Glycobtotogy vol. 1 no. 6 pp. 631-642, 1991
Molecular modelling of protein-carbohydrate interactions. Docking of monosaccharides in the binding site of concanavalin A
Anne Imberty1-5, Karl D.Hardnian2, Jeremy P.Carver3 and Serge Pe"rez* 'Laboratoire de Synthese Organique, Faculti des Sciences et Techniques, 2 me de la Houssimere, 44072 Nantes, France, 2Du Pont Merck, Experimental Station, PO Box 80228, Wilmington, DE 19880-0228, USA, 'Department of Medical Genetics and Medical Biophysics, University of Toronto, Toronto M5S 1A8, Canada and 'Laboratoire de Physicochirrue des Macromolecules, INRA, BP 527, 44072 Nantes, France *To whom correspondence should be addressed
Key words: concanavalin A/force field/glucose/lectin/mannose/ molecular modelling
Introduction An important goal of biology and chemistry is to understand the recognition phenomena occurring between a carbohydrate and a protein receptor. The primary tool for studying such types of interactions is X-ray crystallography. However, the number of crystal structures of such carbohydrate-protein complexes is rather limited [see review by Einspahr (1989)] and their resolution does not always permit a determination of the interactions at the atomic level (Quiocho, 1986). Molecular modelling provides an alternative approach towards the under© Oxford University Press
631
Downloaded from http://glycob.oxfordjournals.org/ at Cambridge University on January 5, 2015
A genera] procedure is described for addressing the computer simulation of protein—carbohydrate interactions. First, a molecular mechanical force field capable of performing conformational analysis of oligosaccharides has been derived by the addition of new parameters to the Tripos force field; it is also compatible with the simulation of protein. Second, a docking procedure which allows for a systematic exploration of the orientations and positions of a ligand into a protein cavity has been designed. This so-called 'crankshaft' method uses rotations and variations about/of virtual bonds connecting, via dummy atoms, the ligand to the protein binding site. Third, calculation of the relative stability of protein ligand complexes is performed. This strategy has been applied to search for all favourable interactions occurring between a lectin [concanavalin A (ConA)] and methyl a-D-mannopyranoside or methyl a-Dglucopyranoside. For each monosaccharide, different stable orientations and positions within the binding site can be distinguished. Among them, one corresponds to very favourable interactions, not only in terms of hydrogen bonding, but also in terms of van der Waals interactions. It corresponds precisely to the binding mode of methyl a-D-mannopyranoside into ConA as revealed by the 2.9 A resolution of the crystalline complex (Derewenda et al., 1989). Some implications of the present modelling study with respect to the molecular basis of the specificity of the interaction of lectins with various monosaccharides are presented.
standing of carbohydrate—protein interactions. However, two major difficulties appear for such computer modelling investigations. First, there is a lack of energy parameters suitable for handling both protein and carbohydrate molecules. Second, no universal docking procedure exists for searching all the possible positions and orientations of the ligand in the protein binding site. Several force field parameterizations appropriate for carbohydrates have appeared in recent years (Niketic and Rasmussen, 1977; Jeffrey and Taylor, 1980; Thogersen et al., 1982; Tvaroska and Pdrez, 1986; Ha etai, 1988). A few attempts to develop energy parameters accounting for both proteins and carbohydrates have also been made in individual laboratories (Mardsen etal, 1988; Rao etai, 1989; Stuikeprill and Meyer, 1990). With the recent increase in modelling, it has become necessary to include in some widely distributed graphics packages appropriate force fields capable of handling carbohydrate molecules. Recently, a parameterization for the study of proteins and carbohydrates has been proposed by Homans (1990) based on the AMBER force field. The SYBYL molecular graphics package includes the Tripos force field (Clark et al., 1989). This program is widely used by molecular modellers and the addition of parameters adequate for handling carbohydrate molecules is highly desirable. 'Computing the energies and structures of substrates binding to receptors is a computational nightmare'; this sentence is extracted from one of the very few publications describing a systematic computer-aided molecular recognition procedure (Lipkowitz and Zegarra, 1989). In most docking studies, a substrate is manually docked in a receptor binding site. The results of such studies must be viewed with scepticism because the preconceived ideas of the scientist will invariably bias the results and 'chemical intuition' will then be misleading. The use of a systematic and automatic procedure is absolutely required in docking studies. Several procedures have been used (Sekharudu and Rao, 1984; Lipkowitz and Zegarra, 1989). These procedures are efficient and yield good results. However, they need a certain amount of software development and cannot be used within a graphics package. The alternate procedure described in the present work makes use only of the SYBYL routine facilities and can be reproduced by any scientist. Here we describe the application of a procedure which allows for a systematic search of the positions and orientation of a carbohydrate ligand within the binding site of a protein receptor, followed by a complete energy minimization. The model chosen to assess the new parameterization is the binding of monosaccharides by concanavalin A (ConA). ConA is a legume seed lectin specific for a-D-mannose and a-D-glucose residues, and exhibits higher affinity for complex carbohydrates with a trimannosidic core compared to the monosaccharides. The structure of the native ConA has been reported from X-ray crystallographic studies at 1.75 A resolution (Hardman et al., 1982). A low-resolution crystalline structure of the complex
A.Imberty et al.
