Synthesis and evaluation of a 18F-curcumin derivate for β-amyloid plaque imaging

Article abstract of DOI:10.1016/j.bmc.2014.03.010

Introduction Curcumin is a neuroprotective compound that inhibits the formation of amyloid oligomers and fibrils and binds to β-amyloid plaques in Alzheimer's disease (AD). We aimed to synthesize an ¹⁸F-labeled curcumin derivate ([¹⁸F]4) and to characterize its positron emission tomography (PET) tracer-binding properties to β-amyloid plaques in a transgenic APP23 mouse model of AD. Methods We utilized facile one-pot synthesis of [¹⁸F]4 using nucleophilic ¹⁸F-fluorination and click chemistry. Binding of [¹⁸F]4 to β-amyloid plaques in the transgenic APP23 mouse brain cryosections was studied in vitro using heterologous competitive binding against PIB. [¹⁸F]4 uptake was studied ex vivo in rodents and in vivo using PET/computed tomography of transgenic APP23 and wild-type control mice. Results The radiochemical yield of [¹⁸F]4 was 21 ± 11%, the specific activity exceeded 1 TBq/μmol, and the radiochemical purity exceeded 99.3% at the end of synthesis. In vitro studies of [¹⁸F]4 with the transgenic APP23 mouse revealed high β-amyloid plaque binding. In vivo and ex vivo studies demonstrated that [¹⁸F]4 has fast clearance from the blood, moderate metabolism but low blood-brain barrier (BBB) penetration. Conclusions [ ¹⁸F]4 was synthesized in high yield and excellent quality. In vitro studies, metabolite profile, and fast clearance from the blood indicated a promising tracer for Aβ imaging. However, [¹⁸F]4 has low in vivo BBB penetration and thus further studies are needed to reveal the reason for this and to possibly overcome this issue.

Full text of DOI:10.1016/j.bmc.2014.03.010

Bioorganic & Medicinal Chemistry 22 (2014) 2753–2762
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
Synthesis and evaluation of a 18F-curcumin derivate for b-amyloid
plaque imaging
Johanna Rokka a, Anniina Snellman b, Cristiano Zona c, Barbara La Ferla c, Francesco Nicotra c,
Mario Salmona d, Gianluigi Forloni d, Merja Haaparanta-Solin b, Juha O. Rinne e, Olof Solin a,f,
⇑
a Turku PET Centre, Radiopharmaceutical Chemistry Laboratory, University of Turku, Porthaninkatu 3, FI-20500 Turku, Finland
b Turku PET Centre, Preclinical Imaging, University of Turku, Tykistökatu 6, FI-20520 Turku, Finland
c Department of Biotechnology and Bioscience, University of Milano-Bicocca, I-20126 Milano, Italy
d Department of Biochemistry and Molecular Pharmacology, Istituto di Ricerche Farmacologiche ‘Mario Negri’, I-20156 Milano, Italy
e Turku PET Centre c/o Turku University Hospital, University of Turku, PO Box 52, FI-20521 Turku, Finland
f Turku PET Centre, Accelerator Laboratory, Åbo Akademi University, Porthansgatan 3, FI-20500 Turku, Finland
a r t i c l e i n f o
a b s t r a c t
Article history:
Introduction: Curcumin is a neuroprotective compound that inhibits the formation of amyloid oligomers
and fibrils and binds to b-amyloid plaques in Alzheimer’s disease (AD). We aimed to synthesize an 18F-
labeled curcumin derivate ([18F]4) and to characterize its positron emission tomography (PET) tracer-
binding properties to b-amyloid plaques in a transgenic APP23 mouse model of AD.
Methods: We utilized facile one-pot synthesis of [18F]4 using nucleophilic 18F-fluorination and click
chemistry. Binding of [18F]4 to b-amyloid plaques in the transgenic APP23 mouse brain cryosections
was studied in vitro using heterologous competitive binding against PIB. [18F]4 uptake was studied
ex vivo in rodents and in vivo using PET/computed tomography of transgenic APP23 and wild-type con-
trol mice.
Received 20 December 2013
Revised 4 March 2014
Accepted 10 March 2014
Available online 18 March 2014
Keywords:
Positron emission tomography (PET)
Nucleophilic 18F-fluorination
Curcumin
Alzheimer’s disease
Click chemistry
Results: The radiochemical yield of [18F]4 was 21 11%, the specific activity exceeded 1 TBq/
lmol, and
the radiochemical purity exceeded 99.3% at the end of synthesis. In vitro studies of [18F]4 with the trans-
genic APP23 mouse revealed high b-amyloid plaque binding. In vivo and ex vivo studies demonstrated
that [18F]4 has fast clearance from the blood, moderate metabolism but low blood–brain barrier (BBB)
penetration.
b-Amyloid
Conclusions: [18F]4 was synthesized in high yield and excellent quality. In vitro studies, metabolite pro-
file, and fast clearance from the blood indicated a promising tracer for Ab imaging. However, [18F]4 has
low in vivo BBB penetration and thus further studies are needed to reveal the reason for this and to pos-
sibly overcome this issue.
Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction
in vivo.3,5 Although curcumin is not toxic (even in large doses), it
has poor physiological properties. In vivo clinical studies show fast
Curcumin (also known by its chemical formula, 1,6-heptadiene-
3,5-dione-1,7bis(4-hydroxy-3-methoxyphenyl)-(1E,6E) or 1,1-
diferuloylmethane) is the active ingredient in the Indian spice
turmeric (Curcuma longa) and is known to be responsible for the
medicinal effects of turmeric.1,2 Curcumin is a neuroprotective
compound that inhibits the formation of amyloid oligomers and
fibrils and binds to b-amyloid (Ab) plaques found in Alzheimer’s
disease (AD) with high binding potential.3,4 Curcumin has also
been shown to reduce Ab deposits in the brains of Tg2576 mice
metabolism and poor absorption of curcumin.1,2,4,6 Therefore, there
has been considerable recent interest in developing curcumin ana-
logues as drug candidates and imaging tracers.7–15
Curcumin has a tautomeric structure with two carbonyl groups
that can interconvert to keto–enol forms (Fig. 1). Studies that
employ curcumin derivates demonstrate that the enol form pre-
dominates when curcumin and its derivates are bound to Ab
aggregates.16–18 Structure–activity relationship studies have
shown that the Ab plaque-binding properties of curcumin are
due to its planar structure, which consists of two terminal phenyl
groups with proper substituents and an appropriate linker length
with double bond conjugation and narrow flexibility.18,19 Also
⇑
Corresponding author. Fax: +358 2 231 8191.
