2) and ROESY experiments indicated the sequence shown in structure 3, which is in agreement with strain genomics.
The 1 H-NMR spectrum of the totally deacylated Ea∆wabG LPS, (G-OSKOH, Figure S3, Supporting Information) showed only five anomeric proton signals. All the 1 H and 13 C chemical shifts of each In addition, the O-substitution at position 2 of I was inferred by the long-range between H1 of L and C2 of I. The value of the coupling constant 3 J H1,H2 of the anomeric proton was found to be 3.5 Hz, thus indicating for residue I an α configuration. The chemical shifts of heptoses E, G, H, L, and M were in agreement with that already found for structure 1, as well as that of terminal β-Glc F. Finally, both the HMBC ( Figure S2) and ROESY experiments indicated the sequence shown in structure 3, which is in agreement with strain genomics.
The 1 H-NMR spectrum of the totally deacylated Ea∆wabG LPS, (G-OS KOH , Figure S3, Supporting Information) showed only five anomeric proton signals. All the 1 H and 13 C chemical shifts of each residue were identified and assigned by two-dimensional NMR experiments (Table 2). NMR data, together with glycosyl analysis, indicated the lack of the trisaccharide α-Hep-(1 → 2)-α-Hep-(1 → 2)-α-GalA. This was in agreement with the hypothesis of assignment of the wabG gene to a galacturonic acid residue glycosyltransferase. In addition, the 1 H, 13 C HSQC-DEPT NMR experiment ( Figure S4) revealed the absence of the inner core heptose residue H, since only two L,D-heptose anomeric signals, E and G, were found (Scheme 4, structure 4).
This fact suggested that the lack of GalA residue may preclude the addition of heptose H to the position O-7 of heptose G.

Bacteria Growth and LPS Isolation
Dried bacteria cells from EaΔwaaL (3.3 g), EaΔwabH (2.63 g), and EaΔwabG (3.58 g) mutants of E. amylovora strain CFBP1430 were all extracted by the PCP method [21] to give 208 mg of LPS (LPSPCP yield 6.3% w/w of dried cells) for the waaL mutant, 124 mg of LPS for the wabH mutant (LPSPCP yield 4.2% w/w of dried cells), and 170 mg of LPS (LPSPCP yield 4.7% w/w of dried cells) for the wabG mutant, respectively.

Sugar Analysis
The sugar analysis was performed as reported [16]. The absolute configurations of the sugars were determined by gas chromatography of the acetylated (S)-2-octyl glycosides [22]. The derivatized samples were injected into an Agilent Technologies gas chromatograph 6850A equipped with a mass selective detector 5973N and a Zebron ZB-5 capillary column (Phenomenex, Bologna, Italy, 30 m × 0.25 mm i.d., flow rate 1 mL/min, He as carrier gas). The following temperature programs were used; 140 °C for 3 min, 140 °C → 240 °C at 3 °C/min (acetylated methyl glycosides), 150 °C for 5 min, 150 °C → 240 °C at 6 °C/min, and 240 °C for 5 min (acetylated octyl glycosides).

Mild Acid Hydrolysis
The LPS of the EaΔwaaL mutant (20 mg) was treated with 5% aqueous CH3COOH (2 mL, 100 °C for 4 h). After centrifugation (7500 rpm, 4 °C, 30 min), the pellet was washed twice with water. Then, the supernatant layers were pooled together and lyophilized. The sample was then fractionated on a Bio-Gel P-10 column (Biorad, 1.5 × 110 cm, flow rate 15 mL/h, fraction volume 2 mL) and eluted with water buffered with 0.05 M pyridine and 0.05 M AcOH. The fractions containing oligosaccharides were pooled and named L-OSAcOH (10.1 mg).

