Mode of Molecular Recognition of l -Fucose by Fucose-Binding Legume Lectins

Biochemical and Biophysical Research Communications 268, 262–267 (2000) doi:10.1006/bbrc.2000.2110, available online at http://www.idealibrary.com on

Mode of Molecular Recognition of L-Fucose by Fucose-Binding Legume Lectins Celestine J. Thomas* and Avadhesha Surolia* ,† ,1 *Molecular Biophysics Unit, Indian Institute of Science, Bangalore 560 012, India; and †Jawaharlal Nehru Centre for Advanced Scientific Research, Jakkur Campus, Jakkur P.O., Bangalore 560 012, India

Received December 28, 1999

Recognition of cell surface carbohydrate moieties by lectins plays a vital role in many a biological process. Fucosyated residues are often implicated as key recognition markers in many cellular processes. In particular, the aspects of molecular recognition of fucose by fucose-bindinglectins UEA 1 and LTA pose a special case because no crystal structure of these lectins is available. The study was conducted to elucidate the process of recognition of L-fucose by UEA1 and LTA by correlating structure-based sequence alignment and other available biochemical/biophysical data. The study points out that the mode of recognition of L-fucose is coordinated by the invariant triad of residues the asparagine 137, glycine 105, and aspartate 87. The major hydrophobic stacking residue in this case is the tyrosine 220. The study also reiterates the key role of the conserved triad of residues in the combining site which is a common feature for all legume lectins whose crystal structures are known. © 2000 Academic Press

Lectins, carbohydrate-specific proteins, have been detected in most organisms ranging from viruses and bacteria to plants and animals (1). Each lectin molecule contains typically two or more carbohydratecombining sites; i.e., they are multivalent (1–3). Therefore, when lectins combine with sugars on cell surfaces, they cross-link and agglutinate cells. Due to their ability to bind to cell surface carbohydrates they are being widely used in cell biology, biochemistry, and histochemistry to characterize cell surface carbohydrates and glycoproteins (4 – 8). Lectins require configurational and structural complementarity of sugars for interaction to occur and have been employed as tools for exploring the structure and dynamics of cell surfaces (2, 3, 9, 10). Among lectins those from legumes have been most extensively studied. Consequently, 1

To whom correspondence should be addressed. E-mail: surolia@ mbu.iisc.ernet.in. 0006-291X/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.

structural information on legume lectins that recognize a wide variety of carbohydrate determinants is now available excepting those with specificities toward L-fucose of which Lotus tetragonolobus (LTA) and Ulex europeus agglutinin 1 (UEA 1) constitute the leading members of this class (11–17). Fucosylated sugars have been found to be expressed often on cells undergoing differentiation or metastasis but not on dormant cells (18). Using UEA 1 and LTA fucosylated carbohydrate structures have been correlated with metastastic potential of the tumor on several carcinomas (19). Though both the lectins recognize L-fucose subtle differences in their recognition mode has been very useful to characterize very fine differences in glycan structures of cell surface glycoconjugates (20). Availability of the high resolution structures of legume lectins complexed with a wide variety of carbohydrate ligands together with insights from a correlation between the organization of their combining site loops with their carbohydrate specificities provides us an excellent platform to attempt to define the framework for molecular recognition of L-fucose by UEA 1 and LTA. MATERIALS AND METHODS The coordinates for the crystal structure of the legume lectins were obtained from PDB (www.rcsb.org) (Table 1). The PDB codes were converted to FASTA format and the primary structures of the lectins were aligned using the CLUSTAL X. The results of the pairwise alignment were compared and the sequences which showed less than 25% identity were ignored. The structure based alignment was obtained by pairwise superimposition of the C ␣ atoms. A pair of structures were considered to be superposed when the rms deviations in the positions of the C ␣ atoms were low. Loop regions that do not correspond to any other portions were aligned manually (Table 2). The molecules were visualised and aligned using INSIGHT II (BioSym Technologies), such that the key conserved residues in the binding site superimpose. The key contacts made by the sugar to the lectin binding site were calculated using INSIGHT II. The figures were generated by using either INSIGHT II or WEBLAB viewer. All calculations were performed on supercomputers platforms, at the Supercomputer Educational Research Facility, Indian Institute of Science, Bangalore, India.

