The interaction of La3+ complexes of DOTA/DTPA glycoconjugates with the RCA120 lectin: a saturation transfer difference NMR spectroscopic study

Abstract

The study of ligand–receptor interactions using high-resolution NMR techniques, namely the saturation transfer difference
(STD), is presented for the recognition process between La(III) complexes of 1,4,7,10-tetrakis(carboxymethyl)-1,4,7,10-tetraazacyclododecane
monoamide and diethylenetriaminepentaacetic acid bisamide glycoconjugates and the galactose-specific lectin Ricinus communis agglutinin (RCA120). This new class of Gd(III)-based potential targeted MRI contrast agents (CAs), bearing one or two terminal sugar (galactosyl
or lactosyl) moieties, has been designed for in vivo binding to the asialoglycoprotein receptor, which is specifically expressed
at the surface of liver hepatocytes, with the aim of leading to a new possible diagnosis of liver diseases. The in vitro affinity
constants for the affinity of the divalent La(III)–glycoconjugate complexes for RCA120, used as a simple, water-soluble receptor model, were higher than those of the monovalent analogues. The combination of the
experimental data obtained from the STD NMR experiments with molecular modelling protocols (Autodock 4.1) allowed us to predict
the mode of binding of monovalent and divalent forms of these CAs to the galactose 1α binding sites of RCA120. The atomic details of the molecular interactions allowed us to corroborate and supported the interaction of both sugar moieties
and the linkers with the surface of the protein and, thus, their contribution to the observed interaction stabilities.

KeywordsLigand–receptor binding–Glycoconjugates–Saturation transfer difference NMR spectroscopy–MRI contrast agents–Protein–ligand interaction

Full text

Page 1

ORIGINAL PAPER
The interaction of La3+complexes of DOTA/DTPA
glycoconjugates with the RCA120lectin: a saturation
transfer difference NMR spectroscopic study
Joa ˜o M. C. Teixeira•David M. Dias•F. Javier Can ˜ada•Jose ´ A. Martins•
Joa ˜o P. Andre ´ •Jesu ´s Jime ´nez-Barbero•Carlos F. G. C. Geraldes
Received: 28 December 2010/Accepted: 19 March 2011
? SBIC 2011
Abstract
high-resolution NMR techniques, namely the saturation
transfer difference (STD), is presented for the recognition
process between La(III) complexes of 1,4,7,10-tetrakis
(carboxymethyl)-1,4,7,10-tetraazacyclododecane
mide and diethylenetriaminepentaacetic acid bisamide gly-
coconjugates and the galactose-specific lectin Ricinus
communis agglutinin (RCA120). This new class of Gd(III)-
based potential targeted MRI contrast agents (CAs), bearing
oneortwoterminalsugar(galactosylorlactosyl)moieties,has
been designed for in vivo binding to the asialoglycoprotein
receptor,whichisspecificallyexpressedatthesurfaceofliver
hepatocytes, with the aim of leading to a new possible
diagnosis of liver diseases. The in vitro affinity constants for
the affinity ofthe divalentLa(III)–glycoconjugate complexes
for RCA120, used as a simple, water-soluble receptor model,
were higher than those of the monovalent analogues.
The study of ligand–receptor interactions using
monoa-
The combination of the experimental data obtained from the
STD NMR experiments with molecular modelling protocols
(Autodock 4.1) allowed us to predict the mode of binding of
monovalent and divalent forms of these CAs to the galactose
1a binding sites of RCA120. The atomic details of the
molecular interactions allowed us to corroborate and sup-
ported the interaction of both sugar moieties and the linkers
with the surface of the protein and, thus, their contribution to
the observed interaction stabilities.
Keywords
Saturation transfer difference NMR spectroscopy ?
MRI contrast agents ? Protein–ligand interaction
Ligand–receptor binding ? Glycoconjugates ?
Introduction
Molecular recognition events are of paramount importance
in chemistry, biology and biomedicine. A large variety of
techniques allow the elucidation of binding events between
a ligand and its receptor. As key examples, enzyme-linked
immunosorbent assay [1], immunoblotting, radioimmuno-
assay [2], affinity chromatography [3] and surface plasmon
resonance experiments (Biacore) [4] are nowadays com-
monly employed for this task. In recent years, NMR-based
techniques [5] have become increasingly popular when
filling in the existing gap for characterization of molecular
binding processes at high resolution. Transferred nuclear
Overhauser effect spectroscopy [6], nuclear Overhauser
effect pumping [7] and water–ligand observed via gradient
spectroscopy (WaterLOGSY) [8, 9] are particular and
powerful examples of such approaches. Among them, the
saturation transfer difference (STD) NMR technique is
probably one of the most popular and robust methods
[5, 10–14]. This technique allows characterization of
J. M. C. Teixeira ? D. M. Dias ? C. F. G. C. Geraldes (&)
Department of Life Sciences,
Faculty of Science and Technology,
Center of Neurosciences and Cell Biology,
University of Coimbra,
P.O. Box 3046,
3001-401 Coimbra, Portugal
e-mail: geraldes@bioq.uc.pt
F. J. Can ˜ada ? J. Jime ´nez-Barbero
Department of Chemical and Physical Biology,
CIB-CSIC, Ramiro de Maeztu 9,
28040 Madrid, Spain
J. A. Martins ? J. P. Andre ´
Centro de Quı ´mica,
Campus de Gualtar,
Universidade do Minho,
Braga, Portugal
123
J Biol Inorg Chem
DOI 10.1007/s00775-011-0773-z

