Glycobiology feature

Protein–carbohydrate interactions
At the heart of biochemistry
Proteins that possess the ability to bind carbohydrates specifically and in the micromolar range. Biologically by Nathan Sharon
reversibly abound in nature, being present in all living organisms, from relevant interaction often requires (The Weizmann Institute of
viruses to humans. Their interactions with their ligands are the basis multivalency of both protein and Science, Rehovot, Israel)
of a myriad of biological processes, both normal and pathological1–3 ligand in order to generate higher
(Table 1). The high selectivity required for these interactions is affinity of binding. Although pro-
provided by a specific stereochemical fit between complementary teins specific for mono-, di- or
molecules, the protein on the one hand and the carbohydrate on the tri-saccharides have small combining
other. This concept has its origins in the lock-and-key hypothesis, sites (Figures 1 and 2), they can read-
introduced by Emil Fischer at the end of the 19th Century to explain ily bind higher oligosaccharides or
the specificity of interactions between enzymes (he studied their derived glycoconjugates. The
glycosidases) and their substrates (carbohydrates), i.e. between combining sites of polysaccharide-
molecules in solution. In time it was extended to describe the specific proteins are not large either,
interactions of cells with soluble molecules and with other cells. and usually do not exceed the size of
approximately six monosaccharide
Complex carbohydrates are found monosaccharides to polysaccharides, units. For example, the binding sites
commonly at the cell surface, where either free (such as starch, cellulose or of lysozyme and of Taka-amylase
they are positioned to interact with hyaluronic acid) or conjugated. The (which hydrolyses amylose) have six Figure 1.
suitable proteinaceous receptors, affinity of such proteins for mono- and seven adjacent subsites respec- N-Acetyllactosamine
primarily lectins, in solution or on saccharides is generally low, in the tively, each complementary to a (Galβ1-4GlcNAc) in the
4,5
the surfaces of other cells . These millimolar range, while for oligosac- monosaccharide; phosphorylase combining site of Erythrina
proteins, originally identified as charides or polysaccharides it may be (which acts on glycogen) has four corallodendron lectin
sugar-specific haemagglutinins, are (a legume seed lectin).
currently known as cell-recognition The essential contact
molecules. They are geared to distin- residues, which include
guish between different oligosaccha- Phe131 (F131) on which
rides, whether as such or as part of the galactose of the ligand
glycoconjugates, most commonly is stacked, are marked. The
glycoproteins and glycolipids. blue sphere denotes the
water molecule that
Combining sites mediates the interaction
between the main chain
Carbohydrate-binding proteins, amide of Leu86 (L86) and
whether enzymes, anti-carbohydrate the O-6 of the galactose.
antibodies, sugar transporters, Broken lines denote
growth factors or lectins, are struc- hydrogen bonds. The cavity
turally diverse, differing markedly in marked with an asterisk is
size and tertiary and quaternary where bulky substituents,
structure. They can bind a wide spec- such as fucose, attached to
trum of ligands, ranging from simple O-2 of the galactose, can
bind (PDB entry 1LTE).
Key words: carbohydrate-binding protein, Courtesy of Miri Eisenstein,
haemagglutinin, ligand recognition, saccharide, Weizmann Institute
X-ray crystallography. of Science

The Biochemist — June 2006. © 2006 Biochemical Society 13

feature Glycobiology
Table 1. Protein–carbonhydrate interactions of biological significance
Interaction
Molecule–molecule
Molecule–cell
Virus–cell
Cell–cell
Figure 2. Combining site of
human influenza virus
haemagglutinin in complex
with the trisaccharide
receptor NeuAcα2-3Galβ1-
4GlcNAc. Hydrogen bonds
are indicated by broken
lines; amino acid residues
making interactions via
main-chain carbonyl groups
are shown as red spheres,
and those interacting via
main chain nitrogens are
shown as blue spheres. The
trisaccharide carbon atoms
are in yellow, its nitrogen
atoms in blue and oxygen
atoms in red. Water
molecules are indicated by
green spheres. Reproduced
with permission from
Gamblin, S.J., Haire, L.F.,
Russell, R.J. et al. (2004)
Science 303, 1838–1842.
Copyright 2004 AAAS.
