Báo cáo khoa học: Structural basis for the recognition of complex-type biantennary oligosaccharides by Pterocarpus angolensis lectin ppt

Structural basis for the recognition of complex-type
biantennary oligosaccharides by Pterocarpus angolensis
lectin
Lieven Buts
1
, Abel Garcia-Pino
1
, Anne Imberty
2
, Nicolas Amiot
3
, Geert-Jan Boons
3
,
Sonia Beeckmans
4
, Wim Verse
´
es
1
, Lode Wyns
1
and Remy Loris
1
1 Laboratorium voor Ultrastructuur, Vrije Universiteit Brussel and Department of Molecular and Cellular Interactions, Vlaams Interuniversitair
Instituut voor Biotechnologie, Belgium
2 Centre de Recherches sur les Macromole
´
cules Ve
´
ge

´
tales (CERMAV) – CNRS (affiliated with Joseph Fourier University), Grenoble, France
3 Complex Carbohydrate Research Center, The University of Georgia, Athens, GA, USA
4 Laboratorium voor Scheikunde der Proteı
¨
nen, Instituut voor Moleculaire Biologie, Vrije Universiteit Brussel, Belgium
Lectins (or lectin domains) represent a specific class of
carbohydrate binding proteins distinct from enzymes
and antibodies. Different lectin families are found in
a wide range of organisms including viruses, bacteria,
plants and animals. Their biological activities are
diverse and include roles in the innate immunity, bac-
terial and viral infection, sorting and trafficking of gly-
coproteins, development and differentiation as well as
defense mechanisms in plants [1].
The lectins from legume plants belong to one of the
best studied lectin families [2]. Members of this family
were initially identified in the seeds of legume plants,
but an increasingly larger number is found in the
vegetative parts. They show strong similarities on the
level of their amino acid sequences and tertiary struc-
tures, and exhibits a wide range of carbohydrate spe-
cificities and quaternary structures. Recently, family
members were discovered in nonlegume plants [3]. Fur-
thermore, the ER-Golgi intermediate (ERGIC) pro-
teins that in animals play a role in glycoprotein
transport though the golgi apparatus (but are absent
in plants) belong to the legume lectin family [4].
Concanavalin A (con A) was the first lectin for
which the crystal structure was determined [5,6]. Since

this early work, the crystal structures of 28 members
of the legume lectin family have been presented either
Keywords
carbohydrate; lectin; legume lectin; protein-
carbohydrate recognition
Correspondence
R. Loris, Laboratorium voor Ultrastructuur,
Vrije Universiteit Brussel and Department of
Molecular and Cellular Interactions,
Vlaams Interuniversitair Instituut voor
Biotechnologie, Pleinlaan 2,
B-1050 Brussels, Belgium
Fax: +32 2 6291963
Tel: +32 2 6291989
E-mail:
(Received 5 January 2006, revised 21 March
2006, accepted 28 March 2006)
doi:10.1111/j.1742-4658.2006.05248.x
The crystal structure of Pterocarpus angolensis lectin is determined in its
ligand-free state, in complex with the fucosylated biantennary complex type
decasaccharide NA2F, and in complex with a series of smaller oligosaccha-
ride constituents of NA2F. These results together with thermodynamic
binding data indicate that the complete oligosaccharide binding site of the
lectin consists of five subsites allowing the specific recognition of the penta-
saccharide GlcNAcb(1–2)Mana(1–3)[GlcNAcb(1–2)Mana(1–6)]Man. The
mannose on the 1–6 arm occupies the monosaccharide binding site while
the GlcNAc residue on this arm occupies a subsite that is almost iden-
tical to that of concanavalin A (con A). The core mannose and the
GlcNAcb(1–2)Man moiety on the 1–3 arm on the other hand occupy a
series of subsites distinct from those of con A.

Abbreviations
Con A, concanavalin A; ITC, isothermal titration calorimetry; LOL, Lathyrus ochrus lectin.
FEBS Journal 273 (2006) 2407–2420 ª 2006 The Authors Journal compilation ª 2006 FEBS 2407
in their unbound form or in complex with mono- or
oligosaccharides ( />From these studies, a detailed picture of the relation-
ship between amino acid sequence and carbohydrate
specificity has emerged (7,8), which is still being refined
[9–22].
Apart from structural studies, the past decade has
also brought a large number of studies involving iso-
thermal titration calorimetry (ITC) (for a recent
review, see [23]). Especially con A and its relatives
from the Diocleinae subtribe have been studied in
detail by different research teams [23]. These include
binding studies of mono- and oligosaccharides, binding
of deoxy sugar analogs, hydroxyethyl analogs and con-
formationally constrained sugars, analysis of solvent
isotope effects, osmotic stress strategies and analysis of
the cluster glycoside-effect. This wealth of data stems
largely from the ease with which even gram quantities
of con A can be produced, which allows the design of
experiments that are not possible with most other car-
bohydrate binding proteins. The body of structural
and thermodynamic data that is available for legume
lectins has also provided an impetus to predict thermo-
dynamic binding parameters using calculations based
on the crystal structures [24–27].
The seeds of the bloodwood tree ( Pterocarpus ango-
lensis) are rich in a Man ⁄ Glc specific lectin (PAL).
This lectin was recently isolated, its encoding gene

sequenced [21] and its crystal structure determined in
complex with nine mono-, di- and trisaccharides
[21,22]. As in other legume lectins, the carbohydrate
binding site consists of five loops which have been des-
ignated A-E [7,15]. These loops form a groove on the
surface of the protein in which an oligosaccharide can
bind (Fig. 1). Central in this groove is the monosac-
charide binding site (M), which is flanked by a series
of subsites. The subsites harboring additional sac-
charide residues linked to O1 of the mannose in the
monosaccharide binding site are designated the ‘down-
stream’ subsites +1, +2, +3, while those linked to
O2 form the ‘upstream’ subsite )1 (Fig. 1).
Here, we present the crystal structures of PAL in its
ligand-free state, in complex with the complex type
decasaccharide NA2F and in complex with four fur-
ther oligosaccharides that are fragments of NA2F. The
structural information is complemented by thermody-
namic parameters of binding determined by means of
isothermal titration calorimetry for the above-men-
tioned complexes as well as other complexes for which
the crystal structures were determined previously [21].
Our results indicate that the complete oligosaccharide
binding site of PAL consists of four extended sites
surrounding the primary monosaccharide binding
sites, conferring a maximal affinity for the pentasaccha-
ride GlcNAcb(1–2)Mana(1–3)[GlcNAcb(1–2)Mana(1–
6)]Man. Although this arrangement bears many simi-
larities to what is observed in con A, there are clear
differences in the conformational details and the

