Radiosynthesis of a 18F-labeled 2,3-diarylsubstituted indole via McMurry coupling for functional characterization of cyclooxygenase-2 (COX-2) in vitro and in vivo

Article abstract of DOI:10.1016/j.bmc.2012.04.022

The radiosynthesis of 3-(4-[¹⁸F]fluorophenyl)-2-(4- methylsulfonylphenyl)-1H-indole [¹⁸F]-3 as potential PET radiotracer for functional characterization of cyclooxygenase-2 (COX-2) in vitro and in vivo is described. [¹⁸F]-3 was prepared by McMurry cyclization of a ¹⁸F-labeled intermediate with low valent titanium and zinc via a two-step procedure in a remote controlled synthesizer unit including HPLC purification and solid phase extraction. In this way [¹⁸F]-3 was synthesized in 80 min synthesis time in 10% total decay corrected yield from [¹⁸F]fluoride in radiochemical purity >98% and a specific activity of 74-91 GBq/μmol (EOS). [¹⁸F]-3 was evaluated in vitro using pro-inflammatory stimulated THP-1 and COX-2 expressing tumor cell lines (FaDu, A2058, HT-29), where the radiotracer uptake was shown to be consistent with up regulated COX-2 expression. The stability of [¹⁸F]-3 was determined by incubation in rat whole blood and plasma in vitro and by metabolite analysis of arterial blood samples in vivo, showing with 75% of original compound after 60 min an acceptable high metabolic stability. However, no substantial tumor accumulation of [¹⁸F]-3 could be observed by dynamic small animal PET studies on HT-29 tumor-bearing mice in vivo. This may be due to the only moderate COX-1/COX-2 selectivity of 3 as demonstrated by both cellular and enzymatic cyclooxygenase inhibition assay in vitro. Nevertheless, the new approach first using McMurry cyclization in ¹⁸F-chemistry gives access to ¹⁸F-labeled diarylsubstituted heterocycles that hold promise as radiolabeled COX-2 inhibitors.

Full text of DOI:10.1016/j.bmc.2012.04.022

Bioorganic & Medicinal Chemistry 20 (2012) 3410–3421
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
Radiosynthesis of a 18F-labeled 2,3-diarylsubstituted indole via McMurry
coupling for functional characterization of cyclooxygenase-2 (COX-2)
in vitro and in vivo
Torsten Kniess a, , Markus Laube a, Ralf Bergmann a, Fabian Sehn a, Franziska Graf a, Joerg Steinbach a,
⇑
Frank Wuest b, Jens Pietzsch a
a Institute of Radiopharmacy, Helmholtz-Zentrum Dresden-Rossendorf, Dresden, Germany
b Department of Oncology, University of Alberta, Edmonton, Canada
a r t i c l e i n f o
a b s t r a c t
The radiosynthesis of 3-(4-[18F]fluorophenyl)-2-(4-methylsulfonylphenyl)-1H-indole [18F]-3 as potential
PET radiotracer for functional characterization of cyclooxygenase-2 (COX-2) in vitro and in vivo is
described. [18F]-3 was prepared by McMurry cyclization of a 18F-labeled intermediate with low valent
titanium and zinc via a two-step procedure in a remote controlled synthesizer unit including HPLC puri-
fication and solid phase extraction. In this way [18F]-3 was synthesized in 80 min synthesis time in 10%
total decay corrected yield from [18F]fluoride in radiochemical purity >98% and a specific activity of 74–
Article history:
Received 8 February 2012
Revised 4 April 2012
Accepted 10 April 2012
Available online 17 April 2012
Keywords:
91 GBq/l
mol (EOS). [18F]-3 was evaluated in vitro using pro-inflammatory stimulated THP-1 and COX-2
Cyclooxygenase-2
COX-2 inhibitor
expressing tumor cell lines (FaDu, A2058, HT-29), where the radiotracer uptake was shown to be consis-
tent with up regulated COX-2 expression. The stability of [18F]-3 was determined by incubation in rat
whole blood and plasma in vitro and by metabolite analysis of arterial blood samples in vivo, showing
with 75% of original compound after 60 min an acceptable high metabolic stability. However, no substan-
tial tumor accumulation of [18F]-3 could be observed by dynamic small animal PET studies on HT-29
tumor-bearing mice in vivo. This may be due to the only moderate COX-1/COX-2 selectivity of 3 as dem-
onstrated by both cellular and enzymatic cyclooxygenase inhibition assay in vitro. Nevertheless, the new
approach first using McMurry cyclization in 18F-chemistry gives access to 18F-labeled diarylsubstituted
heterocycles that hold promise as radiolabeled COX-2 inhibitors.