262

Vol. 268, No. 2, 2000

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS TABLE 1

Sources of the Structure of the Legume Lectins Lectin

Source

Specificity

PDB Code

Reference

WBA1 ECorl GSIV PNA SBA PHAI CON A DGL CON LENL LOLI LOLII FRIL DBL DB58 PSL

Psophocarpus tetragonolobus Erythrina corallodendron Griffonia simplifolia Arachis hypogaea Glycine max Phaseoulus vulgaris Conavalia ensiformis Dioclea grandiflora Conavalia brasiliensis Lens culinaris Lathyrus ochrus Lathyrus ochrus Dolicos lablab Dolicos biflorus Dolicos biflorus Pisum sativum

GalNAc GalNAc GalNAc Gal GalNAc Complex Man/Glc Man/Glc Man/Glc Man/Glc Man/Glc Man/Glc Man GalNAc GalNac Man/Glc

1WBA 1LTE 1LEC 2PEL 1SBA 1FAT 5CAN 1DGL 1AZD 1LES 1LOA 1LGB 1QMO 1BJQ 1LUL 1RIN

Prabu, M. M., et al. (1998) J. Mol. Biol. 276, 787 Shaanan, B., et al. (1991) Science 254, 862 Delbaere, L., et al. (1993) J. Mol. Biol. 68, 1116 Banerjee, R., et al. (1994) Proc. Natl. Acad. Sci. USA 91, 227 Dessen, A., et al. (1995) Biochemistry 34, 4933 Hamelryck, T. W., et al. (1996) J. Biol. Chem. 271, 20479 Hardman, K. D., et al. (1972) Biochemistry 11, 4910 Rozwarski, D. A., et al. (1998) J. Biol. Chem. 273, 32818 Sanz-Aparicio, J., et al. (1997) FEBS Lett. 405, 114 Casset, F., et al. (1995) J. Biol. Chem. 270, 25619 Bourne, Y., et al. (1990) Proteins: Struct. Funct. Genet. 8, 365 Bourne, Y., et al. (1994) Structure 2, 209 Hamelryck, T. W., et al. (1999) To be published Hamelryck, T. W., et al. (1998) J. Mol. Biol. 286, 1161 Hamelryck, T. W., et al. (1998) J. Mol. Biol. 286, 1161 Rini, J. M., et al. (1992) J. Biol. Chem. 268, 10126

Note. All pdb codes were downloaded from www.resb.org/pdb. The pdb files were visualized using Insight II. In cases, where more than one crystal structure is known, the structure solved with the carbohydrate bound lectin was analyzed.

RESULTS AND DISCUSSION General Framework for Legume Lectin–Carbohydrate Interactions The combining site of the legume lectins is made up of four loops (A, B, C, and D) that come from four sequentially disparate regions of these proteins (21, 22). In the recognition of carbohydrates with high specificity by lectins two sets of residues play a pivotal role. These include an aspartate in loop A, glycine or arginine in loop B, an asparagine residue from loop C which hydrogen bond to a distinct pair of sugar hydroxyls and an aromatic residue from loop C that stacks with the pyranosyl ring of the saccharide. In addition, in many legume lectins a backbone NH group from a residue from the loop D hydrogen bonds to a sugar hydroxyl. The binding site is tailored such that a monosaccharide is permitted in an orientation that positions: (a) an equatorial hydroxyl group in the space between the conserved residues from loop A (an aspartate OD2), loop B (glycine NH) and loop C (an asparagine ND2) (b) an axial hydroxyl or a hydroxyl-methyl group as another H-bonding partner to OD1 of the aspartate residue in loop A (23). Additionally, our earlier assertion that the site is designed to exclude recognition of two adjacent equatorially oriented hydroxyl groups in the complementary monosaccharide is strengthened by the structures of lectin-sugar complexes reported so far. Recognition of L-Fucose UEA 1 and LTA bind to L-fucose with differing affinities. LTA exhibits approximately 3.5-fold higher affinity for the sugar over UEA 1 (20). Earlier sequence

alignment analysis shows that in both UEA 1 and LTA loop D has a gap of six residues while the loop C has a gap of one residue in the former and of four residues in the latter. Typically conserved residues in the loops A, B, and C of legume lectins together with the conserved hydrophobic residues in loop C, and the invariant NH group in loop D, are present in both the lectins (Table 2). The hydrophobic moiety in the UEA 1 is an isoleucine instead of the usual aromatic residues and the invariant asparagine (in loop C) is separated from the hydrophobic residue by four residues instead of just one, as exhibited by all the other legume lectins. LTA, on the other hand, follows the general trend noted with other legume lectins in these respects. The longer loop C in UEA 1 might provide a shallower binding pocket for L-fucose in UEA 1. The crystal structure of all the legume lectins solved so far is in consonance with this suggestion. UEA 1 presents a modified binding pocket to fucose. The gap of four residues between the aliphatic hydrophobic residue and the invariant asparagine residue in UEA 1 would necessarily present a shallower binding cleft as compared to that in LTA. This probably might account for the different affinities exhibited by these two lectins for the various sugar ligands. Earlier modeling work has predicted that the serine 132 separated by a residue from isoleucine in loop C is involved in the binding of fucose through a H-bond with O3 of L-fucose (23). Moreover, it was also predicted that the fucose can theoretically bind in four different orientations. Of the four possible orientations two of the solutions are not acceptable as they represent the binding of two adjacent equatorial OH groups inconsistent with the general design of the binding pocket in legume lectins.