E-mail address: olof.solin@utu.fi (O. Solin).
http://dx.doi.org/10.1016/j.bmc.2014.03.010
0968-0896/Ó 2014 Elsevier Ltd. All rights reserved.

2754
J. Rokka et al. / Bioorg. Med. Chem. 22 (2014) 2753–2762
O
OH
O
O
O
O
O
O
H C
CH
3
H C
CH
3
3
3
(b)
(a)
HO
OH
HO
OH
Curcumin
O
O
18F
N
N
N
NH
O
N
N
O
O
CH
3
H C
3
HO
OH
[18F]Curcumin derivate [18F]4
Figure 1. Chemical structures of curcumin, the [18F]curcumin derivate ([18F]4. Curcumin exists in equilibrium between keto (a) and enol (b) forms.
the delocalization of charge through the structure facilitates depro-
tonation of the phenolic group, which improves the Ab binding of
curcumin.18
2.2. Materials for labelling synthesis
Organic solvents were high performance liquid chromatogra-
phy (HPLC) grade from Rathburn Chemicals Ltd (Walkernburn,
Great Britannia). Water was purified using a Milli-Q Plus Ultra Pure
Water system (Millipore, Molsheim, France). Other chemicals were
analytical grade from Merck (Darmstadt, Germany). Thin layer
chromatography (TLC) plates were glass plates pre-coated with
high performanceTLC (HPTLC) Silica gel 60 RP-18 F254s (art. no
113724) and aluminum plates pre-coated with Silica 60 F254s
(art. no 1.05554) from Merck (Darmstadt, Germany).
Because of their Ab plaque-binding properties, curcumin deri-
vates labeled with positron-emitting radionuclides are potential
tracers for positron emission tomography (PET) imaging in
AD.7,8,10,12,14 PET is a noninvasive imaging method that can be used
to visualize Ab plaques in patients with mild cognitive impairment
or AD.20 The 11C-labelled thioflavin derivate [11C]2-(40-(methyl-
amino)phenyl)-6-hydroxybenzothiazole ([11C]PIB) is an exten-
sively studied and used PET tracer because of its highly specific
Ab plaque-binding ability.21,22 Numerous new PET tracers for Ab
imaging in AD are currently under development.23–28 This research
focuses on thioflavin and stilbene derivates.25–28 The potential of
radiolabelled curcumin derivates is often ignored,23,24 although
several studies have shown strong binding of curcumin derivates
to Ab, and even capability to reduce Ab burden.1–5,7,9,11–18
The purpose of this study was to synthesize a 18F-labeled curcu-
min derivate [18F]4 (Fig. 1) and to characterize its tracer binding
properties to Ab plaques. We describe a facile one-pot synthesis
of [18F]4 using nucleophilic 18F-fluorination and click chemistry
(Fig. 2), and show that [18F]4 binds specifically to Ab plaques in
post-mortem brain sections of transgenic APP23 mice. Although
in vitro autoradiography studies, calculated partition coefficient,
metabolite profile, and fast clearance from the blood indicated a
promising tracer for Ab imaging, in vivo studies with [18F]4 showed
little penetration of the blood–brain barrier (BBB).
2.3. Radiosynthesis procedure
2.3.1. 18F-fluoride production
18F-fluoride was produced using the 18O(p,n)18F nuclear reac-
tion. 18O-enriched water (enrichment grade 98%, volume 800
l
l,
Hyox, Rotem Industries Ltd, Israel) was irradiated with 17 MeV
protons of 10 A beam current for 15 min using the MGC-20 cyclo-
l
tron (Efremov Scientific Research Institute for Electrophysical
Apparatuses (NIIEFA), St. Petersburg, Russia). The activity of 18F
was approximately 7 GBq at the start of synthesis.
2.3.2. General procedure for nucleophilic 18F-fluorine labeling
reaction
The aqueous solution of 18F-fluoride from the cyclotron target
chamber was collected into a borosilicate glass reaction vessel
containing 4.7 mg (3.9–5.6 mg) potassium carbonate and
16.8 mg (15.2–18.3 mg) Kryptofix 222. Water was removed using
an azeotropic distillation procedure, during which 1 ml acetoni-
trile was added to the solution. The solution was then heated
to 100 °C for 3 min and evaporated using reduced pressure and
a stream of helium gas. This process was repeated three times.
2. Materials & methods
2.1. Precursors and reference compounds
The nonradioactive compounds 2-[2-(2-azidoethoxy)ethoxy]
ethyl tosylate (1), 2-[2-(2-azidoethoxy)ethoxy]ethylfluoride (2), 2-[3,
5-bis(4-hydroxy-3-methoxystyryl)-1H-pyrazol-1-yl]-N-(prop-2-yn-1-yl)
acetamide (3), and 2-[3,5-bis(4-hydroxy-3-methoxystyryl)-1H-
pyrazol-1-yl]-N-{1-[2-(2-(2-fluoroethoxy)ethoxy)ethyl)-1H-1,2,
3-triazol-4-yl]methyl}acetamide (4) were synthesized at the
Department of Biotechnology and Biosciences, University of
Milano-Bicocca (see Supplementary data).
1.1 mg (0.7–1.3 mg) of precursor
1 was dissolved in 500 ll
DMSO and added to the dry 18F-fluoride–Kryptofix complex.
The reaction mixture was heated at 100 °C for 2 min and then
cooled at room temperature for 3 min. A sample was taken from
the reaction mixture for analysis. Subsequently, 1.8 mg (1.7–
2.0 mg) of compound 3 was dissolved in 250
this solution was added to the reaction vessel. A freshly made
ll of DMSO, and

J. Rokka et al. / Bioorg. Med. Chem. 22 (2014) 2753–2762
2755
18O(p,n)18
F
18
-
18O]H2O (aq)
F (aq)
[
1.) K222, K2CO3
2.) CH3CN, azeotropic distillation
O
(K222/K)+ 18F-
O
O
S
CH
3
O
18F
N
O
N
O
3
3
DMSO, 100 oC
O
1
[18F]2
CH
NH
O
N
N
O
O
CH
3
H C
3
3
HO
OH
1.) DMSO, RT
2.) CuSO4 (aq), Na-ascorbate (aq)
O
O
18F
N
N
N
NH
O
N
N
O
O
CH
3
H C
3
[18F]4
HO
OH
Figure 2. Reactions and reaction conditions of [18F]4 radiosynthesis. Nucleophilic 18F-fluorination using 18F-fluoride–Kryptofix complex was employed to produce [18F]2.
