Gd3+ complexes conjugated to Pittsburgh compound B: Potential MRI markers of β-amyloid plaques Topical Issue on Metal-Based MRI Contrast Agents. Guest editor: Valerie C. Pierre

Article abstract of DOI:10.1007/s00775-013-1055-8

In an effort towards the visualization of β-amyloid (Aβ) plaques by T ₁-weighted magnetic resonance imaging for detection of Alzheimer's disease, we report the synthesis and characterization of stable, noncharged Gd³⁺ complexes of th

Full text of DOI:10.1007/s00775-013-1055-8

J Biol Inorg Chem (2014) 19:281–295
DOI 10.1007/s00775-013-1055-8
ORIGINAL PAPER
Gd3+ complexes conjugated to Pittsburgh compound B: potential
MRI markers of b-amyloid plaques
•
Andre F. Martins Jean-Franc¸ois Morfin
•
´
Carlos F. G. C. Geraldes Eva Toth
´
•
´
Received: 21 August 2013 / Accepted: 9 October 2013 / Published online: 3 December 2013
Ó SBIC 2013
Abstract In an effort towards the visualization of
b-amyloid (Ab) plaques by T1-weighted magnetic reso-
nance imaging for detection of Alzheimer’s disease,
we report the synthesis and characterization of stable,
noncharged Gd3? complexes of three different 1,4,7,10-
tetraazacyclododecane-1,4,7-triacetic acid monoamide
derivatives conjugated to Pittsburgh compound B, a well-
established marker of Ab plaques. The ligands L1, L2, and
L3 differ in the nature and size of the spacer linking the
macrocyclic chelator and the Pittsburgh compound B tar-
geting moiety, which affects their lipophilicity, the octa-
nol–water partition coefficients of the complexes ranging
from -0.15 to 0.32. Given their amphiphilic behavior, the
complexes form micelles in aqueous solution (critical
micellar concentration 1.00–1.49 mM). The parameters
determining the relaxivity, including the water exchange
rate and the rotational correlation times, were assessed for
the monomeric and the micellar form by a combined 17O
NMR and 1H nuclear magnetic relaxation dispersion
(NMRD) study. They are largely influenced by the aggre-
gation state and the hydrophobic character of the linkers.
The analysis of the rotational dynamics for the aggregated
state in terms of local and global motions using the Lipari–
Szabo approach indicates highly flexible, large aggregates.
On binding of the complexes to human serum albumin or to
the amyloid peptide Ab1–40 in solution, they undergo a
fourfold and a twofold relaxivity increase, respectively
(40 MHz). Proton relaxation enhancement studies con-
firmed moderate interaction of Gd(L1) and Gd(L3) with
human serum albumin, with KA values ranging between
250 and 910 M-1
.
Keywords Contrast agents ꢀ Gadolinium ꢀ Magnetic
resonance imaging ꢀ Lanthanides ꢀ Amyloid peptides
Introduction
Responsible Editor: Valerie C. Pierre
Alzheimer’s disease (AD) is the chronic neurodegenerative
disorder that constitutes the most frequent form of intel-
lectual deterioration in elderly individuals [1]. It is char-
acterized by brain deposition of amyloid plaques and
neurofibrillary tangles that constitute representative
neuropathological markers of this disease [2, 3]. Today, the
diagnosis of AD is mainly based on cognitive tests, and it
becomes definitive only in later stages of the disease, when
several sections of the brain are seriously damaged and
progressive cognitive decline has occurred and affected the
person’s ability to perform everyday activities. Although
amyloid deposition is not a reliably quantitative biomarker
for AD, imaging solutions based on innovative imaging
probes that visualize amyloid plaques could contribute to
the early identification of the disease, allowing prompt
Electronic supplementary material The online version of this
article (doi:10.1007/s00775-013-1055-8) contains supplementary
material, which is available to authorized users.
´
´
A. F. Martins ꢀ J.-F. Morfin ꢀ E. Toth (&)
´
Centre de Biophysique Moleculaire,
CNRS, Universite d’Orleans,
´
´
Rue Charles Sadron,
´
45071 Orleans Cedex 2, France
e-mail: eva.jakabtoth@cnrs-orleans.fr
A. F. Martins ꢀ C. F. G. C. Geraldes (&)
Department of Life Sciences,
Faculty of Sciences and Technology,
University of Coimbra,
P.O. Box 3046, 3001-401 Coimbra, Portugal
e-mail: geraldes@ci.uc.pt
123

282
J Biol Inorg Chem (2014) 19:281–295
intervention to slow the progress of the disease, even
without a definitive cure being available today. They could
be also of invaluable help for delineating novel therapies
by facilitating drug development.
in most cases intra-carotid mannitol treatment was also
used to enhance the BBB permeability of the agent.
Given the high electron spin and slow electronic relax-
ation of the Gd3? ion, Gd3? complexes are the most widely
used MRI contrast materials, ensuring positive contrast
[11]. The efficacy of a contrast agent is measured by its
relaxivity, defined as the paramagnetic enhancement of the
water proton relaxation rate normalized to 1 mM concen-
tration of Gd3? [11, 19]. The inner-sphere relaxivity term is
determined by the microscopic parameters of the Gd3?
complex, the most important being the hydration number
(q), the rotational correlation time (sR), the water exchange
rate (kex), and the electron spin relaxation times (T1e, T2e)
[19]. The values of sR and kex of the Gd3? complexes can
be tuned by appropriate ligand design in order to optimize
relaxivity. The maximum relaxivities predicted by the
Solomon–Bloembergen–Morgan (SBM) theory can only be
attained by chelates displaying simultaneous optimization
of kex and sR [11].
Imaging probes with multimodal features (e.g., MRI/
PET, MRI/SPECT, or MRI/optical imaging) would be
particularly useful for the diagnosis of AD. The most
straightforward design is based on metal chelating agents
that can accommodate different metal ions with capabilities
in various imaging modalities, conjugated to specific
amyloid-targeting units. Metal complexes could offer sig-
nificant advantages, including longer half-lives of available
metal-based PET/SPECT tracers and optimal spatial reso-
lution when MRI is concerned. Recently, in a preliminary
communication we have reported the ligand L1 to create
potential metal-based multimodal imaging probes for
the detection of amyloid plaques in AD [20] (Fig. 1). We
have conjugated an optimized derivative of PiB, a well-
established marker of Ab plaques, to 1,4,7,10-tetra-
azacyclododecane-1,4,7-triacetic acid (DO3A) monoamide,
which is capable of forming stable, noncharged complexes
with different trivalent metal ions, including Gd3? for MRI
applications, 111In3? for SPECT applications, and 68Ga3?
for PET applications. In addition, the PiB unit is fluores-
cent, and can be traced with optical microscopy, repre-
senting another possible detection mode of the agent. The
ligand L1 coordinated to Gd3? showed interesting relax-
ivity, which further increased on binding to the amyloid
peptide. Ex vivo immunohistochemical studies showed that
the complexes selectively target Ab plaques on AD human
brain tissue. Ex vivo biodistribution data obtained with the
111In analogue pointed to a moderate BBB penetration in
adult male Swiss mice (without amyloid deposits), with
0.36 % of the injected dose per gram of tissue in the cortex
at 2 min after injection.
The development of in vivo imaging probes for AD has
so far mainly focused on nuclear probes. Many b-amyloid
(Ab)-labeling nuclear imaging agents have been reported,
mostly based on small organic compounds, such as 11C-
and 18F-labeled derivatives of stilbene [4] and Pittsburgh
compound B (PiB) [5, 6]. These are promising PET tracers
of AD, owing to their high in vivo binding affinities for Ab
aggregates and efficient blood–brain barrier (BBB) per-
meation. Some of these compounds are approaching or are
already in clinical application [4, 6, 7]. Their limitations
are general and are mainly associated with the short life-
time of the radioisotope, the use of ionizing radiation, and
the low spatial resolution of the imaging technique,
although the use of co-localized PET–CT images partially
alleviates this problem [8, 9]. Several small, neutral 99mTc-
based Ab binding probes for AD detection using single-
photon-emission CT (SPECT) have also been reported, and
some showed reasonable brain uptake and affinity towards
Ab plaques, such as those based on pyridyl benzofurane
derivatives [10]. However, despite the progress in in vivo
nuclear imaging of AD, there is a need for novel applica-
tions based on less invasive imaging techniques with better
resolution.
MRI is a powerful clinical and biological imaging tool,
offering noninvasive exploration of structure and function
with excellent spatial resolution. In contrast to nuclear
imaging techniques, MRI does not imply any radiation
burden for the patient. The intrinsic MRI contrast can be
largely enhanced by paramagnetic contrast agents, mainly
Gd3? chelates or iron oxide nanoparticles [11]. In an
advanced stage of AD, the iron content of the plaques
allows their detection with specific MRI acquisition
sequences without any contrast agent, leading to hypoin-
tense spots in T2-, T2*-, or susceptibility-weighted images
[12]. However, it requires high magnetic fields (7 T or
more) and long acquisition times, which are impracticable
in clinical MRI. The detection of plaques weakly loaded
with iron requires the use of exogenous contrast agents.
Contrast-agent-aided MRI studies reported to visualize AD
plaques in transgenic mice include the use of 19F-labeled
compounds for 19F MRI [13] and Gd3? complexes or
ultrasmall superparamagnetic iron oxide conjugated with
modified human Ab1–40 or human Ab1–42 peptide (e.g.,
with putrescine) able to cross the BBB [14–17] or ultra-
small superparamagnetic iron oxide conjugated with Ab-
specific peptides selected using the phage display tech-
nique [18]. Owing to the large size of these paramagnetic
probes, several days (weeks) were necessary to label the
amyloid plaques in the transgenic mouse brain in vivo, and
We report the synthesis of two novel DO3A monoamide
derivative ligands conjugated to the PiB moiety, L2 and L3,
that possess linkers of differing length and chemical
123