18F-radiolabeling
McMurry cyclization
Positron emission tomography
Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction
inflammogenesis and radioresistance of tumors, and consequently
blocking the COX-2-mediated synthesis of eicosanoids may
Cyclooxygenase isoforms COX-1 and COX-2, also known as
prostaglandin H synthases or prostaglandin endoperoxide syn-
thases, are the key rate-limiting enzymes converting arachidonic
acid into prostaglandins, prostacyclin, and thromboxanes. These
bioactive lipids are involved in maintaining of physiological condi-
tions but also in certain pathophysiological processes. COX-1 is
constitutively expressed in most mammalian tissues and is respon-
sible for thrombogenesis and homeostasis as well. COX-2 is an
inducible enzyme, whose expression is stimulated by a host of
inflammatory cytokines, mitogens, hormones and growth factors;
consistently, COX-2 overexpression is implicated in a number of
disease processes including pain, inflammation, and atherogene-
sis.1 Furthermore, elevated COX-2 level has been reported in a
wide variety of cancers, including colorectal adenocarcinoma,
breast, and lung cancer.2–4 COX-2 may play a key role in both
provide effective chemoprevention or adjuvant radiosensitizing
therapy.5,6 In this regard, COXs are the target of aspirin-like and
other non-steroidal anti-inflammatory drugs (NSAIDS). Recently,
therapeutic approaches targeting COX-related processes forward
the use of selective COX-2 inhibitors (coxibes) such as celecoxib,
etoricoxib, and lumiracoxib. Determination of COX-2 expression
levels and/or activity in patients can currently only be achieved
by invasive and laborious ex vivo analyses. Characterization of
functional COX-2 expression by means of positron emission
tomography (PET) would non-invasively provide information for
optimization of coxibe therapy and therapy control in vivo.
Furthermore, this would allow a deeper understanding of the role
of COX-2 in various disorders and pathophysiological situations.
Several attempts have been made to develop COX-2 inhibitors
radiolabeled with positron emitters over the last decade. A number
of 18F- and 11C-labeled analogs of celecoxib and related compounds
with heterocyclic core have been described7–9 and recently an
18F-labeled COX-2 inhibitor basing on a celecoxib derivative for
⇑
Corresponding author. Tel./fax: +49 351 2602760.
E-mail address: t.kniess@hzdr.de (T. Kniess).
0968-0896/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmc.2012.04.022

T. Kniess et al. / Bioorg. Med. Chem. 20 (2012) 3410–3421
3411
imaging of inflammation and COX-2 expressing human tumor
xenografts was published.10 We have contributed to this field with
the radiosynthesis and preliminary evaluation of derivatives of 18F-
and 11C-labeled 1,2-diarylcyclopentenes as highly affine and selec-
tive COX-2 inhibitors.11,12 Unfortunately, the results are inconsis-
tent due to the relatively high lipophilicity of the tracers causing
some kind of so-called unspecific binding and the narrow time slot
for pharmacologic investigations originating from the short half-
life of the 11C-label.
Recently, a new series of 2,3-diarylsubstituted indoles with high
affinity and selectivity has been published as supposable COX-2
inhibitors.13 The heterocyclic ring system bearing the two vicinal
aryl moieties substituted with a methylsulfonyl group shows the
classical ‘coxibe-structure’, whereas the indole ring constitutes an
important template for drug design.
resonance spectra were recorded on a Unity 400 MHz spectrome-
ter (Varian). 1H NMR chemical shifts were given in ppm and were
referenced with the residual solvent resonances relative to tetra-
methylsilane (TMS). Mass spectra were obtained on a Quattro/LC
mass spectrometer (Micromass) by electrospray ionization. Flash
chromatography was conducted using MERCK silica gel (mesh size
230–400 ASTM). Thin-layer chromatography (TLC) was performed
on Merck silica gel F-254 aluminum plates with visualization un-
der UV (254 nm). No-carrier-added aqueous [18F]fluoride was pro-
duced in a CYCLONE 18/9 cyclotron (IBA) by irradiation of [18O]H2O
via the 18O(p,n)18F nuclear reaction.
Synthesis of 3-(4-[18F]fluorophenyl)-2-(4-methylsulfonylphe-
nyl)-1H-indole [18F]-3 and semi-preparative purification was per-
formed in an automated nucleophilic synthesizer TracerLabFXN
(GE). Semi-preparative HPLC purifications were carried out with
The published IC50 values of the 2,3-diarylsubstituted indoles
2-(4-aminosulfonylphenyl)-3-(4-methoxyphenyl)-1H-indole
1, 3-(4-methoxyphenyl)-2-(4-methylsulfonylphenyl)-1H-indole 2
a Nucleodur-Isis C18 column (250 ꢀ 10 mm, 5
lm, Macherey-Na-
gel) using an isocratic eluent of acetonitrile/water = 70:30 by a
S1122 HPLC-pump (Sykam) with a flow rate of 4 mL/min. The
product was monitored by a K-2001 filter photometer (Knauer)
at 254 nm and by a gamma-detector integrated in the synthesizer
module. Analytical HPLC analysis was carried out with a C18 col-
and 3-(4-fluorophenyl)-2-(4-methylsulfonylphenyl)-1H-indole
3
obtained in COX-2 enzyme-binding assays were in the low nano-
molar level in addition with high COX-1/COX-2 selectivity ratios
(Fig. 1). The lipophilicity is in acceptable range on account of the
heterocyclic moiety; so for compound 3 a logP was calculated to
be about 2.3 0.7 (ACDLabÒ) a value that is expected to diminish
the unspecific binding of the corresponding radiotracer.
umn (Nucleodur-Isis, 250 ꢀ 4 mm, 5
lm, MN) using an isocratic
eluent of acetonitrile/water 0.1%TFA = 70:30 by a gradient pump
L2500 (Merck, Hitachi) with a flow rate of 1 mL/min. The products
were monitored by an UV detector L4500 (Merck, Hitachi) at
254 nm and by gamma-detection with a scintillation detector GABI
(Raytest).
