[a]Department of Chemistry and Biochemistry, University of Maryland, 5401 Wilkens Avenue, Baltimore, MD 21228-5398
Email:
qiuxu@umbc.edu or bush@umbc.edu[b] Laboratory of Genetics, National Cancer Institute, Bethesda, MD 20892
Click on items in the list below to jump to a particular section.
Click here to return to hyperdocument contents list Click here to go onto next entry
A full paper on the subject area of this paper is in press. If you are interested in a reprint please send me an email with the subject heading 'J22 Antigen Reprint Request' using the form obtained by clicking on my email address below:
qiuxu@umbc.edu or bush@umbc.edu
(** Please make sure you give your correct email address when using this form)
Click here to return to hyperdocument contents list Click here to go onto next entry
We report on the conformation of a tetrasaccharide fragment in the repeating subunit of the cell wall polysaccharide of Streptococcus mitis J22, a receptor of the lectin of Actinomyces viscosus T14V in a bacterial coaggregation which is important in the ecological interactions of oral bacteria. Conventional 2-dimensional NOESY data on the polysaccharide can not fit to a single conformation of the tetrasaccharide fragment and most of the ring protons overlap. Therefore we have prepared a polysaccharide sample fully enriched in 13C from which we have determined accurate NOESY cross peak volumns in a 3-dimensional heteronuclear resolved spectrum which allows accurate determination of many more NOESY cross peaks than does conventional 2-dimensional spectroscopy. We have also used the 13C enriched polysaccharide to measure accurate values of long range 13C-1H coupling constants which can be correlated with glycosidic dihedral angles. Molecular modeling calculations on blood group oligosaccharides in our laboratory using similar methods which showed relatively rigid conformations with little flexibility of the glycosidic linkages. The present NOESY and 3JCH data can be reconsiled with a model for the antigenic tetrasaccharide in which three distinct conformations are in exchange. We propose that some carbohydrate epitopes such as those of the blood group oligosaccharides are relatively rigid while others such as tetrasaccharide fragment in these studies exhibit much greater flexibility.
Click here to return to hyperdocument contents list Click here to go onto next entry
Bacterial colonization generally requires adherence to a surface to provide an environment with a suitable temperature and supply of moisture and nutrients. The mouth is normally inhabited by a wide variety of bacterial species all of which adhere to the soft tissue or to the teeth. The microbial ecology of the mouth is convenient to study and a complex scheme of inter bacterial adherence has been extensively documented. Not only are such studies of importance to oral health but the system also provides an excellent model for adherence of bacteria to other tissues of the host. At the molecular level, the adhesion steps can be identified as specific interactions between proteins and peptides or between proteins and specific carbohydrate epitopes. It is this latter interaction between polysaccharides on one bacterial species with lectins on another which is the focus of the present study.
Research on conformation of oligosaccharide epitopes of glycoproteins and glycolipids has received increased attention in recent years as a result of their importance as lectin receptors in numerous important biological phenomena. Interpretations of the evidence from NMR spectroscopy and molecular modeling calculations are not in complete agreement. Some authors have claimed that certain oligosaccharide epitopes adopt rigid conformations especially in oligosaccharides showing blood group activity while other authors have emphasized equally compelling evidence for the flexibility of glycosidic linkages.
Less effort has been directed toward conformational studies of the polysaccharides found on the surface of bacteria such as capsules, lipopolysaccharides and cell wall polysaccharides than on the oligosaccharides of eukaryotic glycoproteins and glycolipids. This research has used some of the established methods of NOE and molecular modeling to study the cell wall polysaccharide of S. mitis J22 (Abeygunawardana, C., Bush, C. A., and Cisar, J. O. (1990) Biochemistry 29, 234-248), one of the polysaccharides which serves as a receptor for the lectin of A. viscosus T14V in coaggregation of oral bacteria. The polysaccharide is composed of seven repeating monosaccharides, as determined previously. All the sugars are in the D configuration except for the rhamnose residues which are L.
In our preliminary NMR studies, we have not been able to fit our NOE data to a single conformation. Since definition of flexible polysaccharide structure requires much more spectroscopic data than that needed to specify single rigid structure, we report, in addition to simple NOE, 13C edited 3D NOESY experiments on uniformly 13C enriched polysaccharide with greatly improved resolution along with an interpretation of long- range heteronuclear coupling constants (3JCH) which can be reconciled to a model of the antigenic tetrasaccharide epitope from this polysaccharide.
