Glycobiology Advance Access originally published online on July 28, 2006
Glycobiology 2006 16(12):1272-1280; doi:10.1093/glycob/cwl033
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Core oligosaccharide structure from the highly phytopathogenic Agrobacterium tumefaciens TT111 and conformational analysis of the putative rhamnan epitope
Department of Organic Chemistry and Biochemistry, University of Naples, Complesso Universitario Monte Sant’ Angelo, Via Cintia 4, 80126 Napoli, Italy
1 To whom correspondence should be addressed; e-mail: decastro{at}unina.it
Received on June 15, 2006; revised on July 26, 2006; accepted on July 27, 2006
 |
Abstract |
|---|
 Top Abstract IntroductionResults and discussionConclusionsSupplementary dataAcknowledgmentsConflict of interest statementReferences |
|---|
The structure of the complex mixture of the core oligosaccharide components of the lipooligosaccharide (LOS) fraction of Agrobacterium tumefaciens strain TT111 was determined directly on the deacetylated products by means of spectroscopical methods. The rhamnan oligosaccharide elongating the inner Kdo residue shares structural features with other polysaccharides from well-known plant pathogenic bacteria. Its conformation was determined through extensive molecular dynamic (MD) analysis and presents an epitope similar to that recognized from the plant defense system.
Key words: conformation analysis / lipooligosaccharide / NMR spectroscopy / structure elucidation
|
Introduction |
|---|
|
TopAbstractIntroductionResults and discussionConclusionsSupplementary dataAcknowledgmentsConflict of interest statementReferences |
|---|
Agrobacterium tumefaciens is a gram-negative bacterium, belonging to the Rhizobiaceae family; it gains importance as it is widely used as a tool for plant genetic transformation. The bacterium itself incites the crown gall disease in a wide range of dicotyledonous plant species, the disease is characterized by neoplastic transformation at the site of infection, and it results from the transfer and expression of a portion of its Ti plasmid (Hooykaas et al., 1977
), named T-DNA in the susceptible plant cells. This DNA fragment codes mainly for the production of the plant growth hormones and for Agrobacterium nutritive compounds, the opines; its 3' and 5' terminals present special sequences that are the target of the bacterial enzyme machinery dedicated to its excision, transfer, and integration into the plant cell DNA. The genes constitutively present in the T-DNA are nowadays replaced by other genes of interest, and the mutated A. tumefaciens acquires its biotechnological properties.
The infection is a complex process and it is conditioned by the recognition and absorption of the bacterium on the host, an event modulated by the components, both proteins and lipopolysaccharides (LPSs), of the external membrane of the bacterium (Pueppke and Benny, 1984). In this regard, studies on surface polysaccharide from different bacterial sources have pointed out their role in the pathogenesis mechanism, and they are involved in the absorption of the bacterium on the plant cell wall, a process that is not silent for the host but that triggers several events, as demonstrated from in vivo assays (Bedini et al., 2005; Silipo et al., 2005). In an interesting experiment (Silipo et al., 2005), different parts of the lipooligosaccharide (LOS) molecules, the lipid A and the core region, induced the expression of both pathogen response PR-1 and PR-2 genes, as demonstrated in the model system Xanthomonas campestris pv. campestris and Arabidopsis thaliana; focusing on PR-1 gene, the core region was responsible for its early response, maximum after ~12h, whereas lipid A moiety provoked a more intense but delayed induction, after almost 24h, suggesting that the different parts of the same molecule induced the same response but with a different recognition mechanism. It is worthy of note that the PR-1 gene is induced from a large variety of saccharide molecules as demonstrated on the same plant model using synthetic oligorhamnans as inducers (Bedini et al., 2005). The oligosaccharides tested were oligomers of the same repeating unit H-[3)-
-L-Rha-(1
3)--L-Rha-(12)--L-Rha-(1]n-Bn (n=1, 2, or 3; Bn=Benzyl), and the motif 3)--L-Rha-(12)--L-Rha-(13)--L-Rha-(1 was proposed as epitope necessary for the recognition mechanism since increasing the number of repeating units, the corresponding oligosaccharide possessed more epitopes and was more performing both in the PR-1 gene expression and hypersensitive response (HR) assay.