Lately compound 1 has been successfully radiolabeled with car-
bon-11 by reaction of the hydroxyl-precursor with 11C-methyltri-
flate and the radiopharmaceutical behavior has been explicitly
studied.14 The authors found [11C]-1 to be relatively unstable with
54% of intact compound in plasma 30 min post injection, further-
more [11C]-1 was suspected to be a substrate for the P-glycoprotein
efflux pump. With respect on the limitations of 11C-radiochemistry
discussed above, we focused on a labeling of the 2,3-diarylsubsti-
tuted indole system with [18F]fluorine. In this paper we report on
the radiosynthesis of 3-(4-[18F]fluorophenyl)-2-(4-methylsulfonyl-
phenyl)-1H-indole [18F]-3, via McMurry cyclization, a labeling ap-
proach that has not been described so far. Pharmacological
characterization of [18F]-3 was performed in human leukemic
monocyte cell line (THP-1) as well as human solid tumor cell lines
(FaDu, HT-29, and A2058) as models for upregulated COX-2
expression in vitro. Stability experiments were accomplished in
rats and in vivo dynamic small animal PET studies in tumor-bear-
ing mice.
2.2. Synthesis of reference compounds and precursors
2.2.1. 3-(4-Fluorophenyl)-2-(4-methylsulfonylphenyl)-1H-
indole 3
The compound was prepared in a McMurry reaction with 4,
TiCl4 and Zn as previously described in 62% yield.13 mp: 234–
236 °C, (lit.: 224–226 °C), 1H NMR (400 MHz, DMSO-d6): d 3.25
(s, 3H, SO2CH3); 7.08 (t, 3J = 7.2 Hz, 3J = 8.0 Hz, 1H, Hindole); 7.22
(t, 3J = 7.2 Hz, 3J = 8.0 Hz, 1H, Hindole); 7.28 (t, 3J = 8.9 Hz, 2H,
H
F-phenyl); 7.38 (dd, 3J = 8.7 Hz, 4J = 5.7 Hz, 2H, HF-phenyl); 7.45 (d,
3J = 8.0 Hz, 1H, Hindole); 7.49 (d, 3J = 8.0 Hz, 1H, Hindole); 7.66 (d,
3J = 8.5 Hz, 2H, Hphenyl); 7.91 (d, 3J = 8.5 Hz, 2H, Hphenyl); 11.79 (s,
1H, NH) ppm; 19F NMR (376 MHz, DMSO-d6): ꢁ116.32 ppm;
ESI-MS (ES+): m/z = 388 (M+Na).
2.2.2. 2-N(4-Methylsulfonylbenzoyl)-amino-40-fluoro-benzo-
phenone 4
2. Experimental
To a solution of 600 mg (2.79 mmol) 2-amino-40-fluoro-benzo-
phenone13 and 446
lL triethylamine in 5.6 mL of dry THF was
2.1. Materials and methods
added a suspension of 609 mg (2.79 mmol) 4-(methylsulfonyl)ben-
zoyl-chloride15 suspended in 2.8 mL of THF with stirring under a
nitrogen atmosphere. After 2 h at room temperature the mixture
All commercial reagents and solvents were used without
further purification unless otherwise specified. Nuclear magnetic
O
O
F
SO2NH2
SO2Me
SO2Me
NH
NH
NH
2
3
1
IC50 (COX-2): 20 nmol
COX-1/COX-2: 500
IC50 (COX-2): 6 nmol
COX-1/COX-2: 1666
IC50 (COX-2): 20 nmol
COX-1/COX-2: 500
Figure 1. 2,3-Diarylsubstituted indoles as COX-2 inhibitors.13

3412
T. Kniess et al. / Bioorg. Med. Chem. 20 (2012) 3410–3421
was filtrated and evaporated to dryness. The residue was sus-
pended in 15 mL of ethanol, the precipitate filtered off, washed
with 5 mL of hot ethanol and 12 mL of water and purified by col-
umn chromatography (ethyl acetate/petroleum ether = 50:50) to
give compound 4 (663 mg, 60%) as a colorless solid, mp 207–
209 °C. 1H NMR (400 MHz, DMSO-d6): d 3.26 (s, 3H, SO2CH3);
7.30 (t, 3J = 8.8 Hz, 2H, HF-phenyl); 7.36–7.39 (m, 1H, Har); 7.49 (d,
3J = 8.3 Hz, 1H, Har); 7.66–7.68 (m, 2H, 2Har); 7.76 (dd, 3J = 8.8 Hz,
4J = 5.6 Hz, 2H, HF-phenyl); 7.88 (d, 3J = 8.5 Hz, 2H, Hphenyl); 8.01 (d,
3J = 8.5 Hz, 2H, Hphenyl); 10.80 (br s, 1H, NH) ppm; ESI-MS (ES+):
m/z = 420 (M+Na).
d 3.08 (s, 3H, SO2CH3); 3.09 (s, 6H, N(CH3)2); 6.68 (d, 3J = 9.1 Hz, 2H,
phenyl); 7.18 (dd, 3J = 7.6 Hz, 1H, Har); 7.60 (dd, 3J = 7.8 Hz, 1H,
H
Har); 7.67 (d, 3J = 7.8 Hz, 4J = 1.4 Hz, 1H, Har); 7.74 (d, 3J = 9.1 Hz,
2H, Hphenyl); 8.07 (d, 3J = 8.5 Hz, 2H, Hphenyl); 8.20 (d, 3J = 8.5 Hz,
2H, Hphenyl); 8.73 (d, 3J = 8.4 Hz; 1H, Har); 11.78 (br s, 1H, NH);
13C NMR (101 MHz, CDCl3): d 40.2; 44.6; 110.7; 121.6; 122.9;
125.2; 125.3; 128.1; 128.5; 133.0; 133.2; 139.5; 139.9; 143.3;
153.7; 163.9; 197.5 ppm; ESI-MS (ES+): m/z = 445 (M+Na).