Table I. Atom types used for the description of a carbohydrate molecule along with the van der Waals parameters. The atom types indicated by (SYBYL) are the standard SYBYL atom types. The other ones, defined in this work, are specific for carbohydrate molecules
Table III. Angle-bending parameters used for carbohydrate molecules Only the parameters involving new atom types are listed here. The ones involving only standard SYBYL atom types remain untouched
e°n
Angle between Description
r (A)
K Origin (kcal/mol)
C.C C.A C.B C.3 C.2
Carbohydrate carbon sp3 a anomenc carbon (3 anomeric carbon Carbon sp3 Carbon sp2
.7 .7 .7 7 .7
0.107 0.107 0.107 0.107 0.107
This work This work This work [SYBYL] (SYBYL)
O.A O.B O.R O.3 O.2
a anomeric oxygen 0 anomenc oxygen Sugar ring oxygen Oxygen sp3 Oxygen sp2
.52 .52 .52 52 52
0.116 0.116 0.116 0.116 0.116
This work This work This work (SYBYL] (SYBYL)
H.C H
Carbohydrate hydrogen (on C atom) Hydrogen
0.15 0.042
This work [SYBYL]
N.am
Nitrogen amide
1.55 0.095
[SYBYL]
Table D. Bond-stretching parameters used for carbohydrate molecules. Only the parameters involving new atom types are listed here. The ones involving only standard SYBYL atom types remain untouched Bond between Atom 2 Atom 1
d" (A) equilibrium length
k1 (kcal/mol.A2) force constant
C.C C.C C.C C.C CC C.C C.C C.C C.C CC
C.C CA C.B C.2 O.A O.B O.R O.3 H.C N.am
1.523 1.523 .523 522 420 .420 .427 .411 .100 .449
633.6 633.6 633.6 639.0 618.9 618.9 618.9 618.9 662.4 677.6
C.A CA C.A C.A
C2 OA O.R H.C
.522 .411 .420 .100
639.0 618.9 618.9 662.4
CB C.B C.B C.B
C2 O.B O.R H.C
.522 .390 425 100
639.0 618.9 618.9 662.4
O.A OB
H H
0.972 0 972
1007.5 1007.5
ConA/methyl a-mannoside allowed the identification of the binding site (Hardman and Ainsworth, 1976). After the present work was initiated, the crystal structure of this complex was reported at 2.9 A resolution, giving more accurate details about the interactions at the atomic level (Derewenda et al., 1989). Results Tools for studying protein—carbohydrate interactions Force field development. Molecular mechanical force fields proposed in the widely used graphics packages have been 632
Atom 1
Atom 2
Atom 3
equilibrium angle
C.C C.A C.B C.2 OA OB O.R 03 H.C N.am O.A O.B O.3 H.C O.A O B. O.3 HC N am HC HC HC H.C H.C H.C H.C C2 O.A O.R H.C O.A O.R O.R H.C H.C C.2 O.B O.R H.C O.B OR O.R H.C H.C O2 N.am C.A H C.B H C.A C.B H C.2 H
C.C C.C C.C C.C C.C C.C C.C CC
C.C C.C C.C C.C C.C C.C C.C C.C CC C.C C.A C.A C.A C.A C.B C.B C.B C.B C.B C.2 O.A O.B O.R O.3 H.C N am C.C C.C CC
110.7 110.7 110.7 110.7 110 1 110 1 109 4 110.1 108.72 109 7 110.1 110.1 110.1 108.72 110.1 110.1 110.1 108.72 109.7 109.5 109.89 109.89 107.24 109.89 107.85 109.5 110.7 110.1 109 5 108 72 110.1 109.5 111.55 109.89 107.24 110.7 110.1 109.4 108.72 110.1 109.4 107.4 109.89 107.24 120.4 116.6 115.0 109.35 116 4
c.c c.c c.c c.c c.c c.c c.c cc cc c.c c.c c.c c.c c.c c.c CC
c.c c.c CA C.A C.A C.A C.A C A
C.A C.A C.A C.B C.B C.B C.B C.B CB CB C.B C.B C.2 C.2 O.A O.A OB O.B O.R O.R O.3 N.am N.am
c.c C.2 C.2 O.A O.A O.R C.C C.C C.C C.C C.2 C.2 O.B O.B O.R C.C C.C C.C C.A CC CB CC C.C C.C C.C C.C
109 35 113.8 111.9 109.35 121.9 118.4
P (kcal/mol °2) force constant 0.024 0.024 0.024 0.024 0.022 0.022 0.022 0.022 0.016 0.018 0.022 0.022 0.022 0.016 0.022 0.022 0.022 0 016 0 018 0 016 0.016 0.016 0.016 0.016 0.024 0.020 0.018 0.022 0.022 0.016 0.022 0.022 0 020 0 016 0.016 0.018 0.022 0.022 0.016 0 022 0 022 0 020 0 016 0.016 0.026 0.020 0.044 0.020 0044 0.020 0.044 0.044 0.020 0.044 0.020
developed for protein and nucleic acid geometry optimization. Carbohydrate geometry cannot be properly optimized using these energy parameters because of several characteristics due to stereoelectronic effects associated with the anomeric carbon atom and gathered under the name of 'anomeric effects'. These
Downloaded from http://glycob.oxfordjournals.org/ at Cambridge University on January 5, 2015
Atom type
New tools for modelling protein-carbohydrate interactions
effects have been the subject of a recent review (Tvaroska and Bleha, 1989). They consist of energy preference for an axial position for the substituent at the anomeric centre, alteration of C-0 bond length and orientational preference for a gauche conformation for the anomeric substituent, this latter effect being referred to as the 'eco-anomeric effect'. In order to avoid interference with parameters already in place in the 5.3 Tripos force field, new atomic types were created specific for carbohydrates. These new atomic types, along with their van der Waals parameters, are described in Table I. The van der Waals radii and associated force constant were kept identical to the Tripos values, except for the H.C atoms (hydrogen linked to a carbon) which were thought to create unacceptable steric conflicts. The new values were selected to reproduce the distance-dependent energy curve proposed by Scott and Scheraga (1965) which has been successfully tested in the oligosaccharide domain (Tvaroska and Pe"rez, 1986). Bond stretching and angle bending parameters for carbohydrates atoms are displayed in Tables II and IE. Optimal values for bond lengths and valence angles around the anomeric carbon correspond to the mean value in the solid state (Jeffrey et al., 1978). This set of optimal values is the one used in recent molecular mechanics studies of carbohydrates (Tvaroska and Pe"rez, 1986; Ha et al., 1988; Homans, 1990). The force constants were kept unchanged from the ones proposed for the same atom species in a Tripos force field. Because the exoanomeric effect and hydrogen bonding are thought to have important effects on the conformation of carbohydrates, the difficulties in energy parameterization are located in the choice of torsional parameters and in the charge
derivation. For the above-mentioned force fields (Ha et al., 1988; Homans, 1990), the charges used are too high to be compatible with the Pullman charges (Berthod and Pullman, 1965) we used for protein atoms. Therefore, in the present work, the atomic charges were derived from semi-empirical calculations with the use of the MNDO Hamiltonian (Dewar and Thiel, 1977). Calculations were performed on several monosaccharides and disaccharides in the a or the /3 configuration. Averaging over different molecules yielded the atomic charges presented in Figure 1 for the methyl a-Dmannopyranoside molecule. The torsional energy function must be able to reproduce the exo-anomeric effect (Lemieux et al., 1979), i.e. the preference of a gauche conformation for the angle with values of 60° and -60° for a and jS conformations, respectively. However, from theoretical investigations, it has been demonstrated that the magnitude of the effect should not be given too much weight (Tvaroska and Pe"rez, 1986). Other conformations than the gauche one were found experimentally, depending on the environment (Bock et al., 1978; Bonas et al., 1991; Imberty et al., 1991). Energy dependence for the set to zero. The energy curve obtained is subtracted from the quantum 633
Downloaded from http://glycob.oxfordjournals.org/ at Cambridge University on January 5, 2015
Fig. 1. SchemaDc representation of a-D-ManMe along with the labelling of the atoms Atomic charges (e) derived from MNDO calculations are displayed in parentheses under each atom name.