In this article, the structure of the core oligosaccharides from A. tumefaciens DSM 30204 (here referred as TT111), the reference strain for the homonymous group (Lippincott and Lippincott, 1969) is determined; the outer core region is composed of a rhamnose oligosaccharide that shares sequence and conformational similarities with the bioactive rhamnans tested earlier.
|
Results and discussion |
|---|
|
TopAbstractIntroductionResults and discussionConclusionsSupplementary dataAcknowledgmentsConflict of interest statementReferences |
|---|
Bacteria and bacterial LPSs
Dry cells were sequentially extracted according to the petroleum ether–chloroform–phenol (PCP) and the hot phenol/water method. The two organic and the water phases were screened for the presence of LPS material by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE), which was found in both the PCP and the water extracts (3 and 7.5% gLPS/gcells, respectively) and showed the mobility typical of low molecular weight molecules, characteristic of LOS species. Comparing the monosaccharide composition of the two extracts suggested that the water phase was heavily contaminated from a glucan owing to the abundance of the glucose peak; further analysis was performed only on the LOS fraction from the PCP extract.
Combining the information from monosaccharide composition, methylation, and nuclear magnetic resonance (NMR) (residue label in parentheses) analysis, we found that the LOS fraction contained terminal (F and H), 2- (D and E), and 3- (G and I) substituted L-rhamnose, 3-substituted D-mannose (B and C), terminal (O and P), and 2-linked D-galactose (N), and terminal (L) and 6-linked D-glucosamine (A and M), whereas the acidic residues were not detected from the methylation protocol used. The absolute configuration was assigned to the mannose, glucosamine, galactose, and rhamnose residues by analyzing the 2-(+)-octyl derivatives and 2-(+)-butyl derivatives, whereas for the Kdo it was assumed as D.
Fatty acids analysis showed the presence of C14:0 (3-OH), C16:0 (3-OH), C18:1 (3-OH), and C28:0 (27-OH): the lipids distinctive of this bacterial family (Silipo et al., 2004).
Mild alkaline degradation of the LOS mixture (20 mg) led to the de-O-acylated products (11 mg, 55 % yield) that were further subjected to strong alkaline treatment in order to remove the amide-linked acyl residues. The usual work-up led to a mixture of phosphate oligosaccharides (2mg, 10% of the LOS), whose structures were directly elucidated by NMR analysis.
NMR and matrix-assisted laser desorption/ionization analysis of oligosaccharide mixture
Anomeric protons were sequentially labeled with a capital letter in decreasing order of chemical shifts. Starting from the left, 14 anomeric protons were counted (Figure 1), and they could be divided into two groups on the basis of their different intensities: those labeled as B, F, I, L, and O showed a similar intensity amongst them, and they were less intense when compared with the others. The second group contained all the other signals, including M and N, although they appeared less intense in the spectrum due to the solvent suppression effect; in the high field region, the diastereotopic methylene signals of different Kdo units were observed together with several methyl signals diagnostic of 6-deoxysugars, in agreement with the identification of rhamnose residues. The different ratios between the two groups of signals suggested the presence of at least two different oligosaccharide species.
|
The complete assignment of 1H and 13C resonances was achieved (Table I) combining the information from DQ-COSY, TOCSY, ROESY, NOESY, gHSQC, and gHMBC 2-D NMR experiments, according to the general strategy reported in the Supplementary data; the attribution of the two glucosamine residues A and M of the lipid A, of the inner Kdo K3, and that of the two external Kdo units K1 and K2, is reported in the same section as well.
|
NOE contacts (Table II) led to the determination of the sequence of the residues; the anomeric proton of the terminal rhamnose unit H showed a diagnostic weak NOE with the H-1 of E (Figure 2), suggesting the linkage of the residue H to the O-2 of E. Accordingly, the corresponding inter-residue NOE cross-peak of H-1 of H with the H-2 of E was found, although its intensity was overestimated because of the coincidence of the intraresidue NOE of the anomeric proton of H with its own H-2.