2.2.6. 2-N(4-Methylsulfonylbenzoyl)-amino-40-(N,N,N-trimethyl-
ammonium)-benzophenone trifluoromethansulfonate 9
In 10 mL of nitromethane was dissolved 250 mg (0.59 mmol) of
2.2.3. 2-(4-Methylphenylsulfonamido)-40-(N,N-dimethylamino)-
-benzophenone 6
8 under stirring and 88 lL (0.8 mmol) of methyl-trifluormethane-
sulfonate was added. After 12 h at room temperature the nitro-
methane was evaporated in vacuum and the residue dissolved at
1 mL of dichloromethane. The solution was refrigerated overnight;
the precipitate was filtered of, washed with water free methanol
and dried under vacuum to give compound 9 (159 mg, 46%) as a
light yellow solid, mp 162–165 °C., 1H NMR (400 MHz, DMSO-
d6): d 3.26 (s, 3H, SO2CH3); 3.59 (s, 9H, N(CH3)3); 7.31 (dt,
3J = 7.6 Hz, 4J = 1.4 Hz, 1H, Har); 7.47 (d, 3J = 7.8 Hz, 4J = 1,4 Hz, 1H,
Har); 7.65 (dd, 3J = 8.0 Hz, 4J = 1.0 Hz, 1H, Har); 7.89 (dt,
3J = 7.6 Hz, 4J = 1.4 Hz, 1H, Har); 7.92–7.96 (m, 4H, Hphenyl); 8.03
(d, 3J = 8.6 Hz, 2H, Hphenyl); 8.06 (d, 3J = 9.1 Hz, 2H, Hphenyl); 10.90
(br s, 1H, NH); 19F NMR (376 MHz, DMSO-d6): ꢁ78.18 ppm; 13C
NMR (101 MHz, DMSO-d6): d 43.2; 56.4; 120.6; 124.6; 125.2;
127.1; 128.3; 129.9; 131.0; 131.0; 132.5; 136.0; 138.3;138.4;
143.4; 149.6; 164.3; 193.4 ppm; ESI-MS (ES+): m/z = 437 (M+),
(M = 437,53 calcd for C24H25N2O4S).
In a heat dried flask was added 308 mg (2.3 mmol) of water free
AlCl3 and 600 mg (1.93 mmol) of 2-(4-methylphenylsulfonamido)-
benzoyl chloride16 5 in portions under stirring to 10 mL of N,
N-dimethylaniline at 0–5 °C. The mixture was kept at this temper-
ature for 5 h and let come to room temperature overnight. After
quenching with 100 mL of ice water, the solution was acidified
with 1 M HCl to pH 2–3 and extracted with 3 portions of 50 mL
ethyl acetate. The organic layer was washed with water until pH
5, dried with Na2SO4 and evaporated. The excess of N,N-dimethyl-
aniline was removed in high vacuum. The remaining solid was dis-
solved in ethyl acetate and by addition of petroleum ether a yellow
product was precipitated. The precipitate was filtered off, dried
and purified by column chromatography (ethyl acetate/petroleum
ether = 40:60) to give compound 6 (297 mg, 38%) as a yellow solid,
mp 171–175 °C. 1H NMR (400 MHz, CDCl3): d 2.17 (s, 3H, CH3);
3.09 (s, 6H, N(CH3)2; 6.56 (d, 3J = 9.0 Hz, 2H, Hphenyl); 6.94 (d,
3J = 8.4 Hz, 2H, Hphenyl); 7.10 (dd, 3J = 7.6 Hz, 1H, Har); 7.34–7.38
(m, 3H, 2Hphenyl, Har); 7.46 (d, 3J = 7.0 Hz, 1H, Har); 7.49 (d,
3J = 8.3 Hz, 2H, Hphenyl); 7.75 (d, 3J = 8.3 Hz, 1H, Har); 9.53 (br s,
1H, NH) ppm; 13C NMR (101 MHz, CDCl3): d 21.5; 40.2; 110.3;
123.7; 124.2; 124.5; 127.3; 129.0; 129.6; 131.8; 132.3; 133.0;
135.9; 137.9; 143.4; 153.7; 195.3 ppm; ESI-MS (ES+): m/z = 417
(M+Na).
