Synthesis of methyl alpha- and beta-N-dansyl-D-galactosaminides, probes for the combining sites of N-acetyl-D-galactosamine-specific lectins

How proteins bind carbohydrates: lessons from legume lectins

The pioneering studies of Irvin Liener on soybean agglutinin (SBA) in the early 1950s served as the starting point of our involvement in lectin research during the past four decades. Initially we characterized SBA extensively as a glycoprotein and showed that its covalently linked glycan is an oligomannoside commonly present in animal glycoproteins. We have also introduced the use of the lectin to the study of normal and malignant cells and to the purging of bone marrow for transplantation. Our recent work focuses on the combining site of Erythrina corallodendron lectin, closely related to SBA. In this legume lectin, as in essentially all other members of the same protein family, irrespective of their sugar specificity, interactions with a constellation of three invariant residues (aspartic acid, asparagine, and an aromatic residue) are essential for ligand binding. Lectins from other families, whether of plants or animals, also combine with carbohydrates by H-bonds and hydrophobic interactions, but the amino acids involved may differ even if the specificity of the lectins is the same. Therefore, nature finds diverse solutions for the design of binding sites for structurally similar ligands, such as mono- or oligosaccharides. This diversity strongly suggests that lectins are products of convergent evolution.

The Structural Basis for Carbohydrate Recognition By Lectins

1. Different carbohydrate-specific proteins, such as lectins, may combine with the same monosaccharide or oligosaccharide by different H-bonding and hydrophobic side chains. 2. Homologous proteins with distinct specificities may bind different monosaccharides (e.g., for glucose and galactose that differ in the configuration of a single hydroxyl) by the same set of invariant residues that are identically positioned in their tertiary structures. 3. The energetics of protein-carbohydrate interactions cannot be derived from structural information. 4. Nature solves in a variety of different ways the problem of constructing combining sites for carbohydrates, just as it provides diverse solutions for other functions of proteins.

Legume lectin structure

The legume lectins are a large family of homologous carbohydrate binding proteins that are found mainly in the seeds of most legume plants. Despite their strong similarity on the level of their amino acid sequences and tertiary structures, their carbohydrate specificities and quaternary structures vary widely. In this review we will focus on the structural features of legume lectins and their complexes with carbohydrates. These will be discussed in the light of recent mutagenesis results when appropriate. Monosaccharide specificity seems to be achieved by the use of a conserved core of residues that hydrogen bond to the sugar, and a variable loop that determines the exact shape of the monosaccharide binding site. The higher affinity for particular oligosaccharides and monosaccharides containing a hydrophobic aglycon results mainly from a few distinct subsites next to the monosaccharide binding site. These subsites consist of a small number of variable residues and are found in both the mannose and galactose specificity groups. The quaternary structures of these proteins form the basis of a higher level of specificity, where the spacing between individual epitopes of multivalent carbohydrates becomes important. This results in homogeneous cross-linked lattices even in mixed precipitation systems, and is of relevance for their effects on the biological activities of cells such as mitogenic responses. Quaternary structure is also thought to play an important role in the high affinity interaction between some legume lectins and adenine and a series of adenine-derived plant hormones. The molecular basis of the variation in quaternary structure in this group of proteins is poorly understood.

Structural features of the combining site region of Erythrina corallodendron lectin: role of tryptophan 135

The role of Trp 135 and Tyr 108 in the combining site of Erythrina corallodendron lectin (ECorL) was investigated by physicochemical characterization of mutants obtained by site-directed mutagenesis, hemagglutination-inhibition studies, and molecular modeling, including dynamics simulations. The findings demonstrate that Trp 135 in ECorL: (1) is required for the tight binding of Ca2+ and Mn2+ to the lectin because mutation of this residue into alanine results in loss of these ions upon dialysis and concomitant reversible inactivation of the mutant; (2) contributes to the high affinity of methyl alpha-N-dansylgalactosaminide (MealphaGalNDns) to the lectin; and (3) is solely responsible for the fluorescence energy transfer between the aromatic residues of the lectin and the dansyl group in the ECorL-MealphaGalNDns complex. Docking of MealphaGalNDns into the combining site of the lectin reveals that the dansyl moiety is parallel with the indole of Trp 135, as required for efficient fluorescence energy transfer, in one of the two possible conformations that this ligand assumes in the bound state. In the W135A mutant, which still binds MealphaGalNDns strongly, the dansyl group may partially insert itself into the place formerly occupied by Trp 135, a process that from dynamics simulations does not appear to be energetically favored unless the loop containing this residue assumes an open conformation. However, a small fraction of the W135A molecules must be able to bind MealphaGalNDns in order to explain the relatively high affinity, as compared to galactose, still remaining for this ligand. A model for the molecular events leading to inactivation of the W135A mutant upon demetallization is also presented in which the cis-trans isomerization of the Ala 88-Asp 89 peptide bond, observed in high-temperature dynamics simulations, appears not to be a required step.