|
|
Similarly, the anomeric proton of unit E showed inter-residue NOEs with H-1 (weak) and H-3 (medium) of D, suggesting the linkage of residue E to the O-2 of D. This was supported by the strong NOE measured between H-1 of E and a proton in the region of 4.087ppm where the H-2 proton of D occurred. Also in this case, the high intensity of the cross-peak was attributed to the coincidence of inter-residue NOE with the intraresidue one involving its own H-2. In addition to these effects, H-1 of E had two other NOEs involving H-5 proton of both residues H and I; the H-1/H-5 NOE pattern is characteristic of
-L-Rha-(12)--L-Rha disaccharide (De Castro et al., 2004), and it involves the anomeric proton of the reducing unit and the H-5 of the backward residue of the disaccharide. In this case, the NOE with H-5 of H confirmed the presence of this residue on O-2 of E, whereas that with H-5 of I, centered on the left part of E anomeric peak, will be discussed later and deals with one of the minor species of the mixture. Location of residue D at O-3 of G was straightforward; its anomeric proton showed two clear NOEs: an intense one with H-3 of G and a medium with H-5 of the previous unit, that is, E, as expected; in turn, H-1 of G had two intense NOEs with both H-2 and H-3 of C, and the two cross-peaks had a slight asymmetric shape, because they were embracing the analog proton of B. These NOEs clearly located G on residue C (or B), more precisely at O-3 position as deduced from the glycosylation shift of carbon C-3 of this residue.
The mannose unit C was the residue linking the inner core to the outer core region; its H-1 showed an intense NOE with K3 proton H-5 at 4.319ppm, as confirmed from the corresponding C-H long-range correlation.
Similar strategy was applied to the other residues; the terminal glucosamine L was placed at O-2 of Gal N that was located at O-4 of K1 as suggested from two NOEs with this Kdo: a weak one with H-5 and strong one with H-4, and finally from the long-range correlation with C-4 in the HMBC spectrum.
The anomeric proton of the terminal galactose P presented several cross-peaks with the Kdo residue K1 protons, in particular one strong with H-7, a weak correlation with H-8a, and one medium with H-6; the correct location of this residue was inferred from its long-range correlation with the carbon at 72.9ppm, previously attributed to C-8 of this Kdo.
The last terminal galactose O, present in minor amount compared with P residue, was located at O-4 of the last Kdo unit, K2. This residue appeared in the same location of N, but it was not further substituted from the glucosamine L that must then be considered as a nonstoichiometric substituent of this residue; in addition L, named R1 in the comprehensive structure (Figure 3), dictated the different spectroscopical behavior of the external Kdo as well, whose signals were split as K1 or K2, depending on its presence or absence, respectively.
|
Similarly to L, the spectrum showed other two signals in nonstoichiometric ratio, that is, the O-3 linked I and the terminal F units. NOEs’ analysis placed F at O-3 of I, whereas this last "I" was located at O-2 of E, position already occupied from residue H. This contradiction was only apparent, and it must be specified that E was considered so far as a unique residue for simplicity of discussion, but a close inspection of the COSY spectrum (see inset of Figure S1) revealed that both H-1 and H-2 protons of E were composed of two closely spaced signals. In the ROESY spectrum, both E anomeric protons correlated with H-2 of D, but the low field one correlated with H-5 of I only, differently from the high field part that correlated with H-5 of H. Similar to H-1, the H-2 of E was split, the higher field component had an NOE effect with H-1 of I, and the lower field part correlated with H-1 of H. The spectroscopical behavior of this residue was explained considering the nonstoichiometric rhamnose F, named as R2 in Figure 3: when F was absent, I was terminal and resonating at a different chemical shift (H), influenced slightly E resonance as well.
On the basis of the above results, A. tumefaciens TT111 produced a core oligosaccharide decorated with two nonstoichiometric substituents, rhamnose F and glucosamine L; on statistical basis, it is possible to imagine four different structures (Figure 3): the simplest, 1, devoid of both F and L sugars, the second (2) where F is present, the third (3) with L but not F, and the last one (4) with both these units.