2.3. Radiosynthesis of 3-(4-[18F]fluorophenyl)-2-(4-methylsulfo-
nylphenyl)-1H-indole [18F]-3 via McMurry cyclization
Synthesis of [18F]-3 was performed using an automated synthe-
sizer module in a two-step procedure; firstly by reacting [18F]fluo-
ride with the trimethyammonium precursor
9 forming the
intermediate [18F]-4 and secondly by McMurry cyclization with
titanium tetrachloride and zinc. Briefly 15 mg of 9 dissolved at
2.5 mL of acetonitrile were heated with about 8 GBq of the dried
2.2.4. 2-Amino-40-(N,N-dimethylamino)-benzophenone 7
A mixture of 1.3 g (3.3 mmol) 6 in 7.0 mL perchloric acid and
3.0 mL acidic acid was heated at a temperature of 100 °C with stir-
ring for 3 h. The solution was hydrolyzed by careful addition to
150 mL of ice and water. After filtration the acidic mixture was alk-
alized to neutral pH by addition of 25% aqueous NH3-solution, the
yellow precipitate was collected and dried under vacuum. Purifica-
tion occurred by dissolution at ethyl acetate, filtration via a small
silica gel column and evaporation of the solvent to give compound
7 (400 mg, 59%) as a yellow solid, mp 118–123 °C. 1H NMR
(400 MHz, CDCl3): d 3.06 (s, 6H, N(CH3)2); 5.63 (br s, 2H, NH2);
6.64–6.75 (m, 4H, 2Hphenyl, 2Har); 7.22–7.30 (m, 1H, Har); 7.48 (d,
3J = 6.4 Hz, 1H, Har); 7.71 (d, 3J = 8.6 Hz, 2H, Hphenyl) ppm; 13C
NMR (101 MHz, CDCl3): d 40.2; 110.6; 115.7; 116.9; 120.5;
126.7; 132.3; 132.9; 133.5; 149.7; 152.9; 197.2 ppm; ESI-MS
(ES+): m/z = 263 (M+Na).
[
18F]KF/K222 complex at 110 °C for 15 min. After cooling to 60 °C
the acetonitrile was removed in a stream of nitrogen and a freshly
prepared suspension of 20 mg zinc in 2.5 mL dry THF containing
37 lL of TiCl4 was added. The temperature was maintained at
90 °C for 15 min to perform the McMurry cyclization and the THF
was once more removed under a nitrogen stream at 70 °C. The res-
idue was diluted with 3 mL of eluent, passed through a cartridge
filled with sea sand to remove insoluble material and conducted
to semi-preparative HPLC using an isocratic eluent (acetonitrile/
water = 70:30). The radiotracer [18F]-3 was eluted at 8–9 min at a
flow rate of 4 mL/min. The product was separated from the eluent
by means of solid phase extraction (Lichrolut RP18, 500 mg,
Merck), eluted from the cartridge with 0.7 mL of ethanol and
reconstituted for biological investigation with 6.5 mL of 0.9% so-
dium chloride solution. In this way [18F]-3 was synthesized in
80 min synthesis time in 10% total decay corrected yield from
2.2.5. 2-N(4-Methylsulfonylbenzoyl)-amino-40-(N,N-dimethyl-
amino)-benzophenone 8
[
18F]fluoride in radiochemical purity >98% and a specific activity
of 74–91 GBq/ mol at the end of synthesis (EOS).
l
To 350 mg (1.46 mmol) of 7 dissolved at 1.7 mL of dry THF
under a nitrogen atmosphere was added 220
l
L (1.6 mmol) of tri-
2.4. Radiopharmacological characterization
ethylamine and 320 mg (1.46 mmol) of 4-(methylsulfonyl)ben-
zoyl-chloride15 suspended in 1.3 mL of THF. After stirring for 2 h
at room temperature, the suspension was filtrated and the residue
washed with 3 portions of THF. The collected filtrates were evapo-
rated and the residue purified by column chromatography (ethyl
acetate/petroleum ether = 50:50) to give compound 8 (557 mg,
90%) as a yellow solid, mp 188–192 °C. 1H NMR (400 MHz, CDCl3):
2.4.1. Cell uptake studies in vitro
Uptake experiments for evaluation of compound [18F]-3 in vitro
were performed in various cell models following the protocol
published elsewhere with some modifications.12 Therefore, THP-1
(human monocyte/macrophage line; DSMZ ACC 16), FaDu (human
squamous cell carcinoma line; ATCC HTB-43), HT-29 (human

T. Kniess et al. / Bioorg. Med. Chem. 20 (2012) 3410–3421
3413
colorectal adenocarcinoma line; ATCC HTB-38), A2058 (human
malignant melanoma line; ATCC CRL-11147) and A375 (human
malignant melanoma line; ATCC CRL-1619) were used. Cells were
routinely cultivated in RPMI 1640 medium (THP-1, FaDu), McCoy’s
5A medium (HT-29), or Dulbecco’s modified Eagle’s medium
(A2058, A375) supplemented with 10% (v/v) heat-inactivated fetal
Therefore, cells were washed with ice-cold PBS and lysed in lysis
buffer (50 mM Tris–HCl pH 7.5, 150 mM NaCl, 2% Nonidet P40,
0,1% SDS with 1 mM sodium orthovanadate, 1 mM NaF, 10 mM
b-glycerophosphate, 1 mM dithiothreitol (DTT), 1 mM phenyl–
methyl sulfonyl fluoride, and 1
trifuged at 15,000 ꢀ g for 15 min at 4 °C. Then 50
l
g/mL leupeptine). Cells were cen-
g of protein
l
calf serum (FCS), penicillin (100 U/mL), streptomycin (100
lg/mL),
were mixed with sample buffer, denatured at 99 °C for 15 min
and separated on 7.5% SDS–polyacrylamide gels. Proteins were
transferred to nitrocellulose membranes by electroblotting and
blocked for 1 h at room temperature in TBS/Tween (50 mM Tris–
HCl pH 8.0, 150 mM NaCl, 0.05% Tween-20) containing 5%
skimmed milk powder. Membranes were incubated with the pri-
mary antibodies for COX-1 (C-20, sc-1752) or COX-2 (M-19, sc-
1747) (both 1:500, Santa Cruz Biotechnology) at 4 °C overnight,
followed by incubation with secondary peroxidase-conjugated
anti-goat IgG antibodies (1:10,000, A5420, Sigma–Aldrich) for 1 h
at room temperature. After washing with TBS/Tween proteins were
visualized by chemiluminescence using the SuperSignal West Pico
Chemiluminescent Substrate (Pierce, Rockford, USA) and the MF
ChemiBIS bioimager (Biostep). For detection of b-actin primary
polyclonal antibody b-actin IgG (1:1000, A5060, Sigma–Aldrich)
and secondary peroxidase-conjugated antibody anti-rabbit IgG
(1:10,000, A0545, Sigma–Aldrich) were used.