Without intermediate purification, [18F]2 was clicked with 3 to produce [18F]4.
aqueous catalyst mixture containing 5
l
mol of CuSO4ꢀ5 H2O and
2.4. Analytical procedures
6
l
mol of sodium ascorbate was added to the reaction vessel.
The reaction proceeded for 15 min at room temperature. Before
semi-preparative HPLC purification, 1 ml water was added to
the vessel.
After radiofluorination, the radiochemical yield of [18F]2 was
determined with silica 60 F254s thin layer plates developed with
a mixture of petrol ether/ethyl acetate (1:1 v/v). The progression
of the click reaction was followed by analyzing samples from the
reaction with HPTLC plate Silica gel 60 RP-18 F254s developed using
a mixture of 1% acetic acid: methanol (2:8 v/v). The amount of
radioactive compound on the plate was analyzed using photosti-
mulated luminescence (PSL) autoradiography. The developed TLC
plate was placed into an exposure cassette with an imaging plate
(Fuji Imaging Plate BAS-MS2025, Fuji Photo Film Co., Ltd, Tokyo,
Japan). After exposure, the imaging plate was scanned using a Fuji
Analyser BAS 1800 reader (Fujifilm Co., Ltd, Tokyo, Japan) at
The product [18F]4 was separated from the reaction mixture
using semi-preparative HPLC. A Merck Hitachi LaChrom 7000
pump (Merck Hitachi, Darmstadt, Germany) was used together
with a Phenomenex Luna 5u C18(2) 100A (250 ꢁ 10.0 mm) column
(Phenomenex, Allerød, Denmark). UV absorption and radioactivity
were monitored at the outflow of the column using a Merck Hit-
achi LaChrom UV detector and a NaI(Tl) scintillation detector
(2 ꢁ 2 inches, Bicron Corporation, Newbury, OH, USA) connected
in series.
The mobile phase was a gradient of 1% acetic acid containing
200 mg/l ascorbic acid (A) or methanol (B). The gradient profile
was 0–3 min of 100% A, 3.1 min of 60% A: 40% B, 40 min of 10%
A: 90% B, with a flow rate of 5.0 ml/min. The collected fraction of
200 lm resolution. Data were analyzed using the Tina 2.1 program
(Raytest Isotopenmessgeräte GmbH, Straubenhardt, Germany).
The specific radioactivity, radiochemical purity, and chemical
purity of [18F]4 were analyzed using an analytical HPLC method
and a Merck Hitachi LaChrom 7000 system with Hitachi D-7000
HPLC System Manager (HSM) software (version 3.0). The wave-
length of the UV detector was set at 330 nm and the radioactivity
was measured with a similar NaI(TI) scintillation detector as
described above. Analysis was conducted using a Phenomenex
Gemini-NX 3u C18 110A (150 ꢁ 4.60 mm) column and an isocratic
system with an eluent of 1% acetic acid/methanol (51:49 v/v) and a
flow rate of 0.8 ml/min. The concentration of 4 was analyzed using
[
18F]4 was subsequently purified using solid phase extraction
(Sep-PakÒ light tC18 cartridge, Waters). During this procedure,
the HPLC fraction was first diluted with 20 ml buffer A which con-
tained 200 mg/l ascorbic acid in 0.1 M phosphate buffer (pH 7.4).
This solution was passed through a Sep-Pak cartridge. Next, the
cartridge was washed with 20 ml of buffer A. Finally [18F]4 was
eluted from the cartridge using 0.4 ml ethanol, and the fraction
was diluted with 4.0 ml of buffer A.

2756
J. Rokka et al. / Bioorg. Med. Chem. 22 (2014) 2753–2762
linear calibration with an authentic standard of known concentra-
tions. The corresponding peak observed at the radioactivity detec-
tor was collected and the radioactivity of this HPLC fraction was
analyzed. Using this method, the radiochemical purity was also
measured for up to 6 h to ensure the stability of the end product
2.6.2. In vivo studies
In vivo biodistribution of [18F]4 was evaluated in two trans-
genic female APP23 mice29 at the age of 16 months (animal
weights 22.0 g and 23.0 g) and in two female WT mice at the
age of 25 months (animal weights 31.9 g and 25.8 g). In vivo
PET was performed with the Inveon Multimodality PET/com-
puted tomography (CT) device (Siemens Medical Solutions,
Knoxville, TN, USA). Animals were anesthetized using 2.5% iso-
flurane/O2, and CT was conducted for attenuation correction
[
18F]4 dissolved in the formulation solution (ethanol–buffer A,
1:10 v/v).
2.5. ClogP calculations
and anatomical reference. Subsequently, 3.8 1.2 MBq
[
18F]4
The partition coefficients (ClogP) of curcumin, [18F]4, and PIB
were calculated using the Molinspiration property engine soft-
ware, v2011.04 (http://www.molinspiration.com/).
was injected into the tail vein and dynamic 60-min PET scans
were initiated. Data were collected in 3D list mode, divided into
51 time frames (30 ꢁ 10; 15 ꢁ 60; 4 ꢁ 300; 2 ꢁ 600 s), and
reconstructed using the 2D filtered back-projection algorithm.
Dynamic data were analyzed using the Inveon Research Work-
place analysis tool v. 4 (Siemens Medical Solutions). Regions of
interest (ROIs) were drawn to whole brain, frontal cortex,
cerebellum, liver, kidney, intestine, heart, and vena cava for the
estimation of radioactivity in the blood. ROIs were defined to
co-registered CT images used as anatomical reference, and fur-
ther guided by mouse brain atlas for frontal cortex and cerebel-
lum, or radioactivity uptake for the peripheral organs of interest.