A.Imberty et al.
E (Kcal/mol)
3
-•
Fig. 2. PotenUal energy curves of 2-methoxy-tetrahydropyran in the a configuraUon (a) and the 0 configuration (b) as a function of the * angle. The energy calculated using the Tripos 5.3 force field extended to carbohydrates ( ) can be compared with the energy calculated through the quantum chemistry method (—) [from Tvaroska (1984)].
chemistry energy curve and the difference is fitted by linear combination of trigonometric terms. The resulting correcting function for the angle is described by the following equations: >
Ea = 0.35(1—cos4 )+2.2(1—cos3)+ 2.4(1+cos3"f>) for an a conforrner and E0 = 0.7(l-cos*) + 2.22(l-cos3*) + 2.5glucopyranos>de; a-D-ManMe, methyl a-D-mannopyranoside
References Berthod.H. and Pullman,A. (1965) Sur le calcul des caractenstiques du squelette T des molecules conjuguees. J Chem. Phys., 62, 942—946 Bock,K., Meyer.B. and Thiem.J. (1978) Structural elucidation of diastereoisomeric 1,3,6-trioxacyclooctane systems by simultaneous relaxation and double resonance experiments. Angew. Chem. Int. Ed. Engl., 17, 447—449. Bonas.G., Vignon,M.R. and P^rez.S. (1991) Structure and solution conformations of a cyclic trisaccharide from high-resolution n.m r. spectroscopy and molecular modelling. Carbohydr. Res., 211, 191-205. Brewer.C.F., Stemlicht,H , Marcus.D.M. and Grollman.A.P. (1973) Binding of l3C-enriched a-methyl-D-glucopyranoside to concanavalin A as studied by carbon magnetic resonance Proc. Nail. Acad. Sci. USA, 70, 1007-1011 Carver.J.P., Mackenzie.A.E. and Hardman.K.D. (1985) Molecular model for the complex between concanavalin A and a biantennary-complex class glycopeptide. Biopolymers, 24, 49—63. Clark.M., Cramer,R.D.,III and van Opdenbosch.N. (1989) Validation of the general purpose Tripos 5.2 force field. J. Comp. Chem., 10, 982-1012. Derewenda.Z., Yanv.J., Helliwell.J.R., Kalb.A.J., Dodson.E.J., Papiz,M.Z., Wan.T. and CampbellJ. (1989) The structure of the saccharide-binding site of concanavalin A. EMBOJ., 8, 2189-2193. Dewar.M.J.S. and Thiel.W. (1977) Ground states of molecules. 38. The MNDO method. Approximations and parameters. J. Am. Chem. Soc., 99, 4899-4907. Einspahr.H.M. (1989) Protein—carbohydrate interactions in biological systems. Trans. Am. Crystallogr. Assoc., 25, 1—22. Fuhr,B.J., Barber,B.H. and Carver J. P. (1976) Magnetic resonance studies of concanavalin A: Location of the binding site of a-methyl-D-mannopyranoside. Proc. Nail. Acad. Sci. USA, 73, 322-326. Ha.S.N., Giammona.A., Field,M. and BradyJ.W. (1988) A revised potentialenergy surface for molecular mechanics of carbohydrates. Carbohydr. Res., 180, 207-221. Hamodrakas.S.J., Alexandri.E., Troganis.A. and Stassinopoukxi.C.I (1989) Models of binding of 4'-nitrophenyI a-D-mannopyranoside to the lectin concanavalin A. Int. J. Biol. Macromol., 11, 17-22.