and glutamine (4 mM) at 37 °C and 5% CO2 in a humidified incuba-
tor. Cells were passaged twice a week by mild enzymatic dissocia-
tion using 0.25% trypsin/EDTA. Pro-inflammatory stimulation
(differentiation to macrophages) of THP-1 cells was performed by
adding 64 nM phorbol myristate acetate (TPA) for 72 h. Radiotracer
uptake studies were performed in (a) monolayer cultures (FaDu,
HT-29, A2058, A375, TPA-stimulated THP-1), (b) in suspension
cells (unstimulated THP-1), and (c) in multicellular spheroids
(HT-29). The pre-analytical conditions were as follows: (a) cells
were seeded in 24-well plates at a density of 5 ꢀ 104 cells/mL
and grown to confluence, (b) cells were centrifuged and resus-
pended in fresh RPMI medium containing 10% FCS in 24-well
plates at a density of 5 ꢀ 105 cells/mL, and (c) 1 ꢀ 104 cells per well
were seeded in non-adherent round-bottom 96-well plates in the
presence of 0.24% (w/v) methylcellulose and were allowed to form
a single spheroid per well for 3 d without medium exchange and,
after 80% of the medium was renewed, for additional 4 d (total vol-
ume per well was 100
l
L). Spheroid diameters were measured
2.4.3. Inhibition of COX-2 activity
using an inverse microscope equipped with a guidable desk and
digital camera shortly before the addition of radiotracers. In this
The influence of the corresponding nonradioactive reference
compound 3 on prostaglandin E2 (PGE2) synthesis and thus inter-
action with cyclooxygenases was characterized in supernatants
of HT-29, FaDu, and A375 cells by liquid chromatography–tandem
mass spectrometry (LC–MS/MS) according to Cao et al.18 with
some modifications. Therefore, cells were seeded in 12-well plates
(2.5 ꢀ 106 cells/well) and grown for 24 h. The medium was chan-
ged and the three tumor cell lines were incubated with fresh med-
study only spheroids with a diameter of 300 10
600 10 m (7 d) were used. The cell tracer uptake experiments
using compound [18F]-3 (0.4 MBq/mL; specific activity at applica-
tion time, 74 GBq/ mol) were performed in quadruplicate in med-
lm (3 d) and
l
l
ium at 37 °C for 30, 60, and 120 min. For blocking experiments, the
cells were preincubated for 30 min with the corresponding nonra-
dioactive reference compound 3 (100
of the P-glycoprotein efflux pump pretreatment of cells with cyclo-
sporine A (100 M) and verapamil (100 M), respectively was per-
l
M). To investigate the effect
ium containing either 1 or 10 lM of compound 3 for 25 h. Then to
aliquots of cell culture supernatants (1 mL) were added the deuter-
ated internal standard d4-PGE2 (0.7 nM; Cayman Chemical, Ann Ar-
bor, MI, USA) and, as preservatives, citric acid (72 mM) and
butylated hydroxytoluene (BHT, 3.2 mM). The samples were acidi-
fied with glacial acetic acid to pH 4.0, shock frozen with liquid
nitrogen, and stored at ꢁ65 °C. For analysis, samples were thawed
and immediately extracted three times with 1 mL of diethyl ether
under a nitrogen atmosphere to prevent oxidation. The organic lay-
ers were combined and the solvent was evaporated under a gentle
stream of nitrogen at room temperature. The resulting residue was
l
l
formed for 60 min. In order to characterize normoxia in the HT-29
monolayers and intrinsic hypoxia in HT-29 spheroids also uptake
experiments with 0.4 MBq 18F-labeled fluoromisonidazole
([18F]FMISO; specific activity at application time, 115 GBq/
lmol;
incubation time 4 h) were performed. Therefore, [18F]FMISO was
prepared via one pot, two-step synthesis procedure by use of a
modified TracerlabFXN synthesis module (GE).17 After the tracer up-
take was stopped with 1 mL ice-cold PBS, the monolayer and sus-
pension cells were washed three times with PBS and dissolved in
0.5 mL NaOH (0.1 M containing 1% (w/v) sodium dodecylsulfate).