J Biol Inorg Chem (2014) 19:281–295
283
Fig. 1 Structure of the Gd(Lx)
(x = 1, 2, 3) complexes
investigated
structure between the macrocycle and the amyloid-targeting
moiety (Fig. 1). L2 contains an extra 4-acetamidobenzamide
group in the linker and L3 has a longer aliphatic 6-acetami-
dohexanamide spacer. With the introduction of these dif-
ferent linkers, the hydrophobic properties and flexibility of
the ligand can be modulated in order to optimize the target
binding properties, the BBB permeation, and the relaxivity
of the Gd3?-based probe. Besides a gadolinium(III) diethy-
lenetriaminepentaacetic acid (DTPA)–curcumin conjugate
reported recently [21], these are the first low molecular
weight, potential MRI contrast agents bearing a specific unit
that allows targeting of amyloid plaques. We describe the
characterization of the most relevant physicochemical
parameters for the use of these complexes as MRI probes for
Ab detection. These studies include the assessment of (1) the
octanol–water partition coefficients, which characterize the
lipophilicity of the complexes, (2) micelle formation in
aqueous solution, (3) the parameters influencing relaxivity
via a combined 17O NMR and 1H nuclear magnetic relaxa-
tion dispersion (NMRD) study, and (4) the binding of the
complexes to Ab1–40 peptide and to human serum albumin
(HSA).
without further purification. Analytical grade solvents
were used and were not purified further unless
specified.
Reactions were monitored by thin-layer chromatog-
raphy (TLC) on Kieselgel 60 F254 (Merck) on an alu-
minum support, with detection by examination under
UV light (254 nm), by adsorption of iodine vapor, by
spraying with ninhydrin, and by coloration of the
complex with Dragendorff solutions. Polar affinity
chromatography was performed with silica gel (Sigma-
1
Aldrich). H and 13C NMR spectra were recorded with a
Bruker Avance-500 (11.7-T) spectrometer operating at
1
500.132 and 125.769 MHz for H and 13C, respectively.
Chemical shifts (d) are given in parts per million rela-
tive to CDCl3 solvent (1H, d 7.27; 13C, d 77.36) as an
1
internal reference. For H and 13C NMR spectra recor-
ded in D2O, chemical shifts are given in parts per
million relative to trimethylsilylpropionic acid as an
internal reference (1H, d 0.0) and tert-butanol as an
external reference (13C, CH3 d 30.29). Mass spectrom-
etry was performed at the Centre de Biophysique Mol-
´ ´
eculaire du CNRS in Orleans, France.
Synthesis
Materials and methods
Gd(L1), Gd(L2), and Gd(L3) were synthesized according to
Scheme 1. The synthesis of the common intermediates 2–5
and of Gd(L1) was previously reported [20]. The chemical
structures and atom numbering schemes for the synthesized
compounds are shown in Figs. S1–S7.
Reagents and solutions
Chemicals were purchased from Sigma-Aldrich, Alfa
Aesar, and CheMatech (DO3A-t-Bu3) and were used
123

284
J Biol Inorg Chem (2014) 19:281–295
Scheme 1 Synthesis of Gd(Lx).
The reagents and conditions
were as follows:
a tetrahydrofuran, NEt3,
4-nitrobenzoyl chloride, 273 K,
96 %; b chlorobenzene,
O
S
NO2
NO2
H2N
OMe
MeO
NH
MeO
NH
a)
b)
1
2
3
S
MeO
MeO
S
N
5
S
N
4
NO2
NH2
NO2
MeO
NH
d)
c)
Lawesson’s reagent, reflux,
85 %; c 10 % NaOH,
O
PiB
Cl
Cl
6
K3Fe(CN)6, reflux, 53 %;
d SnCl2, EtOH, reflux, 92 %;
e ClCH2COCl (6), CH2Cl2,
NaOHaq, room temperature,
72–86 %; f 1 Pittsburgh
compound B (PiB) (5), K2CO3,
CH3CN or (CH3)2CO, room
temperature, 50–82 %; f 2
SOCl2, Cl3CH, reflux; g DO3A-
t-Bu3 (8), K2CO3, CH3CN,
room temperature, 38–93 %;
h CH2Cl2, trifluoroacetic acid,
318 K, 73–79 %; j GdCl3, pH 7
f) 1)
O
Cl
MeO
O
linker
O
S
N
linker
linker
H2N
HN
NH
O
OH
OH
e)
Cl
7
f) 2)
11
15
12
16
O
13 i)
17
ii)
7
OMe
S
N
BuOt
13 i)
17
HN
O
N HN
linker
ii)
g)
O
O
O
O
N
N
N
N
N
N
O
O
O
O
OtBu
i)
9
BuOt
8
linker
14 i)
18 ii)
O
O
ii)
h)
OMe
OMe
S
N
S
N
HN
O
HN
O
linker
linker
HO
HO
N
N
N
N
N
N
O
O
Gd3+
O
O
O
O
O
O
L1
L2
L3
N
N
j)
OH
HO
OH
HO
Synthesis of L2
vacuum, with heating at 313 K, and the crude product (1.2 g,
5.2 mmol) was added to a 45-mL acetone solution contain-
ing 5 (1.2 g, 4.7 mmol) with potassium carbonate (2.14 g,
15.6 mmol), and the mixture was stirred overnight. The solid
was filtered off and washed with acetone, and then the
solvents were evaporated. The product obtained was
recrystallized in acetonitrile to give 1.05 g of 4-(2-chloro
acetamido)-N-(4-(6-methoxybenzo[d]thiazol-2-yl)phenyl)
benzamide (13) (50 %) (Fig. S9). 1H NMR (dimethyl
sulfoxide, 298 K, 500 MHz): d (ppm) 3.86 (s, 3H, H-23),
4.32 (s, 2H, H-1), 7.123 (d, J = 8.6 Hz, 1H, H-20),
7.7–8.13 (m, 10H, H-18, H-21, H-5, H-4, H-7, H-8, H-12,
H-11, H-14, H-15), 10.44 (s, NH), 10.62 (s, NH). 13C NMR
(dimethyl sulfoxide, 298 K, 126 MHz): d (ppm) 43.6 (C-1),
55.8 (C-23), 104.9 (C-18), 118.6 (C-11, C-15), 119.1 (C-
20), 120.4 (C-4, C-8), 123.2 (C-21), 127.5 (C-12, C-14),
128.1 (C-13), 128.9 (C-5, C-7), 129.6 (C-6), 131.9 (C-17),
135.8 (C-10), 141.6 (C-3), 141.8 (C-22), 148.1 (C-19),
157.4 (C-9), 164.4 (C-2), 165.1 (C-16).
4-(2-Chloroacetamido)benzoic acid
A biphasic reaction of 11 (1.2 g, 8.8 mmol) in 45 mL of
CH2Cl2 and an aqueous 0.1 M NaOH solution was per-
formed by adding a solution of chloroacetyl chloride (6)
(0.9 mL, 11.2 mmol) in 15 mL of CH2Cl2. The mixture
was stirred at room temperature for 3 h and the aqueous
phase was acidified to pH *2 with an aqueous solution of
0.2 M HCl. The solution was extracted with CH2Cl2
(6 9 15 mL). The organic phase was washed with 10 mL
of water, dried over MgSO4, and the solvent was evapo-
rated. The crude product was dissolved in a small amount
of acetone, diethyl ether was added afterwards, and the
solution was stored at 268 K overnight. White crystals
were filtered to give 1.56 g of 4-(2-chloroacetamido)ben-
1
zoic acid (12) (82 %) (Fig. S8). H NMR (CDCl3, 298 K,
500 MHz): d (ppm) 4.29 (s, 2H, H-9), 7.71 (d, J = 8.5 Hz,
2H, H-4, H-6), 7.91 (d, J = 8.5 Hz, 2H, H-3, H-7), 10.66
(s, –OH). 13C NMR (CDCl3, 298 K, 126 MHz): d (ppm)
44.0 (C-9), 119.0 (C-4, C-6), 130.8 (C-3, C-7), 142.9 (C-2),
165.5 (C-8), 167.3 (C-1).
Tri-tert-butyl-2,20,200-(10-(2-((4-(6-methoxybenzo[d]
thiazol-2-yl)phenyl)carbamoyl)phenyl)amino)-2-oxoethyl)-
1,4,7,10-tetraazacyclododecane-1,4,7-triyl)triacetate
4-(2-Chloroacetamido)-N-(4-(6-methoxybenzo[d]thiazol-
2-yl)phenyl)benzamide
To a solution of DO3A-t-Bu3 (8) (300 mg, 0.58 mmol) and
potassium carbonate (226 mg, 2 mmol) in 20 mL of dry
acetonitrile was added a solution of 13 (316 mg, 0.7 mmol)
in 15 mL of dry acetonitrile at room temperature. The
reaction mixture was stirred for 48 h at room temperature
A solution of 12 (1.2 g, 5.6 mmol) was solubilized in SOCl2
for 2 h at 323 K. The SOCl2 was evaporated under a
123