Click here to return to hyperdocument contents list Click here to go onto next entry
Unusually narrow proton linewidth and rather long mixing time (350-400 ms) for developing NOESY cross peaks indicate that this segments of polysaccharide is quite flexible, although the overall tumbling of the whole polymer could be slow. Proton 2D NOESY of natural abundance polysaccharide at 400 ms shows all the proton cross peaks across the linkages as well as intra-residue ones (Figure 1). The cross peaks, however, can not be completely resolved due to ring proton peak overlap within 3- 4 ppm. As a result of the overlap, only twenty five cross peaks from residues a, b, c and g cab be unambiguously measured as shown in Table 1 . Eight of those arise from protons across the glycosidic linkages.
Peak overlap is greatly alleviated in 3D 13C-edited NOESY on 13C enriched polymer. Although the linewidth is broad and peak intensities are small, large 13C chemical shifts lends to improved resolution and almost all the proton cross peaks in NOESY can be clearly assigned.
In principle, the two pulse sequences, HMQC-NOESY and NOESY-HMQC, should give the same spectra. But there are practical differences in applying to carbohydrates. Because of higher digitization along acquisition and somewhat degraded resolution along proton evolution dimension by residual 1JH-C coupling, HMQC- NOESY is generally preferred for carbohydrates (Figure 2a and 2b). Our analysis in the following will be based on HMQC-NOESY experiment.
As shown in Table 1 , 3D HMQC-NOESY provides more than twice in number the cross peaks as does 2D NOESY due to cross peaks separation along the 13C dimension. For example, aH1 has important cross peaks to both bH2 and aH5 which are separated by less than 4.5 Hz. It is impossible to separate the two peaks on 2D NOESY. They are, however, resolved by the 13C chemical shifts of aC5 (70.36 ppm) and bC2 (72.92 ppm) (Figure 3).
Both selective T1's and NOE intensities are smaller for 13C enriched polymer than for the natural abundance polymer. While different samples could have different molecular weights and thus different relaxation properties, we believe the motions responsible for NMR relaxation involved local or segmental effects rather than entire polymer tumbling for this polysaccharide. Therefore, we simulated NOE intensities using a complete relaxation matrix method (Borgias, B. A, Gochin, M., Kerwood, D. J., and James, T. L. Progress in NMR Spectroscopy (1990) 22, 83-100) which includes heteronuclear dipolar relaxation between 13C and 1H, RC-HDD (Equation 1)(Neuhaus D. and Williamson M. P. "The Nuclear Overhauser Effect in Structural and Conformational Analysis" (1989) VCH publisher), in all the diagonal terms.
We searched for an optimum effective rotational correlation time of proton vectors to simulate the experimental NOE intensities between intra-residue protons, under a reasonable assumption that the intra-residue NOE intensities do not strongly depend on polymer conformation (relative residue orientation). For the natural abundance polymer, the optimum value of effective rotational correlation time (tauc) for the proton vectors is 1.4 ns which is similar to the values found for oligosaccharides. Therefore the effective tauc represents segmental or internal motion and not the overall tumbling of the polysaccharide. While the agreement between the calculated and observed intra-residue NOE is not exact, most of the peaks agree within the experimental error suggesting that the anisotropy effect of segmental motion on relaxation is not so serious. If the 13C-1H heteronuclear dipolar relaxation is included, the same effective rotational correlation time can be used to generate the intra-residue NOE for the 13C labeled polysaccharide as determined by HMQC-NOESY Table 1 is obtained from the energy search for the four linked residues (a, b, c and g). While this conformation is adequate for the simulation of the intraresidue NOE which do not depend strongly on conformation, it can not be used to derive all NOE peaks observed.