Page 2

ligand binding through intermolecular saturation transfer
and, moreover, allows screening of ligand libraries [11], as
well as calculation of affinity constants and mapping the
binding epitope [5, 12–14]. In combination with ligand–
protein docking studies, it may also help to derive a con-
sistent 3D model of the intermolecular complex [15–21].
It is obvious that many diseases share a thin line with
molecular recognition events and that targeting specific
receptors is one of the approaches that may be employed to
prevent, understand and control diseases. The development
of magnetic resonance imaging (MRI) contrast agents
(CAs) specifically targeted to different tissues has become
a priority, and is a most profitable approach in this context.
In particular, and within possible targets, the asialoglyco-
protein receptor (ASGPR) is a lectin-type protein only
found at the surface of hepatocytes and macrophages
[22–25], having a determinant role in the targeting of
exogenous compounds to the liver tissues, either for diag-
nosis or for therapy. On the basis of this knowledge, a new
class of CAs has been developed recently with the intent
for them to be selectively taken up by the hepatic ASGPR
[26–28]. 1,4,7,10-Tetrakis(carboxymethyl)-1,4,7,10-tetra-
azacyclododecane (DOTA)-like chelators were attached to
sugar moieties, galactosyl, glucosyl or lactosyl residues, by
pendant arms containing aliphatic chains and amide bonds,
resulting in monovalent or multivalent glycoconjugate
derivative agents. After the development of the DOTA-
based glycoconjugates [26], diethylenetriaminepentaacetic
acid (DTPA) bisamide based glycoconjugates were also
devised and studied [27]. In both types of CAs, DOTA- and
DPTA-based chelates, the structural characteristics are
similar: a central reporter group complexing a paramag-
netic metal centre (Gd3?for MRI,153Sm3?for c scintig-
raphy) with high kinetic and thermodynamic stability and
long linear or branched arms with terminal sugar moieties
as targeting groups (Fig. 1).
Carbohydrate–protein interactions are relatively weak
binding processes. Nevertheless, affinity enhancement is
achieved through multiple and simultaneous interactions of
glycosides (multivalency) with their lectin receptors, a
process known as the cluster glycoside effect [29–32]. In
this way, higher valencies of the glycosides produce a
synergistic effect in affinity constants when they bind to
proteins (i.e. tetraglycosides[triglycosides[diglyco-
sides[monoglycosides) [33, 34]. However, the way in
which the sugar-based ligands interact with their lectin
receptors in order to increase the binding affinity is still
controversial and, most probably, strongly related with the
particular ligand structure [31]. There are two main
mechanisms by which the cluster glycoside effect may take
place: intramolecular or intermolecular interactions. The
intramolecular binding mode is characterized by the
binding of multiple sugar moieties, within the same gly-
coside molecule, to multiple binding sites at the same lectin
receptor. Therefore, this binding mode is also termed
‘‘chelate-type binding’’, as the glycoside simulates a
Fig. 1 Chemical structures and proton numbering scheme of the 1,4,7,10-tetrakis(carboxymethyl)-1,4,7,10-tetraazacyclododecane (DOTA) and
diethylenetriaminepentaacetic acid (DTPA) glycoconjugates
J Biol Inorg Chem
123