J Biol Inorg Chem (2014) 19:281–295
285
and the reaction was followed by TLC. The solid was fil-
tered off and the solvent was evaporated. The crude
product was purified by flash chromatography with
dichloromethane–ethyl acetate (9:2). Tri-tert-butyl-2,20,200-
(10-(2-((4-(6-methoxybenzo[d]thiazol-2-yl)phenyl)carbamoyl)
phenyl)amino)-2-oxoethyl)-1,4,7,10-tetraazacyclododecane-1,
4,7-triyl)triacetate (14) (200 mg) was obtained as an orange
oil (38 %) (Fig. S10). 1H NMR (CDCl3, 298 K, 500 MHz): d
(ppm) 1.42 (bs, 27H, –Ot-Bu, H-34, H-35), 2.80–3.44 (m,
22H, H-25, H-24, H-26, H-27, H-23, H-28, H-29), 3.84 (s,
3H, H-7), 7.02–8.07 (m, 11H, H-1, H-5, H-4, H-10, H-14,
H-11, H-13, H-21, H-17, H-20, H-18), 10.16 (s, NH), 11.19
(s, NH). 13C NMR (CDCl3, 298 K, 126 MHz): d (ppm) 27.9
(C-34, C-35), 42.3 (C-24, C-25, C-26, C-27), 45.4 (C-29,
C-28), 67.3 (C-23), 82.0 (C-32), 82.1 (C-33), 104.1 (C-1),
115.4 (C-5), 119.4 (C-11, C-13), 120.8 (C-20, C-18), 123.3
(C-4), 124.7 (C-12), 127.7 (C-10, C-14), 128.4 (C-16), 128.8
(C-21, C-17), 130.9 (C-2), 136.3 (C-9), 141.7 (C-19), 142.4
(C-3), 148.7 (C-6), 157.5 (C-8), 164.6 (C-15), 165.5 (C-22),
166.1 (C-30), 171.4 (C-31).
was evaporated. The crude product was dissolved in a
small amount of acetone and diethyl ether was added
afterwards, and the solution was stored at 268 K overnight.
White crystals were filtered off to give 1.7 g of 6-(2-
1
chloroacetamido)hexanoic acid (16) (72 %) (Fig. S12). H
NMR (CDCl3, 298 K, 500 MHz): d (ppm) 1.39 (m, 2H,
H-4), 1.57 (m, 2H, H-5), 1.65 (m, 2H, H-3), 2.35 (t, 2H,
H-2), 3.31 (m, 2H, H-6), 6.67 (s, 1H, –NH).
6-(2-Chloroacetamido)-N-(4-(6-methoxybenzo[d]thiazol-
2-yl)phenyl)hexanamide
A solution of 16 (1.2 g, 5.6 mmol) was solubilized in
SOCl2 for 1 h at 298 K. The SOCl2 was evaporated under a
vacuum at 298 K, solubilized in chloroform, and extracted
with a solution of aqueous NaOH (pH *9). The crude
chlorinated product (1.2 g, 5.2 mmol) was added to a
45-mL acetonitrile solution containing 5 (1.2 g, 4.7 mmol)
with potassium carbonate (2.14 g, 15.6 mmol) and the
mixture was stirred overnight at room temperature. The
solid was filtered off and washed with acetonitrile, and then
the solvents were evaporated. The product obtained was
recrystallized in acetonitrile to give 2.04 g of 6-(2-chlo
roacetamido)-N-(4-(6-methoxybenzo[d]thiazol-2-yl)phenyl)
hexanamide (17) (82 %) (Fig. S13). 1H NMR (CDCl3,
298 K, 500 MHz): d (ppm) 1.30 (m, 2H, H-18), 1.45 (m,
2H, H-19), 1.60 (m, 2H, H-17), 2.34 (t, 2H, H-16), 3.10 (m,
2H, H-20), 3.85 (s, 3H, –OCH3, H-7), 4.40 (s, 2H, H-22),
7.11 (bd, 1H, H-5), 7.68 (bs, 1H, H-1), 7.76 (bd, 2H, H-11,
H-13), 7.89 (bd, 1H, H-4), 7.96 (bd, 2H, H-10, H-14), 8.20
(s, –NH), 10.18 (s, –NH).
2,20,200(10-(2-((4-(6-Methoxybenzo[d]thiazol-2-
yl)phenyl)carbamoyl)phenyl)amino)-2-oxoethyl)-1,4,7,10-
tetraazacyclododecane-1,4,7-triyl)triacetic acid (L2)
Compound 14 (200 mg, 0.22 mmol) was dissolved in a
mixture of trifluoroacetic acid and dry dichloromethane
(1:1) and the resulting mixture was heated at 318 K over-
night. The solvents were evaporated and the crude product
was dried under a vacuum. The crude product was dis-
solved in a small amount of absolute ethanol and then
diethyl ether was added. The mixture was allowed to stand
overnight at 268 K. The yellowish precipitate (130 mg)
was filtered off and dried under a vacuum (78 %) (Fig.
S11). 1H NMR (D2O, 323 K, 500 MHz): d (ppm)
3.60–4.13 (bs, 24H, H-23, H-28, H-29, H-25, H-24, H-26,
H-27), 4.44 (s, 3H, H-7), 6.87–7.92 (m, 11H, H-1, H-4,
H-5, H-10, H-14, H-11, H-13, H-21, H-17, H-20, H-18).
High-resolution mass spectrometry (electrospray ioniza-
tion): m/z: calcd for C37H43N7O9S 761.284, found 761.244.
Tri-tert-butyl-2,20,200-(10-(2-((6-((4-(6-methoxybenzo[d]
thiazol-2-yl)phenyl)amino)6-oxohexyl)amino)-2-oxoethyl)-
1,4,7,10-tetraazacyclododecane-1,4,7-triyl)triacetate
To a solution of 8 (600 mg, 1.16 mmol) and potassium
carbonate (552 mg, 4 mmol) in 20 mL of dry acetonitrile
was added a solution of 17 (400 mg, 0.89 mmol) in 15 mL
of dry acetonitrile. The reaction mixture was stirred for
48 h at room temperature and the reaction was followed by
TLC. The solid was filtered off and the solvent was
evaporated. The crude product was purified by flash chro-
matography with dichloromethane–ethyl acetate (9:4).
Tri-tert-butyl-2,20,200-(10-(2-((6-((4-(6-methoxybenzo[d]
thiazol-2-yl)phenyl)amino)6-oxohexyl)amino)-2-oxoethyl)-1,4,
7,10-tetraazacyclododecane-1,4,7-triyl)triacetate (18) (750 mg)
Synthesis of L3
6-(2-Chloroacetamido)hexanoic acid
To a biphasic mixture containing 15 (1.5 g, 11.4 mmol) in
45 mL of CH2Cl2 and an aqueous solution of 0.1 M NaOH,
a solution of 6 (1.05 mL, 13.08 mmol) in 15 mL of CH2Cl2
was added. The mixture was stirred 3 h at room tempera-
ture and the aqueous phase was acidified to pH *2 with an
aqueous solution of 0.2 M HCl. The solution was extracted
with CH2Cl2 (6 9 15 mL). The organic phase was washed
with 10 mL of water, dried over MgSO4, and the solvent
1
was obtained as a yellow oil (93 %) (Fig. S14). H NMR
(CDCl3, 298 K, 500 MHz): d (ppm) 1.25 (t, 27H, –Ot-Bu,
H-33, H-34), 1.46 (bs, 2H, H-18), 1.60 (bs, 4H, H-19,
H-17), 2.43 (bs, 2H, H-16), 2.67 (s, 16H, H-25, H-26,
H-23, H-24), 2.74 (s, 8H, H-22, H-27, H-28), 3.06 (bs, 2H,
H-20), 3.7 (s, 3H, –OCH3, H-7), 6.84 (bd, J = 8.8 Hz, 1H,
123