Subsequently ROIs were transformed to the PET image to obtain
the time-radioactivity curves and results were presented as per-
centages of the injected dose per gram of tissue (%ID/g). Specific
2.6. Animal studies
Animal studies were performed with Sprague Dawley rat,
C57Bl/6N mouse (bred in the animal facility of the University of
Turku), and transgenic APP2329 mice and corresponding wild-type
(WT) control mice (Novartis Pharma, Switzerland). Animal experi-
ments were approved by the Animal Experiment Board of the Prov-
ince of Southern Finland.
2.6.1. In vitro binding studies
The binding of [18F]4 to Ab plaques was studied using 20-
lm,
post-mortem brain cryosections from a transgenic APP23 mouse
at the age of 18 months and from a WT mouse at the age of 28
months. The mice were sacrificed via cardiac puncture in deep
isoflurane anesthesia. The brains were immediately removed and
frozen by immersion in isopentane chilled with dry ice. Twenty-
micrometer brain sections were cut using a cryomicrotome (Leica
Microsystems Nussloch GmbH, Nussloch, Germany) and thaw-
mounted onto glass slides. Sections on glass slides were stored at
ꢂ20 °C until use.
[
18F]4 binding to Ab plaques in APP23 brain was estimated from
frontal cortex-to-cerebellum ratios.
2.6.3. Ex vivo studies
Distribution of [18F]4 in the brain ex vivo was evaluated in
one male C57Bl/6N mouse (animal weight 28.0 g, injected dose
9.8 MBq) and one male Sprague Dawley rat (animal weight
271 g, injected dose 34.4 MBq). The tracer was allowed to dis-
tribute for 10 min. The animals were sacrificed via cardiac punc-
ture in deep isoflurane anesthesia. Cardiac blood samples were
collected in gel-lithium heparin tubes (Terumo Europe N.V., Leu-
ven, Belgium); the brains were rapidly dissected, weighed, and
The binding of [18F]4 to Ab plaques in the cryosections was
studied using heterologous competitive binding against PIB
(Ki = 4.3 nM for Ab plaques 21). Cryosections were thawed to room
temperature 15 min before use and were not pretreated. Sections
were preincubated in 4% human serum albumin in phosphate buf-
fer solution (HSA-solution) for 10 min, followed by 30 min incuba-
tion in 0.5 MBq/ml of [18F]4 in HSA-solution. For competitive
radioactivity was measured using
a NaI(Tl) well counter
(300 ꢁ 300, Bicron, Newbury, OH, USA). Subsequently, the brains
were frozen in chilled isopentane on dry ice, cut into 20-lm
cryosections using a cryomicrotome (Leica Microsystems Nuss-
loch GmbH), air dried, and exposed to an imaging plate (Fuji
Imaging Plate BAS-TR2025, Fuji Photo Film Co.) for approxi-
mately four hours. The plates were scanned with a Fuji BAS
binding studies, various concentrations (0–400 lM) of PIB (ABX
GmbH, Radeberg, Germany) were added to the incubation solution.
Adjacent sections from the mouse brains were used. After incuba-
tion, sections were washed twice for 5 min in the HSA-solution and
with two additional 5-min washes in phosphate buffer solution. Fi-
nally, sections were briefly rinsed with water and rapidly dried in a
stream of air.
5000 Analyzer using 25-lm resolution.
The amounts of unchanged [18F]4 and its radioactive metabo-
lites were analyzed from rat plasma at 10 min post [18F]4 injec-
tion. Plasma and erythrocytes were separated by centrifugation
(1300g, 10 min), and proteins were precipitated from the sample
The brain sections were then apposed to an imaging plate for
3 h. The plates were scanned with a Fuji FLA-5100 laser scanner
with methanol. After centrifugation, 10 ll of the supernatant
(Fuji Photo Film Co., Tokyo, Japan) using 10-
l
m resolution. The
were applied to a HPTLC Silica gel 60 RP-18W plate (Merck,
art. no 1.14296, Darmstadt, Germany) together with a [18F]4
standard sample from the used synthesis batch. The plate was
developed, exposed, and scanned as previously described for
the progression of the click reaction. The images were analyzed
using AIDA Image Analyzer v.4.19 (Raytest Isotopenmessgeräte
GmbH).
digital autoradiography images were analyzed using AIDA image
analysis software, version 4.06 (Raytest Isotopenmessgeräte
GmbH, Straubenhardt, Germany).
The presence and localization of Ab plaques were confirmed
with Thioflavin S staining of the same brain sections that were
used for the [18F]4 binding studies.
Specific uptake of [18F]4 was calculated as PSL per area and was
determined from cortical section areas with high plaque load, as
observed with Thioflavin S staining, a routine histological staining
for Ab plaques. Non-specific uptake was determined in a similar
manner from caudate/putamen areas where no Thioflavin S stain-
ing was observed. A competitive binding curve was constructed
from the data using a sigmoid fit. From this curve, the half maximal
inhibitory concentration (IC50) was determined.
2.7. Statistics
Mean values were calculated from the individual measure-
ments and expressed at a precision of one standard deviation
(mean SD). Saturation binding analyses were performed with
GraphPad Prism, version 2.01 (GraphPad Software, San Diego, CA,
USA).

J. Rokka et al. / Bioorg. Med. Chem. 22 (2014) 2753–2762
2757
(n = 17) from an APP23 and a WT mouse. PSL autoradiography
images of stained APP23 brain sections revealed [18F]4 binding to
fibrillar Ab deposits located in cortical areas. The presence of
fibrillar Ab plaques was verified by staining the same sections with
Thioflavin S (Fig. 4). In brain sections from the WT mouse, Thiofla-
vin S staining detected no plaques and the autoradiographs
detected only non-specific binding of the tracer (Fig. 4).
