Computer modelling studies on the modes of binding of some of the α- glycosidically linked disaccharides to concanavalin A

Transcription

1 Proc. Int. Symp. Biomol. Struct. Interactions, Suppl. J. Biosci., Vol. 8, Nos 1 & 2, August 1985, pp Printed in India. Computer modelling studies on the modes of binding of some of the α- glycosidically linked disaccharides to concanavalin A Y. CHANDRA SEKHARUDU and V. S. R. RAO Molecular Biophysics Unit, Indian Institute of Science, Bangalore , India Abstract. The possible modes of binding of kojibiose, nigerose, maltose and ManPα(1 2)Man to concanavalin A have been investigated using computer modelling studies. While α12 linked disaccharides bind to concanavalin A in two modes, i.e. by placing the reducing as well as non-reducing sugar units in the sugar binding site, nigerose or maltose can bind only in one mode, i.e. by placing the non-reducing sugar unit in the binding site. Though, both the sugar residues in α 12 linked disaccharides can reach the binding site, the preference is high for the non-reducing unit. When the non-reducing residue, in any of these disaccharides, enters the binding site, the allowed orientations and the possible hydrogen bonds with the protein seem to be independent of the glycosidic linkage. However, the number of hydrogen bonds the outward sugar residue forms with the protein are dependent on the type of linkage. Atleast one of the hydroxyl groups adjacent to the glycosidic linkage on the outward sugar residue is involved in the formation of a hydrogen bond with the protein suggesting the presence of an extended binding site. The orientation of the reducing sugar residue in the extended binding site is dependent on the linkage. Its orientation in nigerose is flipped when compared to that found in kojibiose or maltose leading to different non-covalent interactions with the protein which affect their binding affinities. Keywords. Concanavalin A; modes of binding of disaccharides to; computer modelling studies; protein-carbohydrate interactions. Introduction Concanavalin A (Con A) is a widely used lectin for studying various aspects of cell surface carbohydrates (Nicolson, 1976). It can specifically bind to D-mannose (Man) and D-glucose (Glc) and some of their derivatives (Poretz and Goldstein, 1970). There are suggestions that Con A binds to terminal Glc (Goldstein and Hayes, 1978) and terminal and internal Man residues (Goldstein et al., 1973) and that its sugar binding site accommodates a single residue (Brewer and Brown, 1979; Loontiens et al., 1983). However, its binding affinity to oligosaccharides seems to depend significantly on the type of linkages. Among the α-glycosidically linked glucobioses: kojibiose, nigerose and maltose; kojibiose has the highest affinity for Con A and nigerose has the least (Goldstein et al., 1965). The binding affinity of Man Pα(l 2)Man was four fold more than that of methyl-α-d-mannopyranoside (α-me ManP) (Brewer and Brown, 1979). These authors, to explain this observed phenomenon advanced a statistical reason that the increase in the binding affinity is proportional to the available number of binding sugar residues in the oligosaccharide, but not due to an extended binding Abbreviations used: Con A, Concanavalin A; Man, D-mannose; Glc, D-glucose; α-memanp, methyl-α-dmannopyranoside. 389