subsite energetics between both lectins. Thus for the
first time a detailed combined thermodynamic and
crystallographic study can be placed next to the body
of data published on con A. From this we see how
two evolutionarily related proteins can show the same
oligosaccharide specificity but yet show a different pat-
tern of subsite affinities and different interactions in
those subsites.
Fig. 1. Architecture of the carbohydrate bin-
ding site of a Man ⁄ Glc specific lectin (PAL).
Shown is a stereo-view of a CPK (Corey
Pauling Koltun) representation of the carbo-
hydrate binding site. The five different loops
that make up the binding site are colored:
loop A: light blue, loop B: orange, loop C:
yellow, loop D: green, loop E: dark blue. The
monosaccharide binding site is indicated by
a ball-and stick model of mannose. The
relative positions of the upstream ()1) and
downstream (+1, +2, +3) subsites are
indicated as well.
Carbohydrate recognition by PAL L. Buts et al.
2408 FEBS Journal 273 (2006) 2407–2420 ª 2006 The Authors Journal compilation ª 2006 FEBS
Results
Structure of unliganded PAL and its complex
with mannose
A correct understanding of ligand binding not only
requires structural data on the protein-carbohydrate
complexes, but also knowledge of the structure of the
ligand-free protein. As we were unable to obtain crystals

of ligand-free PAL directly, we opted to remove the
bound ligand from a PAL:Mana(1–3)Man complex by
reverse soaking (see Experimental procedures, and [28]).
These crystals contain the PAL dimer in their asymmet-
ric unit, the two subunits of which are labeled A and B.
The structure resulting from this experiment, refined at
1.8 A
˚
to R ¼ 0.185 and R
free
¼ 0.205 (Table 1) shows a
clear Mana(1–3)Man disaccharide bound in the binding
site of subunit A, which is involved in crystal lattice
interactions. It is in all respects identical to subunit A of
the original Mana(1–3)Man complex described earlier
[21]. The binding site of subunit B on the other hand,
which is not involved in crystal packing, is devoid of any
electron density that could be interpreted as carbohy-
drate. Instead the monosaccharide binding site contains
four ordered water molecules (B-factors 22, 29, 49 and
40 A
˚
2
) which are expelled upon binding of mannose
(Fig. 2A). Three of them closely mimic the positions of
hydroxyls OH3, OH4 and OH6 of mannose, which
binds in a lock-and-key fashion without significant
changes in protein conformation.
The binding of mannose to PAL is essentially identi-
cal to what is observed in the ManaMe complex pre-

sented before [21] and is summarized in Table 2. The
anomeric oxygen is entirely in the a-configuration in
the binding site of subunit A, but adopts an a ⁄ b mix-
ture in the binding site of subunit B. Accommodation
of the b-configuration in the binding site of PAL
requires a change in the rotamer conformation of
Glu221 (Fig. 2B). Steric hindrance is observed only if
the b-anomeric oxygen contains an additional substitu-
ent such as a methyl group. The side chain of Glu221
is completely disordered in the ligand-free structure.
Crystal structure of the NA2F decasaccharide
complex
The decasaccharide Galb(1–4)GlcNAcb(1–2)Mana(1–
6)[Galb(1–4)GlcNAcb(1–2)Mana(1–3)]Manb(1–4)Glc-
NAcb(1–4)[Fucb(1–6)]GlcNAc (NA2F) was only
available in small quantities (20 lg) and could be used
for a single soaking experiment. While there is defin-
itely not enough space to accommodate such a large
carbohydrate in the binding site of subunit A in our
PAL crystals, the space around binding site B is ample
and by all means sufficient to accommodate a decasac-
charide. In order not to risk destroing the crystal dur-
ing the experiment, the soak was performed on a
previously ‘desoaked’ Mana(1–3)Man cocrystal, which
as described above still retains the disaccharide in the
binding site of subunit A, but has an empty binding
Table 1. Data collection and refinement statistics.
Ligand-
removed Mannose
GlcNAc

(b1–2)Man
GlcNAc(b1–2)Man
(a1–3)ManOMe M592 NA2F
Space group P2
1
2
1
2
1
P2
1
2
1
2
1
P2
1
2
1
2
1
P2
1
2
1
2
1
P2
1
2

1
2
1
P2
1
2
1
2
1
Beamline X11 X11 X11 X11 X11 ID14-1
Unit cell:
a ¼ 56.86 A
˚
56.73 A
˚
57.28 A
˚
56.80 A
˚
56.71 A
˚
56.78 A
˚
b ¼ 83.21 A
˚
83.36 A
˚
82.68 A
˚
83.14 A

˚
83.59 A
˚
83.32 A
˚
c ¼ 123.23 A
˚
122.94 A
˚
125.00 A
˚
123.02 A
˚
122.37 A
˚
122.38 A
˚
Resolution limits 15.0–1.80 A
˚
15.0–1.70 A
˚
15.0–1.80 A
˚
15.0–1.95 A
˚
15.0–1.80 A
˚
15.0–2.00
Number of measured reflections 681254 305125 214056 170746 285333 123639
Number of unique reflections 54780 53278 54204 43030 54417 34590

Completeness 99.7% 98.1% 98.4% 97.9% 98.8% 85.8%
R
merge
0.080 0.050 0.078 0.091 0.052 0.054
<I⁄ r(I) > 24.2 22.0 13.2 11.7 12.0 12.0
R-factor 0.185 0.170 0.179 0.179 0.176 0.176
R
free
-factor 0.205 0.204 0.212 0.199 0.201 0.216
Ramachandran plot
core 86.9% 88.0% 88.8% 87.1% 87.2% 86.2%
additionally allowed 12.9% 12.0% 11.2% 12.9% 12.1% 13.8%
disallowed 0.2% 0.0% 0.0% 0.0% 0.0% 0.0%
PDB code 1S1A 2ARE 2ARB 2AUY 2AR6 2ARX
L. Buts et al. Carbohydrate recognition by PAL
FEBS Journal 273 (2006) 2407–2420 ª 2006 The Authors Journal compilation ª 2006 FEBS 2409
site for subunit B. The crystal structure indeed shows
the binding site of subunit A to be occupied with
Mana(1–3)Man.
In the binding site of subunit B, electron density for
a heptasaccharide (GlcNAcb(1–2)Mana(1–6)[Galb(1–
4)GlcNAcb(1–2)Mana(1–3)]Manb(1–4)GlcNAcb)is
clearly visible (Fig. 3A). Binding of this heptasaccha-
ride shields a total surface of 1045 A
˚
2
(protein and car-
bohydrate) from the solvent. The Mana(1–3)[Mana(1–
6)]Man moiety is bound as in the previously described
complex with the core trimannose [21], involving three