Page 3

chelate motif imprisoning the metal atom, in this case the
lectin. Moreover, to make this type of interaction possible,
the binding sites on the protein surface should be close
enough to each other to allow simultaneous spanning of the
interacting sites by the ligand binding moieties. At the same
time, the ligand arms must be significantly long to reach the
different binding sites. A conjugation between the protein
and the ligand morphology must support this type of inter-
action. There are also other properties that favour the intra-
molecular binding mode, such as the presence of a
hydrophobic linker, which may promote this binding mode
by enhancing interactions between the linker and the protein
surface [31, 35, 36]. On the other hand, under the same
conditions, a hydrophilic linker could favour an intermo-
lecular binding mode, which takes place when a single
multivalent glycoside molecule binds to more than one
protein molecule, a process that may finally lead to a
precipitate.
In this context, we present the lectin binding features of
this new class of DOTA- and DTPA-based glycoconjugate
CAs and explore the affinity of the multiderivative glyco-
conjugate agents for a model lectin receptor. To accomplish
this task, four different diamagnetic La(III) chelate ana-
logues (diamagnetic chelates were needed in order not to
quench the STD NMR effect) of the Gd(III) compounds of
this class of CAs were studied, namely the monovalent
La(DOTAGal)andLa(DOTALac)
La(DOTALac2) and La(DTPAGal2) (Fig. 1). STD NMR
was chosen to study the binding of the DOTA and DTPA
glycoconjugates to a well-known lectin, Ricinus communis
agglutinin (RCA120), that was used as a model of the hepatic
ASGPR. Although the carbohydrate recognition domain of
the H1 subunit of ASGPR [37] would be a better model
system, RCA120was used as a simple model of the mem-
brane ASGPR in this proof-of-principle study of the method
employed because it has galactose binding affinities in the
same range as ASGPR [38], therefore being largely used for
binding assays of galactose derivatives [39, 40]. RCA120
water solubility also favours the in vitro NMR studies.
RCA120isa dimericlectin, consistingof two non-covalently
boundricin-likemonomers.Inturn,eachricin-likemoietyis
composed of two covalently linked heterochains, chain A
and chain B. Chain A is responsible for the catalytic effect
that gives this protein its toxic character, whereas chain B is
the lectin domain, responsible for sugar affinity. Every
B chainhas,in principle,two sugar binding sites, dubbed 1a
and 2c [41–44]. However, the exact number of accessible
binding sites in each B chain of RCA120was ultimately
confirmed by calorimetric assays to be only one (1a), and its
identity was revealed by site mutations [45–48].
We present a study, at the molecular level, of the mode
of binding of these glycoconjugate derivatives compounds
to RCA120 by using a combination of STD NMR data
andthedivalent
and molecular modelling protocols, namely docking
calculations.
Materials and methods
Samples
The DOTA and DTPA glycoconjugate derivatives and their
La(III) complexes were synthesized and characterized as
described previously [26, 27]. The La(III) complexes were
dissolved as 99.9% D2O/10% phosphate-buffered saline
solutions. RCA120 was isolated as previously described
[49]. The protein was dissolved in 99.96% D2O (purchased
from Sigma-Aldrich) in the absence of buffer. The protein
concentration ranged from 10 to 25 lM depending on the
compound studied and the expected affinity for the protein,
in order to achieve a large range of ligand excess. The
concentrations of the ligands were selected to obtain most
of the ASTDpoints at the beginning of the saturation curve,
with ligand excess raging from 10 to 50, and a point of
large ligand excess, over 200, was also obtained to define
the ‘‘plateau’’ region of the curve. The concentrations of
the protein and ligand were not constant for every com-
pound, and were defined according to the desired ligand
excess ratio and the quantities available.
NMR studies
All1H NMR spectra were acquired using a 5-mm pulse
field gradient (PFG) triple resonance inverse probe using a
Varian VNMRS 600 MHz NMR spectrometer working at
599.72 MHz. For each sample a 1D
obtained, and the spectral assignments from the literature
[12, 26, 27] were used after they had been confirmed by 2D
gradient correlation spectroscopy (gCOSY) spectra (data
not shown). STD NMR spectra were then acquired, where
the double PFG spin echo (DPFGSE) sequence [50] was
used for water suppression. Since in our NMR system the
STD NMR spectra are acquired directly from phase
cycling, the 1D1H NMR spectra were used as off-reso-
nance references in order to calculate the STD amplifica-
tion factor [12]. All spectra were acquired using the same
parameters: equal spectrometer gain value, spectral win-
dow of 8 kHz, number of scans varied between 128 and
256 for 1D1H spectra and between 1,024 and 2,048 for the
STD NMR spectra, a previously calibrated spin-lock filter
(T1q) of 30 ms was used to remove protein resonances, the
acquisition time was 1 s and the repetition time was 3.5 s.
STD experiments were performed using a saturation delay
of 2.5 s. To compare the reference spectra with the STD
NMR spectra, the different number of acquisitions was
normalized according to Eq. 1:
1H spectrum was
J Biol Inorg Chem
123