14
System Function
Antigen-glycan Defence
Enzymes of carbohydrate metabolism and substrates Synthesis and degradation of carbohydrates
Proteoglycans and hyaluronic acid Structural elements of the extracellular matrix
Intracelular lectins and glycoproteins Quality control of glycoprotein synthesis; targeting of lysosmal enzymes
to their subcellular compartment
Bacterial sugar transporters and saccharides Transport and chemotaxis
Sugar antiporters Protein glycosylation
Toxins and cell-surface carbohydrates Cell killing
Cell-surface lectins and glycoproteins Clearance of glycoproteins from the circulatory system and their uptake
into cells
Growth factors and cell-surface proteoglycans Signal transduction into cells
Myxoviruses and cell-surface sialic acid Infection
HIV and dendritic cells Defence
Bacterial surface lectins and epithelial cell-surface saccharides Infection
Lymphocytes and carbohydrates on lymph node endothelial cells Homing of lymphocytes to lymph nodes
Lectins on activated endothelial cells and carbohydrates on leukcocyte Control of leukcocyte extravasation to sites of inflammation
such subsites; the FGF (fibroblast bining sites of carbohydrate-binding detailed atomic picture of the com-
growth factor) combining site fits a proteins was quite limited. It was plexes in the crystal, which appears
particular hexasaccharide sequence of based almost exclusively on speci- to reflect their structure in solution.
heparan sulphate (Figure 3); and the ficity studies of carbohydrate-spe- It was first applied in the 1960s by
combining site of an antibody against cific enzymes and antibodies, David C. Phillips and co-workers
a bacterial polysaccharide possesses complemented by selective chemical to determine the three-dimensional
five monosaccharide-specific subsites modification of the proteins. At structure of complexes of hen egg-
(Figure 4), whereas those of antibod- present, the major source of infor- white lysozyme with oligosaccha-
ies against dextran (a polymer of mation about such sites of carbohy- rides derived from chitin or from
α-glucose) can accommodate up to drate-binding proteins is X-ray bacterial cell wall peptidoglycan,
seven monosaccharide units. crystallography of their complexes both substrates of the enzyme. To
Early information on the com- with ligands. The method gives a date, the structures of hundreds of
protein–carbohydrate complexes
have been elucidated, a large number
Asn-193 of these of being lectins (see
1934-human/human www.cermav.cnrs.fr/lectines)
receptor complex
(Figure 1). Other sources of combin-
ing-site information include binding
190-Helix
GlcNAc-3 experiments with different sugars
Glu-190 and their derivatives (a much wider
range than in the past) using state of
Gal-2 the art techniques, site-directed
2 mutagenesis of the proteins and also,
3 to a limited extent, NMR experi-
Lys-222
4 ments and molecular modelling.
Sialic acid
Such studies have shown that,
227
like the proteins themselves, the
135
combining sites are diverse, even
Asp-225 Gln-226
130-Loop when their specificity is the same
220-Loop Thr-136
137 (although within a given protein
family, the sites may be similar).
The Biochemist — June 2006. © 2006 Biochemical Society

In general, the sites appear to be
preformed, since conformational
changes occur rarely upon ligand
binding (see below).
Two main types of combining
site have been identified in carbo-
hydrate-binding proteins. In
one, the site is located in a cleft
or groove present between two
domains of the molecule, as first
demonstrated with lysozyme
and subsequently found in other
enzymes, such as hexokinase,
and in the bacterial carbohydrate-
binding proteins. The other type,
found in, for example, legume
lectins, has the shape of a pocket
or depression on the protein mole-
cule. It is usually formed by amino
acids from a single polypeptide
chain, but occasionally amino acids
from more than one polypeptide
chain are involved. Thus the bind-
ing sites of wheat germ agglutinin
are located at the interface between
its two subunits and are formed by
amino acids from both.
As is clear from the above
examples, the amino acids that form
the combining sites are not neces-
sarily contiguous, and are brought
together in space by the folding
of the polypeptide chain. The
structural changes needed to form
a carbohydrate combining site
sometimes require the participation
of non-carbohydrate ligands.