641
Downloaded from http://glycob.oxfordjournals.org/ at Cambridge University on January 5, 2015
The Tripos force field treats hydrogen bonds as non- directional and electrostatic in nature. During force field evaluation, H bond energies are accounted for by ignoring the van der Waals interactions between the H bond acceptors and hydrogens bonded to H bond donors. From such a treatment, it is difficult to property evaluate the magnitude of binding energy occurring from van der Waals interactions and hydrogen bond interactions, respectively. In order to do so, the non-bonded interactions between the lectin-monosacchande complex were reassessed using a different force field where the magnitude of the hydrogen bond energy was
Haneef.I. (1990) Calculation of binding energies using a robust molecular mechanics technique: Application to an antibody—antigen complex. J. Mol. Graph., 8, 45-51. Hardman.K.D. and Ainsworth,C.F. (1976) Structure of the concanavalin A—methyl a-omannopyranoside complex at 6-A resolution. Biochemistry, IS, 1120-1128. Hardman.K.D., Agarwal.R.C. and Freiser.M.J. (1982) Manganese and calcium binding sites of concanavalin A. J. Biol. Chem., 157, 69—86. Homans,S.W. (1990) A molecular mechanical force field for the conformational analysis of oligosaccharides: Comparison of theoretical and crystal structures of manal-3Man£l-4GlcNAc. Biochemistry, 29, 9110-9118. Hopfinger.A.J. and Pearistein,R.A. (1984) Molecular mechanics force-field parametrization procedure. J. Comp. Chem., 5, 486-499. Imberty,A., Delage.M.M., Boume.Y., Cambillau.C. and Pe>ez,S (1991) Data bank of three-dimensional structures of disaccharides: Part II, N-acetyllactosaminic type A'-glycans. Comparison with the crystal structure of a biantennary octosaccharide. Glycoconjugate J., in press. IUPAC-IUB Commission on Biochemical Nomenclature (1971) Abbreviations and symbols for the description of the conformation of polypeptide chain. Arch. Biochem. Biophys., 145, 4 0 5 ^ 2 1 . Jeffrey,G.A., PopleJ.A., BirkleyJ S. and Vishveshwana,S. (1978) Application of ab iniuo molecular orbital calculations to the structural moieties of carbohydrates 3. J Am. Chem. Soc., 100, 373-379. Jeffrey,G.A. and Taylor.R. (1980) The application of molecular mechanics to the structures of carbohydrates J. Comp. Chem , 1, 99-109. Lemieux.R.U., Koto.S. and Voisin.D. (1979) In Szarek.W.A. and Horton.D. (eds), Anomenc Effect. Origins and Consequences. ACS Monograph 87, Chapter 2, American Chemical Society, Washington, DC. Lipkowitz.K B. and Zegarra.R. (1989) Theoretical studies in molecular recognition: Rebek's cleft. J. Comp Chem., 10, 595-602. Marchessault.R H and Perez,S. (1979) Conformations of the hydroxymethyl group in crystalline aldohexopyranoses. Biopolymers, 18, 2369-2374. Mardsen.A., Robson,B. and ThompsonJ.C (1988) Derivations of parameters for conformational calculations of carbohydrate systems, including bacterial cell wall peptidoglycan. /. Chem. Soc., Faraday Trans. 1, 88, 2519-2536. Mayer.D., Naylor.C B , Motoc.l. and Marshall.G.R. (1987) A unique geometry of the active site of angiotensin-converting enzyme consistent with structure-activity studies. J. Comput.-Aided Mol. Design, 1, 3-16. Niketic.S.R. and Rasmussen,K. (1977) The consistent force field In Lecture Notes in Chemistry. Vol.3, Spnnger-Verlag, Berlin. Perez,S. and Delage,M.M. (1991) A database of three-dimensional structures of monosaccharides from molecular-mechanics calculations. Carbohydr. Res., 212, 523-529. Pe>ez,S. and Vergelati.C. (1987) Solid state and solution features of amylose and amylosic fragment Polym. Bull, 17, 141-148. Pe>ez,S., Imberty.A. and Meyer.C (1991) Practical tools for accurate molecular modelling of complex carbohydrates. In Reinhold.V. and Cumming.D. (eds), Separation and Characterization of Complex Oligosacchandes: Determination by MS and NMR Methods. Academic Press, Orlando, FL, in press. Poretz.R.D. and Goldstein,I.J. (1970) An examination of the topography of the saccharide binding sites of concanavalin A and of the forces involved in complexation. Biochemistry, 9, 2890—2896. Press,W.H., Flannery,B.P , Teukolsky.S A. and Vetterling.W.T. (1986) In Numerical Recipes, the Art of Scientific Computing. Cambridge University Press. Quiocho.F.A. (1986) Carbohydrate-binding proteins. Tertiary structures and protein—sugar interactions. Annu. Rev. Biochem., 55, 287-315. Rao.V.S.R., Biswas.M., Mukhopadhyay.C. and Balaji,P.V. (1989) Computer simulation of protein-carbohydrate complexes. Application to arabinosebinding protein and pea lectin. J. Mol. Struct., 194, 203-214. Rao.V.S.R., Reddy.B.V.S., Mukhopadhyay.C. and Biswas.M. (1990) Computer simulation of protein-carbohydrate complexes: Application to concanavalin A and L-arabinose-binding protein. In BradyJ.W. and French,A.D. (eds), Computer Modeling of Carbohydrate Molecules. ACS Series 430, American Chemical Society, Washington, DC, pp. 361-376. Scott,R.A and Scheraga,H.A. (1965) Methods for calculating internal rotation barriers. J. Chem. Phys., 42, 2209-2215. Sekharudu.Y.C. and Rao.V.S.R. (1984) Theoretical studies on the modes of binding of some of the derivatives of D-mannose to concanavalin A. Int J. Biol. Macromol., 6, 337-345. Stuikeprill.R. and Meyer.B (1990) A new force-field program for the calculation of glycopeptides and its application to a heptacosapeptidedecasaccharide of immunoglobulin G,. Eur. J. Biochem., 194, 903-919
A.Imberty el at. Thogersen.H., Lemieux.R.U., Bock.K. and Meyer.B. (1982) Further justification for the exo-anomeric effect. Conformational analysis based on nuclear magnetic resonance spectroscopy of oligosaccharides. Can. J. Chem., 60, 44-57. Tnpos Associates (1988) 1699 S. Hanley Road, Suite 303, St. Louis, MO 63144. SYBYL 5.2 Tvaroska.I. (1984) An attempt to derive the potential function for evaluation of the energy associated with the exo-anomeric effect. Carbohydr. Res., 125, 155-160. Tvaroska,I. and Bleha.T. (1989) Anomeric and exo-anomenc effects in carbohydrate chemistry Adv. Carbohydr. Chem. Biochem., 47, 45—123. Tvaroska.I. and PeYez.S. (1986) Conformational-energy calculations for oligosaccharides: A comparison of methods and a strategy of calculation.
Carbohydr. Res., 149, 389^10.