direct hydrogen bonds with the core mannose (+1
subsite) and two with the 1–3 linked mannose (+2
subsite) (Fig. 3B and Table 2). The mannose on the
1–6 arm occupies the monosaccharide binding site.
The GlcNAc residue on the 1–6 arm is bound in the
)1 subsite and makes two hydrogen bonds with the
protein. The GlcNAc residue on the 1–3 arm binds in
the +3 subsite, making two hydrogen bonds with the
side chain of Asn83 as well as van der Waals contacts
with the side chains of Leu81 and Gln222. The galac-
tose on the 1–3 arm has defined electron density, but
does not make any direct hydrogen bonds with the
protein while the galactose on the 1–6 arm is com-
pletely disordered. Electron density is also visible for
an additional GlcNAc residue linked b1–4 to the core
mannose, but this sugar residue does not contact the
protein. The electron density for this GlcNAc residue
is well defined, probably due to conformational restric-
tions. The N-linked GlcNAc and the fucose branch are
completely disordered. All observed glycosidic linkages
are near their global low energy conformation with the
exception of the GlcNAcb(1–2)Man linkage on the 1–6
arm, which occupies a secondary minimum (Fig. 4).
Fig. 2. The monosaccharide binding site of
PAL. (A) Stereo view of the monosaccharide
binding site of PAL in absence of bound car-
bohydrate. The electron density for four
water molecules that occupy the binding
site is shown, together with the hydrogen
bond network these waters make with the

protein. Superimposed in black are the
equivalent residues of the mannose com-
plex (subunit B in the asymmetric unit). The
water molecules clearly mimic the oxygen
atoms of bound mannose (black, thick lines).
(B) Stereo view of the interactions of
mannose in the monosaccharide binding
site. The beta-anomeric oxygen position of
mannose is drawn in light green, as is the
corresponding orientation of the side chain
of Glu221. The second conformation of the
side chain of Glu221, which also adopts two
conformations but not correlated with the
anomeric form of the bound mannose is
shown in dark green.
Carbohydrate recognition by PAL L. Buts et al.
2410 FEBS Journal 273 (2006) 2407–2420 ª 2006 The Authors Journal compilation ª 2006 FEBS
Structure of PAL in complex with
GlcNAcb(1–2)Man
GlcNAcb(1–2)Man binds with mannose in the mono-
saccharide binding site and with its nonreducing
GlcNAc in the upstream )1 subsite (Table 2 and
Fig. 5), however, using a binding mode that is distinct
from the one observed in the NA2F complex. In this
binding mode, a total surface of 615 A
˚
2
is shiel-
ded from the solvent, compared to 565 A
˚

2
for the
Table 2. Interactions between lectin and carbohydrates (all distances in A
˚
). NP, not present. The values for the binding sites of subunit A
and subunit B are separated by ( ⁄ ).
H-bond Mannose GlcNAcb(1–2)Man M592 NA2F
Site )1 Gly102(O)

GlcNAc(O4) NP 2.81 ⁄ 2.76 2.77 ⁄ 2.65 NP
Ser45(OG)

GlcNAc(O7) NP 3.48 ⁄ 3.45 3.28 ⁄ 3.38 NP
Gly104(O)

GlcNAc(N2) NP NP NP NP ⁄ 3.46
Glu221(OE2)

GlcNAc(O6) NP NP NP NP ⁄ 3.22
Site M Asp86(OD1)

Man(O4) 2.60 ⁄ 2.58 2.60 ⁄ 2.54 2.56 ⁄ 2.50 2.56 ⁄ 2.78
Asp86(OD2)

Man(O6) 2.85 ⁄ 2.78 2.85 ⁄ 2.75 2.78 ⁄ 2.77 2.75 ⁄ 2.84
Gly106(N)

Man(O3) 2.86 ⁄ 2.94 2.97 ⁄ 2.99 2.83 ⁄ 2.80 2.96 ⁄ 2.84
Asn138(ND2)

Man(O4) 3.00 ⁄ 2.92 2.93 ⁄ 2.93 2.95 ⁄ 2.93 3.15 ⁄ 2.98
Glu221(N)

Man(O5) 3.00 ⁄ 3.01 3.01 ⁄ 3.01 3.12 ⁄ 3.05 3.18 ⁄ 2.99
Gln222(N)

Man(O6) 2.99 ⁄ 3.06 3.01 ⁄ 3.07 3.03 ⁄ 3.01 3.01 ⁄ 3.05
Site +1 Asp136(OD2)

Man(O2) NP NP NP ⁄ 2.58 NP ⁄ 2.62
Ser137(OG)

Man(O2) NP NP NP ⁄ 2.74 NP ⁄ 2.64
Gln222(NE2)

Man(O4) NP NP NP ⁄ 2.62 NP ⁄ 2.46
Site +2 Asn83(ND2)

Man(O3) NP NP NP ⁄ 2.84 NP ⁄ 3.02
Asp136(OD1)

Man(O6) NP NP NP ⁄ 2.69 NP ⁄ 2.75
Site +3 Asn83(OD1)

GlcNAc(O6) NP NP NP ⁄ 2.71 NP ⁄ 2.81
Asn83(ND2)

GlcNAc(O5) NP NP NP ⁄ 2.84 NP ⁄ 2.88
Fig. 3. (A) Electron density for NA2F in the binding site of subunit B of PAL. The map shown is a simulated-annealing OMIT map calculated
after removal of the carbohydrate from the model and contoured at 3 sigma. Clear density is seen for seven out of 10 sugar residues. The

final model of the carbohydrate is superimposed. (B) Stereo view of the interactions between NA2F and PAL. The heptasaccharide moiety
of NA2F that is visible in the electron density map is drawn in green ball-and-stick. Protein residues that are part of the )1, +1, +2 and +3
subsites are coloured accoring to atom type. Hydrogen bonds are shown as dotted lines. The amino acids that make up the monosaccharide
binding site are omitted for clarity.
L. Buts et al. Carbohydrate recognition by PAL
FEBS Journal 273 (2006) 2407–2420 ª 2006 The Authors Journal compilation ª 2006 FEBS 2411
GlcNAcb-(1–2)Man moiety from NA2F. The confor-
mation of the b(1–2) linkage is near the global energy
minimum. Extensive van der Waals contacts are made
between GlcNAc and the backbone atoms of loop B
(as defined by Sharma & Surolia, [7]). The conforma-
tion of the GlcNAcb(1–2)Man disaccharide is stabil-
ized by an intramolecular hydrogen bond from
O3(Man) to O5(GlcNAc). GlcNAc makes a direct
hydrogen bond with its O4 hydroxyl to the carbonyl
group of Gly102. A second potential, but weaker
hydrogen bond may be present between O7 of the
N-acetyl group and the side chain hydroxyl of Ser45.
O3 of GlcNAc is bridged via a water molecule
(Wat116) to the backbone NH group of Gly104 while
another water (Wat117) bridges O7 of GlcNAc to the
backbone NH of Gly105 and the backbone carbonyl
of Gly219. Both waters are conserved in the sugar-free
protein as well as in all structures where the )1 subsite
is not occupied [21,22].
Crystal structure of the M592 pentasaccharide
complex
Clear electron density is seen for the complete penta-
saccharide GlcNAc b(1–2)Mana(1–3)[GlcNAcb(1–
2)Mana(1–6)]Man (M592) in the binding site of sub-