2 390 Chandra Sekharudu and Rao site. Loontiens et al. (1983) from fluroscence quenching studies on Con A-4- methylumbelliferyl-α-(manp) n complexes suggested that the binding site accommodates a single sugar residue and the increase in the binding affinities of α12 linked oligomers may be due to unspecific interactions, some of which could originate from aglycon. These authors suggested a large binding preference for terminal ManP residue in 4-methylumbelliferyl-α-(ManP) n. On the other hand, Williams et al. (1981) suggested from the kinetic studies of Con A with some of the derivatives of 2-O-substituted p- nitrophenyl-α-d-mannopyranoside, that the ManP residue at the reducing end of the disaccharide occupies the primary binding site, and attributed the increase in the binding affinity to interactions in the extended binding site. Thus, the modes of binding of these disaccharides and the effect of the linkage on their affinities for Con A and the size of the binding site are not clearly understood. Such information can generally be obtained from the three dimensional structures of Con Α-saccharide complexes. However, no attempt has been made to determine the three dimensional structure of these complexes using experimental or theoretical methods. Recently, Sekharudu and Rao (1984a, b, c) using computer modelling studies determined the modes of binding of α-memanp, α-meglcp and some of their derivatives to Con A and also the interactions between the sugar and the lectin. These studies have now been extended to α- glycosidically linked disaccharides which help not only in understanding their binding mechanism but also provide a theoretical basis for the differential binding affinities of the disaccharides differing in linkage. This study helps also in the elucidation of carbohydrate structures on cell surfaces since differences in the binding affinities of Con A to oligosaccharides of ManP and GlcP varying in the type of linkage are being utilized to detect the architecture of cell surface carbohydrates. Method of calculations The numbering of atoms and dihedral angles which define the conformation of the disaccharides: kojibiose, nigerose, maltose and ManPα(1 2)Man are shown in figure 1. Total conformational energy of these molecules was computed using empirical potential energy functions by varying the two dihedral angles (φ, ψ) at 10 interval. The functions used, for computing nonbonded, electrostatic and torsional energy, are the same as described earlier (Momany et al., 1975). For calculating the anomeric energy, the function (Prakash, 1980) VANOMERIC(α-linkage) = 0 9[1 cos(φ + 60º)] was used. Iso-energy contours were drawn at intervals of 1 Kcal and are displayed in figures 2, 3 and 4. The co-ordinates of various atoms in the binding site of Con A, from the Protein Data Bank, deposited by Hardman and Anisworth (Hardman and Anisworth, 1976) were used for the computer modelling of the binding site. The 75 amino acid residues (Sekharudu and Rao, 1984b) which fall within a sphere of 20 Å radius around the reference point (the coordinates for the center of the pyranose ring in the binding site) and the bivalent metal ions were used in the calculations. The atoms in the amino acid side chains beyond C ß were treated as flexible. However, the side chains which are

3 Computer modelling studies 391 Figure 1. Numbering of atoms and dihedral angles in disaccharides, (a) Kojibiose (b) nigerose (c) maltose (d) ManPα(1 2)Man. Hydrogen atoms are not shown. involved in the formation of co-ordination bonds with the metal ions(hardman et al., 1982) or in dimerization (Reeke et al., 1975) were treated as rigid. While fitting the disaccharides in the binding site of Con A, the center of the sugar residue (non-reducing or reducing unit, as the case may be) was placed at the positions determined previously for methyl-α(and ß)-D-glucopyranosides (mannopyranosides) and the orientations were also restricted to the allowed orientations of the corresponding methyl-α (and ß)- D-gluco(manno) pyranosides (Sekharudu and Rao, 1984a, b). For each of these orientations, the sugar residue in the binding site was fixed and the other sugar residue was allowed to take up all conformations within 5 Kcal/mol of its global minimum. Stereochemical contact criteria (Ramachandran and Sasisekharan, 1968) were followed

4 392 Chandra Sekharudu and Rao Figure 2. Iso-energy contour diagrams for kojibiose. The numbers indicate the level of contour in Kcal/mol. Shaded area describes the probable conformations of kojibiose that enter into the binding site (a) when the non-reducing glucose unit is placed in the binding site (b) when the reducing glucose unit is placed in the binding site. to identify the allowed orientations in the binding site for every probable conformation of the disaccharides under study. Whenever a conformation fitted into the binding site without any severe steric overlap, then a search was made for possible non-covalent interactions between the sugar and the protein. These included hydrogen bonds, hydrophobic and charge interactions. Results and discussion From figures 2 and 3 it can be seen that, most of the probable conformations of kojibiose, nigerose and maltose may be placed in the binding site of Con A without much steric overlap. Figure 5 represents the possible binding orientations for the non-

5 Computer modelling studies 393 Figure 3. Iso-energy contour diagrams for (a) nigerose (b) maltose. Shaded area describes the probable conformations of the respective disaccharides which can reach the sugar binding site. reducing units of kojibiose, nigerose and maltose in the binding site of Con A. These allowed orientations are characteristic of α-anomers of GlcP (Sekharudu and Rao, 1984a) to bind to Con A, suggesting that the addition of a second Glc residue through α-linkages may not significantly alter the possible binding orientations of the sugar that enters the binding site. Representative stereoscopic projections of Con A-disaccharide complexes (figures 6, 7 and 8) suggest that the mode of binding of the non-reducing sugar unit is the same for all the three disaccharides. However, the reducing sugar units are placed differently with respect to the protein. The most striking observation is the different arrangement of the reducing unit of nigerose when compared to kojibiose and maltose. It can be seen from the figures 6, 7 and 8, that one of the sides of the reducing sugar residue is nigerose, which contains Η-1B (if the reducing sugar unit is in β-form), H-3B, H-5B, H-6'B and H-6B is flipped when compared to the reducing sugar units of kojibiose and maltose. This may affect the non-covalent interactions between the sugar and the protein. Since, the non-reducing residues of these disaccharides assume orientations similar to those of α-meglcp (Sekharudu and Rao, 1984a) in the binding