Villafranca.J.J. and Viola.R.E. (1974) The use of 13C spin lattice relaxation times to study the interaction of or-methyl-D-glucopyranoside with concanavalin A. Arch. Biochem Biophys., 160, 465—468. White,D.N.J. andGuy.M.H.P (1975) The molecular conformation of cyclodi/3-alanyl. J. Chem. Soc. Perkin Trans. 2, 43^t6. Received on August 15, 1991; accepted on September 18, 1991 Downloaded from http://glycob.oxfordjournals.org/ at Cambridge University on January 5, 2015
642
Molecular modelling of protein-carbohydrate interactions. Docking of monosaccharides in the binding site of concanavalin A
Anne Imberty1-5, Karl D.Hardnian2, Jeremy P.Carver3 and Serge Pe"rez* 'Laboratoire de Synthese Organique, Faculti des Sciences et Techniques, 2 me de la Houssimere, 44072 Nantes, France, 2Du Pont Merck, Experimental Station, PO Box 80228, Wilmington, DE 19880-0228, USA, 'Department of Medical Genetics and Medical Biophysics, University of Toronto, Toronto M5S 1A8, Canada and 'Laboratoire de Physicochirrue des Macromolecules, INRA, BP 527, 44072 Nantes, France *To whom correspondence should be addressed
Key words: concanavalin A/force field/glucose/lectin/mannose/ molecular modelling
Introduction An important goal of biology and chemistry is to understand the recognition phenomena occurring between a carbohydrate and a protein receptor. The primary tool for studying such types of interactions is X-ray crystallography. However, the number of crystal structures of such carbohydrate-protein complexes is rather limited [see review by Einspahr (1989)] and their resolution does not always permit a determination of the interactions at the atomic level (Quiocho, 1986). Molecular modelling provides an alternative approach towards the under© Oxford University Press
631
Downloaded from http://glycob.oxfordjournals.org/ at Cambridge University on January 5, 2015
A genera] procedure is described for addressing the computer simulation of protein—carbohydrate interactions. First, a molecular mechanical force field capable of performing conformational analysis of oligosaccharides has been derived by the addition of new parameters to the Tripos force field; it is also compatible with the simulation of protein. Second, a docking procedure which allows for a systematic exploration of the orientations and positions of a ligand into a protein cavity has been designed. This so-called 'crankshaft' method uses rotations and variations about/of virtual bonds connecting, via dummy atoms, the ligand to the protein binding site. Third, calculation of the relative stability of protein ligand complexes is performed. This strategy has been applied to search for all favourable interactions occurring between a lectin [concanavalin A (ConA)] and methyl a-D-mannopyranoside or methyl a-Dglucopyranoside. For each monosaccharide, different stable orientations and positions within the binding site can be distinguished. Among them, one corresponds to very favourable interactions, not only in terms of hydrogen bonding, but also in terms of van der Waals interactions. It corresponds precisely to the binding mode of methyl a-D-mannopyranoside into ConA as revealed by the 2.9 A resolution of the crystalline complex (Derewenda et al., 1989). Some implications of the present modelling study with respect to the molecular basis of the specificity of the interaction of lectins with various monosaccharides are presented.
standing of carbohydrate—protein interactions. However, two major difficulties appear for such computer modelling investigations. First, there is a lack of energy parameters suitable for handling both protein and carbohydrate molecules. Second, no universal docking procedure exists for searching all the possible positions and orientations of the ligand in the protein binding site. Several force field parameterizations appropriate for carbohydrates have appeared in recent years (Niketic and Rasmussen, 1977; Jeffrey and Taylor, 1980; Thogersen et al., 1982; Tvaroska and Pdrez, 1986; Ha etai, 1988). A few attempts to develop energy parameters accounting for both proteins and carbohydrates have also been made in individual laboratories (Mardsen etal, 1988; Rao etai, 1989; Stuikeprill and Meyer, 1990). With the recent increase in modelling, it has become necessary to include in some widely distributed graphics packages appropriate force fields capable of handling carbohydrate molecules. Recently, a parameterization for the study of proteins and carbohydrates has been proposed by Homans (1990) based on the AMBER force field. The SYBYL molecular graphics package includes the Tripos force field (Clark et al., 1989). This program is widely used by molecular modellers and the addition of parameters adequate for handling carbohydrate molecules is highly desirable. 'Computing the energies and structures of substrates binding to receptors is a computational nightmare'; this sentence is extracted from one of the very few publications describing a systematic computer-aided molecular recognition procedure (Lipkowitz and Zegarra, 1989). In most docking studies, a substrate is manually docked in a receptor binding site. The results of such studies must be viewed with scepticism because the preconceived ideas of the scientist will invariably bias the results and 'chemical intuition' will then be misleading. The use of a systematic and automatic procedure is absolutely required in docking studies. Several procedures have been used (Sekharudu and Rao, 1984; Lipkowitz and Zegarra, 1989). These procedures are efficient and yield good results. However, they need a certain amount of software development and cannot be used within a graphics package. The alternate procedure described in the present work makes use only of the SYBYL routine facilities and can be reproduced by any scientist. Here we describe the application of a procedure which allows for a systematic search of the positions and orientation of a carbohydrate ligand within the binding site of a protein receptor, followed by a complete energy minimization. The model chosen to assess the new parameterization is the binding of monosaccharides by concanavalin A (ConA). ConA is a legume seed lectin specific for a-D-mannose and a-D-glucose residues, and exhibits higher affinity for complex carbohydrates with a trimannosidic core compared to the monosaccharides. The structure of the native ConA has been reported from X-ray crystallographic studies at 1.75 A resolution (Hardman et al., 1982). A low-resolution crystalline structure of the complex
A.Imberty et al.