Page 4

Relative STDð%Þ ¼ISTD? 2 ? scansreference
I0? scansSTD
;
ð1Þ
where ISTDis the peak intensity of the STD NMR spectra
and I0is the intensity of the peaks in the
spectra. Then, the peak intensities were normalized to the
STD amplification factor (ASTD) (Eq. 2):
1H reference
ASTD¼ relativeSTD ? ligandexcess:
Binding studies
ð2Þ
Affinity constant (KD) estimation was performed by
studying the build-up behaviour of the STD NMR spectra
in conditions of constant protein concentration and
increasing ligand concentration. The KDvalues were esti-
mated by fitting the plotted data points to a one-site binding
model [13, 16] (Eq. 3):
ASTD¼aSTD? L ½ ?
L ½ ? þ KD
where aSTDis the maximum ASTDand [L] is the total ligand
concentration. Plots and fits were obtained using GnuPlot
version 4.2-3.
;
ð3Þ
Docking calculations
Automated docking was performed using Autodock4.1 [51]
and the Lamarckian genetic algorithm [52] as a searching
procedure. The Protein Data Bank (PDB) file corresponding
to the protein used, RCA120, was 1RZO. The protein model
was kept rigid, and torsions were allowed only at the ligand
level. Owing to the large number of torsions and the size of
theligands,onlythesugarmoietyandtheadjacentarmswere
docked. The reporter groups (DOTA and DTPA) were
removed and, in the case of divalent ligands, just one of the
arms was considered. La(III) ligand chelates were designed
in three dimensions using Maestro (Schro ¨dinger) [53]. For
glycoconjugate derivatives, grid maps were constructed
using 50 9 50 9 50 points, with a grid box point spacing of
0.303 A˚and centred some points below the binding site.
The size of the initial random population was set differently
for each compound, 150 for La(DOTAGal) and 50 for
La(DTPAGal2). The other parameters were set common for
all runs, the maximum number of generations was 27,000,
theelitismwas1,theprobabilitythatagenewouldundergoa
random change was 0.02 and the crossover probability was
0.80. Fifty docking runs were performed. The maximum
number of generations was reached for these calculations
and the total number of evaluations was kept around
1.1 9 107. For galactose calculations, grid maps were con-
structed using 40 9 40 9 40 points, with a grid box point
spacing of 0.336 A˚and centred at the ligand that was placed
in the binding site in the PDB file by default. The size of the
initial random population was 50 individuals, the maxi-
mum number of energy evaluations was 1.5 9 107, the
maximum number of generations was 40,000, the elitism
was 1, the probability that a gene would undergo a
random change was 0.02 and the crossover probability
was 0.80. Fifty docking runs were performed. The results
wereclusteredusinga
(RMSD) cutoff of 0.5 A˚.
root-mean-squaredeviation
Results and discussion
Figure 2 represents both the 1D
spectra of the four La(III)-complexed glycoconjugates in
the presence of RCA120. Resonances from the1H spectra
were assigned on the basis of data in previous publications
[12, 26, 27] and on gCOSY analysis. The resonances are
identified in Fig. 2, following the proton numbering
schemes shown in Fig. 1. The sugar resonances are the
main visible resonances in the STD NMR spectra, thus
proving that these DOTA/DTPA branched glycoconjugate
derivatives specifically interact with RCA120through the
sugar moieties. Owing to the nature of the STD experi-
ment, it is possible to characterize the binding epitope of
the ligand for a particular interaction. Table 1 summarizes
the saturation profiles (relative STD) determined for the
sugar and linker protons of the four compounds studied
H protons refer to the protons from the galactosyl residue,
and H0protons refer to those of the glucosyl residues, in
the case of lactosyl derivatives. The evaluation of the sat-
uration profile of the six (12) groups of protons from the
galactosyl (lactosyl) residues proves that the sugar protons
that remain closer to the protein are always H3 and H2 of
galactosyl residues, with H4 also experiencing great
transfer of saturation [49]. On the other hand, protons H5
and H6/60seem to remain somehow further from the pro-
tein surface, since weaker STD effects were observed. The
percentage of transferred saturation for protons H5/50,
H66a/606a0and H3/30in the case of lactosyl derivatives was
not considered because these peaks are superimposed on
each other. The binding epitope revealed for these glyco-
conjugates is in agreement with the expected epitope on the
basis of previous studies for this type of interaction
[12, 29]. The anomeric protons were not considered in the
group epitope mapping evaluations because the DPFGSE
water suppression scheme dramatically affected its inten-
sity. It is noteworthy that additional STD effects were
observed for the linker protons (see Fig. 2), which will be
discussed later.
After normalization of the acquired data points to the
STD amplification factors (ASTD) [12], the plots of ASTD
versus ligand concentration (lM) were drawn and fit to a
one-site binding model. Figure 3 represents the two plots
1H and the STD NMR
J Biol Inorg Chem
123

Page 5

Fig. 2 One-dimensional
difference (STD) NMR (bottom) spectra of a 0.82 mM La(DOTA-
Gal), b 6.2 mM La(DOTALac), c 1.1 mM La(DOTALac2) and
d 0.58 mM La(DTPAGal2) in the presence of Ricinus communis
1H NMR (top) and saturation transfer
agglutinin (RCA120) with the following concentrations: a 25 lM,
b 10 lM, c 15 lM and d 15 lM. Selective saturation (2.5 s) was
performed at the aromatic region of the protein. The spin-lock pulse
was calibrated to avoid unwanted protein resonances
J Biol Inorg Chem
123