6 394 Chandra Sekharudu and Rao Figure 4. Iso-energy contour diagram for ManP α(1 2)Man. Shaded area describes the probable conformations of ManP α (1 2)Man that enter into the binding site, when the nonreducing mannose unit is placed in the binding site. Figure 5. Steric map representing the allowed orientations for the nonreducing glucose units of kojibiose, nigerose and maltose for entering into the binding site. The Eulerian angles Φ, Θ and ψ can be read from the figure in the following way. The length of the line joining the dot. and the point on the grid gives the value of the angle ψ. The grid point gives the values of Φ and Θ.

7 Computer modelling studies 395 Figure 6. Representative stereoscopic projection of kojibiose with its non-reducing glucose unit placed in the binding site in one of the allowed orientation (Φ =330; Θ =150; ψ =200) with the dihedral angles (φ, ψ) as (30, 30). The dashed lines represent possible hydrogen bonds between the sugar and the protein. The hydrogen bonds described here are permitted to the orientation given above. The hydrogen bonding scheme, however, varies with the conformation of the disaccharide and its orientation (see text). The same thing applies for the hydrogen bonding schemes shown in figures 7, 8, 10 and 12. Figure 7. Representative stereoscopic projection of nigerose with its non-reducing glucose unit placed in the binding site in one of the allowed orientation (Φ = 330; Θ = 150; ψ = 200) with the dihedral angles (φ, ψ) as (30, 30).

8 396 Chandra Sekharudu and Rao Figure 8. Representative stereoscopic projection of maltose with its non-reducing glucose unit placed in the binding site in one of the allowed orientation (Φ = 330; Θ = 150; ψ = 200) with the dihedral angles (φ, ψ) as (30, 30). site, the hydrogen bonding scheme for these residues with the protein is the same as that earlier discussed for methyl-α-d-glucopyranoside (Sekharudu and Rao 1984a). Thus, in this mode of binding, the sugar residue in the binding site can form four hydrogen bonds with the protein. In the case of kojibiose, the hydroxyl group, O-1B or O-3B of the reducing unit shows more propencity to form a hydrogen bond in the conformations around (φ =20; ψ = 20)and (φ = 40; ψ = 20). Depending on the orientation of the non-reducing sugar unit in the binding site, the O-1Β or O-3B hydroxyl group of the reducing unit may form a hydrogen bond with OH of Tyr(l00) or with guanidyl group of Arg(228). The O-3B or O-6B hydroxyl group of maltose in conformations around φ =20, ψ = 20 may form a hydrogen bond with OH of Tyr(l00) or with guanidyl group of Arg(228). The O-2B or O-4B hydroxyl group of nigerose in conformations (φ =20, ψ = 20) shows increased tendency to form a hydrogen bond with the guanidyl group Arg(228). To a small extent, the hydroxyl group O-6B or O-2B in conformations φ = 40, ψ = 20 may also form a hydrogen bond with the protein in a few orientations. Interestingly, in the mode of binding wherein the non-reducing sugar units are placed in the primary binding site, the side of the reducing unit of kojibiose or maltose which contains Η-1B, H-3B, H-5B, H-6B and H-6'B hydrogens is placed close to the hydrophobic side chains of Leu (99) and Tyr(l00) in several of the allowed orientations. This may lead to favourable hydrophobic interactions between the protein and kojibiose or maltose. On the other hand, for nigerose, instead of this hydrophobic side of the reducing unit, the other side which is less hydrophobic in nature, is placed close to the hydrophobic residues of the protein. This suggests that the