Matrix-assisted laser desorption/ionization (MALDI) analysis of the oligosaccharide mixture (Figure 4) confirmed the above hypothesis; the spectrum contained five peaks with different intensities; the peak at m/z 2170, labeled as A, was the most intense and consistent with the molecular species 3, having the composition Rha4Hex3HexN3Kdo2P2; in this species, the inner Kdo bore an oligosaccharide chain with one mannose and four rhamnose units, whereas the external one was highly substituted and the glucosamine L appeared on galactose N; the second main peak at m/z 2316, named B, was consistent with structure 4, and it differed from peak A for the additional rhamnose F unit, elongating the core oligosaccharide from the internal Kdo side. The less intense peaks, C and D, at m/z 2009 and 2155, respectively, were consistent with the last two species, 1 and 2, that were Rha4Hex3HexN2Kdo2P2 and Rha5Hex3HexN2Kdo2P2, respectively. The peak at m/z 2074 was an artefact of peak A, and it was due to a phosphate loss.
|
Analysis of molecular mechanics and dynamics calculations
The similarity among the rhamnan moiety of the LOSs with the motif 3)-
-L-Rha-(12)--L-Rha-(13)--L-Rha-(1, suggested as the epitope necessary to induce the PR-1 gene expression (Bedini et al., 2005), prompted the conformational study of the following two molecules, 5 and 6 (Figure 3b and c), representative of the two rhamnan moieties of the core oligosaccharide 1, 3 and 2, 4, respectively. In particular, the I–E–D sequence of molecule 6 resembled quite well the proposed epitope, except for the linkage modality of residue D, which was substituted at O-2 instead at O-3.
Molecular mechanics (MM) approach allowed to evaluate the optimal dihedral angles for D–G glycosidic junction; these values, together with those previously calculated for rhamnose glycosidic linkages (Bedini et al., 2005), were used to build molecules 5 and 6, needed for the successive computational phase.
In the molecular dynamics (MD) simulations, each molecule was kept in a thermal bath at 303K for 4000ps, and ensemble average interproton distances for each molecule were extracted from the simulations and translated into NOE contacts according to a full matrix relaxation approach (Espinosa et al., 1995; Poveda et al., 1997). The predicted NOEs (Table III; 3-D model of 6 in Figure 5) showed good agreement with those collected experimentally proving the reliability of the simulation data. More precisely, the NOESY spectrum was selected for the calculation of the interatomic distances, because the ROESY experiment is less reliable for their estimation: in this case, the effects depend not only on the internuclear distance but also on the offset between the resonance and the frequency at which the spin-lock is applied (Weimar and Woods, 2003). Conformity among MM and MD results was assumed on the basis of the similarity among the relaxed maps of the MM data (Bedini et al., 2005) and those scattered obtained through the extraction of 
and 
values from dynamic simulation (section B of Figure 6 and Figure 7). Statistical analysis of these data (Table IV) provided other rather useful parameters and pointed out the following features: all trajectories were dispersed around one minimum as attested from the similarity among the averaged value and the minimum of the corresponding flexible map; accordingly, standard deviation (SD) values were rather low. On the contrary, angles could be divided in two families, the first, belonging the Rha-(13)-Rha junction, behaved similarly to the angles, whereas the second, describing the Rha-(12)-Rha junction, reflected the presence of two minima, because its averaged value diverged substantially from the corresponding minimum value of the flexible map and had an higher SD.
|
|
|
|
|
Analysis of the
and trajectories of the two molecules provided further information. In the case of compound 5 (section C of Figure 6), the graphs suggested that each glycosidic angle equilibrated rapidly among the two possible conformations, including that belonging to the E–D connection, allowing the simplification that this molecule occupied the preferred values most of the time and that the lifetime of the maximum energy conformation was short compared with that of the minimum.
The dynamic behavior of molecule 6 resembled only in part that of 5; homolog and trajectories showed the same behavior with the exception of E (section C, Figure 7). In 6, this glycosidic angle equilibrated again between the two allowed values, but the lifetime of the second conformer, named 6b, was not as short as found in 5. It is worthy of note that, although the shape of E trajectory varied among the two molecules, the statistical values (average and SD in Table IV) were similar, as well as the frequency count graphs (Supplementary data; Figure S4), suggesting that the proportion among the different conformers around E in the two molecules was the same, whereas their exchange rate was altered, being very fast in 5 and slow in 6. The reason behind this behavior lies in the different structure of the two molecules; 6 presented the additional rhamnose F unit that evidently did not change the equilibrium population ratio around the E–D junction but increased the interconversion energy barrier, increasing the lifetime of the conformer with higher energy.