286
J Biol Inorg Chem (2014) 19:281–295
H-5), 7.10 (bs, 1H, H-1), 7.67 (d, J = 8.8 Hz, 1H, H-4),
7.77 (d, J = 8.3 Hz, 2H, H-11, H-13), 7.89 (d, J = 8.3 Hz,
2H, H-14, H-10), 8.26 (s, NH), 10.18 (s, NH). 13C NMR
(CDCl3, 298 K, 126 MHz): d (ppm) 25.0 (C-18), 25.6 (C-
17), 27.8 (C-34), 27.9 (C-33), 27.9 (C-22), 28.1 (C-19),
37.1 (C-16), 38.4 (C-20), 55.4 (C-25, C-26), 55.6 (C-27,
C-28), 55.8 (C-24, C-23), 56.2 (C-7), 81.7 (C-32), 81.8 (C-
31), 104.2 (C-1), 115.3 (C-5), 112.0 (C-11, C-13), 123.2
(C-4), 127.5 (C-10, C-14), 128.0 (C-9), 136.1 (C-2), 142.4
(C-12), 148.7 (C-3), 157.5 (C-6), 165.7 (C-21), 171.6 (C-
8), 172.3 (C-29, C-30), 173.5 (C-15).
from the bottle to an equimolar solution of Gd(Lx)
(200 lM) in 0.05 M N-(2-hydroxyethyl)piperazine-N0-eth-
anesulfonic acid buffer at pH 7 and the sample was soni-
cated. Milli-Q water was always used to avoid metal
contamination. The sample containing the reconstituted
peptide was used immediately to avoid degradation in
solution.
Determination of the octanol–water partition coefficient
The partition coefficient was determined as the ratio of the
concentration of the compound in octanol and concentra-
tion of the compound in the aqueous phase (Eq. 1):
2,20,200-(10-(2-((6-((4-(6-Methoxybenzo[d]thiazol-2-
yl)phenyl)amino)-6-oxohexyl)amino)-2-oxoethyl)-1,4,7,10-
tetraazacyclododecane-1,4,7-triyl)triacetic acid (L3)
½soluteꢁ octanol phase
Partition coefficient (PÞ ¼
:
ð1Þ
½soluteꢁ aqueous phase
The logarithm of the partition coefficient is referred to as
the log P value.
Compound 18 (750 mg, 0.81 mmol) was dissolved in a
mixture of trifluoroacetic acid and dry dichloromethane
(1:1) and the resulting mixture was heated at 318 K over-
night. The solvents were evaporated and the crude product
was dried under a vacuum. The crude product was dis-
solved in a small amount of CH2Cl2, and then acetone was
added to the mixture and the resulting mixture was left
overnight at 268 K for recrystallization. Yellowish crystals
(450 mg) were filtered off and dried under a vacuum
The ‘‘shake flask’’ method was used for the determination
of log P [23]. Water saturated with octanol and octanol sat-
urated with water were used in the experiments. The ben-
zothiazol ring absorbs strongly at approximately 330 nm;
therefore, the partition was quantified using UV spectro-
photometry with a PerkinElmer Lambda 19 UV–vis spec-
trophotometer [24]. For each phase, the maximum
wavelength was verified. A 1:1 volume ratio was used for the
partitioning of the solution with Gd(Lx). In a 2-mL Eppen-
dorf tube, 0.5 mL of a 100 lM solution of Gd(Lx) was added
to 0.5 mL of the saturated phase of 1-octanol. Each sample
was centrifuged for 30 min. Gd(Lx) concentrations were
determined in each phase using standard curves.
1
(73 %) (Fig. S15). H NMR (D2O, 323 K, 500 MHz): d
(ppm) 1.72 (bs, 2H, H-18), 1.91 (bs, 2H, H-19), 1.98 (bs,
2H, H-17), 2.67 (s, 2H, H-20), 3.6 (bs, 12H, H-25, H-26,
H-16), 3.70 (bs, 4H, H-24), 3.89 (bs, 3H, H-7), 4.10 (bs,
4H, H-27), 4.26 (bs, 2H, H-28), 4.31 (bs, 2H, H-22), 6.98
(bs, 1H, H-5), 7.17 (bs, 1H, H-1), 7.59 (bs, 4H, H-10, H-14,
H-11, H-13), 7.87 (bs, 1H, H-4). 13C NMR (D2O, 323 K,
126 MHz): d (ppm) 25.4 (C-17), 26.6 (C-18), 28.8 (C-19),
37.3 (C-16), 40.1 (C-20), 49.9 (C-23, C-24), 51.3 (C-25,
C-26), 54.5 (C-28), 55.4 (C-22), 55.9 (–O–CH3, C-7), 56.2
(C-27), 104.8 (C-1), 116.5 (C-5), 120.6 (C-11,13), 122.6
(C-4), 127.3 (C-9), 128.0 (C-10,14), 135.3 (C-2), 141.2 (C-
12), 146.3 (C-3), 157.9 (C-6), 163.0 (C-21), 163.3 (C-8),
166.5 (C-15), 172.6 (C-30), 174.8 (C-29). High-resolution
mass spectrometry (electrospray ionization): m/z: calcd for
C36H49N7O9S 755.331, found 755.265.
1H NMRD measurements
1H NMRD profiles were recorded with a Stelar SMARtr-
acer fast field cycling NMR relaxometer (0.01–10 MHz)
and a Bruker WP80 NMR electromagnet (20, 40, 60, and
80 MHz) adapted to variable-field measurements and
controlled by a SMARtracer PC-NMR console. The tem-
perature was monitored by a VTC91 temperature control
unit and was maintained by a gas flow. The temperature
was determined by previous calibration with a platinum
resistance temperature probe. The longitudinal relaxation
rates (1/T1) were determined in water. Measurements were
performed at 298 and 310 K.
Sample preparation
Gd(Lx) (x = 1, 2, 3) complexes were prepared by mixing
solutions of GdCl3 and the ligand Lx in equimolar quanti-
ties and adjusting the pH to 7 with aqueous NaOH
(0.1 mM). The solutions were allowed to react for 24 h at
333 K while the pH was regularly controlled. The absence
of free metal ion was checked in each sample by using the
xylenol orange test [22]. In the relaxometric HSA binding
studies, 0.6 mM HSA (4 %) solutions were used. For the
Ab1–40 binding studies, the peptide was added directly
17O NMR experiments
Variable-temperature 17O NMR measurements were per-
formed with a Bruker Avance-500 (11.7-T) spectrometer,
and a BVT-3000 temperature control unit was used to
stabilize the temperature. The temperature was calculated
according to a previous calibration with ethylene glycol
123

J Biol Inorg Chem (2014) 19:281–295
287
and methanol [25]. The samples were sealed in glass
spheres that fitted into 10 mm outer diameter NMR tubes,
to eliminate susceptibility corrections to the chemical shifts
[19, 26]. Longitudinal relaxation rates (1/T1) were obtained
by the inversion recovery method and transverse relaxation
rates (1/T2) were obtained by the Carr–Purcell–Meiboom–
Gill spin-echo technique. Acidified water of pH 3.4 was
used as an external reference. 17O-enriched water (10 %
H217O, CortectNet) was added to the solutions to reach
around 1 % enrichment.
from the aniline moiety undergoes a fast reaction with the
chloroacyl group, forming product 7 with good yield after
recrystallization. To obtain the intermediate 13, the reagent
4-aminobenzoic acid (11) was reacted with 6 under
Schotten–Baumann conditions [30, 31], giving the corre-
sponding amide 12. Compound 12 was then activated by
chlorination with thionyl chloride (SOCl2). Consequently,
4-(2-chloroacetamido)benzoyl chloride was introduced in a
fast reaction with compound 5 using acetone as the solvent,
to achieve the condensation and obtain the final intermediate
4-(2-chloroactetamido)-N-(4-(6-methoxybenzo[d]thiazol-2-
yl)phenyl)benzamide) (13).
Determination of the affinity constants for HSA binding
of Gd(L1) and Gd(L3)
The preparation of the longer linker 6-acetoamidohex-
anamide required the acylation of compound 15. The
intermediate product 6-(2-chloroacetoamido)hexanoic acid
(16) obtained was reacted with SOCl2 to achieve the
chlorination of the carboxylic acid in very mild conditions
(CH2Cl2, 298 K) and to minimize side reactions (cycliza-
tion). The very reactive 2-chloro-6-acetoamidohexanamide
chloride obtained before was reacted with compound 5
with the primary free amine available and made possible
the condensation, giving 6-2-chloroactetamido)-N-(4-(6-
methoxybenzo[d]thiazol-2-yl)phenyl)hexanamide (17) in
high yield.
Affinity constants with regard to HSA (defatted, from
Sigma-Aldrich, 0.01 % or less fatty acids and 1 % or less
globulins) were assessed by proton relaxation enhancement
(PRE) measurements. The proton relaxation rates at
increasing concentrations of the protein or the metal che-
late were measured using a Bruker WP80 NMR electro-
magnet adapted to variable-field measurements and
controlled by a SMARtracer PC-NMR console (40 MHz,
310 K). For the E-titration [the concentration of Gd(L1) or
Gd(L3) is constant, and the protein concentration is varied],
the Gd(L1) and Gd(L3) concentrations were 0.1 mM, and
for the M-titration (the protein concentration is constant
and the complex concentration is varied), the HSA con-
centration was 0.6 mM.
Ligands L1, L2, and L3 were obtained by the mono-
alkylation of the functionalized compound 8 with com-
pounds 7, 13, and 17. The reaction proceeded in
acetonitrile at room temperature for 48 h. The acid-sensi-
tive tert-butyl protecting groups were then removed with
trifluoroacetic acid and the ligands with the free carboxylic
acid available to form Ln3? complexes were obtained with
global yields over the two last steps of 70, 30, and 68 % for
L1, L2, and L3, respectively.
Results and discussion
Synthesis
The synthesis of the benzothiazole targeting moiety was
done according to Mathis et al. [5, 27, 28]. The general
strategy for the synthesis of the ligands is outlined in
Scheme 1. We performed the amide formation by the
acylation of 4-methoxyaniline (1) with 4-nitrobenzoyl
chloride to form the product 2 in very high yield (96 %).
Subsequently, compound 2 was made to react with half
equivalents of Lawesson’s reagent and the thiation occur-
red with a good yield, giving the corresponding thioamide
3. The following reaction was the cyclization of compound
3 in the ortho position of the methoxyphenyl ring to form
6-methoxy-2-(4-nitrophenyl)benzo[d]thiazole (4) [29].
Compound 4 was reacted with stannous chloride to reduce
the nitrophenyl group, to obtain, with a remarkably high
yield, the envisaged benzothiazol derivative 4-(6-meth-
oxybenzol[d]thiazol-2-yl)aniline (5), containing a free ter-
minal primary amine function.
Lipophilicity of the complexes as determined
by the partition coefficient
A contrast agent designed to detect amyloid plaques in the
brain should optimally pass the BBB. The BBB perme-
ability of a compound is determined by different factors,
including lipophilicity, often expressed by the water–oct-
anol partition coefficient, log Poct/water, molecular weight,
and plasma pharmacokinetics [32]. A low molecular
weight and an amphiphilic character of the molecule are
known to favor BBB permeability. Typically, compounds
with log Poct/water *2 have optimal BBB penetration. The
log Poct/water values obtained for Gd(L2) and Gd(L3), 0.32
and 0.03, respectively, are considerably higher than the
value obtained for Gd(L1), -0.15 [20], owing to the more
hydrophobic nature of their linker (Table 1). As expected,
all these values are lower than those of the highly lipophilic
PiB molecule or other phenylbenzothiazole derivatives
[5, 32, 33].
Acylation of compound 5 with chloroacetyl chloride (6)
was performed in acetone. The available primary amine
123