Click here to return to hyperdocument contents list Click here to go onto next entry
Molecular modeling calculations were carried out with QUANTA/CHARMm using potential function and charges developed for carbohydrates (Rassmussen, K. (1982) Acta Chem. Scand. A 36 323-327) and dielectric constant of 3. Disaccharide (a-b, b-c, g- b) maps were constructed by restraining the glycosidic angles phi (O5-C1-O-Cx)and psi (C1-O-Cx-Cx-1) from 0 to 350 degree in 10 degree step. and minimizing with respect to all the other degrees of freedom, except for keeping sugar pucker in chair conformation. Minimization was carried out for 100 steps of Adopted Newton Raphson methods (ABNR) at each point on the grid. At each of the low energy points, local energy minima were sought by unconstrained minimization.
All the unique minimum energy conformations of the disaccharides ( Table 2 ) were combined to form the tetrasaccharide. The energy of these tetrasaccharides was then minimized (using ABNR and Powell methods) allowing the glycosidic dihedral angles to relax. Minimization was carried out until RMS deviation of the energy gradient was no more than 0.001 kcal/mole. Fifteen unique tetrasaccharide conformers of energy within 10 kcal/mole of the global minimum were chosen for further analysis ( Table 3 ). The eight lowest energy conformers within 5 kcal/mole of global minimum were selected for molecular dynamics simulations. Simulations were started by assigning random initial velocities at 0K. The system was thermalized in 15K increments at 0.25 ps intervals until room temperature (300K). The atom velocities were periodically scaled if the temperature drifted more than 5K and the molecule was equilibrated at that temperature for 25 ps. The production run was carried out for a period of 1ns during which the energy of the system was well conserved. Integration, using Verlet's leap frog algorithm, was done every 1 fs and coordinates were stored for every 100 steps, leading to 10,000 data points. The bond length were constrained using SHAKE algorithm with an error tolerance of 10-6. Several of molecular dynamics simulations for the eight conformers having relative energies within 5 kcal/mole (Figure 4) show transitions to other regions of low energy. Simulations show that several distinct conformations are available for a-b and g- b linkages, but b-c linkage has access to a fairly broad contiguous conformational region of phi and psi. The conformations in aqueous solution were located with NOESY and 3JCH. NOE cross peaks volumes were calculated for each tetrasaccharide as a function of the glycosidic dihedral angles. NOE cross peaks from protons across the three glycosidic linkages, such as aH1-bH3, bH1-cH4 and gH1-bH2, can be fit by a variety of different conformations. Inclusion of aH1-bH2, aH1-gH1 etc. make it impossible to find a single conformation to accommodate all the observations.
The failure of above approach suggests that no single conformation can be fit to NOE data and a model which includes multiple conformations in exchange will be required. In Table 4 are tabulated the three bond coupling constants 3JCH across glycosidic linkages as computed for all the tetrasaccharides listed in Table 3 according to Tvaroska et al. (Tvaroska, I., Hricovini, M, and Petrakova, E (1989) Carbohydr. Res., 189, 359-362). It is clear that no single conformation can be used to simulate all the coupling constants. A linear combination of conformation 3 and 4 with statistical weights of 0.63 and 0.37 gives the best match between simulated and experimental values of 3JCH.
Calculation of an average relaxation matrix from conformers 3 and 4 with the above statistical weights give reasonable consistence between simulation and experimental 2D NOE data.
This "virtual conformation" from those two conformers can not fit cross peaks between bH1-cH5 and bH1-cH3+cH4. Figure 5 shows that bH1 is close to cH3 in conformer 3 (Figure 5a) and close to cH4 in conformer 4 (Figure 5b). The only low energy conformation with close vicinity of bH1 and cH5 is conformer 8 (Figure 5c). Therefore, the combination of 15% conformer 8, 50% of conformer 3 and 35% of conformer 4 reproduces the 3D NOESY data ( Table 1 ). The statistical average 3JCH's from those three conformers are listed in Table 4 .
Click here to return to hyperdocument contents list Click here to go onto next entry
(1) 13C edited NOESY on 13C enriched polysaccharide provide much more constraints for searching polysaccharide conformation.
(2) Both NOESY data and 3JCH indicate that the antigenic tetrasaccharide of the polysaccharide from cell wall of S. mitis J22 exist in multiple conformation in aqueous solution.
(3) Molecular modeling shows that there is considerable flexibility of glycosidic linkages for this tetrasaccharide. There are a number of distinct low energy conformations and molecular trajectories show transitions among the multiple conformations.
Click here to return to hyperdocument contents list