Page 6

from the galactosyl and lactosyl derivatives, each one
representing both monovalent and divalent forms. Table 2
summarizes the values of the affinity constants estimated
for the corresponding interactions with the lectin. To esti-
mate the affinity constants for the different compounds, the
data points obtained from the H3 protons of the galactosyl
derivatives were used for the calculations since they
showed the highest degree of saturation, thus allowing a
more accurate KDestimation [16, 17]. Also, because of
degeneration of different proton resonances in the spectra
of the lactosyl derivatives, the affinity constant for these
compounds was estimated according to the data obtained
from the galactosyl H4 proton, which remained fairly
isolated. However, taking into account the recently pub-
lished detailed analysis of the factors affecting the deter-
mination of ligand–receptor dissociation constants by STD
NMR titration experiments [17], one has to consider the KD
values presented here as approximations of the real values.
The calculated KDvalues (Table 2) reflect an increase in
binding affinity for both divalent compounds relative to the
monovalent ones, also evidenced by the lower ASTDvalues
output by the fitting curves, with a substantial decrease of
the dissociation rate (koff) with respect to that of the
monovalent analogues, as it is rather unlikely that, when
completely bound, both sugar residues of the divalent
compounds simultaneously dissociate from the two binding
sites [5].
The stronger binding of the divalent compounds to the
lectin clearly observed in the present experiments is in
agreement with findings of previous in vivo studies, where
the higher affinity of the multivalent forms of glycocon-
jugate derivatives of this type showed a more rapid
incorporation by the liver than the monovalent forms [54].
The STD NMR spectra provided additional information
on the slightly different interaction features for the mono-
valent and divalent molecules with RCA120. For the
monovalent compounds, La(DOTAGal) and La(DOT-
ALac) (Fig. 2a, b), the resonances from protons j and k in
the linker arms are barely visible. On the other hand,
several proton resonances are clearly visible in the STD
NMR spectra of the divalent compounds, and it is possible
to measure the saturation profiles of resonances n and o for
La(DOTALac2)andresonances
La(DTPAGal2). The calculated STD values for protons n
and o of La(DOTALac2) were 37 and 44%, respectively,
normalized to H2. In the case of La(DTPAGal2) resonance
l was measured to have 29% saturation and resonances gj
and hi had 37 and 28%, respectively, normalized to H3
(Table 1). Although we have to consider that the gj and hi
signal contribution comes from eight protons, we cannot
ignore that the linkers of these divalent compounds interact
with the lectin surface.
A hydrophobic linker is more prone to establish inter-
actions with a lectin protein surface than a hydrophilic one
[55]. In fact, the La(DTPAGal2) linker is longer and more
hydrophobic, when compared with the linker (protons g–j)
of La(DOTALac2). It also displays more torsional degrees
of freedom in solution, thus facilitating the interaction with
the surface of the protein. That might explain why we do
not observe interaction of protons g–j of La(DOTALac2)
with the protein.
The presence of hydrophobic interactions between a
given ligand and the neighbouring regions of the sugar
binding site in the lectin surface has already been reported
[31, 35]. Although very frequently binding features of this
type have been associated with intramolecular, or chelate-
type, binding modes [55, 56], the 3D structure of the
RCA120protein does not allow such a type of binding. The
X-ray diffraction structure available for RCA120(PDB file
1RZO) shows a dimer of AB ricin heterodimers in the
crystal. In fact, the distance between the two galactose
binding sites within one heterodimer is 36 A˚, too far for the
l,gjand hifor
Table 1 Summary of the
characterized binding epitopes
with saturation transfer
difference (STD) values relative
to those of H2 or H3 as a
percentage
See Fig. 1 for the structures of
the epitopes
Protons La(DOTAGal) La(DPTAGal)2
La(DOTALac)La(DOTALac2)
H294 95 100100
H3100 100––
H4
H5 H6a/6b
H20
H40
k
9364 68 94
6465––
7280
3734
11
j 10
hi28
gj37
l29
n37
o44
J Biol Inorg Chem
123

Page 7

two arms of the divalent ligands to span them and account
for a possible simultaneous intramolecular effect. More-
over, the distance between the two closest galactose
binding sites, each one on each B-chain subunit of the two
dimers, is even larger, more than 50 A˚. Thus, the tenfold
increased affinity of the divalent compounds for the lectin
relative to the monovalent compounds cannot be explained
by an intramolecular mechanism of a cluster glycoside
effect. In the case of an intermolecular mechanism, lectin–
lectin interactions and finally precipitation should occur to
produce the increased affinity, which was not observed in
the present case. Therefore, with the available data, the
statistical effect of the multiple carbohydrate epitopes
present together with the interaction of the linkers with the
protein surface (see later) is considered to be responsible
for the observed increased affinity.
The empirical results of the STD experiments were
substantiated with a 3D model of the complex by using
molecular modelling calculations based on Autodock 4.1
[51]. Docking calculations were then performed for
La(DTPAGal2) with RCA120binding site 1a, considering
only one of the sugar arms, following the protocol descri-
bed in ‘‘Docking calculations’’. The highest ranking cluster
encompassed eight possible binding conformations, with
the output geometries clustered using a RMSD of 2 A˚.
Three of the eight calculated conformers were selected
according to the orientation of the sugar moiety inside the
binding pocket, and considering they keep conformity with
the STD NMR results and with the chemical nature of the
molecule itself. Indeed, the binding mode obtained was
completely in agreement with that obtained for an isolated
galactosyl moiety, thus validating the orientation of the
saccharide residue of La(DTPAGal2) within the binding
site. Figure 4a shows one representative structure of the
selected cluster, and Fig. 4b represents three superimposed
structures from the above-mentioned calculations. The
galactosyl residues from the different runs, including that
for a single galactose moiety, are oriented in a similar
manner, although they are not perfectly superimposed.
Nonetheless, all the intermolecular hydrogen bonds that
occur for galactose binding also occur for the different
solutions for this glycoconjugate. With respect to the
interactions of the linker with the protein surface, the
models obtained set the long hydrophobic linker of
the La(DTPAGal2) chelate close to a hydrophobic region of
the protein surface, and thus it interacts with the side chains
of different amino acids. Owing to the size of the docked
ligand, the structures obtained can be considered as a good
approximation of the interaction mode, which cannot be
seen within the concept of a rigid, static representation.
Very probably, different orientations of the linker may be
adopted for it to properly interact with the lectin, as sug-
gested by the docking calculations.
Docking studies of the La(DOTAGal) single-arm mol-
ecule with RCA120binding site 1a were also performed, as
described in ‘‘Docking calculations’’. Only the structures
which fit the STD data were selected for further analysis
(Fig. 5). In this case, the STD data suggested very weak
interactions between the linker of these monovalent
derivatives and the protein surface. Again, the same region
Fig. 3 Direct determination of the KDvalues of La(III) complexes of
a DOTAGal (circles) and DTPAGal2(triangles) and b DOTALac
(circles) and DOTALac2(triangles) binding to RCA120by fitting the
acquired data points to Eq. 2
Table 2 Individual KDvalues for the protons of the glycoconjugate
compounds obtained from the STD NMR experiments
ProtonsIndividual KD
value (mM)
Maximum STD
amplification factor
(La)DOTAGal–H31.666.5
(La)DTPAGal2–H3
(La)DOTALac–H4
0.15 1.6
1.16 5.8
(La)DOTALac2–H4 0.12 2.2
J Biol Inorg Chem
123