3. Results
3.1. Radiosynthesis of [18F]4
The 18F-curcumin derivate [18F]4 was synthesized using a one-
pot synthesis procedure (Fig 2). After radiofluorination, the radio-
chemical yield of [18F]2 was 77 10%, decay corrected to the end
of cyclotron bombardment (n = 9, Fig. 3A and Supplementary data
Fig. 1A). Without intermediate purification, [18F]2 and 3 were
clicked together using in situ prepared Cu(I) as the catalyst for
the reaction (Fig. 3B and Supplementary data Fig. 1B). The end
product [18F]4 was purified using semipreparative HPLC. The
non-optimized radiochemical yield of [18F]4 was 21 11% at the
end of synthesis (EOS) (n = 9, Supplementary data Fig. 1C), and
In a separate experiment specific [18F]4 uptake was displaced
by incubating APP23 mouse brain tissue slices (n = 4) with various
concentrations of PIB. Nonspecific binding of [18F]4 was defined as
binding at the caudate/putamen brain area of the same section,
where no Ab plaques were found. Figure 5 shows Ab binding of
[
18F]4 as a function of increasing PIB concentration. The IC50 value
for this displacement was 10ꢂ4 M.
the specific radioactivity exceeded 1 TBq/lmol at the EOS. The
radiochemical purity of [18F]4 was 99.3% (96.5–100%, SD 1.2%).
The radiochemical purity of the final product remained unchanged
over the course of 6 h after synthesis.
3.3.2. Biodistibution of [18F]4
The in vivo biodistribution of [18F]4 in a representative WT
mouse during a 60-min dynamic PET scan is presented in Figure 6A.
Rapid clearance of 18F-radioactivity from blood and immediate
uptake into the liver was seen in the initial phase of the scan. At
a later phase, high and persistent uptake occurred in the intestine
(Fig. 6A). 18F-radioactivity uptake to the brains of representative
transgenic APP23 and WT mice is presented in Figure 6B. Immedi-
ately after injection (0–1 min), 18F-radioactivity in the blood was
high, and the blood circulation in the brain was visible. However,
as 18F-radioactivity in the blood decreased (1–5 min), very low
18F-radioactivity was detected in the brains of APP23 and WT mice.
Increased retention of [18F]4 was not detected in the APP23 brain,
which harbored abundant Ab deposition in comparison to the WT
brain, which lacked binding sites for [18F]4. No increase in frontal
cortex-to-cerebellum ratios were seen in APP23 brain. The ratios
calculated for the last imaging frame (40–60 min p.i.,) were 1.00
and 0.98 for APP23 TG mice, and 0.82 and 0.96 for WT mice.
Time-radioactivity curves for blood and brain for APP23 and WT
mice are presented in Figure 6B.
3.2. CLogP
The ClogPs for curcumin, [18F]4, and PIB were 2.3, 3.0, and 3.6,
respectively.
3.3. Animal studies
3.3.1. In vitro binding of [18F]4
To evaluate the binding of [18F]4 to fibrillar Ab, in vitro autora-
diography was performed using post mortem brain cryosections
A
3000
[18F]2
3.3.3. Ex vivo studies
Entrance of [18F]4 to the brain was further investigated using
ex vivo digital autoradiography and tissue counting of C57Bl/6N
mouse and Sprague Dawley rat brains at 10 min post [18F]4 injec-
tion. 18F-radioactivity in the whole brain was low in the mouse
(0.04% ID/g) and rat brains (0.03% ID/g); digital autoradiography
indicated that the highest 18F-radioactivity occurred in the brain
ventricles (Fig. 7A, B).
1500
0
-0.2
0
0.2
0.4
Rf
0.6
0.8
1
3.3.4. Metabolism of [18F]4
The parent compound [18F]4 (Rf = 0.6) and two polar radioactive
metabolites, M1 (Rf = 0.8) and M2 (Rf = 0.85), were detected in rat
plasma 10 min after [18F]4 injection (Fig. 8). At 10 min after tracer
injection, 45% of the total 18F-radioactivity in the plasma was un-
changed [18F]4. M1 and M2 comprised 17% and 38% of the total
18F-radioactivity, respectively. These metabolites were not charac-
terized further.
B
6000
3000
0
4. Discussion
[
18F]4
In this study we describe a one-pot synthesis of the 18F-curcu-
min derivate [18F]4. Initial radiofluorination of the labeling precur-
sor [18F]2 was conducted using nucleophilic 18F-fluorination. The
reaction was efficient: the reaction mixture was heated for only
2 min at 100 °C to obtain a high yield of [18F]2. Extending the
reaction time did not improve the yield; on the contrary, it caused
-0.2
0
0.2
0.4
Rf
0.6
0.8
1
Figure 3. TLC analysis of reaction mixture after 18F-radiofluorination (A) using
silica 60 F254s thin layer plates with a mixture of petrol ether/ethyl acetate (1:1, v/
v). TLC analysis of the reaction mixture after the click reaction (B) using HPTLC Silica
gel 60 RP-18 F254s thin layer plates with a mixture of 1% acetic acid: methanol (2:8,
v/v) and autoradiography.
[
18F]2 to degrade.
Click-reaction components were added to the reaction vessel
without intermediate purification of [18F]2. The reaction vessel

2758
J. Rokka et al. / Bioorg. Med. Chem. 22 (2014) 2753–2762
Figure 4. Representative in vitro autoradiographic images of brain sections from an APP23 mouse (A) and a WT mouse (B). The color bar on the right indicates the relative
uptake of radioactivity. Images of the same brain sections after Thioflavin S staining (C, D). Thioflavin S staining appears as bright green.
4000
3000
2000
1000
0
1.E-10
1.E-08
1.E-06
1.E-04
1.E-02
1.E+00
CPIB [M]
Figure 5. Competition binding curve fitted to displacement data. An IC50 of 10ꢂ4 M was determined for [18F]4 versus PIB.
was cooled to room temperature before addition of the other reac-
result, produced [18F]4 had high radiochemical purity that did
not change at least over the course of 6 h after end of synthesis.
The ClogP values of curcumin, [18F]4, and PIB were calculated to
estimate the lipophilicity of [18F]4. These calculations showed that
tion components, as heat can cause decomposition of compounds 3
and [18F]4. For the same reason, the click-reaction was performed
at room temperature. The click-reaction catalyst, Cu(I), was pre-
pared in situ through the reaction of copper sulfate with sodium
ascorbate. A water solution of copper sulfate and sodium ascorbate
was mixed and added to the click-reaction solution. This small
amount of water was well tolerated by the click-reaction.