9 Computer modelling studies 397 reducing units of kojibiose and maltose may be involved in hydrophobic interactions, unlike the reducing unit of nigerose. This, perhaps may provide an explanation for the relatively lower binding affinity of α13 glycosidically linked glucobiose when compared to α12 or α14 linked glucobiose (Goldstein et al., 1965). Furthermore, it is of great interest, that only in the case of kojibiose the reducing sugar residue could also reach the carbohydrate specific combining site (figures 9 and 10). However, in this mode of binding, the allowed orientations for the reducing unit in the binding site (figure 9) and the conformations which can reach the binding site (figure 2b) are relatively less. This suggests that even kojibiose may predominantly reach the sugar binding site of Con A with its non-reducing residue. However, one cannot rule out the possibility of kojibiose binding to Con A by placing its reducing sugar unit in the binding site. In this series kojibiose has the unique binding property that its non-reducing as well as reducing glucose units can reach the binding site to different extents and in both of these modes the residue outwards from the binding site forms one or two hydrogen bonds with the protein. The higher number of available binding residues in kojibiose may explain its higher affinity to Con A than maltose (Goldstein et al., 1965) which can bind in only one mode. Thus, the linkage between glucose units not only affects their modes of binding but also their interactions with the protein and thereby their relative binding affinities. ManPα (l 2)Man can reach the binding site of Con A with its non-reducing unit placed in the binding site in several of the orientations (figure 11) of α-memanp (Sekharudu and Rao, 1984b). In these orientations, all the probable conformers of ManPα (l 2)Man (figure 4) can reach the binding site and the hydrogen bonding Figure 9. Steric map representing the allowed orientations for the reducing glucose unit of kojibiose.

10 398 Chandra Sekharudu and Rao Figure. 10. Representative stereoscopic projection of kojibiose its reducing glucose unit placed in the binding site in one of the allowed orientation (Φ = 350; Θ = 90; ψ = 330) in the conformation (φ = 20; ψ = 20). Figure 11. Steric map representing the allowed orientations for the non-reducing residue in ManPα(1 2)Man.

11 Computer modelling studies 399 scheme for this non-reducing sugar unit is similar to that of α-memanp, i.e. the hydroxyl groups at C-2, C-3, C-4 and C-6 are hydrogen bonded to the protein (Sekharudu and Rao, 1984b). In addition, at least two of the O-3B, O-4B and O-6B hydroxyl groups of the reducing sugar residue which is outwards from the binding site, also form hydrogen bonds with Arg(228), Thr(226), OH of Tyr(l00) or Gly(98). The stereoview (figure 12) of ManPα(1 2)Man when its non-reducing unit is placed in the binding site, shows that the reducing unit is outwards from the binding site. Figure 12. Representative stereoscopic projection of ManPα(1 2) Man with its nonreducing glucose unit placed in the binding site in one of the allowed orientation ( Φ = 340; Θ = 90; ψ = 350) with the conformational angles (φ, ψ) as (20, 40). The reducing sugar unit of ManPα(l 2)Man, in some of the favoured conformations (φ = 60, ψ = 60 to 0) can reach the binding site in the orientations (Φ = 355; Θ = 90; ψ = 325), (Φ = 355; Θ = 90; ψ = 340). Figure 13 describes the placement of the reducing unit of ManPα(l 2)Man in the binding site. In this mode of binding, the residue which is in the binding site has to be slightly pulled outwards from the binding site to relieve some of the unfavourable steric contacts encounted by the non-reducing unit. However, the hydrogen bonding scheme for the reducing unit with the protein is also similar to that of α-memanp. The O-2A hydroxyl group of the non-reducing sugar unit which is outwards from the binding site can form a hydrogen bond with Thr(226). Thus, ManPα(1 2)Man can also reach the binding site in two modes, i.e. by placing its non-reducing as well as reducing sugar units in the binding site. Comparison of figures 12 and 13 throw much light on the different directions of

12 400 Chandra Sekharudu and Rao Figure 13. Representative stereoscopic projection of ManPα(1 2) Man with its reducing glucose unit placed in the binding site in one of the allowed orientation (Φ = 355; Θ = 90; ψ = 340) with the conformational angles (φ, ψ) as ( 60, 20). propagation of the outward residue in the two modes of binding of ManPα(l 2)Man. However, the probability for ManPα(l 2)Man to reach the binding site in the former mode of binding, i.e. when the non-reducing unit is the occupant of the binding site, is much higher when compared to the latter, because of the possibility of more number of hydrogen bonds between the sugar and the protein and also as its more probable conformers can get into the binding site without any steric overlap. Thus, ManPα(l 2)Man can reach the binding site in two modes similar to kojibiose and the former one forms more number of hydrogen bonds with the protein. This may explain the higher binding affinity of ManPα(l 2)Man over kojibiose. Goldstein and coworkers (Goldstein et al., 1965), while explaining the higher binding affinity of kojibiose over nigerose and maltose assumed that the α12 glycosidic linkage might allow a closer approach of the sugar to the protein surface than the αl3 glycosidic linkage. However, these authors neither considered the two possible modes of binding for kojibiose nor the differences in the orientation of the sugar residue which is outwards from the binding site and hence its interactions with Con A in explaining the binding affinities of kojibiose, nigerose and maltose as revealed in the present study. The present computer modelling studies on the αl2 linked gluco(manno) bioses have shown that, the non-reducing as well as reducing sugar units which possess free 3-, 4- and 6-hydroxyl groups can reach the binding site of Con A. Secondly, these studies suggest much higher preference for the non-reducing sugar residues of αl2 linked gluco(manno) bioses to bind to Con A, than for the reducing sugar residue. Lastly, they also demonstrate that at least one of the hydroxyl groups adjacent to the glycosidic linkage on the sugar residue which is outwards from the binding site also forms a