263

Vol. 268, No. 2, 2000

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS TABLE 2

Structure-Based Alignment of the Loop Regions of the Legume Lectins of Different Monosaccharide Specificities

Note. The structure-based alignment was obtained by pairwise superimposition of the C ␣ atoms. A pair of structures was considered to be superposed when the rms deviations in the positions of the C atoms were low. Loop regions that do not correspond to any other structural components were aligned manually. * Because the crystal structures are not available for UEAI and LTA, residues in their combining sites are italicized.

The role of serine suggested in the above work, in the binding process is at best trivial in comparison to asparagine in loop C as explained below. Replacement of asparagine in loop C even by aspartic acid in legume lectins leads to a complete loss of activity. Moreover, occurrence of serine at the position of the conserved asparagine in UEA 1 does not preclude the possibility of the participation of the latter in the binding process, as the addition of the four residue in loop C, GSPV, still should allow this residue to loop back and come close enough to the sugar in the binding pocket to make contacts with the sugar moiety as a consequence of a sharp turn introduced by glycine and proline preceding it. A representative diagram of the sugar binding site of WBA 1 (24) (Fig. 1) interacting with Me␣Gal shows the critical role played by the asparagine from the loop C. The ASN 128 ND2 binds to the O3 of galactose through a direct H-bond and through a water-mediated H-bond to the O2 of the sugar. Table 3 shows the key contacts made by this critical residue, in loop C, for

binding to sugars in various legume lectins seen in the crystal structures. UEA 1 also has another unique feature, loop C lacks an aromatic hydrophobic residue common to most legume lectins. This residue stacks with the hydrophobic B face providing for the stability of complexes with monosaccharides. As suggested earlier Phaseolus lectins are unable to bind to a monosaccharide as they have a leucine instead of the usual aromatic residue (25). While the lectins from Dolichos biflorus fail to recognize galactose because of the leucine there (26). The presence of isoleucine in place of aromatic residue in loop C of UEA 1 would again result in weak stacking interaction with the monosaccharide. The fact that UEA 1 is able to bind to monosaccharide suggests that additional hydrophobic interactions between a hydrophobic residue and groups in the saccharide should occur. Hydrophobic interactions between tyrosine 220 with the C-6 methyl group of L-fucose fulfills such a role. Another crucial locus for hydrogen bonding to C3/C4OOH group is provided by hydrogen

264

Vol. 268, No. 2, 2000

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

FIG. 1. Schematic representation of the monosaccharide binding site of WBA 1. The key residues and the vital hydrogen bonding patterns that are involved in the monosaccharide binding are shown.

bonding with the NH group of isoleucine 221 in loop D which would appear consistent with an analogous role for this residue found in several legume lectins. Consequently, the possibility of a hydrophobic interaction between the side chain of isoleucine 130 and C6-CH 3 group of L-fucose can also be discounted. Based on the above the following picture of UEA 1–L-fucose interaction emerges. The equatorial C-3 hydroxyl group covers the space amidst the conserved triad of residues hydrogen bonding with aspartate 87 OD2 from loop A, glycine 105 N from loop B, and ND2 of asparagine 135 from loop C while the isoleucine 130 stacks with the hydrophobic face of L-fucose. The axially oriented hydroxyl group consequently would then be positioned to donate a hydrogen bond to OD1 of aspartate 87 in loop A. This orientation also permits the projection of C6 methyl group in close vicinity of tyrosine 220, for the hydrophobic interaction, necessary for the stabilization of the binding of the monosaccharide with UEA 1. The C-3 hydroxyl group’s vital role is proven by the fact that structures in which the hydroxyl group had been removed or substituted

TABLE 3

Key Contacts Made by Sugar Residues in Binding Site of Legume Lectins Position of the sugar hydroxyls Lectin