[
18F]4 is expected to be more lipophilic than curcumin, but less
lipophilic than PIB. Values from the calculations are estimates;
however, they do provide an indication of the relative lipophilici-
ties of these compounds. The measured logP value for PIB is
1.2.21 For 18F-labelled curcumin derivates measured logP values
have varied from 1.8 to 3.7.7,12,14 The ClogP for [18F]4 is in line with
the earlier measured values for 18F-curcumin derivates and in the
range that is usually considered close to optimal for in vivo BBB
penetration for small molecules.23,32,33
As previous studies 1,4,30 have shown, the low stability of curcu-
min and its derivates is a shortcoming for these compounds. [18F]4
has a structure that has been shown to improve the water solubil-
ity and chemical stability of the compound.31 We further stabilized
[
18F]4 during the production process by carefully choosing the
purification conditions, formulation procedure, and formulation
solution. Curcumin and its derivates are usually more stabile in
acidic than in neutral or basic aqueous solutions 30 and acidic con-
ditions were favored in the purification procedure. During the syn-
thesis development we noticed that the key element for stabilizing
this curcumin derivate was the combination of alcohol and ascor-
bic acid. These additives drastically decreased the oxidative
decomposition and radiolysis of [18F]4. Therefore ascorbic acid
was added in each solution used for purification and formulation
of [18F]4, methanol was the organic component of the preparative
HPLC solvent and formulation solution contained ethanol. As a
Curcumin has been shown to bind to aggregated Ab oligomers
and fibrils, i.e. not to monomeric structures.3–5 In vitro binding
studies of [18F]4 with brain cryosections from APP23 and WT mice
were used to evaluate tracer affinity to amyloidal plaques. Tissues
were not treated beforehand because pretreatment might change
nonspecific [18F]4 binding in white matter areas of the brain
sections. The high logP value presumed high [18F]4 uptake in fatty
tissue. Therefore, the low non-specific uptake of [18F]4 in fatty
brain tissue in vitro is surprising. This property is of value for
PET-tracers, as high non-specific uptake easily confounds interpre-
tation of clinical PET images and data.

J. Rokka et al. / Bioorg. Med. Chem. 22 (2014) 2753–2762
2759
A
brain
heart
liver
Intestine
Intestine
kidney
0-1 min p.i.
1-5 min p.i.
5-30 min p.i.
30-60 min p.i.
25
20
15
10
5
140
120
100
80
Heart
Liver
Kidney
Brain
Kidney
Liver
50
60
Intestine
40
20
0
0
0
10
20
30
40
50
60
0
10
20
30
40
60
Time p.i. [min]
Time p.i. [min]
B
0-1 min p.i.
1-5 min p.i.
0-1 min p.i.
1-5 min p.i.
WT
APP23 TG
25
25
20
15
10
5
APP23 TG
WT
Brain
Blood
20
15
10
5
Brain
Blood
0
0
0
10
20
30
40
50
60
0
10
20
30
40
50
60
Time p.i. [min]
Time p.i. [min]
Figure 6. In vivo biodistribution and elimination of [18F]4 in a WT and APP23 mice. (A) In vivo biodistribution of [18F]4 in a representative WT mouse (25 months old, injected
dose 2.6 MBq, anesthetized using 2.5% isoflurane/O2) and time-radioactivity curves for 0–60 min post injection (p.i.) for regions of interest. (B) Uptake of [18F]4 in the brain of
a representative transgenic APP23 mouse (15 months old, injected dose 5.2 MBq, anesthetized using 2.5% isoflurane/O2) and a representative WT mouse (25 months old,
injected dose 4.4 MBq, anesthetized using 2.5% isoflurane/O2), and time-radioactivity curves for blood and whole brain of the same mice.
[
18F]4 binding was high in cortical areas of transgenic mouse
affinity (Ki 4.3 nM,21), and is therefore a proper displacement
substance. Displacement with PIB did not change white matter
binding in either the transgenic or the WT mouse. The combination
of effective PIB displacement in the cortical areas of transgenic
mouse brain and matching histochemical staining further confirms
that [18F]4 binds mainly to Ab plaques.
brain and histochemical staining indicated that the Ab plaque pat-
terns were compatible with [18F]4 binding. These findings suggest
that [18F]4 binds to Ab plaques. In the transgenic mouse brain sec-
tions, cortical binding of [18F]4 was displaced by adding nonradio-
active PIB to the incubation solution. PIB has high Ab binding

2760
J. Rokka et al. / Bioorg. Med. Chem. 22 (2014) 2753–2762
A
C57Bl/6N mouse, 10 min p.i.
Sprague Dawley rat, 10 min p.i.
B
Figure 7. Ex vivo autoradiographic images of coronal ex vivo brain cryosections from a C57Bl/6N mouse (A) and a Sprague Dawley rat (B) 10 min after [18F]4 injection. Images
from both striatal (left) and cerebellar (right) planes are presented.
We measured an IC50 value of PIB for Ab plaques in vitro using
18F]4. We have no data for the actual binding site for curcumin or
facilitate the BBB penetration.17–19,31 A similar pyrazole derivate
of curcumin is the drug candidate CNB-001 that has been shown
to penetrate BBB and to bind Ab.15 The co-planarity created by
the pyrazole moiety creates a unique structure of [18F]4 that differs
from earlier curcumin derivates developed as PET tracers.7,10,12
However, in addition to the curcumin-like structure [18F]4
has the {1-[2-(2-(2-fluoroethoxy)ethoxy)ethyl)-1H-1,2,3-triazol-
4-yl]methyl}acetamide group attached to the pyrazole ring. In
[
PIB on Ab in the APP23 mouse. In Supplementary data, table 1 is
presented various determinations of Kd and Ki for curcumin, curcu-
min derivates and Ab dye derivates. As can be seen, there is a very
wide range for curcumin binding constant (KD and Ki) against Ab,
ranging from 0.2 to 46,000 nM. This wide range can at least partly
be explained by the different inhibitors used in the determinations
or by methodological issues. In competition binding assays studies
with curcumin and its 18F-labelled derivates against the Thiofla-
vine T derivate [125I] IMPY or the Congo Red derivate [125I] IMSB,
the Ki values for curcumin and its derivates were in the low nano-
molar range (see Supplementary data, table 1).7,12,14 We chose to
use PIB, a thioflavine derivate, as the inhibitor in the [18F]4
displacement studies. [11C]PIB is by far the best characterized
15
CBN-001 there is a phenyl group attached to the pyrazole ring
and in the fluorinated curcumin analogs a fluoropegylated tail is
attached to the aromatic structure.7,14 These compounds all show
ready passage through the BBB. In [18F]4 the acetamide group is
in the centre of the molecule and the fluoropegylated tail is
attached to the azide ring; the presence of this rather large non-
aromatic group may cause hindrance in the BBB passage of the
molecule. Also the molecular weight (635 Da) of [18F]4 is in the
range where BBB passage is limited due to the high molecular
weight.32,33 Compounds that have molecular weight more than
400 Da but less than 657 Da have shown to have restricted ability
to cross the BBB.32,33 The overall large size and high molecular
weight of [18F]4 are possible reasons for poor brain uptake.