Synthesis of soybean agglutinin in bacterial and mammalian cells

The cDNA of soybean agglutinin (SBA), a glycoprotein lectin, obtained from the mRNA of soybean seeds at mid-maturation, was cloned in a lambda gt 10 phage and subcloned in a pUC-8 plasmid. Probing with a fragment of the lectin gene [Vodkin, L. O., Rhodes, P. R. & Goldberg, R. B. (1983) Cell 34, 1023-1031] afforded a clone of 1012 nucleotides containing the complete coding region of 858 nucleotides for the precursor to soybean agglutinin. The deduced amino acid sequence contains the 253 residues of the mature lectin and an hydrophobic N-terminal signal peptide of 32 amino acids. Expression in Escherichia coli of the cDNA coding for the precursor to the lectin or for the mature lectin led to the accumulation of large quantities of inclusion bodies, from which mature SBA was isolated in small yield (up to 1 mg/l). It was identical with the native lectin in the hemagglutinating activity and carbohydrate specificity, N-terminal sequence and oligomeric structure, but, because it was not glycosylated, its subunit mass was lower by 2 kDa. Our findings show that pre-SBA is processed into the mature form in the bacteria, and that, contrary to what has been suggested [Nagai, K. & Yamaguchi, H. (1993) J. Biochem. (Tokyo) 113, 123-125], glycosylation is not essential for the folding of the lectin, nor for its subunit assembly into a biologically active tetramer. To obtain recombinant SBA in secreted form, the pre-SBA cDNA was subcloned in pTM1 vector and the construct inserted into vaccinia virus. When monkey BS-C-1 cells were infected by the virus, using a double expression protocol, recombinant lectin was secreted into the growth medium, from which it was isolated by immunoaffinity chromatography at a yield similar to that from the bacteria. Except for its lower hemagglutinating activity, the product was indistinguishable from native SBA in all properties tested. It was also susceptible to digestion by endo-beta-N-acetylglucosaminidase H or N-glycanase which caused a decrease of 2 kDa in its subunit mass and gave the same results on lectin blot analysis, indicating that it too is a glycoprotein with a single oligomannose unit.

Mutational studies of the amino acid residues in the combining site of Erythrina corallodendron lectin

High-resolution X-ray crystallography of the complex of the Gal/GalNAc-specific Erythrina corallodendron lectin with lactose identified the amino acid side chains that form contacts with the galactose moiety of the disaccharide. The contribution of these amino acids to the binding of different monosaccharides and oligosaccharides by the lectin was examined by site-directed mutagenesis. Replacement of Phe131, on which the galactose is stacked, by tyrosine, gave a mutant with the same hemagglutinating activity and carbohydrate specificity as the parent lectin, but replacement by alanine or valine resulted in loss of activity. Mutations of Ala88, Asp89, and Asn133 produced mutants that were also inactive whereas those of the other combining site residues, Tyr106, Ala218, and Gln219, were biologically active. None of the active mutants interacted with mannose or glucose. Thus, contrary to an earlier assumption. Ala218 is not responsible for the inability of E. corallodendron lectin to bind these sugars. Our findings also demonstrate that Gln219 is not involved in galactose binding in solution, even though this is implicated by the crystal data. Instead, our data suggest that Gln219 assists in the ligation of N-acetyllactosamine to the lectin, by interacting with the acetamide group of the disaccharide. Comparison with other legume lectins specific for mannose/glucose, galactose, N-acetylgalactosamine, L-fucose or N-acetylglucosamine, shows that only three of the combining site residues of E. corallodendron lectin occupy invariant positions both in their primary and tertiary structures. These residues are an aspartic acid and an asparagine corresponding to positions 89 and 133, respectively, in E. corallodendron lectin, and an aromatic residue, either phenylalanine (as Phe131 in this lectin), tyrosine or tryptophan. We therefore postulate that these three residues are essential for ligand binding by all such lectins, irrespective of their specificity.