288
J Biol Inorg Chem (2014) 19:281–295
Table 1 Molecular weight (MW) and lipophilicity (log Poct=water) of
phenylbenzothiazole derivatives
log Poct=water
MW
References
Thioflavin T
0.57
2.6
319
240
254
256
256
602
616
559
573
587
842
918
912
[5]
R1 is CH3, R2 is NH2
R1 is CH3, R2 is NHCH3
R1 is OCH3, R2 is NH2
R1 is OH, R2 is NHCH3 (PiB)
R1 is OH, R2 is ReO-TEEDA
R1 is OCH3, R2 is ReO-TEEDA
99mTcO-BAT-Bp-1
99mTcO-BAT-Bp-2
99mTcO-BAT-Bp-3
Gd(L1)
[5, 27]
[5]
2.6
1.9
[5]
1.23
1.21
2.52
0.68
1.35
2.09
-0.15
0.32
0.03
[33]
[33]
[33]
[10]
[10]
[10]
Fig. 2 Paramagnetic contribution to the water 1H longitudinal
relaxation rates as a function of the Gd(L3) concentration at
40 MHz and 298 K
[20]
Gd(L2)
This work
This work
Gd(L3)
BAT-Bp-1, bis(aminoethanethiol)-benzofuranpyridyl-NH2; BAT-Bp-2, bis
(aminoethanethiol)-benzofuranpyridyl-NH(CH3); BAT-Bp-3, bis(amino-
ethanethiol)-benzofuranpyridyl-N(CH3)2; PiB, Pittsburgh compound B;
TEEDA, ethylthiol-diethylenetriamine
determined the critical micellar concentration (cmc) for the
Gd(L3) complex by relaxometric measurements [38]. The
cmc for Gd(L1) was reported previously [20]. The low
solubility of Gd(L2) (0.133 mM) does not allow assessment
of the cmc. To obtain the cmc, the paramagnetic relaxation
rates, R1p, were plotted versus the Gd(L3) concentration at
40 MHz, a magnetic field where the effect of slower
rotation on relaxivity is the most pronounced and thus the
relaxivity difference between the monomer and the
aggregated state is important (Fig. 2).
Some rhenium and technetium complexes proposed as
nuclear imaging probes for AD detection have been repor-
ted to have relatively high, close-to-optimal log Poct/water
values, and correspondingly, they had interesting BBB
passage in vivo [10]. These complexes are, however,
structurally very different from our Gd3? chelates. In fact, it
is not possible to create stable lanthanide complexes that are
sterically as compact as those of technetium or rhenium,
since Ln3? cations are much larger than M(V) (M is Tc,
Re) cations in the MO3? oxocations [34] and need con-
siderably more chelating functions to form stable and inert
complexes. Moreover, the chemical nature of lanthanide
ions requires strongly ionic coordinating functions. These
factors all contribute to an increased hydrophilic character
of the chelates. The addition of bulky lipophilic groups
could increase the overall lipophilicity of the complex, but
this would also increase the molecular weight, which dis-
favors BBB permeability. Indeed, low molecular weight
compounds cross the BBB more efficiently with rates that
seem to correlate inversely with the square root of their
molecular weight [35], at least for compounds with a
molecular weight 600. The molecular weights of Gd(L1),
Gd(L2), and Gd(L3) are 842, 918, and 912, respectively,
which are above the optimal values. However, peptide
derivatives with a molecular weight over 1,000 were shown
to cross the BBB by passive diffusion [36, 37].
At concentrations below the cmc no aggregates form,
and under these conditions, only the monomeric chelate
1
contributes to the paramagnetic H relaxation rate mea-
sured in solution, which is given by Eq. 2:
R1p ¼ Ro1bsꢂRd1 ¼ r1na ꢃ CGd
ð2Þ
Here R1d is the diamagnetic contribution to the longitudinal
relaxation rate (the relaxation rate of pure water), r1na rep-
resents the relaxivity of the free, nonaggregated Gd3?
chelate (mM-1 s-1), and CGd is the analytical Gd3?
concentration.
At concentrations above the cmc, the measured relaxa-
tion rate is the sum of two contributions, one due to the
chelate as a monomer (free surfactant) present at a con-
centration given by the cmc, and the other due to the
1
aggregated form (micelles). The water H relaxation rate
measured for the paramagnetic solution can be then
expressed as in Eq. 3:
R1obs ꢂ R1d ¼ ðr1n:aꢂr1aÞ ꢃ cmc þ r1a ꢃ CGd
ð3Þ
where r1a is the relaxivity of the micellar (aggregated) form.
The cmc is determined from a plot of the paramagnetic
relaxation rate versus the Gd3? concentration by a simul-
taneous least-squares fit of the two straight lines (Fig. 2).
The slopes of these two lines define r1na and r1a, below and
Determination of the critical micellar concentration
1
by H relaxivity measurements
Given the amphiphilic character of the Gd(Lx) chelates,
they form micellar aggregates in aqueous solution. We
123

J Biol Inorg Chem (2014) 19:281–295
289
Table 2 Relaxivities of the monomer (r1na) and the aggregated (r1a)
form and critical micellar concentrations obtained from the 1H re-
laxivity study for Gd(Lx) complexes (40 MHz, 298 K) compared with
those for other micellar systems from the literature
Gd(Lx) chelates exhibit approximately 50–60 % higher re-
laxivities than clinical contrast agents [39, 40]: r1 = 6.30,
6.48, and 6.46 mM-1 s-1, respectively, for x = 1, 2, and 3
(20 MHz, 298 K). These higher r1 values are the conse-
quence of a slower rotation, resulting from the presence of
the bulky PiB moiety. When the concentration is increased
to 5 mM, the NMRD profiles change considerably, and
correspond to slowly tumbling systems with a typical high-
field peak around 40 MHz (Fig. 3c, d). The features of the
NMRD curve are influenced by the water exchange rate,
electron relaxation parameters, and rotational correlation
times. Therefore, NMRD measurements are usually com-
bined with 17O NMR measurements to assess all the
parameters that determine proton relaxivity. By performing
variable-temperature 17O T2 measurements, one can accu-
rately determine the water exchange rate. The rotational
correlation time can be assessed by performing variable-
temperature 17O T1 measurements. On the other hand, var-
iable-temperature measurements of the chemical shift dif-
ference between bulk and bound water (Dxr), give an
indication of the q value. A reliable determination of the
Complex
r1na
r1a
Critical
(mM-1 s-1
)
(mM-1 s-1
)
micellar
concentration
(mM)
Gd(L1)a
6.4
13.9
1.49
Gd(L3)b
6.1 0.2
5.3
13.8 0.4
10.8
1.00 0.02
7.20
-c,d
-c,d
-c,d
Gd(DOTAC10
Gd(DOTAC12
Gd(DOTAC14
)
)
)
5.5
17.9
4.45
5.4
22.0
0.87
-c,d
Gd(DOTASAC18
)
10.5
17.0
0.06
See Fig. 1 for the structure of Gd(L1), Gd(L2), and Gd(L3)
DOTAC10, 1,4,7,10-tetraaza-1-(1-carboxymethylundecane)-4,7,10-triacetic acid
cyclododecane; DOTAC12 1,4,7,10-[tetraaza-1-(1-carboxymethyltridecane)-
,
4,7,10-triacetic acid cyclododecane; DOTAC14, 1,4,7,10-tetraaza-1-(1-carboxy-
tert-butoxymethylquintodecane)-4,7,10-triacetic acid; DOTASAC18, [2-(methy-
loctadecylamino)-2-oxoethyl]-1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraace-
tic acid
a
From [20]
b
This work
c
parameters common to H NMRD and 17O NMR is often
Parameters obtained at 60 MHz and 298 K
1
d
From [38]
performed through the simultaneous least-squares fitting of
all the data obtained. Nevertheless, given the different
concentrations typically used in NMRD and 17O NMR
measurements, such a simultaneous fit is not possible when
concentration-dependent phenomena such as micellar
aggregation occur in the system.
We performed a variable-temperature 17O NMR study
on an aqueous solution of Gd(L3). The low water solubility
of the other Gd3? complexes prevented 17O NMR mea-
surements, which require concentrations typically above
10 mM. Figure 4 shows the temperature dependency of the
reduced 17O chemical shifts (Dxr), the transverse relaxa-
tion rates (1/T2r), and the longitudinal relaxation rates
(1/T1r). For Gd(L3), the transverse 17O relaxation rates
increase with decreasing temperatures above 320 K, indi-
cating that this complex is in the fast exchange regime. At
lower temperatures, the transverse relaxation rates corre-
spond to a slow exchange regime. The reduced chemical
shifts are in accordance with this trend.
above the cmc, respectively. The values obtained for
Gd(L3) were r1na = 6.1 mM-1 s-1 and r1a = 13.8 mM-1
s-1 (298 K, 40 MHz), similar to those for Gd(L1) (Table 2)
[20]. The cmc for Gd(L3) is 1.00 0.02 mM, slightly
lower than that for Gd(L1) (Table 2), owing to the presence
of the C6 lipophilic spacer chain. These values, when
compared with those for previously studied hydrocarbon
chain amphiphilic Gd3? complexes (Table 2) are similar to
the cmc of complexes comprising relatively long, C12–C14
lipophilic tails, showing that the aromatic structures in
Gd(L1) and Gd(L3) efficiently promote aggregation. The
cmc has been previously determined at variable tempera-
tures for other systems and it was found to be identical
within the error at 298 and 310 K [38]. Therefore, we can
consider that the cmc remains constant for our systems
within this temperature range.
1
17O NMR and H NMRD measurements
In the slow exchange regime, the reduced transverse
relaxation rates are directly determined by the water
exchange rate. In the fast exchange regime, they are also
influenced by the longitudinal electronic relaxation rate
(1/T1e) and the scalar coupling constant (A/hꢀ). The reduced
NMRD profiles reflect the magnetic field dependency of the
proton relaxivity (r1) and are commonly used to characterize
1
MRI contrast agents. H NMRD profiles were recorded at
17O chemical shifts are determined by A/ꢀh. Transverse 17
O
298 and 310 K over the frequency range 0.01–100 MHz for
the Gd3? complexes at concentrations below and above the
cmc, except for Gd(L2), for which low solubility prevented
working in the aggregated state. The NMRD curves recor-
ded at low, 0.2 mM (L1, L3), or 0.1 mM (L2) concentration
are characteristic of low molecular weight complexes
(Figs. 3a, b, S19). Even at these low concentrations, the
relaxation is governed by the scalar relaxation mechanism,
and thus contains no information on the rotational motion
of the system. In contrast to 1/T2r, the longitudinal 17O
relaxation rates (1/T1r) are determined by dipole–dipole
and quadrupolar relaxation mechanisms, both related to
rotation.
123