Table I. Atom types used for the description of a carbohydrate molecule along with the van der Waals parameters. The atom types indicated by (SYBYL) are the standard SYBYL atom types. The other ones, defined in this work, are specific for carbohydrate molecules
Table III. Angle-bending parameters used for carbohydrate molecules Only the parameters involving new atom types are listed here. The ones involving only standard SYBYL atom types remain untouched
e°n
Angle between Description
r (A)
K Origin (kcal/mol)
C.C C.A C.B C.3 C.2
Carbohydrate carbon sp3 a anomenc carbon (3 anomeric carbon Carbon sp3 Carbon sp2
.7 .7 .7 7 .7
0.107 0.107 0.107 0.107 0.107
This work This work This work [SYBYL] (SYBYL)
O.A O.B O.R O.3 O.2
a anomeric oxygen 0 anomenc oxygen Sugar ring oxygen Oxygen sp3 Oxygen sp2
.52 .52 .52 52 52
0.116 0.116 0.116 0.116 0.116
This work This work This work (SYBYL] (SYBYL)
H.C H
Carbohydrate hydrogen (on C atom) Hydrogen
0.15 0.042
This work [SYBYL]
N.am
Nitrogen amide
1.55 0.095
[SYBYL]
Table D. Bond-stretching parameters used for carbohydrate molecules. Only the parameters involving new atom types are listed here. The ones involving only standard SYBYL atom types remain untouched Bond between Atom 2 Atom 1
d" (A) equilibrium length
k1 (kcal/mol.A2) force constant
C.C C.C C.C C.C CC C.C C.C C.C C.C CC
C.C CA C.B C.2 O.A O.B O.R O.3 H.C N.am
1.523 1.523 .523 522 420 .420 .427 .411 .100 .449
633.6 633.6 633.6 639.0 618.9 618.9 618.9 618.9 662.4 677.6
C.A CA C.A C.A
C2 OA O.R H.C
.522 .411 .420 .100
639.0 618.9 618.9 662.4
CB C.B C.B C.B
C2 O.B O.R H.C
.522 .390 425 100
639.0 618.9 618.9 662.4
O.A OB
H H
0.972 0 972
1007.5 1007.5
ConA/methyl a-mannoside allowed the identification of the binding site (Hardman and Ainsworth, 1976). After the present work was initiated, the crystal structure of this complex was reported at 2.9 A resolution, giving more accurate details about the interactions at the atomic level (Derewenda et al., 1989). Results Tools for studying protein—carbohydrate interactions Force field development. Molecular mechanical force fields proposed in the widely used graphics packages have been 632
Atom 1
Atom 2
Atom 3
equilibrium angle
C.C C.A C.B C.2 OA OB O.R 03 H.C N.am O.A O.B O.3 H.C O.A O B. O.3 HC N am HC HC HC H.C H.C H.C H.C C2 O.A O.R H.C O.A O.R O.R H.C H.C C.2 O.B O.R H.C O.B OR O.R H.C H.C O2 N.am C.A H C.B H C.A C.B H C.2 H
C.C C.C C.C C.C C.C C.C C.C CC
C.C C.C C.C C.C C.C C.C C.C C.C CC C.C C.A C.A C.A C.A C.B C.B C.B C.B C.B C.2 O.A O.B O.R O.3 H.C N am C.C C.C CC
110.7 110.7 110.7 110.7 110 1 110 1 109 4 110.1 108.72 109 7 110.1 110.1 110.1 108.72 110.1 110.1 110.1 108.72 109.7 109.5 109.89 109.89 107.24 109.89 107.85 109.5 110.7 110.1 109 5 108 72 110.1 109.5 111.55 109.89 107.24 110.7 110.1 109.4 108.72 110.1 109.4 107.4 109.89 107.24 120.4 116.6 115.0 109.35 116 4
c.c c.c c.c c.c c.c c.c c.c cc cc c.c c.c c.c c.c c.c c.c CC
c.c c.c CA C.A C.A C.A C.A C A
C.A C.A C.A C.B C.B C.B C.B C.B CB CB C.B C.B C.2 C.2 O.A O.A OB O.B O.R O.R O.3 N.am N.am
c.c C.2 C.2 O.A O.A O.R C.C C.C C.C C.C C.2 C.2 O.B O.B O.R C.C C.C C.C C.A CC CB CC C.C C.C C.C C.C
109 35 113.8 111.9 109.35 121.9 118.4
P (kcal/mol °2) force constant 0.024 0.024 0.024 0.024 0.022 0.022 0.022 0.022 0.016 0.018 0.022 0.022 0.022 0.016 0.022 0.022 0.022 0 016 0 018 0 016 0.016 0.016 0.016 0.016 0.024 0.020 0.018 0.022 0.022 0.016 0.022 0.022 0 020 0 016 0.016 0.018 0.022 0.022 0.016 0 022 0 022 0 020 0 016 0.016 0.026 0.020 0.044 0.020 0044 0.020 0.044 0.044 0.020 0.044 0.020
developed for protein and nucleic acid geometry optimization. Carbohydrate geometry cannot be properly optimized using these energy parameters because of several characteristics due to stereoelectronic effects associated with the anomeric carbon atom and gathered under the name of 'anomeric effects'. These
Downloaded from http://glycob.oxfordjournals.org/ at Cambridge University on January 5, 2015
Atom type
New tools for modelling protein-carbohydrate interactions
effects have been the subject of a recent review (Tvaroska and Bleha, 1989). They consist of energy preference for an axial position for the substituent at the anomeric centre, alteration of C-0 bond length and orientational preference for a gauche conformation for the anomeric substituent, this latter effect being referred to as the 'eco-anomeric effect'. In order to avoid interference with parameters already in place in the 5.3 Tripos force field, new atomic types were created specific for carbohydrates. These new atomic types, along with their van der Waals parameters, are described in Table I. The van der Waals radii and associated force constant were kept identical to the Tripos values, except for the H.C atoms (hydrogen linked to a carbon) which were thought to create unacceptable steric conflicts. The new values were selected to reproduce the distance-dependent energy curve proposed by Scott and Scheraga (1965) which has been successfully tested in the oligosaccharide domain (Tvaroska and Pe"rez, 1986). Bond stretching and angle bending parameters for carbohydrates atoms are displayed in Tables II and IE. Optimal values for bond lengths and valence angles around the anomeric carbon correspond to the mean value in the solid state (Jeffrey et al., 1978). This set of optimal values is the one used in recent molecular mechanics studies of carbohydrates (Tvaroska and Pe"rez, 1986; Ha et al., 1988; Homans, 1990). The force constants were kept unchanged from the ones proposed for the same atom species in a Tripos force field. Because the exoanomeric effect and hydrogen bonding are thought to have important effects on the conformation of carbohydrates, the difficulties in energy parameterization are located in the choice of torsional parameters and in the charge
derivation. For the above-mentioned force fields (Ha et al., 1988; Homans, 1990), the charges used are too high to be compatible with the Pullman charges (Berthod and Pullman, 1965) we used for protein atoms. Therefore, in the present work, the atomic charges were derived from semi-empirical calculations with the use of the MNDO Hamiltonian (Dewar and Thiel, 1977). Calculations were performed on several monosaccharides and disaccharides in the a or the /3 configuration. Averaging over different molecules yielded the atomic charges presented in Figure 1 for the methyl a-Dmannopyranoside molecule. The torsional energy function must be able to reproduce the exo-anomeric effect (Lemieux et al., 1979), i.e. the preference of a gauche conformation for the angle with values of 60° and -60° for a and jS conformations, respectively. However, from theoretical investigations, it has been demonstrated that the magnitude of the effect should not be given too much weight (Tvaroska and Pe"rez, 1986). Other conformations than the gauche one were found experimentally, depending on the environment (Bock et al., 1978; Bonas et al., 1991; Imberty et al., 1991). Energy dependence for the set to zero. The energy curve obtained is subtracted from the quantum 633
Downloaded from http://glycob.oxfordjournals.org/ at Cambridge University on January 5, 2015
Fig. 1. SchemaDc representation of a-D-ManMe along with the labelling of the atoms Atomic charges (e) derived from MNDO calculations are displayed in parentheses under each atom name.