In contrast to 5, E–D junction in 6 defined two distinct conformers labeled as 6a and 6b, the first one being the more stable.
The secondary structure of 5 (Figure 8a) and 6a and 6b (Figure 8c and d, respectively) were elaborated according to the Connelly surface method and compared with that of the epitope in the synthetic nonasaccharide 7 (structure in Figure 3 and Connelly surface in Figure 8b) previously published (Bedini et al., 2005).
|
In order to simplify the comparison, different colors were used; the yellow surface was calculated from sugar residues common to all the compounds; the blue code was used for the nonasaccharide 7 only, and it was associated to the diversifying residue of the sequence, that is, the O-3 linked rhamnose belonging to the proposed epitope; the O-2 linked units in the homolog sequence position of 5 and of the low energy conformer 6a are red; the same O-2 linked units of 6b is orange; the rest of each molecule is not pertinent to the discussion and is gray colored.
The yellow areas of both the nonasaccharides, 6a, 6b and 7 matched each other as expected from the sequence homology, taking in account the next residue the model compound displayed a clear bending in the carbohydrate backbone that was half way from that less pronounced in 6a and that rather tight of 6b; clearly this difference was because of the replacement of the O-3 rhamnose with a O-2 linked unit; in any case, this new conformation was not drastically different from that of the model compound 7. As far as oligosaccharide 5 is regarded, it presented some similarities with 6a, but it was difficult to relate its conformation to the key area of the nonasaccharide owing to its reduced length.
|
Conclusions |
|---|
|
TopAbstractIntroductionResults and discussionConclusionsSupplementary dataAcknowledgmentsConflict of interest statementReferences |
|---|
In this study, the chemical and spectroscopical analysis has led to the determination of four different oligosaccharides, 1–4, that are representative of the carbohydrate composition of the membrane from the phytopathogenic bacterium A. tumefaciens TT111.
These species descend from a common architecture, oligosaccharide 1, further substituted from the nonstoichiometric residues L and F, namely the substituent R1 and R2 of Figure 3.
The structures of these LOS species acquire importance when the conformation of rhamnan tails, 5 and 6, are compared with the synthetic nonasaccharide 7 (Figure 8). In this regard, molecule 6 is well described from two different conformers, 6a and 6b, differing for the loose or tight state of their coil, respectively; these conformational motifs are rather similar to those described for the synthetic nonasaccharide and suggest that core oligosaccharides 2 and 4 are prone to be recognized from the plant defense systems. The response modality and intensity of the core fraction from A. tumefaciens TT111 are the topic of future studies.
On the basis of the above data, some consideration can be drawn for the recently published A. tumefaciens A1 (De Castro et al., 2006), member of the same group of TT111 but less virulent; this strain produces very low molecular weight LOSs, with the heptasaccharide 8 (Figure 3) as the most complex structure.
This oligosaccharide has some points in common with those from A. tumefaciens TT111; the external Kdo is substituted at O-8 with a ß-Gal unit, and the outer core is connected to the inner Kdo through a mannose residue; but differently from the reference strain, this last region is constituted from just one sugar residue, a mannose; clearly, one of the factors responsible of the low pathogenic profile of A. tumefaciens A1 might relate with its LOS core structure that lacks the conformational motif peculiar of TT111, dictating its adsorption properties.
|
Supplementary data |
|---|
|
TopAbstractIntroductionResults and discussionConclusionsSupplementary dataAcknowledgmentsConflict of interest statementReferences |
|---|
Supplementary data are available at Glycobiology online (http://glycob.oxfordjournals.org/).
|
Acknowledgments |
|---|
|
TopAbstractIntroductionResults and discussionConclusionsSupplementary dataAcknowledgmentsConflict of interest statementReferences |
|---|
The authors thank the ‘Centro di Metodologie Chimico-Fisiche’ of the University Federico II of Naples for NMR facilities, PRIN 2004 (M.P.) for financial support.