Deacylation of the LOSs
The LOS from each mutant (100 mg) was dried over phosphorus anhydride under a vacuum and treated with hydrazine (5 mL, at 37 °C for 2 h) [23]. Cold acetone was added, and the pellet was recovered after centrifugation at 4 °C and 7000 rpm for 30 min. After being washed three times with acetone, it was suspended in water and lyophilized, obtaining LLOS-OH (68.6 mg), HLOS-OH (72.4 mg), and GLOS-OH (65.9 mg).
The partially deacylated LPS (LOS-OH) from each mutant was dissolved in KOH 4 M and incubated at 120 °C for 16 h. The KOH was neutralized with HCl, and the mixture was extracted three times with CHCl3. The aqueous phases were recovered and desalted on a Sephadex G-10 column (Amersham Biosciences, Little Chalfont, UK, 2.5 × 43 cm, 31 mL·h −1 , fraction volume 2.5 mL, eluent NH4HCO3 10 mM). The eluted oligosaccharides mixture was then lyophilized (L-OSKOH 19 mg; H-OSKOH 13 mg; G-OSKOH 20 mg) Scheme 4. Structures of the totally deacylated LPS fraction (G-OS KOH ) isolated from the E. amylowora Ea∆wabG mutant.

Bacteria Growth and LPS Isolation
Dried bacteria cells from Ea∆waaL (3.3 g), Ea∆wabH (2.63 g), and Ea∆wabG (3.58 g) mutants of E. amylovora strain CFBP1430 were all extracted by the PCP method [21] to give 208 mg of LPS (LPS PCP yield 6.3% w/w of dried cells) for the waaL mutant, 124 mg of LPS for the wabH mutant (LPS PCP yield 4.2% w/w of dried cells), and 170 mg of LPS (LPS PCP yield 4.7% w/w of dried cells) for the wabG mutant, respectively.

Sugar Analysis
The sugar analysis was performed as reported [16]. The absolute configurations of the sugars were determined by gas chromatography of the acetylated (S)-2-octyl glycosides [22]. The derivatized samples were injected into an Agilent Technologies gas chromatograph 6850A equipped with a mass selective detector 5973N and a Zebron ZB-5 capillary column (Phenomenex, Bologna, Italy, 30 m × 0.25 mm i.d., flow rate 1 mL/min, He as carrier gas). The following temperature programs were used; 140 • C for 3 min, 140 • C → 240 • C at 3 • C/min (acetylated methyl glycosides), 150 • C for 5 min, 150 • C → 240 • C at 6 • C/min, and 240 • C for 5 min (acetylated octyl glycosides).

Mild Acid Hydrolysis
The LPS of the Ea∆waaL mutant (20 mg) was treated with 5% aqueous CH 3 COOH (2 mL, 100 • C for 4 h). After centrifugation (7500 rpm, 4 • C, 30 min), the pellet was washed twice with water. Then, the supernatant layers were pooled together and lyophilized. The sample was then fractionated on a Bio-Gel P-10 column (Biorad, 1.5 × 110 cm, flow rate 15 mL/h, fraction volume 2 mL) and eluted with water buffered with 0.05 M pyridine and 0.05 M AcOH. The fractions containing oligosaccharides were pooled and named L-OS AcOH (10.1 mg).

Deacylation of the LOSs
The LOS from each mutant (100 mg) was dried over phosphorus anhydride under a vacuum and treated with hydrazine (5 mL, at 37 • C for 2 h) [23]. Cold acetone was added, and the pellet was recovered after centrifugation at 4 • C and 7000 rpm for 30 min. After being washed three times with acetone, it was suspended in water and lyophilized, obtaining L LOS-OH (68.6 mg), H LOS-OH (72.4 mg), and G LOS-OH (65.9 mg).
The partially deacylated LPS (LOS-OH) from each mutant was dissolved in KOH 4 M and incubated at 120 • C for 16 h. The KOH was neutralized with HCl, and the mixture was extracted three times with CHCl 3 . The aqueous phases were recovered and desalted on a Sephadex G-10 column (Amersham Biosciences, Little Chalfont, UK, 2.5 × 43 cm, 31 mL·h −1 , fraction volume 2.5 mL, eluent NH 4 HCO 3 10 mM). The eluted oligosaccharides mixture was then lyophilized (L-OS KOH 19 mg; H-OS KOH 13 mg; G-OS KOH 20 mg).