O2

O3

WBA 1

H 2O

OD1 Asp 87 N Asp 212

ND2 Asn 128

OD1 Asp 89 NH Ala 218

NE2 Gln 219 H 2O

CON A

OD2 Asp 87 NE2 Asn 128 NH Gly 105 OD2 Asp 89 NE2 Asn 133 NH Gly 107 OD2 Asp 89 ND2 Asn 135 NH Gly 107 OD2 Asp 83 ND2 Asn 127 NH Gly 104 OD2 Asp 88 ND2 Asn 130 CO Ala 105 OD1 Asp 88 N Arg 228

DGL

N Arg 228

ND2 Asn 14 OD2 Asp 208 N Arg 228 ND2 Asn 125 OD2 Asp 81 OD1 Asp 81 ND2 Asn 125

EcorL

GSVI

H 2O

PNA

H 2O H 2O

SBA

LENL

H 2O

N Gly 99 H 2O

LOL I

DBL

PSL

OD1 Asp 85 N Gly 103 ND2 Asn 129 N Gly 99

O4

O5

O6

OD1 Asp 89

OD1 Asp 83 OG Ser 211

OD1 Asp 80

OD1 Asp 88 NH Leu 214

OD1 Asp 215 CD1 Ile 216

ND2 Asn 14

N Leu 99 N Try 100 N Leu 99 N Try 100 OD1 Asp 208 N Glu 31 N Ala 30 OD2 Asp 81 N Gly 209 N Glu 211

OD2 Asp 85 N Leu 214 OD1 Asp 81 ND2 Asn 125

265

N Leu 99

N Als 30 N Ala 210

OG Ser 215 OD2 Asp 81 N Ala 218 N Ala 217

Vol. 268, No. 2, 2000

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

showed no binding to the lectin, whereas modifications of the hydroxyl at C-2 resulted only in a slight diminution of the affinity (27, 28). This orientation of the monosaccharide also explains the preference of the lectin for Me␣ L-fucose over its ␤-counterpart as it is only with the former that one can have non-polar interactions with tyrosine 220. Binding of L-fucose in a reversed orientation would in addition to positioning the axially oriented hydroxyl group amidst the conserved triad would also lead to the projection of C6 methyl group outside the binding pocket and into the solvent—a situation incompatible with the binding of the monosaccharide. Hence, this orientation is not favored for binding to UEA 1. A speculative model of the binding site of Lotus lectin based on these predictions would involve, the C3 and C4 hydroxyl groups of L-fucose in interactions akin to UEA 1 excepting that the loop C of LTA is shorter in comparison and asparagine 130 in it is at a position similar to that observed for most other legume lectins. Further, the presence of tyrosine 128 in loop C of LTA as the stacking residue provides with sufficient energy for its stable binding to the monosaccharide. Consequently, the interaction of L-fucose with LTA do not need additional hydrophobic interaction between C-6 methyl group of the sugar and the corresponding loci in the protein. Indeed the residue corresponding to tyrosine 220 of UEA 1 in LTA is a glycine, i.e., glycine 205. Stacking between tyrosine 128 and the pyranose ring of L-fucose also explains the higher affinity of the monosaccharide with LTA than it is observed for UEA 1 where isoleucine 130 abutts the pyranosyl ring of L-fucose. The proposed model for the interaction of L-fucose with UEA 1 and LTA explains important facets of their saccharide specificities. UEA 1 binds very well to allyl␣-L-fucose and reasonably so to methyl-␣-D-mannose (Fig. 2b) (29). Its better binding to allyl-␣-L-fucose as compared to methyl-␣-L-fucose is readily explained by a relatively stronger nonpolar interaction between the allyl group of the former and tyrosine 220 in loop D. Absence of an aromatic residue at the corresponding position in LTA would not allow such a discrimination between allyl-␣-L-fucose and methyl-␣-L-fucose as observed experimentally. Mannose by itself binds weakly to both LTA and UEA 1 while methyl-␣-mannose binds reasonably well to the latter. However, the binding of methyl-␣-mannose to UEA 1 is made plausible by the non-polar interaction between the methyl group of this saccharide with tyrosine 220. In LTA again lack of an aromatic residue at the corresponding position would make the interaction with methyl-␣-D-mannose too weak to be observed specially as the orientation of this saccharide in the combining site of these lectins is such that it can stack only with its polar face viz. the A-face to the hydrophobic residue in loop C. These studies also allow us to posit about the role of the lengths of loop C in UEA 1 and LTA on their ligand

FIG. 2. (a) Schematic representation of the UEA 1 binding to The key binding residues and the loops in which they are present are shown. The model also indicates the hydrogen bonding pattern and the invariant water molecule ( ). (b) Schematic representation of the UEA 1 binding to Me␣Man. The key binding residues and the loops in which they are present are shown. The model also indicates the hydrogen bonding pattern and the invarient water molecule ( ). L-fucose.