For example, concanavalin A must
2+ 2+
combine with Mn and Ca ions
in order to be able to bind saccha-
rides. X-ray crystallography has
revealed that several of the amino
acids in the combining site of con-
canavalin A occupy different posi-
tions in the metal-containing and
metal-free lectin. Further changes
in the spatial position of the amino
acid residues occur occasionally
upon binding of the carbohydrate,
bringing them to an orientation
The Biochemist — June 2006. © 2006 Biochemical Society
GlcNS,6S
IdoA2S
Gln134
Lys125
that improves stereocomplementar-
ity with the ligand, a phenomenon
known as ‘induced fit’.
The fit between the binding site
on the protein and the carbohy-
drate is also affected by the shape
of the ligand. Oligosaccharides are
flexible molecules with consider-
able freedom of rotation around
the glycosidic bonds connecting
the individual monosaccharide
constituents. As a result, they may
assume different shapes, only one
of which may fit the combining
sites. Cases are known in which
oligosaccharides that differ in
their chemical structure may have
common topographic features,
and as a result interact effectively
with the same protein.
The participation of a particular
amino acid of a protein and of a
specific group of the carbohydrate
ligand in the interaction between
the two can be assessed by differ-
ent methods. In the case of the
amino acids, it is done mainly by
site-directed mutagenesis, as
mentioned above. This technique,
combined with ligand-binding
experiments, also provides infor-
mation on the relative contribution
of individual residues to the pro-
tein–carbohydrate interaction.
Glycobiology feature
Figure 3. Heparan sulphate
IdoA2S pentasaccharide in the
combining site of FGF.
GlcNS,6S Heparan sulphate binding
GlcNS,6S
site of FGF is shown with
the critical contact amino
acids lysine and arginine
Arg120
(blue), and asparagine
and glutamine (light blue).
Lys26
Lys135 The heparin-derived
Asn101
hexasaccharide (pink),
with the critical groups
Asn27 coloured by atom (C,
green; O, red; N, blue; S,
yellow), is shown. Red type
indicates monosaccharides
Certain combining site residues with interacting sulphate
are essential to ligand binding, groups. Monosaccharide
and their replacement leads to nomenclature: ∆U2S, 2-O-
complete loss of this activity sulphated uronic acid with
(Figure 1). Interestingly, homolo- ∆4,5 unsaturated linkage;
gous proteins with distinct speci- IdoA2S, 2-O-sulphated α-
ficities, for example for galactose L-iduronic acid; GlcNS6S,
and mannose, often bind these N-sulpho-6-O-sulphate-α-
monosaccharides by the same set glucosamine. The broken
of invariant residues that are posi- lines denote electrostatic
tioned identically in their tertiary bonds. Courtesy of Dr Ram
structures. On the other hand, Sasisekharan, Massachusetts
structurally different proteins with Institute of Technology,
similar specificities may bind the MA, USA.
same saccharide by different sets
of combining site residues.
Concerning the carbohydrate
ligand, its bonding pattern in the
complex can be mapped using
deoxy-, methoxy-, deoxyfluoro- or
other sugar analogues 6. However,
it is only X-ray crystallography
that provides detailed information
at the atomic level of the interplay
between the two molecules.
Because crystallization conditions
may induce conformational
changes in the protein and alter
the mode of ligand binding, it is
essential to complement the data
obtained by this method with
information from other sources
(e.g. NMR) on the solution struc-
ture of the protein–ligand complex.
15
feature Glycobiology

Figure 4. A pentasaccharide best possible linear bond with an

sequence L-Rhaα1-2L- acceptor group, which is important in
Rhaα1-3Rhaα1- imparting specificity. When each of
3GlcNAcβ1-2Rhaα of a two adjacent hydroxy groups of a
bacterial lipopolysaccharide monosaccharide interacts with differ-
bound by a monoclonal ent atoms of the same amino acid (e.g.
antibody. The antibody the two oxygens of the carboxylate of
recognizes the trisaccharide glutamic or aspartic acid), they form
L-Rhaα1-3L-Rhaα1-3Glc- bidentate hydrogen bonds. Another
NAcβ of the sequence, so type of hydrogen bond characteristic
that the L-rhamnose of protein–carbohydrate complexes
residues A and A’ at both is the co-operative bond, in which the
ends of the pentasaccha- hydroxy group acts simultaneously
ride make little contri- as donor and acceptor (see Figure 1).