unit B (Fig. 6A), which is adjacent to a large solvent
channel in the crystal and far away from any lattice
contact. Therefore, it is assumed that the interactions
observed in binding site B correspond to the situation
in solution. Binding of the pentasaccharide shields a
total surface of 1030 A
˚
2
from the solvent, a value
almost identical to that of NA2F. The tetrasaccharide
moiety GlcNAcb(1–2)Mana(1–3))[Mana(1–6)]Man
binds in an identical way as seen in the complex with
NA2F (Fig. 6B). The GlcNAc residue in the )1 sub-
site, however, is oriented as is the GlcNAcb(1–2)Man
complex (Fig. 6B). All glycosidic bonds adopt confor-
mations that correspond to energy minima in the cal-
culated energy landscapes (Fig. 4).
The same binding mode is not possible in binding
site A due to steric overlap with a symmetry-related
molecule of PAL. As a consequence, only the )1 and
primary sites are occupied by a GlcNAcb(1–2)Man
moiety in the A subunit, with some ill-defined density
indicating the disordered binding of the remaining
three monosaccharides.
Fig. 4. Rigid conformational energy maps in function of Phi and Psi
for the different linkages observed in the pentasaccharide M592
and NA2F bound to PAL, con A and LOL. (upper) a(1–6) linkages.
(middle) a(1–3) linkages. (lower) b(1–2) linkages. In each case,
energy levels are contoured at 5 kcal ⁄ mol intervals starting from
the minimum energy. The omega angle of Mana(1–6)Man was

fixed in the gauche ⁄ trans (gt) conformation. The conformations
observed in the crystal structures are indicated.
Carbohydrate recognition by PAL L. Buts et al.
2412 FEBS Journal 273 (2006) 2407–2420 ª 2006 The Authors Journal compilation ª 2006 FEBS
Structure of PAL in complex with
GlcNAcb(1–2)Mana(1–3)ManaMe
GlcNAcb(1–2)Mana(1–3)ManaMe corresponds to the
1–3 arm of the pentasaccharide M592. In the M592
and NA2F complexes this trisaccharide moiety occu-
pies the +1, +2 and +3 subsites, whereas by itself it
adopts a different binding mode and occupies the )1
(GlcNAc), M (Man) and +1 (ManaMe) subsites
(Fig. 7). The binding of GlcNAcb(1–2)Mana(1–3)Man-
Fig. 5. (A) Electron density for GlcNAcb(1–2)Man in the binding site of subunit A of PAL. (B) Interactions of PAL with GlcNAcb(1–2)Man. The
protein is coloured according to atom type. The disaccharide is colored green. Selected residues as well as the sugar residues occupying
subsites M and )1 are labeled. The co-ordinates of the GlcNAcb(1–2)Man moiety on the 1–6 arm of the decasaccharide NA2F is superim-
posed in thin black lines.
Fig. 6. (A) Electron density for the pentasaccharide M592 in the binding site of subunit B of PAL. The map is calculated and drawn as in
Fig. 3. (B) Interactions of M592 with PAL. The pentasaccharide is shown in green, protein atoms are coloured accoring to atom type. The
equivalent pentasaccharide from NA2F is superimosed in thin black lines.
L. Buts et al. Carbohydrate recognition by PAL
FEBS Journal 273 (2006) 2407–2420 ª 2006 The Authors Journal compilation ª 2006 FEBS 2413
aMe to PAL is thus a linear combination of what
is observed for Mana(1–3)Man21 and GlcNAcb
(1–2)Man, shielding a total surface of 785 A
˚
2
from the
solvent. This is in agreement with the general observa-
tion on lectins that there is a primary monosaccharide

binding site that needs to be occupied first before bind-
ing to adjacent subsites will occur.
Thermodynamics of carbohydrate binding to PAL
To complement our structural studies, the thermody-
namic parameters for the binding to PAL for man-
nose, GlcNAcb(1–2)Man, the pentasaccharide M592 as
well a number of other mono-, di- and trisaccharide
constituents of M592 were measured by isothermal
titration calorimetry. The results are summarized in
Table 3. As is usually observed for protein:carbohy-
drate interactions, the binding constants are in the mil-
limolar range.
Binding of mannose in the primary binding site is
enthalpy-driven. Also for most oligosaccharides, the
entropy contributions remain unfavorable. Ligand
binding in the downstream +1 subsites is not mirrored
by a significantly enhanced affinity compared to man-
nose and the differences in their thermodynamic
parameters are close to or perhaps within the error
limits of the measurements. This contrasts with the
well-defined carbohydrate conformations observed in
the crystal structures of the Mana(1–3)Man, Mana(1–
4)Man and Mana(1–6)Man complexes [21], a situation
also observed in con A [29] (often carbohydrate resi-
dues have a well defined conformation only if they
contribute to affinity). In the case of the core trisac-
charide Mana(1–3)[Mana(1–6)]Man there is a small
increase in affinity and the additional carbohy-
drate:protein contacts are mirrored by a more favora-
ble enthalpy of binding.

On the other hand, binding of N-acetylglucosamine
to the upstream )1 subsite contributes to a modest (15
fold) increase of affinity (from 1.9·10
3
m
)1
to 26·10
3
m
)1
, Table 3). Surprisingly, for GlcNAcb(1–2)Man,
occupation of the upstream binding site by GlcNAc is
entropy-driven. A loss of 2.7 kcalÆmol
)1
in binding
enthalpy is compensated by a 4.3 kcalÆmol
)1
gain in
TDS°. Although odd, this result is most likely real and
not a consequence of the correlation of fitting parame-
ters often observed for weak binding events. Evidence
for this stems from the reproducibility of this result as
well as from the c-value used in the titration (34.7)
which allows a meaningful separation of DG° into DH°
and TDS° terms.
Discussion
Thermodynamics versus structure
The most surprising observation in the present study
is the entropy-favorable binding in the )1 subsite of
Fig. 7. (A) Electron density for GlcNAcb(1–2)Mana(1–3)ManaMe in the binding site of subunit A of PAL. (B) Interactions of PAL with Glc-

NAcb(1–2)Mana(1–3)ManaMe. The trisaccharide is shown in green, protein atoms are coloured accoring to atom type. The equivalent trisac-
charide from the NA2F complex is superimosed in thin black lines.
Carbohydrate recognition by PAL L. Buts et al.
2414 FEBS Journal 273 (2006) 2407–2420 ª 2006 The Authors Journal compilation ª 2006 FEBS
PAL, which is only rarely seen in protein-carbohy-
drate recognition systems. One potential explanation
for this unexpected observation would be a sliding
mechanism of binding. Such a mechanism was pro-
posed to explain thermodynamic data for the binding
of Mana(1–2)Man and GlcNAcb(1–2)Man to con A
[30], and was later confirmed for Mana(1–2)Man by
X-ray crystallography [31]. In the case of PAL such a
sliding binding mechanism seems unlikely. GlcNAc is
always seen to occupy the )1 subsite and never the
monosaccharide binding site. When GlcNAc is mode-
led in the monosaccharide binding site, severe steric
clashes are observed between the mannose residue and
the protein for all conformations due to the b(1–2)
linkage. A second possibility is that the b(1–2) linkage
adopts two different conformations, both of which
result in favorable interactions between GlcNAc and
the protein. Indeed, a secondary minimum is occupied
in the NA2F complex, but this is most likely due to
the presence of an additional galactose residue that
prevents binding in the global minimum conforma-
tion. Neither in the GlcNAcb(1–2)Man complex, nor
in the complexes with M592 or GlcNAcb(1–2)
Mana(1–3)ManaMe is there any evidence for two
conformations despite that they are not prevented by
crystal lattice contacts. Thus, if the ligand in the con-