Page 8

of the protein was targeted by the linker, as for
La(DTPAGal2). It can be observed that the charged regions
of the linker were placed near the charged atoms of the
surface amino acids, which could stabilize the conforma-
tion by polar interactions. H–H distances between the
linkers of both docking results for La(DTPAGal2) and
La(DOTAGal) and the surface of RCA120binding site 1a
Fig. 4 Automated docking structures of one arm of La(DTPAGal2)
in binding site 1a of RCA120. a One of resulting runs is shown in
green and b the three most reliable runs. The single docked galactose
molecule was also superimposed to allow a better comparison, and it
is displayed in orange. Marked in yellow are the hydrogen bonds
considered between the ligand and the protein, involving OH2 and
Lys40, OH3 and Asn46, OH4 and Gly25, and OH6 and Gln35. CH–p
stacking interactions with Trp37 also occur [52]
Fig. 5 Docking results of the sugar-linker moiety of La(DOTAGal).
a One of the docked results and b the superimposed view of all the
selected possible structures. The number of torsional bonds was 13
J Biol Inorg Chem
123

Page 9

were measured using Autodock 4 (Table 3). On average,
protons f–k of the hydrophobic patch of the La(DTPAGal2)
linker are closer to the protein surface protons (calculated
average distance of 2.9 A˚) than protons i–h of the polar
linker of La(DOTAGal) (calculated average distance of
5.8 A˚). In fact, as one moves from the sugar moiety of the
molecule, the polar linker of La(DOTAGal) tends to move
away from the protein surface, whereas the hydrophobic
parts of the La(DTPAGal2) linker far from the sugar
moiety stay quite close to the protein surface.
Finally, Table 4 shows the output values for the calcu-
lated binding structures of La(DOTAGal) and La(DTPA-
Gal2). The intermolecular energy is lower in La(DOTAGal)
and La(DTPAGal2) runs when compared with the single
galactose molecule. As expected from the higher number of
torsions,theaverageinternalenergyuponbindingislowerin
theLa(DTPAGal2)runsthaninthe La(DOTAGal) runs.The
values should be regarded as merely qualitative, given the
simplification of the model employed.The affinity of gal-
actoseforRCA120is2.2 9 103M-1[46],avaluewhichlies
between the calculated values for the monovalent and
divalent compounds, in qualitative agreement with the esti-
mated free energies obtained by docking calculations.
Conclusion
An STD NMR analysis has shown that the divalent
La(DTPAGal2) and La(DOTALac2) glycoconjugate deriv-
atives have higher affinity for the RCA120lectin than their
monovalent La(DOTAGal) and La(DOTALac) analogues.
This effect is therefore concordant with the results
observed in in vivo binding studies with hepatocyte cells
and the corresponding153Sm3?chelates [54]. The so-called
cluster glycoside effect may be invoked to explain the
observations. Our studies have tried to clarify the binding
mode of this new class of potential liver imaging agents,
using the RCA120lectin as a simple model receptor, in
order to provide new insights into the development of lead
compounds and optimization of those already developed.
The STD NMR data, assisted by docking calculations,
suggest the existence of interactions between the linkers of
the divalent compounds and the protein surface.
The structural features of RCA120and the glycoconju-
gate imaging agents used in this work preclude the
existence of an intramolecular binding process. An inter-
molecular type of binding cannot be considered, as it
would imply protein clustering and precipitation, which did
not occur in the experimental conditions used. Taking into
consideration the STD NMR data and the docking results
obtained, we can conclude that the main interaction
between these ligands and the lectin protein occurs through
the sugar residues, through a combination of hydrogen
bonds, van der Waals forces and CH–p stacking interac-
tions [57, 58], but the hydrophobic linker arms also interact
with the protein surface, especially for the divalent agents.
These interactions, together with a statistical effect of the
presence of multiple carbohydrate epitopes, are considered
to be responsible for the increased affinity of the divalent
compounds for the lectin. We believe that the approach to
study CA–target protein interactions combining NMR and
modelling tools, proposed and exemplified in this work for
Table 3 H–H
La(DOTAGal) and La(DTPAGal2)
linker–proteindistanceofdockedarmsof
La(DOTAGal)La(DTPAGal2)
ProtonsDistance (A˚)ProtonsDistance (A˚)
k3.2m2.8
j4.1l3
i5.2hi2.9
h6 gj3.1
Distances were measured relative to Ca G25, Cb E26 and Cb E27 and
are presented as average values
Table 4 Calculated energies
for the ‘‘single arm’’ of
La(DOTAGal) and
La(DTPAGal2) (kcal mol-1)
The first set of runs 1–4 and
second set of runs 1–3 refer
to La(DOTAGal) and
La(DTPAGal2), respectively
Run La(DOTAGal)/
La(DTPAGal2)
Intermolecular
energy
Internal
energy
Torsional
free
energy
Unbound
system
energy
Estimated
free
energy
Run 1-8.62-1.55?3.57-0.46-6.14
Run 2-7.01-2.48?3.57-0.46-5.47
Run 3-8.16-1.45?3.57-0.46-5.58
Run 4-8.73-1.56?3.57-0.46-6.25
Run 1-8.60-1.82?4.39-0.60-5.43
Run 2-9.13-1.62?4.39-0.60-5.76
Run 3-7.86-1.78?4.39-0.60-4.65
Galactose-6.69-1.491.65-0.37-6.16
J Biol Inorg Chem
123