bution to the energy of
the binding. Hydrogen Hydrophobic interactions
bonds are indicated by Although carbohydrates are highly
broken lines. Hydrophobic polar molecules, the steric disposition
interactions are not of the hydroxy groups creates on
marked and involve their surfaces hydrophobic patches
primarily the methyl group that can form contacts with
of the acetamido residue Protein–carbohydrate tein–carbohydrate interactions, as hydrophobic side chains of the pro-
of GlcNAc with TyrL32, the bonds well as contributing to their affinity. tein. Such contacts are generally non-
base of the site formed by They are approximately 2.5–3 Å long specific and not directional. One
ThrL91 and a methionine The types of bonds involved in the and their energy is relatively high, up widely occurring interaction of this
(not shown), both of which formation of protein–carbohydrate to 5 kcal/mol (1k cal≡4.184k J). kind is the stacking of a monosaccha-
interact with the α complexes are in principle not differ- However, their contribution to the ride on a side chain of an aromatic
face of GlcNAc and the ent from those involved in the forma- total energy of binding, which for amino acid (see Figures 1 and 4). It is
β face of RhaC. Courtesy tion of the complexes of proteins carbohydrate–protein complexes is present in the sugar complexes of
of Dr David R. Bundle, with other ligands, such as peptides, usually 5–10 kcal/mol, is much lower almost all legume lectins, of the
University of Alberta, oligonucleotides or various small than the sum of their interactions, for galectins (a class of galactose-specific
Edmonton, Canada. molecules. Binding between proteins reasons discussed below. These bonds animal lectins) and of several glycosi-
and carbohydrates is stabilized pri- occur largely between the hydroxy dases, as well as those of bacterial
marily by a network of hydrogen groups of the carbohydrate and the toxins (e.g. Escherichia coli lytic toxin
bonds and hydrophobic interactions; amino acid side chains of the protein, and tetanus toxin). In addition, the
in rare cases, electrostatic interactions most frequently of aspartic acid, methyl moiety of the acetamide of
(or ion pairing) and co-ordination aparagine, glutamic acid, glutamine, acetamido sugars often interacts with
7,8
with metal ions also play a role . arginine and serine residues. Main aromatic residues of influenza virus
Bonding is sometimes mediated by chain NH and CO groups also con- haemagglutinin (Figure 2) and wheat
one or more water molecules (see tribute to hydrogen bonding, but to a germ agglutinin that are specific for
below). Although in a single protein a lesser extent. such sugars, as does the methyl group
limited set of amino acid residues A sugar hydroxy group has the of fucose. Hydrophobic contacts also
contribute to the interactions with capacity to interact with a protein occur with side chains of aliphatic
the ligand, in general most of the side both as a hydrogen bond donor and amino acids, such as valine or leucine.
chains of the 20 standard amino acids as an acceptor, by way of lone elec-
can participate in ligand binding. tron pairs. Moreover, as a donor, the Other interactions
hydroxy group possesses the added Most saccharides are uncharged and
Hydrogen bonds advantage of having rotational free- therefore ionic, i.e. charge–charge
Hydrogen bonds are directional and dom about the C-OH torsional interactions, do not commonly par-
therefore confer specificity to pro- angle, thus enabling it to attain the ticipate in the formation of

16 The Biochemist — June 2006. © 2006 Biochemical Society


protein–carbohydrate complexes.
However, such interactions do occur
in complexes with sialic acids, for
example with influenza virus
haemagglutinin (Figure 1), and are
common in complexes with gly-
coaminoglycans, such as those of
FGF (Figure 4) or foot and mouth
disease virus. In C-type lectins, an
unusual bond is present between the
protein-bound Ca2+ and hydroxy
groups of the sugar ligand.
Role of water
Contacts between the protein and
its carbohydrate ligands are some-
times mediated by water bridges 6.
Water acts as a molecular mortar,
its small size and ability to serve as
both a hydrogen donor and accep-
tor endowing it with ideal proper-
ties for this purpose. Such bridges,
which consist of a single water
molecule or chains of several such
molecules, may be important for
ligand recognition. Comparison of
the structures of unligated proteins
from the same family with those
of their complexes with the same
or different sugars has shown that
certain ordered water molecules are
conserved in all such structures,
and several of these make a signifi-
cant contribution to ligand binding.