290
J Biol Inorg Chem (2014) 19:281–295
Fig. 3 1H nuclear magnetic relaxation dispersion (NMRD) profiles at
298 K (squares) and 310 K (triangles) of a Gd(L1) at 0.2 mM
concentration (momomer form), b Gd(L3) at 0.2 mM concentration
(momomer form), c Gd(L1) at 5.0 mM concentration (micellar form),
and d Gd(L3) at 5.0 mM concentration (micellar form). The solid
lines correspond to the fits as explained in the text
Since the 17O NMR and the NMRD data were obtained
at different concentrations where the aggregation state of
the Gd(L3) complex is different, they were analyzed sep-
arately using the traditional SBM theory [11, 19]. Mono-
hydration was assumed for all complexes—Gd(Lx) stands
for [Gd(Lx)(H2O)] throughout the text. For Gd(L3), par-
tially in the slow water exchange regime, the reduced 17O
chemical shifts, and consequently the scalar coupling
constant calculated, give direct indication of the hydration
state and they clearly evidence one inner-sphere water
molecule. For the NMRD data, we evaluated the curves
both below and above the cmc (Figs. 3, S19). The theo-
retical equations and the details of the analysis are shown
in the electronic supplementary material and the parame-
ters obtained are given in Tables 3 and 4.
in this way for the micellar form were fitted to the SBM
theory by including the Lipari–Szabo treatment for the
description of the rotational motion. In this approach, two
kinds of motion are assumed to modulate the interaction
causing the relaxation, namely a rapid, local motion which
lies in the extreme narrowing limit and a slower, global
motion. We calculated, therefore, sg, the correlation time
for the global motion (common to the whole micelle), and
sl, the correlation time for the fast local motion, which is
specific for the individual relaxation axis, and is thus
related to the motion of the individual Gd3? chelate units.
The generalized order parameter, S2, is a model-indepen-
dent measure of the degree of spatial restriction of the local
motion, with S2 = 0 if the internal motion is isotropic and
S2 = 1 if the motion is completely restricted. In the fit of
the micellar form, we used only the relaxivities above
4 MHz, where the validity of the SBM theory for slowly
rotating systems is respected. The water exchange rate and
the water exchange enthalpy were fixed for Gd(L3) to the
values obtained in the 17O NMR study (Table 3, Fig. 4).
The Gd(L1) complex in its micellar form exhibits a distinct
and unusual temperature dependency: the relaxivities at
Above the cmc, the proton relaxation rates measured
represent the sum of the contributions of the monomer
complex, which is present at a concentration equal to the
cmc, and of the aggregated state. Therefore, to calculate the
relaxivity of the aggregated form, we subtracted the re-
laxivity contribution of the monomer from the relaxation
rates measured above the cmc. The relaxivities calculated
123

J Biol Inorg Chem (2014) 19:281–295
291
Table 4 Best-fit parameters obtained for the monomer forms of
[Gd(L1)(H2O)], [Gd(L2)(H2O)], and [Gd(L3)(H2O)] from the analysis
of NMR dispersion (NMRD) data below the critical micellar
concentration
Parameters
[Gd(L1)(H2O)] [Gd(L2)(H2O)] [Gd(L3)(H2O)]
(106 s-1
)
298
ex
2.8
2.8
2.8
k
DHà (kJ mol-1
)
32.0
32.0
32.0
ER (kJ mol-1
)
18.2 0.2
130 10
1.0
37.2 0.2
150 10
1.0
32.3 0.8
141 11
1.0
298
RH
s
(ps)
EV (kJ mol-1
)
s2V98 (ps)
50
5
27
3
25
3
D2 (1020 s-2
)
0.11 0.01
0.40 0.01
0.26 0.02
Values in italics were fixed in the fitting
Table 5 Best-fit parameters obtained for the micellar forms of
[Gd(L1)(H2O)] and [Gd(L3)(H2O)] from the analysis of NMRD data
above the critical micellar concentration, after subtraction of the
monomer relaxivities
Parameters
[Gd(L1)(H2O)]
[Gd(L3)(H2O)]
298
ex
1.0 0.1
30.0 0.2
16 0.2
105 10
16 0.2
12.0 0.8
2.8
k
[106 s-1
]
DHà (kJ mol-1
El (kJ mol-1
)
32.0
Fig. 4 Temperature dependence of the reduced longitudinal (a,
squares) and transverse (a, triangles) 17O relaxation rates and reduced
chemical shifts (b) of Gd(L3) in aqueous solution (B = 11.7 T,
c = 10.28 mM)
)
16 0.2
102 10
16 0.2
8.0 0.5
298
lH
s
(ps)
Eg (kJ mol-1
)
298
gH
s
(ns)
Table 3 Best-fit parameters obtained for [Gd(L3)(H2O)] from the
analysis of 17O NMR data
S2
0.12 0.03
0.09 0.03
EV (kJ mol-1
)
1.0
1.0
Parameters
[Gd(L3)(H2O)]
[Gd(DOTA)(H2O)]- [39]
s2V98 (ps)
20
5
20
5
D2 (1019 s-2
)
0.13 0.01
0.19 0.01
k
(106 s-1
)
298
ex
2.8 0.4
4.1
DHà (kJ mol-1
DSà (J mol-1 K-1
)
32.0 1.0
49.8
48.5
16.1
77
Values in italics were fixed in the fitting
)
14
4
ER (kJ mol-1
)
14.2 0.2
601 10
1.0
298
RO
s
(ps)
Water exchange rate and rotational dynamics
EV (kJ mol-1
)
1.0
s2V98 (ps)
D2 (1020 s-2
A/ꢀh (MHz)
The water exchange rate, which was determined directly
17
8.4
1
11
from the O T2 data for Gd(L3), ke2x98 = 2.8 9 106 s-1, is
)
0.28 0.01
0.16
-3.7
in good agreement with water exchange rates reported for
similar DO3A monoamide derivative Gd3? complexes
[11]. According to the empirical rule that has been
observed for a large number of amide derivative Gd3?
complexes of 1,4,7,10-tetraazacyclododecane-1,4,7,10-tet-
raacetic acid (DOTA) or DTPA, the water exchange rate is
decreased to about one half or one third by the replacement
of each carboxylate by an amide function (Table 3) [11].
An amide group is coordinated less strongly with the lan-
thanide ion than a carboxylate, the bond distances are
longer, and as a consequence, the inner sphere is less
crowded in amide than in carboxylate complexes [11]. In
dissociatively activated water exchange processes, the
steric crowding is of primary importance, i.e., a less tightly
-3.7 0.2
Values in italics were fixed in the fitting
DOTA, 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid
310 K are higher than those at 298 K. This shows that the
slow water exchange rate is a limiting factor for this
complex. Indeed, the proton relaxivities for Gd(L1) could
298
ex
not be fitted with a k
value fixed to that of Gd(L3); a
reasonable fit could be obtained only when smaller values
were considered. We obtained an acceptable fit for
= 1.0 9 106 s-1, a value about one third of that
determined for Gd(L3) by 17O NMR measurements
298
ex
k
(Table 5, Fig. 3c, d).
123