Thermodynamics of monosaccharide and disaccharide binding to Erythrina corallodendron lectin

Isothermal titration calorimetry measurements of the binding of 2'-fucosyllactose, lactose, N-acetyllactosamine, galactopyranose, 2-acetamido-2-deoxygalactopyranoside, methyl alpha-N-dansylgalactosaminide (Me-alpha-DNS-GalN), methyl alpha-D-galactopyranoside, methyl beta-D-galactopyranoside, and fucose to Erythrina corallodendron lectin (ECorL), a dimer with one binding site per subunit, were performed at 283-286 and 297-299 K. The site binding enthalpies, DeltaHb, with the exception of Me-alpha-DNS-GalN, are the same at both temperatures and range from -47.1 +/- 1.0 kJ mol-1 for N-acetyllactosamine to -4.4 +/- 0.3 kJ mol-1 for fucose, and the site binding constants range from 3.82 +/- 0.9 x 10(5)M-1 for Me-alpha-DNS-GalN at 283.2 K to 0.46 +/- 0.05 x 10(3) M-1 for fucose at 297.2 K. The binding reactions are mainly enthalpically driven except for fucose and exhibit enthalpy-entropy compensation. The binding enthalpies of the disaccharides are about twice the binding enthalpies of the monosaccharides in contrast to concanavalin A where the binding enthalpies do not double for the disaccharides. Differential scanning calorimetry measurements show that denaturation of the ECorL dimer results in dissociation into its monomer subunits. The binding constants from the increase in denaturation temperature of ECorL in the presence of saccharides are in agreement with values from isothermal titration calorimetry results. The thermal denaturation of ECorL occurs around 333 K, well below the 344-360 K denaturation temperature of other legume lectins of similar size and tertiary structure, undoubtedly due to the difference in its quaternary structure relative to other legume lectins. This is also apparent from the independent unfolding of its two domains.

Modification by site-directed mutagenesis of the specificity of Erythrina corallodendron lectin for galactose derivatives with bulky substituents at C-2

Examination of the three-dimensional structure of Erythrina corallodendron lectin (ECorL) in complex with a ligand (lactose), the first of its kind for a Gal/GalNAc-specific lectin [(1991) Science 254, 862-866], revealed the presence of a hydrophobic cavity, surrounded by Tyr108 and Pro134-Trp135, which can accommodate bulky substituents such as acetamido or dansylamido (NDns) at C-2 of the lectin-bound galactose. Comparison of the primary sequence of ECorL with that of soybean agglutinin, specific for galactose and its C-2 substituted derivatives, and of peanut agglutinin, specific for galactose only, showed that in soybean agglutinin, Tyr108 is retained, and Pro134-Trp135 is replaced by Ser-Trp, whereas in peanut agglutinin, the former residue is replaced by Thr and the dipeptide by Ser-Glu- Tyr-Asn. Three mutants of ECorL were therefore constructed: L2, in which Pro134-Trp135 was replaced by Ser-Glu-Tyr-Asn; Y108T, in which Tyr108 was replaced by Thr and the double mutant L2; Y108T. They were expressed in Escherichia coli, as done for recombinant ECorL [(1992) Eur. J. Biochem. 205, 575-581]. The mutants had the same hemagglutinating activity as native or rECorL. Their specificity for galactose, GalNAc and Me beta GalNDns was examined by inhibition of hemagglutination and of the binding of the lectin to immobilized asialofetuin; in addition, their association constants with Me alpha GalNDns and Me beta GalNDns were measured by spectrofluorimetric titration. The results showed that Y108T had essentially similar specificity as the native and recombinant lectins. The affinity of L2 and L2;Y108T for galactose was also the same as ECorL, but they had a lower affinity for GalNAc and markedly diminished affinity for the dansyl sugars (up to 43 times, or 2 kcal, less). This appears to be largely due to steric hindrance by the two additional amino acids present in the cavity region in these mutants. Our findings also provide an explanation for the inability of PNA to accommodate C-2-substituted galactose derivatives at its primary subsite.