13 Computer modelling studies 401 hydrogen bond with the protein, indicating that the interactions in the binding site are extended slightly beyond one sugar residue. Thus, the results obtained in the present study agree in parts with the conclusions of the earlier workers (Brewer and Brown, 1979, Loontiens et al., 1983; Williams et al., 1981). Brewer and Brown (1979) have suggested that a necessary requirement for the enhanced binding of a saccharide is that it should contain multiple glucose or mannose residues which possess free 3-, 4- and 6- hydroxyl groups, which is in agreement with the present results. However, these authors have not considered the different preferences for the different binding residues in a saccharide as revealed in the present study. Loontiens et al. (1983) have suggested that the binding site of Con A interacts with one ManP group at a time, and that it can interact differently with any, out of several ManP groups in an oligosaccharide with higher affinity to the terminal ManP in the Me Umb-(ManP) 2 which is in agreement with the results of our present study. However, these authors attributed the increase in the binding affinity to unspecific interactions which are in disagreement to the extended interactions as discussed earlier. However, the presence of such extended interactions have been postulated by Williams et al. (1981). Thus, the present investigation not only rationalizes the available experimental data but also presents an insight into the stereochemistry of Con A-carbohydrate interactions. Acknowledgements The authors wish to thank Dr. Margaret Biswas for her useful suggestions. This work was partially supported by the Department of Science and Technology, New Delhi. References Brewer, C. F. and Brown, III, R. D. (1979) Biochemistry, 18, Debray, H. and Montreuil, J. (1983) J Biosci., 5, 93 Goldstein, I. J., Hollerman, C. E. and Smith, E. E. (1965) Biochemistry, 4, 876. Goldstein, I. J., Reichart, C. Μ., Misaki, A. and Gorin, P. A. J. (1973) Biochim. Biophys. Acta, 317, 500. Goldstein, I. J. and Hayes, C. E. (1978) Adv. Carbohydr. Chem. Biochem., 35, 127. Hardman, K. D. and Anisworth, C. F. (1976) Biochemistry, 15, Hardman, K. D., Agarwal, R. C. and Freiser, M. J. (1982) J. Mol. Biol., 157, 69 Kornfeld, Κ., Reitman, Μ. L. and Kornfeld, R. (1981) J. Biol. Chem., 256, Loontiens, F. G., Clegg, R. M. and Landschoot, A. V. (1983) J. Biosci., 5, 105. Momany, F. A., McGuire, R. F., Burgess, A. W. and Scheraga, H. A. (1975) J. Phys. Chem., 79, Nicolson, G. L. (1976) in Concanavalin A as a Tool (eds H. Bittiger and H. P. Schnebli) (New York: Wiley) p. 1. Poretz, R. D. and Goldstein, I. J. (1970) Biochemistry, 9, Prakash, S. (l980) Theoretical studies on the influence of exo-anomeric effect, geometry of the sugar and linkage on the unperturbed dimensions of polysaccharides, Ph.D. Thesis, Indian Institute of Science, Bangalore. Ramachandran, G. N. and Sasisekharan, V. (1968) Adv. Protein Chem., 23, 283. Reeke Jr, G. N., Becker, J. W. and Edelman, G. M. (1975) J. Biol. Chem., 250, Sekharudu, Y. C. and Rao, V. S. R. (1984a) J. Biomol. Struct. Dyn., 2, 41. Sekharudu, Y. C. and Rao, V. S. R. (1984b) Int. J. Biol. Macromol., 6, 337. Sekharudu, Υ. C. and Rao, V. S. R. (1984c) Curr. Sci., 5, 93. Williams, Τ. J., Homer, L. D., Shafer, J. Α., Goldstein, I. J., Garegg, P. J., Hultberg, Hans., Iversen, T. and Johansson, Rolf. (1981) Arch. Biochem. Biophys., 209, 555. Yathindra, Ν. and Rao, V. S. R. (1970) Biopolymers, 9, 783.