292
J Biol Inorg Chem (2014) 19:281–295
coordinating ligand disfavors the dissociative activation
step, leading to a lower water exchange rate.
The rotational dynamics was assessed directly for both
the monomer species and the aggregated state by NMRD
measurements. The rotational correlation time calculated
from the 17O T1 data is an average value, since it has
contributions from both the monomer and the aggregated
state, which cannot be separated. In addition, one cannot
use the Lipari–Szabo treatment to analyze the 17O NMR
data, since they were acquired at only one magnetic field
strength, which does not allow the separation of local and
global motions. Nevertheless, s calculated from 17O T1
298
RO
data clearly shows slow rotational motion: 601 ps for
Gd(L3) at 10 mM concentration versus 77 ps for
Gd(DOTA)- (Table 3). On the other hand, the NMRD of
the aggregated form could be analyzed in terms of global
and local rotation, and these values characterize directly
the micellar state. The rotational correlation times obtained
from the NMRD curves for the Gd(L1), Gd(L2), and Gd(3)
monomers are in accordance with their larger size with
respect to Gd(DOTA)-. The small variations between the
three systems are also coherent with the increasing size of
298
RH
the side chain and its flexibility. Indeed, the smallest s
value is determined for Gd(L1), which has the shortest
linker between the DO3A unit and the PiB unit. Gd(L3) has
the longest linker; however, it is more flexible with the C5
Fig. 5 1H NMRD profiles at a 298 K and b 310 K of Gd(L3) at
0.2 mM (squares), 5 mM (inverted triangles), 0.2 mM in the
presence of 0.2 mM Ab1–40 (circles), and 0.2 mM in the presence
of 0.2 mM human serum albumin (HSA) (upright triangles) [0.05 M
N-(2-hydroxyethyl)piperazine-N0-ethanesulfonic acid; pH 7.4]. cmc
critical micellar concentration
alkyl chain than the benzene derivative L2, which therefore
298
RH
has the highest s
value. As expected, these rotational
correlation times are also in accordance with the monomer
relaxivities measured at high field.
For the micellar state (Table 5), the global rotational
correlation times are in the range of a few nanoseconds
[12.0 and 8.0 ns for Gd(L1) and Gd(L3), respectively],
whereas the local rotational correlation times are rather
short (approximately 100 ps). The generalized order
parameter, S2, also has very low values, implying a large
flexibility of the system [38].
amyloid peptide. Instead, amyloid binding affinities for
Gd(L1) and Gd(L3) have been investigated in detail by
other methods, including surface plasmon resonance and
saturation transfer difference NMR, which are more sen-
sitive than relaxometry. These results will be reported
elsewhere. The binding affinity of Gd(L1) for Ab1–40
,
determined from surface plasmon resonance measure-
ments, was published previously (Kd = 180 lM) [20].
Similarly to Gd(L1), Gd(L3) interacts with HSA as well,
causing a remarkable increase of relaxivity at intermediate
fields (Fig. 5). In general, albumin binding leads to a pro-
longed lifetime of the agent in the blood pool, which, given
the slower elimination from the body, can be useful for an
MRI probe. However, strong HSA binding can be detri-
mental for the BBB permeability of the agent [32].
Relaxometric assessment of the interaction
of the complexes with HSA and Ab1–40 peptide
When the monomeric form of Gd(Lx) binds to Ab plaques,
higher relaxivity is expected, in particular at intermediate
fields, since the complex becomes immobilized. Indeed, in
the presence of the amyloid peptide Ab1–40 ðcGd ¼ cAb
1ꢂ40
The binding affinity of Gd(L3) for HSA was assessed by
PRE measurements. which are commonly used to deter-
mine affinity constants of Gd3? complexes and HSA. The
PRE method is a nonseparative technique in which the
binding parameters can be obtained by exploiting the dif-
ferences in the NMR water solvent relaxation rates
between the bound and the unbound substrates. Since the
¼ 0:2 mMÞ, the relaxivity of Gd(L1) and Gd(L3) increases
considerably at magnetic fields where the effect of slower
rotation is most pronounced (80 % increase at 40 MHz).
This is interesting for imaging applications, where the
amyloid binding specifically contributes to a higher MRI
efficiency of the agent. The relaxivity increase could be
used to assess the binding affinity of the probe for the
123

J Biol Inorg Chem (2014) 19:281–295
293
Fig. 6 Proton relaxation enhancement data to assess HSA binding of Gd(L3): a M-titration at 0.6 mM HSA; b E-titration at 0.1 mM Gd(L3)
(40 MHz, 310 K, pH 7.4)
relaxation rate is markedly increased in the presence of a
paramagnetic substrate interacting with the protein, this
method is perfectly tailored to investigate the binding of
paramagnetic metal chelates. It consists of measuring the
proton relaxation rates R1obs at increasing concentrations of
the protein and the metal chelate (Fig. 6). To obtain the
affinity constant KA, the data were fitted to Eq. 4:
and a relaxivity of the non-covalently bound complex r1c of
77 5.3 s-1 mM-1 were obtained for Gd(L3) in com-
parison with KA = 910 M-1 and r1c = 55.5 s-1 mM-1 for
Gd(L1) (310 K, 40 MHz) (Table 6) [20]. HSA has multiple
binding sites. Among the numerous Gd3? complexes with
HSA-binding capability that have been investigated in the
past, some were reported to bind to more than one inde-
8 ꢀ
<
ꢁ
ꢀ
ꢁ
9
=
1
2
r1f ꢀ c1
þ
r1c ꢂ r1f ꢃ
ꢂ
ꢃ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Rp1obs ¼ 103 ꢃ
;
ð4Þ
ꢀ
ꢁ
2
:
ncHSA þ c1 þ KAꢂ1
ꢂ
ncHSA þ c1 þ KAꢂ1 ꢂ4ncHSA ꢀ C1
;
where r1f and r1c are the proton relaxivities of the free and
the bound state, cHSA and c1 are the concentration of HSA
and the complex, respectively, and n is the number of
binding sites on the protein.
pendent site with different binding constants [41, 42].
Typically, these different binding constants are derived
from ultrafiltration experiments. Most often, relaxometric
data alone do not allow one to distinguish between dif-
ferent binding models. On the other hand, the stepwise
binding constants calculated from the ultrafiltration data
Under the assumption that there is one binding site in
HSA (n = 1), a binding constant KA of 250 18 M-1
,
Table 6 Parameters obtained from relaxometric titrations of [Gd(L1)(H2O)] and [Gd(L3)(H2O)] complexes with human serum albumin
(40 MHz), compared with literature data for other complexes (310 K)
r1f (s-1 mM-1
)
Complex
KA (103 M-1
)
n
r1c (s-1 mM-1
)
References
Gd(L1)
0.91 0.09
0.25 0.02
1.5
1
1
1
1
1
55 8.6
77 5.3
42.9
5.3 0.2
4.9 0.1
5.2
[20]
Gd(L3)
Gd(BOPTA)2-
This work
[44]
MS-325
Gd(Bz-DTTA)-
6.1
48.9
5.6
[44]
0.71
45.5
7.1
[45]
Gd(BOPTA)2-, (4RS)-[4-carboxy-5,8,11-tris(carboxymethyl)-1-phenyl-2-oxa-5,8,11-triazatridecan-13-oato(5-)] gadolinate(2-); Bz-DTTA, N0-
benzyl-diethylene-triamine-N,N,N00,N00-tetraacetic acid; MS-325, trisodium {(2-(R)-[(4,4-diphenylcyclohexyl) phosphonooxymethyl]-diethyle-
netriaminepentaacetato)(aquo) gadolinium(III)
123