34
PET-tracer for imaging of Ab. Based on earlier published data
we conclude that [18F]4 and PIB share the same high affinity bind-
ing site, since Congo Red and curcumine share the same high affin-
ity binding site as do Congo Red and Thioflavine T. However, at
least Congo Red and Thioflavine T have other low affinity binding
sites as well as one other high affinity site for Thioflavine.34 The
IC50 of 100
l
M of PIB determined for fibrilar Ab using [18F]4 sug-
Evaluation of [18F]4 ex vivo and in vivo revealed low brain up-
take of the compound; this uptake was lower than that previously
described for other 18F-labeled curcumin derivates.7,12 However,
the healthy animals used for ex vivo autoradiography lack Ab
plaques and thus have no binding sites for [18F]4 in the brain. To
further investigate brain uptake, in vivo studies were conducted
with old transgenic animals with abundant Ab deposition in the
brain. Due to the high-affinity binding of [18F]4 to the Ab plaques
in this model that was observed in vitro, increased retention of
gest that the binding site of [18F]4 is not strictly associated with
the high-affinity binding site of PIB, although matching binding
of [18F]4 and Thioflavin S in the APP23 tissue sections indicated
otherwise.
With regard to the curcumin-like moiety of [18F]4, the structure
matches the properties shown to be important for BBB penetration
and Ab plaque binding. [18F]4 has two phenyl rings at appropriate
distance and an enol-type arrangement due to the pyrazole moiety
that creates coplanarity and extended double-bond conjugation
through the center of the molecule; a structure shown to be impor-
tant for Ab plaque binding of curcumin derivates and should also
[
18F]4 in the APP23 brain was expected if the tracer crosses the
BBB. However, no differences in brain uptake and retention were
detected between APP23 and WT mice.

J. Rokka et al. / Bioorg. Med. Chem. 22 (2014) 2753–2762
2761
derivates in order to increase affinity and enhance transport
through the BBB.
A
[
18F]4
7000
3500
0
5. Conclusions
The [18F]curcumin derivate [18F]4 was successfully synthesized
with good radiochemical yield in a one-pot synthesis using nucle-
ophilic 18F-fluorination and click chemistry. In vitro studies of
[
18F]4 showed specific binding to Ab in post mortem transgenic
mouse brain; Ab is considered a hallmark of AD pathology. How-
ever, [18F]4 has low in vivo BBB penetration and thus future studies
are needed to reveal the reason for this and to possibly overcome
this issue.
-0.2
0
0.2
0.4
Rf
0.6
0.8
1
Acknowledgements
This study received funding from the European Community’s
Seventh Framework Programme (FP7/2007-2013) under Grant
agreement no. 212043 and from the Academy of Finland Grant
agreement 266891. The APP23 mice were used with the kind
permission of Novartis Pharma, Swizerland.
B
600
300
0
[
18F]4
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmc.2014.03.010.
References and notes
1. Goel, A.; Kunnumakkara, A.; Aggarwal, B. Biochem. Pharmacol. 2008, 75, 787.
2. Hatcher, H.; Planalp, R.; Cho, J.; Torti, F.; Torti, S. Cell. Mol. Life Sci. 2008, 65,
1631.
3. Yang, F.; Lim, G.; Begum, A.; Ubenda, O.; Simmons, M.; Ambegaokar, S.; Chen,
P.; Kayed, R.; Glabe, C.; Frautschy, S.; Cole, G. J. Biol. Chem. 2005, 7, 5892.
4. Begum, A.; Jones, M.; Lim, G.; Morihara, T.; Kim, P.; Heath, D.; Rock, C.; Pruitt,
M.; Yang, F.; Hudspeth, B.; Hu, S.; Faull, K.; Teter, B.; Cole, G.; Frautschy, S. J.
Pharmacol. Exp. Ther. 2008, 326, 196.
5. Lim, G.; Chu, T.; Yang, F.; Beech, W.; Frautschy, A.; Cole, G. J. Neurosci. 2001, 21,
8370.
6. Pan, M.-H.; Huang, T.-M.; Lin, J.-K. Drug Metab. Dispos. 1999, 1, 486.
7. Ryu, E.; Choe, Y.; Lee, K.-H.; Choi, Y.; Kim, B.-T. J. Med. Chem. 2006, 49, 6111.
8. Patel, P.; Collins, D.; Nistor, M.; Easwaramoorthy, B.; Head, E.; Mukherjee, J. J.
Nucl. Med. 2007, 48, 22P.
-0.2
0
0.2
0.4
Rf
0.6
0.8
1
Figure 8. TLC analysis of radiometabolites of
[
18F]4 from rat plasma using
autoradiography. Unchanged [18F]4 was used as reference standard. (A) Radiom-
etabolic profile of [18F]4 from Sprague Dawley rat plasma 10 min after [18F]4
injection. (B) Plates were HPTLC Silica gel 60 RP-18W developed with a mixture of
1% acetic acid/methanol (2:8, v/v).
9. Ran, C.; Xu, X.; Raymond, S.; Ferrara, B.; Neal, K.; Bacskai, B.; Medarova, Z.;
Moore, A. J. Am. Chem. Soc. 2009, 131, 15257.
10. Al Jamaz, A.; Al-Otaibi, B.; Aboussekhra, A. Q. J. Nucl. Med. Mol. Imaging 2010,
54, 17.
In previous studies, where low uptake in the brain was seen, it
was suggested to be caused by rapid metabolism of the tracers in
the liver, similar to that of curcumin.6,12 In this study, unchanged
11. Mourtas, S.; Canovi, M.; Zona, C.; Aurilia, D.; Niarakis, A.; La Ferla, B.; Salmona,
M.; Nicotra, F.; Gobbi, M.; Antimisiaris, S. Biomaterials 2011, 32, 1635.