A.Imberty et al.
E (Kcal/mol)
3
-•
Fig. 2. PotenUal energy curves of 2-methoxy-tetrahydropyran in the a configuraUon (a) and the 0 configuration (b) as a function of the * angle. The energy calculated using the Tripos 5.3 force field extended to carbohydrates ( ) can be compared with the energy calculated through the quantum chemistry method (—) [from Tvaroska (1984)].
chemistry energy curve and the difference is fitted by linear combination of trigonometric terms. The resulting correcting function for the angle is described by the following equations: >
Ea = 0.35(1—cos4 )+2.2(1—cos3)+ 2.4(1+cos3"f>) for an a conforrner and E0 = 0.7(l-cos*) + 2.22(l-cos3*) + 2.5glucopyranos>de; a-D-ManMe, methyl a-D-mannopyranoside
References Berthod.H. and Pullman,A. (1965) Sur le calcul des caractenstiques du squelette T des molecules conjuguees. J Chem. Phys., 62, 942—946 Bock,K., Meyer.B. and Thiem.J. (1978) Structural elucidation of diastereoisomeric 1,3,6-trioxacyclooctane systems by simultaneous relaxation and double resonance experiments. Angew. Chem. Int. Ed. Engl., 17, 447—449. Bonas.G., Vignon,M.R. and P^rez.S. (1991) Structure and solution conformations of a cyclic trisaccharide from high-resolution n.m r. spectroscopy and molecular modelling. Carbohydr. Res., 211, 191-205. Brewer.C.F., Stemlicht,H , Marcus.D.M. and Grollman.A.P. (1973) Binding of l3C-enriched a-methyl-D-glucopyranoside to concanavalin A as studied by carbon magnetic resonance Proc. Nail. Acad. Sci. USA, 70, 1007-1011 Carver.J.P., Mackenzie.A.E. and Hardman.K.D. (1985) Molecular model for the complex between concanavalin A and a biantennary-complex class glycopeptide. Biopolymers, 24, 49—63. Clark.M., Cramer,R.D.,III and van Opdenbosch.N. (1989) Validation of the general purpose Tripos 5.2 force field. J. Comp. Chem., 10, 982-1012. Derewenda.Z., Yanv.J., Helliwell.J.R., Kalb.A.J., Dodson.E.J., Papiz,M.Z., Wan.T. and CampbellJ. (1989) The structure of the saccharide-binding site of concanavalin A. EMBOJ., 8, 2189-2193. Dewar.M.J.S. and Thiel.W. (1977) Ground states of molecules. 38. The MNDO method. Approximations and parameters. J. Am. Chem. Soc., 99, 4899-4907. Einspahr.H.M. (1989) Protein—carbohydrate interactions in biological systems. Trans. Am. Crystallogr. Assoc., 25, 1—22. Fuhr,B.J., Barber,B.H. and Carver J. P. (1976) Magnetic resonance studies of concanavalin A: Location of the binding site of a-methyl-D-mannopyranoside. Proc. Nail. Acad. Sci. USA, 73, 322-326. Ha.S.N., Giammona.A., Field,M. and BradyJ.W. (1988) A revised potentialenergy surface for molecular mechanics of carbohydrates. Carbohydr. Res., 180, 207-221. Hamodrakas.S.J., Alexandri.E., Troganis.A. and Stassinopoukxi.C.I (1989) Models of binding of 4'-nitrophenyI a-D-mannopyranoside to the lectin concanavalin A. Int. J. Biol. Macromol., 11, 17-22.