294
J Biol Inorg Chem (2014) 19:281–295
project. This work was carried out in the frame of the European
Actions TD1004 ‘‘Theragnostics Imaging and Therapy’’ and TD1007
‘‘PET-MRI’’.
typically show that one binding site is much stronger than
the others (Ka1/Ka2 between 15 and 30); therefore, a 1:1
binding isotherm is used most commonly for the albumin
binding of related Gd3? complexes [43].
Our complexes show moderate binding as compared
with Gd(BOPTA)2- or MS-325, which were specifically
developed as blood pool agents exhibiting strong HSA
binding [44]. The affinity is similar to that determined for
Gd(Bz-DTTA)-, a linear DTPA-type chelate bearing a
benzyl substituent. We have not found any literature data
concerning the binding of the small molecule PiB to
HSA.
References
1. Fratiglioni L, Winblad B, von Strauss E (2007) Physiol Behav
92:98–104. doi:10.1016/j.physbeh.2007.05.059
2. Hardy J, Selkoe DJ (2002) Science 297:353–356. doi:10.1126/
science.1072994
3. Rauk A (2009) Chem Soc Rev 38:2698. doi:10.1039/b807980n
´
4. Camus V, Payoux P, Barre L et al (2012) Eur J Nucl Med Mol
Imaging 39:621–631. doi:10.1007/s00259-011-2021-8
5. Mathis CA, Bacskai BJ, Kajdasz ST, McLellan ME, Frosch MP,
Hyman BT, Holt DP, Wang Y, Huang G-F, Debnath ML, Klunk
WE (2002) Bioorg Med Chem Lett 12:295–298. doi:10.1016/
S0960-894X(01)00734-X
6. Vandenberghe R, Van Laere K, Ivanoiu A, Salmon E, Bastin C,
Triau E, Hasselbalch S, Law I, Andersen A, Korner A, Minthon
L, Garraux G, Nelissen N, Bormans G, Buckley C, Owenius R,
Thurfjell L, Farrar G, Brooks DJ (2010) Ann Neurol 68:319–329.
doi:10.1002/ana.22068
Conclusions
We have synthesized a family of metal-based molecular
imaging probes for the detection of amyloid plaques
associated with AD. The ligands are obtained by linking
the PiB moiety, an efficient amyloid marker, to DO3A
monoamide, which forms stable and inert complexes
with various metal ions adapted to different imaging
modalities, including Gd3? for MRI, 111In3? for SPECT,
or 68Ga3? for PET. The ligands differ in the chemical
structure and the length of the spacer between the PiB
unit and the DO3A unit, which allows hydrophobicity of
the complexes to be modulated. The Gd(Lx) chelates
have amphiphilic behavior as shown by their octanol–
water partition coefficients and form micelles in solu-
tion. The cmcs, which were obtained by relaxometric
measurements, are in the low millimolar range. The
parameters that influence water proton relaxivity,
including the water exchange rate and the rotational
correlation times, were determined in a combined 17O
NMR and NMRD study. The analysis of the rotational
dynamics for the aggregated micellar state in terms of
local and global motions using the Lipari–Szabo
approach indicated highly flexible, large aggregates. The
Gd(Lx) complexes bind the amyloid peptide Ab1–40 and
HSA, yielding increased relaxivities owing to reduced
rotational motion. HSA binding affinity constants (KA)
of 250 and 910 M-1 were calculated for Gd(L1) and
Gd(L3), respectively, from PRE experiments.
7. Barthel H, Gertz H-J, Dresel S, Peters O, Bartenstein P, Buerger
K, Hiemeyer F, Wittemer-Rump SM, Seibyl J, Reininger C, Sabri
O (2011) Lancet Neurol 10:424–435. doi:10.1016/S1474-
4422(11)70077-1
8. Verhoeff NPLG, Wilson AA, Takeshita S, Trop L, Hussey D,
Singh K, Kung HF, Kung M-P, Houle S (2004) Am J Geriatr
Psychiatry 12:584–595. doi:10.1176/appi.ajgp.12.6.584
9. Nordberg A (2004) Lancet Neurol 3:519–527. doi:10.1016/
S1474-4422(04)00853-1
10. Cheng Y, Ono M, Kimura H, Ueda M, Saji H (2012) J Med Chem
55:2279–2286. doi:10.1021/jm201513c
11. Merbach AE, Helm L, Toth E (2013) The chemistry of contrast
agents in medical magnetic resonance imaging, 2nd edn. Wiley,
New York
12. Jack CR Jr, Garwood M, Wengenack TM, Borowski B, Curran
¨
GL, Lin J, Adriany G, Grohn OHJ, Grimm R, Poduslo JF (2004)
Magn Reson Med 52:1263–1271. doi:10.1002/mrm.20266
13. Higuchi M, Iwata N, Matsuba Y, Sato K, Sasamoto K, Saido TC
(2005) Nat Neurosci 8:527–533. doi:10.1038/nn1422
14. Poduslo JF, Wengenack TM, Curran GL, Wisniewski T, Sig-
urdsson EM, Macura SI, Borowski BJ, Jack CR (2002) Neurobiol
Dis 11:315–329. doi:10.1006/nbdi 2002.0550
15. Wadghiri YZ, Sigurdsson EM, Sadowski M, Elliott JI, Li Y,
Scholtzova H, Tang CY, Aguinaldo G, Pappolla M, Duff K,
Wisniewski T, Turnbull DH (2003) Magn Reson Med
50:293–302. doi:10.1002/mrm.10529
16. Poduslo JF, Curran GL, Peterson JA, McCormick DJ, Fauq AH,
Khan MA, Wengenack TM (2004) Biochemistry (Mosc)
43:6064–6075. doi:10.1021/bi0359574
17. Yang J, Wadghiri YZ, Hoang DM, Tsui W, Sun Y, Chung E, Li
Y, Wang A, de Leon M, Wisniewski T (2011) Neuroimage
55:1600–1609. doi:10.1016/j.neuroimage.2011.01.023
18. Larbanoix L, Burtea C, Laurent S, Van Leuven F, Toubeau G,
Elst LV, Muller RN (2010) Neurobiol Aging 31:1679–1689.
doi:10.1016/j.neurobiolaging.2008.09.021
These Gd3? complexes will likely not pass the BBB in
sufficient amount for MRI detection. Nevertheless, they
might be still useful in animal studies with BBB opening,
in intracerebroventricular injection protocols, or in ex vivo
MRI to stain AD brain tissues.
19. Solomon I (1955) Phys Rev 99:559. doi:10.1103/PhysRev.99.559
Acknowledgments This work was financially supported by Fun-
ˆ
´ˇ
´
´ˇ
20. Martins AF, Morfin J-F, Kubıckova A, Kubıcek V, Buron F,
Suzenet F, Salerno M, Lazar AN, Duyckaerts C, Arlicot N,
dac¸a˜o para a Ciencia e a Tecnologia, Portugal (PhD grant SFRH/BD/
46370/2008 to AFM) and Rede Nacional de RMN (project REDE/
1517/RMN/2005) for the acquisition of the Varian VNMRS 600
NMR spectrometer in Coimbra and the French-Portuguese PESSOA
´
´
Guilloteau D, Geraldes CFGC, Toth E (2013) ACS Med Chem
Lett 5:436–440. doi:10.1021/ml400042w
123

J Biol Inorg Chem (2014) 19:281–295
295
21. Vithanarachchi SM, Allen MJ (2013) Chem Commun
49:4148–4150. doi:10.1039/C2CC36583A
22. Barge A, Cravotto G, Gianolio E, Fedeli F (2006) Contrast Media
Mol Imaging 1:184–188. doi:10.1002/cmmi.110
23. Leo A, Hansch C, Elkins D (1971) Chem Rev 71:525–616.
doi:10.1021/cr60274a001
35. Levin VA (1980) J Med Chem 23:682–684
36. Bornebroek M, Verzijlbergen JF, Haan J, Van Scheyen EJ,
Verhoeff NP, Pauwels EK, Roos RA (1996) Nucl Med Commun
17:929–933
37. Saito Y, Buciak J, Yang J, Pardridge WM (1995) Proc Natl Acad
Sci USA 92:10227–10231
¨
24. Anthony JL, Maginn EJ, Brennecke JF (2001) J Phys Chem B
105:10942–10949. doi:10.1021/jp0112368
25. Raiford DS, Fisk CL, Becker ED (1979) Anal Chem
51:2050–2051. doi:10.1021/ac50048a040
38. Nicolle GM, Toth E, Eisenwiener K-P, Macke HR, Merbach AE
(2002) J Biol Inorg Chem 7:757–769. doi:10.1007/s00775-002-
0353-3
39. Powell DH, Dhubhghaill OMN, Pubanz D, Helm L, Lebedev YS,
26. Hugi AD, Helm L, Merbach AE (1985) Helv Chim Acta
68:508–521. doi:10.1002/hlca.19850680224
Schlaepfer W, Merbach AE (1996)
118:9333–9346. doi:10.1021/ja961743g
J Am Chem Soc
27. Mathis CA, Wang Y, Holt DP, Huang G-F, Debnath ML, Klunk
WE (2003) J Med Chem 46:2740–2754. doi:10.1021/jm030026b
28. Wang Y, Klunk WE, Huang G-F, Debnath ML, Holt DP, Mathis
CA (2002) J Mol Neurosci 19:11–16. doi:10.1007/s12031-002-
0004-8
29. Thyagarajan BS (1958) Chem Rev 58:439–460. doi:10.1021/
cr50021a001
30. Schotten C (1884) Ber Dtsch Chem Ges 17:2544–2547. doi:10.
1002/cber.188401702178
31. Baumann E (1886) Ber Dtsch Chem Ges 19:3218–3222. doi:10.
1002/cber.188601902348
32. Mensch J, Oyarzabal J, Mackie C, Augustijns P (2009) J Pharm
Sci 98:4429–4468. doi:10.1002/jps.21745
33. Lin K-S, Debnath ML, Mathis CA, Klunk WE (2009) Bioorg
Med Chem Lett 19:2258–2262. doi:10.1016/j.bmcl.2009.02.096
34. Earnshaw A, Greenwood N (1997) Chemistry of the elements.
Elsevier, Amsterdam
40. Caravan P, Ellison JJ, McMurry TJ, Lauffer RB (1999) Chem
Rev 99:2293–2352. doi:10.1021/cr980440x
41. Caravan P, Cloutier NJ, Greenfield MT, McDermid SA, Dunham
SU, Bulte JWM, Amedio JC Jr, Looby RJ, Supkowski RM,
Horrocks WD Jr, McMurry TJ, Lauffer RB (2002) J Am Chem
Soc 124:3152–3162
42. Moriggi L, Yaseen MA, Helm L, Caravan P (2012) Chem Eur J
18:3675–3686. doi:10.1002/chem.201103344
43. Giardiello M, Botta M, Lowe MP (2011) J Incl Phenom Mac-
rocycl Chem 71:435–444. doi:10.1007/s10847-011-0009-4
44. Henrotte V, Vander Elst L, Laurent S, Muller RN (2007) JBIC J
Biol Inorg Chem 12:929–937. doi:10.1007/s00775-007-0247-5
45. Silverio S, Torres S, Martins AF, Martins JA, Andre JP, Helm L,
Prata MIM, Santos AC, Geraldes CFGC (2009) Dalton Trans
4656–4670
123