|
Conflict of interest statement |
|---|
|
TopAbstractIntroductionResults and discussionConclusionsSupplementary dataAcknowledgmentsConflict of interest statementReferences |
|---|
None declared.
|
Abbreviations |
|---|
LOS, lipooligosaccharide; LPS, lipopolysaccharide; MALDI, matrix-assisted laser desorption/ionization; MD, molecular dynamics; MM, molecular mechanics; NMR, nuclear magnetic resonance; PCP, petroleum ether–chloroform–phenol; PR, pathogen response
|
References |
|---|
|
TopAbstractIntroductionResults and discussionConclusionsSupplementary dataAcknowledgmentsConflict of interest statementReferences |
|---|
Bedini E., De Castro C., Erbs G., Mangoni L., Dow J.M., Newman M., Parrilli M., Unverzagt C. (2005) Structure-dependent modulation of a pathogen response in plants by synthetic O-antigen polysaccharides. J. Am. Chem. Soc. 127:2414–2416.[CrossRef][Web of Science][Medline]
De Castro C., Bedini E., Garozzo D., Sturiale L., Parrilli M. (2004) Structural determination of the O-chain moieties of the lipopolysaccharide fraction from Agrobacterium radiobacter DSM 30147. Eur. J. Org. Chem. 3842–3849.
De Castro C., Carannante A., Lanzetta R., Lindner B., Nunziata R., Parrilli M., Holst O. (2006) Structural characterisation of the core oligosaccharides isolated from the lipooligosaccharide fraction of Agrobacterium tumefaciens A1. Chem. Eur. J. 12:4668–4674.[CrossRef]
Espinosa J.F., Martín Pastor M., Asensio J.L., Dietrich H., Martín Lomas M., Smidtch R.R., Jiménez-Barbero J. (1995) Experimental and theoretical evidence of conformational flexibility of C-glycosides. NMR analysis and molecular mechanics calculations of C-lactose and its O-analogue. Tetrahedron Lett. 36:6329–6332.[CrossRef]
Hooykaas P.J.J., Klapwjik P.M., Nuti M.P., Schilperoort R.A., Rorsch A. (1977) Transfer of the A. tumefaciens Ti plasmid to avirulent Agrobacteria and Rhizobium ex planta. J. Gen. Microbiol. 98:477–484.
Lippincott J.A. and Lippincott B.B. (1969) Tumour-initiating ability and nutrition in the genus Agrobacterium. J. Gen. Microbiol. 59:57–75.
Poveda A., Asensio J.L., Martín Pastor M., Jiménez-Barbero J. (1997) Solution conformation and dynamics of a tetrasaccharide related to the Lewis X antigen deduced by NMR relaxation measurements. J. Biomol. NMR 10:29–43.[CrossRef][Web of Science][Medline]
Pueppke S.G. and Benny U.K. (1984) Adsorption of tumorigenic Agrobacterium tumefaciens cells to susceptible potato tuber tissues. Can. J. Microbiol. 30:1030–1037.[Web of Science][Medline]
Silipo A., De Castro C., Lanzetta R., Molinaro A., Parrilli M. (2004) Full structural characterization of the lipid A components from the Agrobacterium tumefaciens strain C58 lipopolysaccharide fraction. Glycobiology 14:805–815.
Silipo A., Molinaro A., Sturiale L., Dow J.M., Erbs G., Lanzetta R., Newman M., Parrilli M. (2005) The elicitation of plant innate immunity by lipooligosaccharide of Xanthomonas campestris. J. Biol. Chem. 280:33660–33668.
Weimar T. and Woods R.J. (2003) Combining NMR and simulation methods in oligosaccharide conformational analysis. In Jimenez-Barbero J. and Peters T. (Eds.). NMR Spectroscopy of Glycoconjugates(Wiley-VCH, Weinheim, DE) pp. 111–142.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M.-A. Newman, J. M. Dow, A. Molinaro, and M. Parrilli Invited review: Priming, induction and modulation of plant defence responses by bacterial lipopolysaccharides Innate Immunity, April 1, 2007; 13(2): 69 - 84. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||