Crystallography and Lectin Structure Database - CNRS

36 U. Krengel and A. Imberty
one observed in animal C-type lectins [109]. PA-IIL, the fucose-specific lectin, is
specifically inhibited by human milk, which is unusually rich in fucosylated
oligosaccharides and protects infants against microbial infections. Lewis a (Le a ),
a trisaccharide determinant present in the human milk of most women, is the best
ligand of PA-IIL, with an affinity constant of 2.2 10 7 M. The crystal structure
of PA-IIL in complex with Le a trisaccharide revealed the presence of two additional
hydrogen bonds, hence explaining the higher affinity compared to the
fucose monosaccharide [110].
4.2.1. Bacterial lectins – A case study
X-ray crystallography is a very powerful technique to determine the high-resolution
structure of proteins and protein–ligand complexes. But more than that, the insights
gained often reach far beyond the structure itself and can help to solve important
biological questions. Take for example a study on bacterial toxins. Two researchers,
Mike Lebens and Susann Teneberg from the University of Gothenburg, were
intrigued by the different binding properties of two highly homologous toxins,
cholera toxin and the heat-labile enterotoxin from enterotoxigenic E. coli. The
receptor-binding B-subunits, CTB and LTB, respectively, share 83% sequence
identity, yet cholera toxin binds with very high specificity only to its primary receptor,
the GM1 ganglioside, whereas heat-labile enterotoxin is much more promiscuous
and binds in addition to several other glycoconjugates [111–114]. In order to
find out which of the amino acids are responsible for the broader binding specificity
of heat-labile enterotoxin, the researchers constructed hybrids between the
two carbohydrate-binding domains and tested their binding specificities [115].
Eventually, they created a hybrid that was indistinguishable in its binding properties
from the native LTB, but had half of its residues from CTB. When substituting
amino acids back one by one to the CTB sequence, the researchers were surprised:
substitution of amino acid residue 4 (SerAsn) resulted in a hybrid with previously
undescribed binding specificity to blood group antigens. Molecular modeling
suggested binding to a site distinct from the primary GM1 binding site [116] –
apparently, a new binding site had been created. But modeling of such complex
systems is difficult and experimental verification needed. The crystal structure (at
1.9Å resolution) indeed revealed the presence of a second binding site (Fig. 7), but
with unexpected characteristics [82]. First of all, there were only very few direct
protein–carbohydrate interactions; most of the hydrogen bonds were indirect, via
water molecules. Second, even the interaction to the critical residue Asn4 turned
out to be an indirect water-mediated interaction. And finally, except for Asn4, all
amino acid residues in this new binding site were conserved also in native LTB. In
fact, a single Ser4Asn mutant of LTB showed equally strong binding to blood
group antigens as did the structurally characterized hybrid [82]. This gave strong
indications that the binding site must already have been present in the native toxin
and the mutation only served to enhance the affinity such that it became detectable