formation corresponding to the secondary minimum
of the b(1–2) linkage binds to PAL in solution, it is
almost certainly a minority binding mode (< 10%)
and would not significantly affect the outcome of the
ITC titration experiments.
The biantennary pentasaccharide GlcNAcb(1–
2)Mana(1–3)[GlcNAcb(1–2)Mana(1–6)]Man (M592)
shows the highest affinity of all carbohydrates tested.
It is potentially bivalent, but analysis of the ITC titra-
tion data indicates a 1 : 1 stoichiometry, in agreement
with crystallographic data. Here, the formal possibility
of a backwards binding mode [with the a(1–3) arm
occupying the )1 and M subsites as also observed for
GlcNAcb(1–2)Mana(1–3)ManaMe] needs to be consid-
ered. Again, no evidence for this is seen in the crystal
structure despite that this would not be sterically hin-
dered by lattice interactions in the binding site of sub-
unit B. Thus, most likely, the backwards binding mode
will again be a minority species in solution and not sig-
nificantly influence the binding data.
Comparison with related systems
The complex of PAL with NA2F is one of the few
complexes between a lectin and a large biantennary
glycan that have been studied using X-ray crystallogra-
phy. The Man ⁄ Glc-specific lectins from the Viciae tribe
have their highest affinity for N-acetyllactosamine type
N-glycans bearing a fucose a(1–6) linked to the Asn-
linked GlcNAc such as NA2F [32,33]. In contrast to
PAL, the lectin from Lathyrus ochrus (LOL) binds
with its a1–3 arm in the monosaccharide binding site

[34]. The conformation of the GlcNAcb(1–2)Man
entity occupying the )1 and M subsites of LOL
roughly resembles the conformation seen in the NA2F
complex of PAL (Fig. 4). LOL binds the fucosylated
Table 3. Thermodynamic parameters of saccharide binding to PAL.
Carbohydrate
No. of
expts c-value Stoichiometry*
K
ass
(10
3
M
)1
)
DG
0
(kcalÆmol
)1
)†
DH
0
(kcalÆmol
)1
)†
TDS
0
(kcalÆmol
)1
)†

Mannose 4 3.8 0.90 1.9 )4.4 )6.3 )1.9
ManaMe 3 5.4 0.95 3.4 )4.8 )6.5 )1.7
Mana(1–2)Man 3 13.0 0.92 15.4 )5.7 )7.0 )1.3
Mana(1–3)Man 4 2.1 0.97 2.3 )4.6 )5.7 )1.1
Mana(1–4)Man 3 2.1 0.92 3.4 )4.8 )6.4 )1.6
Mana(1–6)Man 3 1.6 1.04 2.0 )4.5 )6.5 )2.0
Mana(1–3)[Man 3 1.4 0.95 5.6 )5.1 )7.1 )2.0
(a1–6)]Man
Mana(1–3)[Man 3 2.3 0.95 6.6 )5.2 )8.7 )3.5
a(1–3)[Mana(1–6)
]Mana(1–6)]Man
GlcNAc(b1–2)Man 3 34.7 0.90 26.0 )6.0 )3.6 2.4
GlcNAcb(1–2)Man 3 59.0 1.10 63.0 )6.5 )7.9 )1.4
a(1–3)[GlcNAcb(1–2)
Mana(1–6)]Man
* Obtained from fitting the ITC data. The reported values for K
ass
, DG
0
, DH
0
and TDS
0
are determined with n fixed at 1.0. These values do
not differ significantly from those obtained when treating the number of binding sites as a variable. † The errors on DG
0
and DH
0
are of the
order of magnitude of 0.1 kcalÆmol

)1
while for TDS
0
they are 0.2 kcalÆmol
)1
.
L. Buts et al. Carbohydrate recognition by PAL
FEBS Journal 273 (2006) 2407–2420 ª 2006 The Authors Journal compilation ª 2006 FEBS 2415
chitobiose stem in its downstream subsites, while the
a1–6 arm points away from the protein towards the
solvent. PAL apparently lacks the fucose-recognizing
subsite of LOL.
Within NA2F, the pentasaccharide GlcNAcb(1–
2)Mana(1–3)[GlcNAcb(1–2)Mana(1–6)]Man (M592)
seems to be the largest entity that is specifically recog-
nized by PAL as evidenced from our combined crystal-
lographic and thermodynamic results. The same
pentasaccharide is the largest monovalent determinant
specifically recognized by con A [30,35]. Significant dif-
ferences are observed between PAL and con A from the
structural as well as the thermodynamic viewpoint. Both
proteins have the a(1–6)-linked mannose in their mono-
saccharide binding site as well as a GlcNAc residue in
the same orientation in the )1 subsite. The )1 subsite,
which mainly consists of loop B (as defined by Sharma
& Surolia [7]), is well-conserved between con A and
PAL. The only relevant substitution concerns Gly104
of PAL which is Thr226 in con A. The side chain of
Thr226 occupies the space taken by the conserved
waters 116 and 117 of PAL and makes a hydrogen bond

to O3 of GlcNAc (Fig. 8A). Binding of GlcNAc in the
)1 subsite of PAL is energetically favorable and contri-
butes significantly to the higher affinity of PAL for the
pentasaccharide M592 (Table 3). In the case of con A
on the other hand, binding of GlcNAc to the )1 subsite
does not affect affinity [30]. This was attributed to a
strained conformation of the disaccharide in the com-
plex of con A with pentasaccharide M592 [35]. These
conclusions were, however, based on older, less accurate
energy maps [36], which later have been updated [37].
As can be seen in Fig. 4, this linkage conformation
observed in the con A complex is very close to those
observed in all PAL complexes, except for the NA2F
complex and corresponds to a low energy conformation.
Fig. 8. Comparison between PAL and con
A. (A) Comparison between con A and PAL.
Stereo view of the binding of M592 (green)
to the )1 and M subsites of con A (colored
according to atom type). The corresponding
situation in PAL is superimposed as shown
in thin black lines. (B) Downstream binding
sites. Stereo view of the binding of M592
(green) to the +1, +2 and +3 subsites of
con A (colored according to atom type). The
corresponding situation in PAL is super-
imposed as shown in thin black lines.
Carbohydrate recognition by PAL L. Buts et al.
2416 FEBS Journal 273 (2006) 2407–2420 ª 2006 The Authors Journal compilation ª 2006 FEBS
In the latter complex this conformation is made imposs-
ible by the additional galactose substituted on O4 of