binding specificities. LTA binds equally well to the type 1 and type 2 H-antigenic determinants while UEA 1 prefers strongly the latter structure. Poor affinity of UEA 1 for the type H-antigenic determinant has been proven to be related to the steric hindrance experienced by the bulky acetamido group of N-acetylglucosamine in ␤1-3 linkage to galactose. We believe that one of the major determinant of this steric hindrance is the long C loop, which does not leave enough space in the combining site of UEA 1. ACKNOWLEDGMENTS C.J.T. thanks R. Gosu and N. Balagi for their kind help and valuable suggestions in the usage of model building software. This work was supported by a grant to A.S. from the Department of Biotechnology, Government of India.

266

Vol. 268, No. 2, 2000

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

REFERENCES 1. Sharon, N., and Lis, H. (1989) Lectins, Chapman and Hall, London. 2. Sharon, N,., and Lis, H. (1990) FASEB J. 4, 3198 –3208. 3. Liener, I. E., Sharon, N., and Goldstein, I. J. (Eds.) (1986) The Lectins: Properties, Functions and Applications in Biology and Medicine, Academic Press, London. 4. Schwarz, F. P., Puri, K. D., Bhatt, R. G., and Surolia, A. (1993) J. Biol. Chem 268, 7668 –7677. 5. Toone, E. J. (1994) Curr. Opin. Struct. Biol. 4, 719 –728. 6. Bookbinder, L. H., Cheng, A., and Bliel, J. D. (1995) Science 269, 86 – 89. 7. Sastry, K., and Ezekowitz, R. A. (1993) Curr. Opin. Immunol. 5, 59 – 66. 8. Chrispeels, M. J., and Raikhel, N. V. (1991) Plant Cell 3, 1–3. 9. Lasky, L. A. (1992) Science 258, 946 –969. 10. Sharon, N., and Lis, H (1995) Essays Biochem. 30, 59 –75. 11. Matsumoto, I., and Osawa. T. (1969) Biochem. Biophys. Acta 194,180. 12. Horejsi, V., and Kocourek, J. (1974) Biochem. Biophys. Acta 336, 329. 13. Allen, H. J., and Johnson, E. A. Z.(1977) Carbohydr. Res. 58, 253. 14. Konami, Y., Yamamoto, K., and Osawa, T. (1991) J. Biochem. (Tokyo) 109, 650 – 658. 15. Konami, Y., Yamamoto, K., and Osawa, T. (1990) FEBS Lett. 268, 281–286.

16. Debray, H., Decout, D., Strecker, G. Spik, G., and Montreuil, J. (1981) Eur. J. Biochem. 117, 41–55. 17. Thomas,C. J., and Surolia, A. (1999) Arch. Biochem. Biophys., in press. 18. Ellinger, A., and Pavelka, M. (1988) Eur. J. Cell Biol. 47, 62–71 19. Shirahama, T., et al. (1993) Cancer 72, 1329 –1334. 20. Allen, H. J., Johnson, A. Z., and Matta, K. L. (1977) Immun. Commun. 6, 585– 602. 21. Young, N. M., and Oomen, R. P., (1992) J. Mol. Biol. 228, 924 – 934. 22. Sharma, V., and Surolia, A. (1997) J. Mol. Biol. 267, 433– 445. 23. Gohier, A., Espinosa, J. F., Jimenez-Barbero, J., Carrupt, P. A., Perez, S., and Imberty, A. (1996) J. Mol. Graph. 14, 322–7, 363– 4. 24. Prabhu, M. M., Sankarnarayanan, R., Puri, K. D., Sharma, V., Surolia, A., Vijayan, M., and Suguna, M. (1998) J. Mol. Biol. 276, 787–797. 25. Hamelryck, T. W., Dao-Thi, M. H., Chrispeels, M. J., Poortmans, F., Wyns, L., Loris, R. (1996) J. Biol. Chem. 271, 20479 –20491. 26. Hamelryck, T. W., Loris R., Bouckaert, J., Dao-Thi, M. H., Stecker, G., Imberty, A., Fernandez, E., Wyns, L., and Etzler, M. E. (1999) J. Mol. Biol. 286, 1161–1177. 27. Hindsgaul, O., Norberg, T., Pendu, J. L., and Lemieux, R. U. (1982) Carbohydr. Res. 109, 109 –142. 28. Lemieux, R. U., (1996) Acc. Chem. 29, 373–380. 29. Goldstein, I. J., and Hayes, C. E. (1978) Adv. Carbohydr. Chem. Biochem. 35, 127–340.

267