The radioactivity in cell extracts was then measured with a Cobra
II gamma counter (Canberra-Packard, Meriden, CT, USA). Spheroids
were harvested and washed five times with PBS on filters using a
commercial cell harvester (Filtermate 196, Canberra-Packard, Mer-
iden, CT, USA). The filters were then transferred into tubes contain-
reconstituted in 100 lL of mobile phase (composition at starting
point) for the chromatographic separation. Analyses were carried
out on an Agilent 1100 HPLC system (Agilent Technologies, Santa
Clara, CA, US) coupled to a Micromass Quattro LC mass spectrom-
eter (Micromass, Manchester, UK). Therefore, sample aliquots of
15
l
L were injected onto a HypercarbÒ column (100 ꢀ 2.1 mm,
5 lm; Thermo Fisher Scientific, Waltham, MA, US) and chromato-
ing 350
l
L distilled water and radiotracer uptake was measured
graphed at 35 °C using an eluent comprising of mobile phase A
with a Cobra II gamma counter. Total protein concentration in cell
extracts was determined by the bicinchoninic acid assay (BCA;
Pierce, Rockford, Ill, USA) using bovine serum albumin as protein
standard. For determination of mean protein concentration in
spheroids additional 96-well plates were prepared in parallel. Sub-
sequently, 8 spheroids of each target diameter were pooled in
quintuplicate, disintegrated in 0.5 mL NaOH (0.1 M containing 1%
(w/v) sodium dodecylsulfate and measured using BCA method. Up-
take data for all experiments are expressed as percent of injected
(10 mM NH4Ac, pH 10.5) and mobile phase B (methanol/acetoni-
trile 60/40, v/v) at a flow rate of 150 lL/min. The gradient was as
follows: %A; held 95% for 10 min, decreased to 5% in 15 min, held
5% for 20 min, decreased to 1% in 0.5 min, held 1% for 5 min, in-
creased to 95% in 5 min. Electrospray mass spectrometry was per-
formed in the negative ion mode due to the acidic character of the
investigated eicosanoids using selected reaction monitoring (SRM).
The electrospray capillary voltage and the source temperature
were set at 4 kV and 100 °C. Desolvation gas (nitrogen) tempera-
ture was 150 °C. The specific mass transitions (quantitation transi-
tion; Cone ꢁ18 V, Ecoll 12 V) used for PGE2 determination ([MꢁH]ꢁ;
PGE2 m/z 351; PGE2-d4, m/z 355) were PGE2 m/z 351 ? m/z 333
and PGE2-d4 m/z 355 ? m/z 337 corresponding to the [MꢁH–
H2O]ꢁ ions. A dwell time of 600 ms was applied to monitor the
dose per lg protein (%ID/lg protein).
2.4.2. Western blot analysis
In order to characterize protein expression of COX-1 and COX-2
in the models used, cells were analyzed by Western blotting.

3414
T. Kniess et al. / Bioorg. Med. Chem. 20 (2012) 3410–3421
two SRM transitions. Calibration curves were obtained by plotting
the peak area ratio for PGE2—d4-PGE2 ion pair versus the mass ratio
of the analytes and the internal standard, respectively. Calibration
curves were analyzed by unweighted least-squares linear regres-
sion analysis and were found to be linear over the range studied
(0.03–1000 nM; R2 >0.996). The limit of determination of this ap-
proach was ascertained at 15 pM (220 fmol/injection).
Moreover, the ability of the corresponding nonradioactive refer-
ence compound 3 to inhibit ovine COX-1 and recombinant human
COX-2 was determined using an enzymatic fluorescence-based
cyclooxygenase inhibitor assay (catalog number 700100, Cayman
Chemical, Ann Arbor, MI, USA) according to the manufacturer’s
protocol. The potent and selective COX-2 inhibitor celecoxib was
used as reference compound. The COX inhibitors were assayed in
concentrations ranging from 10ꢁ10 M to 10ꢁ3 M. PRISM5 software
was used for the calculation of the IC50 values.
Experimentelles Zentrum der Medizinischen Fakultät Carl Gustav
Carus, Dresden, Germany) were subcutaneously xenotransplanted
into the right hind leg with human colorectal adenocarcinoma
HT-29 according to a protocol published elsewhere.19 When tumor
size reached a diameter of about of 7–9 mm, imaging studies were
performed. General anesthesia of HT-29 tumor-bearing mice (n = 8,
body weight 40 3 g) was induced and maintained with inhalation
of 10% desflurane in 30% oxygen/air. Mice were positioned and
immobilized prone with their medial axis parallel to the axial axis
of the scanner (microPETÒ P4, Siemens preclinical solutions, Knox-
ville, TN, USA). In the PET experiments, 20 MBq of [18F]-3 (in 0.5 mL
of E153/10% ethanol) was administered intravenously as an infu-
sion (1 mL/min) with a syringe pump (Harvard Apparatus, Hollis-
ton, Ma, USA) using
a needle catheter into a tail vein. A
transmission scan was carried out prior to the injection of [18F]-3
using a rotating 57Co point source. PET acquisition was started with
a delay of 30 s after beginning of the radiotracer infusion. Emission
data were collected continuously for 120 min after injection of
2.4.4. Stability studies in vitro and in vivo
2.4.4.1. In vitro.
The stability of radiotracer [18F]-3 in vitro
[
18F]-3. The 3D list mode data were sorted into sinograms with
after incubation with both rat whole blood and plasma was evalu-
ated by radio-HPLC. Therefore, 0.5 MBq of radiotracer were added
to 400 lL of blood sample taken from male Wistar-Unilever rats
38 frames (15 ꢀ 10 s, 5 ꢀ 30 s, 5 ꢀ 60 s, 4 ꢀ 300s, 9 ꢀ 600s). The
data were decay, scatter and attenuation corrected. The data were
normalized to the injected radioactivity by using 18F-standards
(strain HsdCpb:WU, Harlan Winkelmann, Borchen, Germany) and
incubated in a thermo-mixer at 37 °C at 600 rpm for 5, 30 and
60 min. Plasma was separated by centrifugation (3 min,
13.000 ꢀ g) followed by precipitation of plasma proteins with ace-
tonitrile/water/trifluoroacetic acid (50:45:5) in a 1:2 ratio. The
clear supernatant separated by second centrifugation (3 min,
13.000 ꢀ g) was used for analysis. For evaluation of in vitro stabil-
ity in rat plasma the blood cells have been removed by centrifuga-
tion before addition of the radiotracer.