GlcNAc, as is also observed in LOL:biantennary gly-
cans complexes [34]. Therefore, we conclude that the
reason for the difference in energetics of the )1 subsite
between PAL and con A must be due to details in the
interaction between the protein and the carbohydrate
and not to the carbohydrate conformation.
Downstream from the monosaccharide binding site,
both energetics and oligosaccharide conformations dif-
fer between PAL and con A. The +1, +2 and +3
subsites are dominated in both proteins by interactions
with the metal binding loop C (as defined by Sharma
& Surolia, [7]). These differ in length, sequence and
conformation between both lectins, making the down-
stream subsites nonsuperimposable (Fig. 8B). In both
PAL and con A [30], the mannose residue bound in
the +1 subsite does not contribute to affinity but func-
tions as a hinge. For con A, interactions of mannose
in the +2 subsite contribute dominantly to its higher
affinity for the pentasaccharide M592 compared to
methyl-a-d-mannopyrannoside [30]. For PAL on the
other hand, binding of mannose this subsite, although
favorable, contributes only moderate to the overall
affinity of M592 to the lectin. For con A, a small
improvement in interaction energy is also observed for
GlcNAc binding in the +3 subsite [30], while for the
lectin:ligand complexes of PAL this is neutral.
In conclusion, the combination of structural biology
and thermodynamics has allowed us to obtain new
insights in the mechanisms that govern protein–oligo-
saccharide interactions. To explain thermodynamics

from structure nevertheless remains a challenging
objective. The detailed characterization of two closely
related systems, PAL and con A, can be used as a
basis for theoretical calculations and force-field valid-
ation and may help towards correct prediction and a
true understanding of biological recognition events.
Experimental procedures
Protein and sugars
Extracts from uncoated ground and defatted P. angolensis
seeds were fractionated with ammonium sulfate. The
30–60% ammonium sulfate fraction was suspended in
100 mm NaCl +10 mm phosphate buffer pH 7.4 and then
dialyzed extensively against the same buffer. After removal
of all remaining insoluble material by centrifugation
(30 min at 24 000 g), the resulting solution was applied to a
fetuin-sepharose column. After extensive washing to remove
all unbound protein, the lectin was eluted using 0.3 m man-
nose in 100 mm NaCl +10 mm phosphate buffer pH 7.4.
The protein was subsequently extensively dialyzed against
50 mm phosphate pH 7.2 +150 mm NaCl. Protein concen-
trations were determined from UV absorption spectroscopy
at 280 nm using the specific absorption coefficient e of
37410 m
)1
Æcm
)1
as calculated according to Gill and von
Hippel [38].
The trisaccharide GlcNAcb(1–2)Mana(1–3)ManaMe was
synthesized as described [39]. All other carbohydrates were

purchased from DEXTRA laboratories and are claimed to
be 98% pure by the manufacturer. Solutions for ITC titra-
tions were prepared by weighting the total amount of sugar
on a microbalance prior to dissolving it into the buffer used
for the final dialysis of the protein. The errors on the sugar
concentrations should be smaller than 2% for the oligosac-
charides and negligible for mannose and methyl-a-d-man-
nopyrannoside.
Crystal structure determination
Attempts to crystallize PAL in absence of any carbohydrate
ligands failed. To obtain crystals of ligand-free PAL, a co-
crystal of the Mana(1–3)Man complex [21] was transferred
to artificial mother liquor (100 mm Na-cacodylate pH 6.5,
200 mm Ca-acetate, 20% (w ⁄ v) PEG-8000) devoid of car-
bohydrate ligand as described [28]. This reverse-soaking
experiment was continued for four weeks while transferring
the crystal to fresh solution every week. The crystal did not
show any visible signs of damage after this procedure.
To obtain data for the complexes of PAL with different
sugars, crystals of the Mana(1–3)Man complex [21] were
soaked for 1 h in artificial mother liquor enriched with
100 mm of the desired sugar as described [28]. In the case
of the decasaccharide, the soaking procedure was per-
formed on a previously ‘desoaked’ crystal. Crystals were
mounted in glass capillaries for data collection at room
temperature on EMBL beamlines X11 (DESY, Hamburg,
Germany) and ID14-1 (ESRF, Grenoble, France). The data
were processed with denzo and scalepack [40]. Data were
integrated with denzo, merged with scalepack and conver-
ted to structure factor amplitudes using the CCP4 program

truncate [41]. The statistics of the data collections are
given in Table 1.
The crystal structure of the PAL:Mana(1–3)Man com-
plex, stripped of water molecules, metal ions and carbo-
hydrate ligands, was used as the starting model in
refinement with CNS 1.0 [42]. After an initial rigid body
refinement, a slow cool stage was used to uncouple R
and R
free
. From then on restrained positional and B-fac-
tor refinements were alternated with manual fitting in
electron density maps using turbo [43]. The refinement
statistics are given in Table 1. Superpositions of crystal
structures were performed using turbo. All were pro-
duced using molscript [44], raster3d [45] and pymol
( />L. Buts et al. Carbohydrate recognition by PAL
FEBS Journal 273 (2006) 2407–2420 ª 2006 The Authors Journal compilation ª 2006 FEBS 2417
Isothermal titration calorimetry
ITC measurements were performed using a Microcal
OMEGA microcalorimeter. The instrument was calibrated
by standard heat injections according to the user’s manual
and by titrating the RNase A ⁄ 2¢CMP binding reaction.
Protein (0.6–1.0 mm) and sugar (10–50 mm) were dissolved
in 50 mm Na-phosphate pH 7.2, 150 mm NaCl, 1 mm
CaCl
2
and 1 mm MnCl
2.
In individual titrations, 10 lL
injections were added at 180–240 s intervals from a compu-

ter-controlled 250 lL syringe to a working protein volume
of 1.3 mL under constant stirring at 350 rpm.
Experimental data were fitted using Microcal’s origin
5.0 software package to obtain the binding constant K
b
and
binding enthalpy DH
b
from the measured heat release Q
i
on each injection using the Wiseman isotherm [46]. Free
energy and entropy of binding are calculated from the well-
known relationships
DG
b
¼ÀRT lnðK
b
Þ
and
DG
b
¼ DH
b
À TDS
b
Errors on the thermodynamic parameters of binding were
derived from a statistical analysis of at least three repeti-
tions for each carbohydrate. The initial protein concentra-
tion [C]
0

, the carbohydrate concentrations in the syringe
[S]
0
and the volume V of the calorimeter cell were consid-
ered to be known constants. The experiments were carried
out under conditions where the c-value (n.K
a
.[C]
0
) varied
between 1 and 60. The heat of dilution of each ligand was
measured by running a series of analogous injections in
buffer solution. They were found to be negligible in all
cases. The number of binding sites n per protein subunit
was initially allowed to vary, but converged to values
between 0.9 and 1.05. As a consequence, n was fixed at one
site per subunit. For the pentasaccharide M592 and the tri-
saccharide Mana(1–3)[Mana(1–6)]Man, which are poten-
tially bivalent, the data were also fit assuming n ¼ 0.5.
These fits were clearly unsatisfactory as judged both visu-
ally and from the v
2
of the fits, confirming the 1 : 1 stoichi-
ometry that is also in agreement with our crystallographic
data. As an additional check to evaluate the effect of
potential errors on protein and ligand concentrations, DG
and DH were calculated assuming 5% lower or higher con-
centrations of protein and ligand (highly conservative esti-
mates for the errors on these concentrations). This led to
variations in those two parameters that were within