Both the protein and the ligand
in aqueous solutions are normally
hydrogen-bonded to water mole-
cules. Therefore the binding
process is an exchange reaction in
which both partners exchange their
bonds to water for bonds to each
other and release water to the bulk
6,9
solvent . The net interaction
energy represents the differences
between the hydrogen bond ener-
gies and entropies of the protein
and of the carbohydrate with
water, and those of the protein and
carbohydrate with each other.
The Biochemist — June 2006. © 2006 Biochemical Society
Solvation–desolvation energies
are very large, due to entropy, and
cannot be reliably calculated for
hydrophilic compounds such as
sugars. Thus, although the ener-
getic contributions of various
bonding interactions can be esti-
mated, errors in the estimation
of the attendant solvation energy
changes can be very large, making
the overall calculations of binding
energy difficult.
Conclusion
The knowledge about protein–carbo-
hydrate interactions not only is of
theoretical interest, but also has prac-
tical implications, for example for
development of novel drugs for dif-
ferent diseases. Influenza is initiated
by the binding of the virus to host
cells, and so the details of the
influenza virus haemagglutinin–N-
acetylneuraminic acid interactions
provide a basis for the design of
antiviral drugs that would act by
blocking the attachment of the virus
to target cells. Fludase®, one such
inhibitor of the haemagglutinin, is
scheduled for clinical trials this year.
Similarly, many bacteria, such as E.
coli, Helicobacter pylori and
Streptococcus pneumoniae, initiate
infection by sticking to sugars on
host tissues. Experiments in different
animals have shown that inhibitors of
such sugar-mediated adherence can
block infection. Here, too, a better
knowledge of the carbohydrate-
binding sites on the bacterial surface
may be useful for the development of
suitable carbohydrates or their deriv-
atives, for anti-adhesion therapy of
such infections10. Last but not least,
further investigation of the combin-
ing sites and specificity requirements
of the selectins, a family of animal
lectins, is expected to lead to the
development of drugs for the preven-
Glycobiology feature
tion of adverse effects of extravasa-
tion, such as those that accompany
myocardial infarctions and strokes11.
I am grateful to my long-time col-
league Dr Halina Lis for her help in
the preparation of this manuscript.
References
1. Lee, Y.C. and Lee, R.T. (1995) Acc. Chem. Res.
28, 321–327
2. Varki, A., Cummings, R., Esko, J. et al. (eds)
(1999) Essentials of Glycobiology, pp. 653,
Cold Spring Harbor Laboratory Press, Cold
Spring Harbor
3. Taylor, M. and Drickamer, K. (2003)
Introduction to Glycobiology, Oxford
University Press, Oxford
4. Sharon, N. and Lis, H. (1993) Sci. Am. 268,
82–89
5. Sharon, N. and Lis, H. (2003) Lectins, 2nd Ed.,
pp. 457, Kluwer Academic Publishers,
Dodrecht/Springer, Amsterdam
6. Lemieux, R.U. (1996) Acc. Chem. Res. 29,
373–380
7. Bundle, D. (1999) In Carbohydrates (Hecht, S.,
ed.), pp. 370–440, Oxford University Press,
Oxford
8. Quiocho, F.A. (1999) In Carbohydrates (Hecht,
S., ed.), pp. 441–457, Oxford University
Press, Oxford
9. Toone, E.J. (1994) Curr. Opin. Struct. Biol. 4,
719–728
10. Sharon, N. (2006) Biochim. Biophys. Acta
1760, doi:10.1016/j.bbagen.2005.12.008
11. Ley, K. (2003) Trends Mol. Med. 9, 263–268
Nathan Sharon is
Emeritus Professor at
Weizmann Institute of
Science, which he joined
in 1954, after receiving
his PhD from Hebrew
University, Jerusalem. He
has published extensively
on carbohydrates and
lectins and was instru-
mental in formulating and widely disseminating the
concept that these substances function in cell
recognition. His work has stimulated attempts at
the development of anti-adhesion therapy of infec-
tious diseases, and is the basis of a lectin-purging
technique of bone marrow used clinically for
treatment of ‘bubble children’.
email: nathan.sharon@weizmann.ac.il
17