641
Downloaded from http://glycob.oxfordjournals.org/ at Cambridge University on January 5, 2015
The Tripos force field treats hydrogen bonds as non- directional and electrostatic in nature. During force field evaluation, H bond energies are accounted for by ignoring the van der Waals interactions between the H bond acceptors and hydrogens bonded to H bond donors. From such a treatment, it is difficult to property evaluate the magnitude of binding energy occurring from van der Waals interactions and hydrogen bond interactions, respectively. In order to do so, the non-bonded interactions between the lectin-monosacchande complex were reassessed using a different force field where the magnitude of the hydrogen bond energy was
Haneef.I. (1990) Calculation of binding energies using a robust molecular mechanics technique: Application to an antibody—antigen complex. J. Mol. Graph., 8, 45-51. Hardman.K.D. and Ainsworth,C.F. (1976) Structure of the concanavalin A—methyl a-omannopyranoside complex at 6-A resolution. Biochemistry, IS, 1120-1128. Hardman.K.D., Agarwal.R.C. and Freiser.M.J. (1982) Manganese and calcium binding sites of concanavalin A. J. Biol. Chem., 157, 69—86. Homans,S.W. (1990) A molecular mechanical force field for the conformational analysis of oligosaccharides: Comparison of theoretical and crystal structures of manal-3Man£l-4GlcNAc. Biochemistry, 29, 9110-9118. Hopfinger.A.J. and Pearistein,R.A. (1984) Molecular mechanics force-field parametrization procedure. J. Comp. Chem., 5, 486-499. Imberty,A., Delage.M.M., Boume.Y., Cambillau.C. and Pe>ez,S (1991) Data bank of three-dimensional structures of disaccharides: Part II, N-acetyllactosaminic type A'-glycans. Comparison with the crystal structure of a biantennary octosaccharide. Glycoconjugate J., in press. IUPAC-IUB Commission on Biochemical Nomenclature (1971) Abbreviations and symbols for the description of the conformation of polypeptide chain. Arch. Biochem. Biophys., 145, 4 0 5 ^ 2 1 . Jeffrey,G.A., PopleJ.A., BirkleyJ S. and Vishveshwana,S. (1978) Application of ab iniuo molecular orbital calculations to the structural moieties of carbohydrates 3. J Am. Chem. Soc., 100, 373-379. Jeffrey,G.A. and Taylor.R. (1980) The application of molecular mechanics to the structures of carbohydrates J. Comp. Chem , 1, 99-109. Lemieux.R.U., Koto.S. and Voisin.D. (1979) In Szarek.W.A. and Horton.D. (eds), Anomenc Effect. Origins and Consequences. ACS Monograph 87, Chapter 2, American Chemical Society, Washington, DC. Lipkowitz.K B. and Zegarra.R. (1989) Theoretical studies in molecular recognition: Rebek's cleft. J. Comp Chem., 10, 595-602. Marchessault.R H and Perez,S. (1979) Conformations of the hydroxymethyl group in crystalline aldohexopyranoses. Biopolymers, 18, 2369-2374. Mardsen.A., Robson,B. and ThompsonJ.C (1988) Derivations of parameters for conformational calculations of carbohydrate systems, including bacterial cell wall peptidoglycan. /. Chem. Soc., Faraday Trans. 1, 88, 2519-2536. Mayer.D., Naylor.C B , Motoc.l. and Marshall.G.R. (1987) A unique geometry of the active site of angiotensin-converting enzyme consistent with structure-activity studies. J. Comput.-Aided Mol. Design, 1, 3-16. Niketic.S.R. and Rasmussen,K. (1977) The consistent force field In Lecture Notes in Chemistry. Vol.3, Spnnger-Verlag, Berlin. Perez,S. and Delage,M.M. (1991) A database of three-dimensional structures of monosaccharides from molecular-mechanics calculations. Carbohydr. Res., 212, 523-529. Pe>ez,S. and Vergelati.C. (1987) Solid state and solution features of amylose and amylosic fragment Polym. Bull, 17, 141-148. Pe>ez,S., Imberty.A. and Meyer.C (1991) Practical tools for accurate molecular modelling of complex carbohydrates. In Reinhold.V. and Cumming.D. (eds), Separation and Characterization of Complex Oligosacchandes: Determination by MS and NMR Methods. Academic Press, Orlando, FL, in press. Poretz.R.D. and Goldstein,I.J. (1970) An examination of the topography of the saccharide binding sites of concanavalin A and of the forces involved in complexation. Biochemistry, 9, 2890—2896. Press,W.H., Flannery,B.P , Teukolsky.S A. and Vetterling.W.T. (1986) In Numerical Recipes, the Art of Scientific Computing. Cambridge University Press. Quiocho.F.A. (1986) Carbohydrate-binding proteins. Tertiary structures and protein—sugar interactions. Annu. Rev. Biochem., 55, 287-315. Rao.V.S.R., Biswas.M., Mukhopadhyay.C. and Balaji,P.V. (1989) Computer simulation of protein-carbohydrate complexes. Application to arabinosebinding protein and pea lectin. J. Mol. Struct., 194, 203-214. Rao.V.S.R., Reddy.B.V.S., Mukhopadhyay.C. and Biswas.M. (1990) Computer simulation of protein-carbohydrate complexes: Application to concanavalin A and L-arabinose-binding protein. In BradyJ.W. and French,A.D. (eds), Computer Modeling of Carbohydrate Molecules. ACS Series 430, American Chemical Society, Washington, DC, pp. 361-376. Scott,R.A and Scheraga,H.A. (1965) Methods for calculating internal rotation barriers. J. Chem. Phys., 42, 2209-2215. Sekharudu.Y.C. and Rao.V.S.R. (1984) Theoretical studies on the modes of binding of some of the derivatives of D-mannose to concanavalin A. Int J. Biol. Macromol., 6, 337-345. Stuikeprill.R. and Meyer.B (1990) A new force-field program for the calculation of glycopeptides and its application to a heptacosapeptidedecasaccharide of immunoglobulin G,. Eur. J. Biochem., 194, 903-919
A.Imberty el at. Thogersen.H., Lemieux.R.U., Bock.K. and Meyer.B. (1982) Further justification for the exo-anomeric effect. Conformational analysis based on nuclear magnetic resonance spectroscopy of oligosaccharides. Can. J. Chem., 60, 44-57. Tnpos Associates (1988) 1699 S. Hanley Road, Suite 303, St. Louis, MO 63144. SYBYL 5.2 Tvaroska.I. (1984) An attempt to derive the potential function for evaluation of the energy associated with the exo-anomeric effect. Carbohydr. Res., 125, 155-160. Tvaroska,I. and Bleha.T. (1989) Anomeric and exo-anomenc effects in carbohydrate chemistry Adv. Carbohydr. Chem. Biochem., 47, 45—123. Tvaroska.I. and PeYez.S. (1986) Conformational-energy calculations for oligosaccharides: A comparison of methods and a strategy of calculation.
Carbohydr. Res., 149, 389^10.
Villafranca.J.J. and Viola.R.E. (1974) The use of 13C spin lattice relaxation times to study the interaction of or-methyl-D-glucopyranoside with concanavalin A. Arch. Biochem Biophys., 160, 465—468. White,D.N.J. andGuy.M.H.P (1975) The molecular conformation of cyclodi/3-alanyl. J. Chem. Soc. Perkin Trans. 2, 43^t6. Received on August 15, 1991; accepted on September 18, 1991 Downloaded from http://glycob.oxfordjournals.org/ at Cambridge University on January 5, 2015
642