0.1 kcalÆmol
)1
.
Calculation of conformational energy maps
Energy maps are calculated using the tripos force-field [47]
together with the PIM parameterization [37] developed for
carbohydrates as a function of the F and Y dihedral angles
defined as: F ¼ O5-C1-O1-Cx¢ and Y ¼ C1-O1-Cx¢-CC(x
+1) for the b(1–2) and a(1–3) linkages. For a(1–6) link-
ages, three dihedral angles F ¼ O5-C1-O1-C6¢, Y ¼ C1-
O1-C6¢-C5¢-and x ¼ O1-C6¢-C5¢-O5¢ were used.
Acknowledgements
This work was made possible thanks to financial
support from the Vlaams Interuniversitair Instituut
voor Biotechnologie (VIB), the Fonds voor Wet-
enschappelijk Onderzoek Vlaanderen (FWO) and the
Onderzoeksraad of the Vrije Universiteit Brussel. The
authors acknowledge the use of synchrotron beamtime
at the EMBL beamlines at the DORIS storage ring
(Hamburg, Germany) and the ESRF (Grenoble
France). GJB and NA were supported by the NIH
Research Resource Center for Biomedical Complex
Carbohydrates (P41-RR-5351).
References
1 Taylor ME & Drickamer K (2003) Introduction to Gly-
cobiology. Oxford University Press, Oxford, UK.
2 Sharon N & Lis H (1990) Legume lectins: a large
family of homologous proteins. FASEB J 4, 3198–
3208.
3 Wang W, Peumans WJ, Rouge

´
P, Rossi C, Proost P,
Chen J & Van Damme EJ (2003) The leaves of the
lamiaceae species Glechoma hederacea (ground ivy) con-
tains a lectin that is structurally and evolutionary simi-
lar to the legume lectins. Plant J 33, 293–304.
4 Velloso LM, Svensson K, Pettersson RF & Lindqvist Y
(2003) The crystal structure of the carbohydrate-recog-
nition domain of the glycoprotein sorting receptor
p58 ⁄ ERGIC-53 reveals an unpredicted metal-binding
site and conformational changes associated with calcium
ion binding. J Mol Biol 334, 845–851.
5 Edelman GM, Cunningham BA, Reeke GN, Becker
JW, Waxdal MJ & Wang JL (1972) The covalent and
three-dimensional structure of concanavalin A. Proc
Natl Acad Sci USA 69, 2580–2584.
6 Hardman KD & Ainsworth CF (1972) structure of con-
canavalin at 2.4 A
˚
resolution. Biochemistry 11, 4910–
4919.
7 Sharma V & Surolia A (1997) Analyses of carbohydrate
recognition by legume lectins: size of the combining site
loops and their primary specificity. J Mol Biol 267, 433–
445.
8 Loris R, Hamelryck T, Bouckaert J & Wyns L (1998)
Legume lectin structure. Biochim Biophys Acta 1383,
9–36.
9 Prabu MM, Sankaranarayanan R, Puri KD, Sharma V,
Surolia A, Vijayan M & Suguna K (1998) Carbohydrate

specificity and quaternary association in basic winged
Carbohydrate recognition by PAL L. Buts et al.
2418 FEBS Journal 273 (2006) 2407–2420 ª 2006 The Authors Journal compilation ª 2006 FEBS
bean lectin: X-ray analysis of the lectin at 2.5 A
˚
resolu-
tion. J Mol Biol 276, 787–796.
10 Rozwarski DA, Swami BM, Brewer CF & Sacchettini
JC (1998) Crystal structure of the lectin from Dioclea
grandiflora complexed with core trimannoside of aspara-
gine-linked carbohydrates. J Biol Chem 273, 32818–
32825.
11 Hamelryck TW, Loris R, Bouckaert J, Dao-Thi MH,
Strecker G, Imberty A, Fernandez E, Wyns L & Etzler
ME (1999) Carbohydrate binding, quaternary structure
and a novel hydrophobic binding site in two legume lec-
tin oligomers from Dolichos biflorus. J Mol Biol 286,
1161–1177.
12 Manoj N, Srinivas VR, Surolia A, Vijayan M & Suguna
K (2000) Carbohydrate specificity and salt-bridge
mediated conformational change in acidic winged bean
agglutinin. J Mol Biol 302, 1129–1137.
13 Audette GF, Vandonselaar M & Delbaere LT (2000)
The 2.2 A resolution structure of the O(H) blood-
group-specific lectin I from Ulex europaeus. J Mol Biol
304, 423–433.
14 Hamelryck TW, Moore JG, Chrispeels MJ, Loris R &
Wyns L (2000) The role of weak protein–protein inter-
actions in multivalent lectin-carbohydrate binding: crys-
tal structure of cross-linked FRIL. J Mol Biol 299, 875–

883.
15 Imberty A, Gautier C, Lescar J, Perez S, Wyns L &
Loris R (2000) An unusual carbohydrate binding site
revealed by the structures of two Maackia amurensis lec-
tins complexed with sialic acid-containing oligosacchar-
ides. J Biol Chem 275, 17541–17548.
16 Loris R, De Greve H, Dao-Thi M-H, Messens J, Imber-
ty A & Wyns L (2000) Structural basis of carbohydrate
recognition by lectin II from Ulex europaeus, a protein
with a promiscuous carbohydrate-binding site. J Mol
Biol 301, 987–1002.
17 Rabijns A, Verboven C, Rouge P, Barre A, Van Dam-
me EJ, Peumans WJ & De Ranter CJ (2001) Structure
of a legume lectin from the bark of Robinia pseudo-
acacia and its complex with N-acetylgalactosamine.
Proteins 44, 470–478.
18 Svensson C, Teneberg S, Nilsson CL, Kjellberg A,
Schwarz FP, Sharon N & Krengel U (2002) High-reso-
lution crystal structures of Erythrina cristagalli lectin in
complex with lactose and 2¢-a-l-fucosyllactose and cor-
relation with thermodynamic binding data. J Mol Biol
321, 69–83.
19 Tempel W, Tschampel S & Woods RJ (2002) The xeno-
graft antigen bound to Griffonia simplicifolia lectin
1-B(4). X-ray crystal structure of the complex and
molecular dynamics characterization of the binding site.
J Biol Chem 277, 6615–6621.
20 Lescar J, Loris R, Mitchell E, Gautier C, Chazalet V,
Cox V, Wyns L, Pe
´