12. Cui, M.; Ono, M.; Kimura, H.; Liu, B.; Saji, H. J. Med. Chem. 2011, 54, 2225.
13. Yangisawa, D.; Amatsubo, T.; Morikawa, S.; Taguchi, H.; Urushitani, M.; Shirai,
N. Neuroscience 2011, 184, 120.
14. Lee, I.; Yang, J.; Lee, J.-H.; Choe, Y.-S. Bioorg. Med. Chem. Lett. 2011, 21, 5765.
15. Liu, Y.; Dargusch, R.; Maher, P.; Schubert, D. J. Neurochem. 2008, 105, 1336.
16. Narlawar, R.; Pichardt, M.; Leuchtenberger, S.; Baumann, K.; Krause, S.; Dyrks,
T.; Weggen, S.; Mandelkow, E.; Schmidt, B. ChemMedChem 2008, 3, 165.
17. Yanagisawa, D.; Shirai, N.; Amatsubo, T.; Taguchi, H.; Hirao, K.; Urushitani, M.;
Morikawa, S.; Inubushi, T.; Kato, M.; Kato, F.; Morino, K.; Kimura, H.; Nakano, I.;
Yoshida, C.; Okada, T.; Sano, M.; Wada, Y.; Wada, K.; Yamamoto, A.; Tooyama, I.
Biomaterials 2010, 21, 4179.
18. Balasubramanian, K. J. Agric. Food Chem. 2006, 54, 3512.
19. Reinke, A.; Gestwicki, J. Chem. Biol. Drug Des. 2007, 70, 6944.
20. Nordberg, A.; Rinne, J.; Kadir, A.; Långström, B. Nat. Rev. Neurol. 2010, 6, 78.
21. Mathis, C.; Wang, Y.; Holt, D.; Huang, G.-F.; Depnath, M.; Klunk, W. J. Med.
Chem. 2003, 46, 2740.
22. Klunk, W.; Engler, H.; Nordberg, A.; Wang, Y.; Blomqvist, B.; Holt, D.;
Bergström, M.; Savitcheva, I.; Huang, G.; Estrada, S.; Ausén, B.; Debnath, M.;
Barletta, J.; Price, J.; Sandell, J.; Lopresti, B.; Wall, A.; Koivisto, P.; Antoni, G.;
Mathis, C.; Långström, B. Ann. Neurol. 2004, 55, 306.
[
18F]4 still comprised 45% of the total 18F-radioactivity in rat
plasma 10 min post injection; rapid metabolism is therefore
not expected to explain the low uptake in the brain. [18F]4
exhibited fast clearance from the blood to the liver, and fast hepa-
tobiliary excretion to the intestine, where 18F-radioactivity accu-
mulated at a later phase. Similar excretion via the hepatobiliary
system was previously observed with curcumin,6 18F-fluoropegy-
7
lated curcumin derivates and 18F-labeled fluoropropyl-substi-
tuted curcumin.12 This indicates that [18F]4 metabolizes using
hexahydrocurcumin-glucuroniside conjugate route,6 although the
metabolites of [18F]4 were not identified. The number of animals
used for this study was low, but as low brain uptake was observed,
further ex vivo experiments in animals were not considered neces-
sary to the current investigation.
Although [18F]4 as such is probably not a workable PET-tracer,
this study shows that further development of analogues of curcu-
min is a promising route for in vivo imaging of Ab load. As non-
toxic compounds, with strong Ab binding as well as inhibition
capacity towards both plaques as well as oligomers, curcumin
analogs may have potential as tracers for early onset AD. Thus
we plan to modify the structure of the curcumin analogue and to
combine functionalized nanoliposomes with 18F-labelled curcumin
23. Eckroat, T.; Mayhoub, A.; Garneau-Tsodikova, S. Beilstein. J. Org. Chem. 2013,
1012, 9.
24. Svedberg, M.; Rahman, O.; Hall, H. Nucl. Med. Biol. 2012, 39, 484.
25. Snellman, A.; Rokka, J.; Lopez-Picon, F.; Eskola, O.; Wilson, I.; Farrar, G.;
Scheinin, M.; Solin, O.; Rinne, J.; Haaparanta-Solin, M. Eur. J. Nucl. Med. Mol.
Imaging 2012, 39, 1784.
26. Cselényi, Z.; Jönhagen, M.; Forsberg, A.; Halldin, C.; Julin, P.; Schou, M.;
Johnström, P.; Varnäs, K.; Svensson, S.; Farde, L. J. Nucl. Med. 2012, 53, 415.

2762
J. Rokka et al. / Bioorg. Med. Chem. 22 (2014) 2753–2762
27. Zhang, W.; Oya, S.; Kung, M.-P.; Hou, C.; Maier, D.; Kung, H. Nucl. Med. Biol.
2005, 32, 799.
30. Wang, Y.-J.; Pan, M.-H.; Cheng, A.-L.; Lin, L.-I.; Ho, Y.-S.; Hsieh, C.-Y.; Lin, J.-K. J.
Pharm. Biomed. Anal. 1867, 1997, 15.
28. Choi, S.; Golding, G.; Zhuang, Z.; Zhang, W.; Lim, N.; Hefti, F.; Benedum, T.;
Kilbourn, M.; Skovronsky, D.; Kung, H. J. Nucl. Med. 1887, 2009, 50.
29. Sturchler-Pierrat, C.; Abramowski, D.; Duke, M.; Wiederhold, K.; Mistl, C.;
Rothacher, S.; Ledermann, B.; Bürki, K.; Frey, P.; Paganetti, P.; Waridel, C.;
Calhoun, M.; Jucker, M.; Probst, A.; Staufenbiel, M.; Sommer, B. Proc. Natl. Acad.
Sci. U.S.A. 1997, 94, 13287.
31. Airoldi, C.; Zona, C.; Sironi, E.; Colombo, L.; Messa, M.; Aurilia, D.; Gregori, M.;
Masserini, M.; Salmona, M.; Nicotra, F.; La Ferla, B. J. Biotechnol. 2011, 156,
317.
32. Levin, V. J. Med. Chem. 1980, 23, 682.
33. Partridge, W. M. J. Cereb. Blood Flow Metab. 1959, 2012, 32.
34. Reinke, A.; Gestwicki, J. Chem. Biol. Drug Des. 2011, 77, 399.