Methylation Analysis
The linkage positions of the monosaccharides were determined by GC-MS analysis of the partially methylated alditol acetates (PMAAs). The acetic acid oligosaccharides fraction of Ea∆waaL mutant (1 mg) was first reduced with NaBD 4 and then methylated with CH 3 I (100 µL) and NaOH powder in dimethyl sulfoxide (DMSO) (300 µL) for 20 h [24].
The reduction of the carboxymethyl groups was obtained by treating the sample with sodium boro deuteride (NaBD 4 ). After the reaction working up, the sample was totally hydrolyzed with 2 M trifuoroacetic acid (TFA) at 120 • C for 2 h, reduced again with NaBD 4 , and acetylated with Ac 2 O and pyridine (50 µL each, 100 • C for 30 min) [25]. The PMAA mixture was analyzed by GC-MS with the following temperature program: 90 • C for 1 min, 90 → 140 • C at 25 • C/min, 140 • C → 200 • C at 5 • C/min, 200 → 280 • C at 10 • C/min, and 280 • C for 10 min.

Mass Spectrometry Analysis
Mass spectra were (ESI FT-ICR) were performed in negative ion mode using an APEX QE (Bruker Daltonics GmbH, Bremen, Germany) equipped with a 7 Tesla actively shielded magnet. The sample concentration was~10 ng/µL. The solutions were sprayed at a flow rate of 2 µL/min and analyzed. The mass spectra obtained were charge-deconvoluted, and the mass numbers reported refer to the monoisotopic masses of the neutral molecules.

NMR Spectroscopy
1 H and 13 C NMR spectra were performed using a Bruker Avance 600 MHz spectrometer (Milano, Italy) equipped with a cryoprobe. All 2D homo-and heteronuclear experiments (double quantum-filtered correlation spectroscopy, DQF-COSY; total correlation spectroscopy TOCSY; rotating-frame nuclear Overhauser enhancement spectroscopy, ROESY; nuclear Overhauser effect spectroscopy, 1 H, 13 C HSQC-DEPT; and heteronuclear multiple bond correlation, 1 H, 13 C HMBC) were obtained using the standard pulse sequences available in the Bruker software (TopSpin 3.1 version). The TOCSY and ROESY experiments were obtained with mixing times of 100 ms. Chemical shifts were measured at 298 K in D 2 O.

Conclusions
In this work, for the first time, we have characterised the complete structure of the core oligosaccharide from E. amylowora strain CFBP1430 LPS (Scheme 5, structure 5). To this aim, we prepared three mutants, i.e. waaL, wabH, and wabG, the purified lipopolysaccharides of which were characterised. The reported data showed that the core oligosaccharides here reported share structural fragments with those of Klebsiella pneumoniae and Serratia marcescens. In particular, by comparison with the core of S. marcescens the present structure lacks the non-stoichiometric Ko residue, a feature considered useful to distinguish between the genera Burkholderia and Pseudomonas.

Conclusions
In this work, for the first time, we have characterised the complete structure of the core oligosaccharide from E. amylowora strain CFBP1430 LPS (Scheme 5, structure 5). To this aim, we prepared three mutants, i.e. waaL, wabH, and wabG, the purified lipopolysaccharides of which were characterised. The reported data showed that the core oligosaccharides here reported share structural fragments with those of Klebsiella pneumoniae and Serratia marcescens. In particular, by comparison with the core of S. marcescens the present structure lacks the non-stoichiometric Ko residue, a feature considered useful to distinguish between the genera Burkholderia and Pseudomonas.