rez S, Breton C & Imberty A (2002)
Isolectins I-A and I-B of Griffonia (Bandeiria) simplici-
folia: crystal structure of metal-free GSI-B4 and molecu-
lar bases for metal binding and monosaccharide
specificity. J Biol Chem 277, 6608–6614.
21 Loris R, Van Walle I, De Greve H, Beeckmans S, Deb-
oeck F, Wyns L & Bouckaert J (2004) Structural basis
of oligomannose recognition by the Pterocarpus ango-
lensis seed lectin. J Mol Biol 335, 1227–1240.
22 Loris R, Imberty A, Beeckmans S, Van Driessche E,
Read JS, Bouckaert J, De Greve H, Buts L & Wyns L
(2003) Crystal structure of Pterocarpus angolensis lectin
in complex with glucose, sucrose and turanose. J Biol
Chem 278, 16297–16303.
23 Dam TK & Brewer CF (2002) Thermodynamic studies
of lectin–carbohydrate interactions by isothermal titra-
tion calorimetry. Chem Rev 102, 387–429.
24 Imberty A & Pe
´
rez S (1994) Molecular modeling of pro-
tein–carbohydrate interactions. Understanding the spe-
cificities of two legume lectins towards oligosaccharides.
Glycobiology 4, 351–366.
25 Bradbrook GM, Gleichmann T, Harrop SJ, Habash J,
Raftery J, Kalb J, Yariv J, Hillier IH & Helliwell JR
(1998) X-ray and molecular dynamics studies of conca-
navalin-A glucoside and mannoside complexes. Relating
structure to thermodynamics of binding. J Chem Soc
Faraday Trans 94, 1603–1611.
26 Bradbrook GM, Forshaw JR & Pe

´
rez S (2000) Struc-
ture-thermodynamics relationships of lectin-saccharide
complexes. The Erythrina corallodendron case. Eur
J Biochem 267, 4545–4555.
27 Bryce RA, Hillier IH & Naismith JH (2001) Carbohy-
drate-protein recognition: molecular dynamics simula-
tions and free energy analysis of oligosaccharide binding
to concanavalin A. Biophys J 81, 1373–1388.
28 Loris R, Garcia-Pino A, Buts L, Bouckaert J,
Beeckmans S, De Greve H & Wyns L (2005)
Crystallisation and crystal manipulation of the Ptero-
carpus angolensis seed lectin. Acta Crystallogr D61,
685–689.
29 Bouckaert J, Hamelryck TW, Wyns L & Loris R (1999)
The crystal structures of Man(a1–3)Man(a1-O)Me and
Man(a1–6)Man(a1-O)Me in complex with concanavalin
A. J Biol Chem 274, 29188–29195.
30 Mandal DK, Kishore N & Brewer CF (1994) Thermo-
dynamics of lectin–carbohydrate interactions. Titration
microcalorimetry measurements of the binding of
N-linked carbohydrates and ovalbumin to concanavalin
A. Biochemistry 33, 1149–1156.
31 Moothoo DN, Canan B, Field RA & Naismith JH
(1999) Mana(1–2)Mana-OMe-concanavalin A complex
reveals a balance of forces involved in carbohydrate
recognition. Glycobiology 9, 539–545.
32 Debray H, Decout D, Strecker G, Spik G & Montreuil
J (1981) Specificity of twelve lectins towards oligosac-
charides and glycopeptides related to N-glycosylpro-

teins. Eur J Biochem 117, 41–55.
L. Buts et al. Carbohydrate recognition by PAL
FEBS Journal 273 (2006) 2407–2420 ª 2006 The Authors Journal compilation ª 2006 FEBS 2419
33 Kornfeld K, Reitman ML & Kornfeld R (1981) The
carbohydrate-binding specificity of pea and lentil lectins.
Fucose is an important determinant. J Biol Chem 256,
6633–6640.
34 Bourne Y, Mazurier J, Legrand D, Rouge P, Montreuil
J, Spik G & Cambillau C (1994) Structures of a legume
lectin complexed with the human lactotransferrin N2
fragment, and with an isolated biantennary glycopep-
tide: role of the fucose moiety. Structure 2, 209–219.
35 Moothoo DN & Naismith JH (1998) Concanavalin A
distorts the bGlcNAc(1–2)Man linkage of bGlcNAc(1–
2)aMan(1–3)[bGlcNAc(1–2)aMan(1–6)]Man upon bind-
ing. Glycobiology 8, 173–181.
36 Imberty A, Delage MM, Bourne Y, Cambillau C &
Pe
´
rez S (1991) Data bank of three-dimensional struc-
tures of disaccharides: Part II, N-acetyllactosaminic type
N-glycans. Comparison with the crystal structure of a
biantennary octasaccharide. Glyconconj J 8, 456–483.
37 Imberty A & Pe
´
rez S (2000) Structure, conformation,
and dynamics of bioactive oligosaccharides: theoretical
approaches and experimental validations. Chem Rev
100, 4567–4588.
38 Gill SC & von Hippel PH (1989) Calculation of protein

extinction coefficients from amino acid sequence data.
Anal Biochem 182, 319–326.
39 Navarre N, Amiot N, van Oijen A, Imberty A, Poveda
A, Jime
´
nez-Barbero J, Cooper A, Nutley M & Boons
G-J (1999) Synthesis and conformational analysis of
a conformationally constrained trisaccharide, and
complexation properties with concanavalin A. Chem
Eur J 5, 2281–2294.
40 Otwinowski Z & Minor W (1997) Processing of X-ray
diffraction data collected in oscillation mode. Methods
Enzymol 276, 307–326.
41 Collaborative Computational Project, Number 4 (1994)
The CCP4 Suite: Programs for Protein Crystallography.
Acta Cryst D50, 760–763.
42 Bru
¨
nger AT, Adams PD, Clore GM, DeLano WL, Gros
P, Grosse-Kunstleve RW, Jiang JS, Kuszewski J, Nilges
M, Pannu NS, et al. (1998) Crystallography and NMR
system: a new software suite for macromolecular struc-
ture determination. Acta Cryst D54, 905–921.
43 Roussel A & Cambillau C (1989) TURBO-FRODO.
Silicon Graphic Geometry Partner Directory. pp. 71–78.
Silicon Graphics, Mountain View, CA, USA.
44 Kraulis PJ (1991) MOLSCRIPT: a program to produce
both detailed and schematic plots of protein structures.
J Appl Crystallogr 24, 946–950.
45 Merritt EA & Bacon DJ (1997) Raster3D: photorealistic

molecular graphics. Methods Enzymol 277, 505–524.
46 Wiseman T, Williston S, Brandts JF & Lung-Nan L
(1989) Rapid measurement of binding constants and
heats of binding using a new titration calorimeter. Anal
Biochem 197, 131–137.
47 Clark M, Cramer RDI & van den Opdenbosch N
(1989) Validation of the general purpose Tripos 5.2
force field. J Comput Chem 10, 982–1012.
Carbohydrate recognition by PAL L. Buts et al.
2420 FEBS Journal 273 (2006) 2407–2420 ª 2006 The Authors Journal compilation ª 2006 FEBS