Page 10

the first time, can be very useful in the design of novel
targeted MRI CAs. In particular, novel design and pro-
duction of high-affinity glycoconjugates, such as ones
investigated here, to interact with lectins should focus on
the optimization of the linker arms as a protein binding
complement to the sugar residues, regarding their length,
flexibility and chemical nature. Such an approach will aim
at increasing entropic and enthalpic savings that derive
from the linker. Good knowledge of the structure of the
target receptor is also of extreme importance in order to
design specific and protein-directed ligands.
Acknowledgments
a Cie ˆncia e a Tecnologia (FCT), Portugal (project PTDC/QUI/70063/
2006) and FEDER. The Varian VNMRS 600 MHz NMR spectrom-
eter in Coimbra was acquired with the support of the Programa
Nacional de Reequipamento Cientı ´fico of FCT, Portugal, contract
REDE/1517/RMN/2005—as part of Rede Nacional de RMN
(RNRMN). This work was carried out in the framework of the COST
D38 Action. The group in Madrid thanks the Ministery of Science and
Innovation of Spain for financial support (grant CTQ2009-08536).
We also thank Eurico Cabrita for useful discussions.
This work was supported by the Fundac ¸a ˜o para
References
1. Engvall E, Perlmann P (1971) Immunochemistry 8:871–874
2. Yalow BRS, Berson SA (1960) J Clin Invest 39:11–13
3. Cuatrecasas P, Wilchek M, Anfinsen CB (1968) Biochemistry
61:636–643
4. Homola J, Yee SS, Gauglitz G (1999) Sens Actuators B Chem
54:3–15
5. Meyer B, Peters T (2003) Angew Chem Int Ed 42:864–890
6. Ni F (1994) Progr Nucl Magn Reson Spectrosc 26:517–606
7. Chen A, Shapiro MJ (1998) J Am Chem Soc 120:10258–10259
8. Dalvit C, Pevarello P, Tato ` M, Veronesi M, Vulpetti A, Sun-
dstro ¨m M (2000) J Biomol NMR 18:65–68
9. Dalvit C, Fogliatto G, Stewart A, Veronesi M, Stockman B
(2001) J Biomol NMR 21:349–359
10. Mayer M, Meyer B (1999) Angew Chem Int Ed 38:1784–1788
11. Vogtherr M, Peters T (2000) J Am Chem Soc 122:6093–6099
12. Mayer M, Meyer B (2001) J Am Chem Soc 123:6108–6117
13. Lepre C, Moore JM, Peng JW (2004) Chem Rev 104:3641–3676
14. Fielding L (2007) Progr Nucl Magn Reson Spectrosc 51:219–242
15. Meinecke R, Meyer B (2001) J Med Chem 44:3059–3065
16. Neffe AT, Bilang M, Gru ¨neberg I, Meyer B (2007) J Med Chem
50:3482–3488
17. Angulo J, Enrı ´quez-Navas PM, Nieto PM (2010) Chem Eur J
16:7803–7812
18. Hajduk PJ, Mack JC, Olejniczak ET, Park C, Dandliker PJ,
Beutel B (2004) J Am Chem Soc 126:2390–2398
19. Klein J, Meinecke R, Mayer M, Meyer B (1999) J Am Chem Soc
121:5336–5337
20. Kolympadi M, Fontanella M, Venturi C, Andre ´ S, Gabius H-J,
Jime ´nez-Barbero J, Vogel P (2009) Chem Eur J 15:2861–2873
21. Jime ´nez-Barbero J, Dragoni E, Venturi C, Nannucci F, Arda ´ A,
Fontanella M, Andre ´ S, Can ˜ada FJ, Gabius H-J, Nativi C (2009)
Chem Eur J 15:10423–10431
22. Leffler H, Carlsson S, Hedlund M, Qian Y, Poirier F (2004)
Glycoconj J 19:433–440
23. Almkvist J, Karlsson A (2004) Glycoconj J 19:575–581
24. Critchley P, Willand MN, Rullay AK, Crout DH (2003) Org
Biomol Chem 1:928–938
25. Ashwell G, Harford J (1982) Ann Rev Biochem 51:531–544
26. Andre ´ JP, Geraldes CFGC, Martins JA, Merbach AE, Prata MIM,
Santos AC, de Lima JJP, To ´th E (2004) Chem Eur J
10:5804–5816
27. Baı ´a P, Andre ´ JP, Geraldes CFGC, Martins JA, Merbach AE,
To ´th E (2005) Eur J Inorg Chem 2110–2119
28. Fulton DA, Elemento EM, Aime S, Chaabane L, Botta M, Parker
D (2006) Chem Commun 1064–1066
29. Lee YC (1992) FASEB J 6:3193–3200
30. Lee RT, Lee YC (2001) Glycoconj J 17:543–551
31. Lundquist J, Toone EJ (2002) Chem Rev 102:555–578
32. Lee YC, Lee RT (1995) Acc Chem Res 28:321–327
33. Lee YC, Townsend RR, Hardy MR, Lo ¨nngren J, Arnarp J, Har-
aldsson M, Lo ¨nn H (1983) J Biol Chem 258:199–202
34. Biessen EA, Broxterman H, van Boom JH, van Berkel TJ (1995)
J Med Chem 38:1846–1852
35. Lundquist JJ, Debenham SD, Toone EJ (2000) J Org Chem
65:8245–8250
36. Corbell J (2000) Tetrahedron Asymmetry 11:95–111
37. Meier M, Bider MD, Malashkevich VN, Spiess M, Burkhard P
(2000) J Mol Biol 300:857–865
38. Naha ´lkova ´ J, Svitel J, Gemeiner P, Danielsson P, Pribulova ´ B,
Petrus L (2002) J Biochem Biophys Methods 52:11–18
39. D’Agata R, Grasso G, Iacono G, Spoto G, Vecchio G (2006) Org
Biomol Chem 4:610–612
40. Lee M, Park S, Shin I (2006) Bioorg Med Chem Lett
16:5132–5135
41. Nicolson GL, Blaustein J, Etzler ME (1974) Biochemistry
13:196–204
42. Olsnes S, Saltvedt E, Pihl A (1974) J Biol Chem 249:803–810
43. Endo Y, Tsurugi K (1987) J Biol Chem 262:8128–8130
44. Lamb FI, Roberts LM, Lord JM (1985) Eur J Biochem
148:265–270
45. Houston LL, Dooley TP (1982) J Biol Chem 257:4147–4151
46. Sharma S, Bharadwaj S, Surolia A, Podder SK (1998) Biochem J
333(3):539–542
47. Dam TK, Brewer CF (2002) Chem Rev 102:387–429
48. Sphyris N, Lord JM, Wales R, Roberts LM (1995) J Biol Chem
270:20292–20297
49. Rivera-Sagredo A, Jime ´nez-Barbero J, Martı ´n-Lomas M, Solı ´s D,
Dı ´az-Maurin ˜o T (1992) Carbohydr Res 232:207–226
50. Hwang T-L, Shaka AJ (1995) J Magn Reson A 112:275–279
51. Goodsell DS, Morris GM, Olson AJ (1996) J Mol Recognit 9:1–5
52. Morris GM, Goodsell DS, Halliday RS, Huey R, Hart WE, Belew
RK, Olson AJ (1998) J Comput Chem 19:1639–1662
53. Schro ¨dinger (2008) Maestro, version 8.5. Schro ¨dinger, New York
54. Prata MIM, Santos AC, Torres S, Andre ´ JP, Martins JA, Neves
M, Garcı ´a-Martı ´n ML, Rodrigues TB, Lo ´pez-Larrubia P, Cerda ´n
S, GeraldesCFGC (2006)
258:246–258
55. Kanai M, Mortell KH, Kiessling LL (1997) J Am Chem Soc
119:9931–9932
56. Kramer RH, Karpen JW (1998) Nature 395:710–713
57. Bernardi A, Arosio D, Potenza D, Sa ´nchez-Medina I, Mari S,
Can ˜ada FJ, Jime ´nez-Barbero J (2004) Chem Eur J 10:4395
58. Terraneo G, Potenza D, Canales A, Jime ´nez-Barbero J, Baldridge
KK, Bernardi A (2007) J Am Chem Soc 129:2890–2900
Contrast Med MolImaging
J Biol Inorg Chem
123