from the injection solution measured in a c-well counter (Isomed
2000, Germany) cross calibrated to the PET scanner and expressed
in percent of injected dose per cubic centimeter (%ID/cm3). The
frames were reconstructed by Ordered Subset Expectation Maxi-
mization applied to 3D sinograms (OSEM3D) with 14 subsets, 15
OSEM3D iterations, 25 maximum a posteriori (MAP) iterations,
and 1.8 mm resolution using the FastMAP algorithm (Siemens Pre-
clinical Solutions, Knoxville, TN). The pixel size was 0.07 by 0.07 by
0.12 cm, and the spatial resolution in the center of field of view
was 1.8 mm. No correction for partial volume effects was applied.
The image volume data were converted to Siemens ECAT7 format
further processed using the ROVER software (ABX GmbH, Rade-
berg, Germany). Masks for defining three-dimensional regions of
interest (ROI) were set with ROI’s defined by thresholding and
ROI time activity curves (TAC) were derived for the subsequent
data analysis. The standardized uptake values (SUVPET) were calcu-
lated over the ROI as the ratio of activity concentration at time t
and injected dose at the time of injection divided by body weight.
2.4.4.2. In vivo.
Male Wistar-Unilever rats (n = 2; body
weight 150 12 g) were anesthetized with desflurane (9–10% v/v,
30% oxygen/air). The threshold value for breathing frequency was
65 breaths/min. Animals were put in supine position and placed
on a heating pad to maintain body temperature. The spontaneously
breathing rats were heparinized with 100 units/kg heparin (Hepa-
rin-Natrium 25.000-ratiopharmÒ, ratiopharm GmbH, Germany) by
subcutaneous injection to prevent blood clotting on intravascular
catheters. After local anesthesia with lignocain (1%; XylocitinÒ
loc, mibe, Jena, Germany) into the right groin, a catheter (0.8 mm
Umbilical Vessel Catheter, Tyco Healthcare, Tullamore, Ireland)
was introduced into the right femoral artery for arterial blood sam-
pling. A second needle catheter (35 G) was placed into a tail vein
and was used for [18F]-3 radiotracer injection (30 MBq in 0.5 mL
of E153/10% ethanol, infusion 1 mL/min). Arterial blood samples
were taken 1, 3, 5, 10, 20, 30 and 60 min after injection. Arterial
plasma was separated by centrifugation followed by precipitation
and removal of plasma proteins as described above. The radio-
HPLC system (Agilent 1100 series) applied for metabolite analysis
was equipped with UV detection (254 nm) and an external
radiochemical detector (RAMONA, Raytest GmbH, Straubenhardt,
3. Results and discussion
3.1. Synthesis of labeling precursors and reference compounds
Our primary synthetic efforts were directed on the synthesis of
the nonradioactive reference compound 3-(4-fluorophenyl)-2-(4-
methylsulfonylphenyl)-1H-indole 3. The synthesis sequence
started from 2-amino-benzoic acid, the amino group was protected
by tosylation, followed by a Friedel–Crafts acylation with fluoro-
benzene. After deprotection the amino function was reacted with
4-methylsulfonyl-benzoic acid chloride to give the key intermedi-
ate 2-N(4-methylsulfonylbenzoyl)-amino-40-fluoro-benzophenone
4. The final step was the cyclization of 4 with low-valent titanium
and zinc in THF to form the indole skeleton by the McMurry cycli-
zation (Scheme 1).20
Germany). Analysis was performed on
a Zorbax C18 300SB
(250 ꢀ 9.4 mm; 4 m) column with an eluent system C (water +
l
0.1%TFA) and D (acetonitrile + 0.1% TFA) in a gradient 5 min 95%
C, 10 min to 95% D and 5 min at 95% D at a flow rate of 3 mL/
min. All animal procedures and experiments were carried out
according to the guidelines of the German Regulations for Animal
Welfare. The protocols were approved by the local Ethical Commit-
tee for Animal Experiments.
Introduction of fluorine-18 onto the aromatic system is gener-
ally realized by nucleophilic substitution of a suitable leaving
group, preferred a trimethylamino- or nitro- functionality with
[
18F]fluoride. Due to the weak nucleophilic properties of the anio-
nic [18F]fluoride, an activation of the aromatic system by strong
electron withdrawing groups such as nitrile and carbonyl group
is essential for facilitating and accelerating the reaction. In the
present case of a diarylsubstituted indole a direct nucleophilic at-
tack with [18F]fluoride appeared to be not very promising. Consid-
ering this we decided to develop a precursor molecule having an
2.4.5. Small animal PET studies in vivo
Dynamic small animal PET studies for evaluation of compound
[
18F]-3 in vivo were performed in HT-29 tumor-bearing mice.
Therefore, NMRI nu/nu mice (Technische Universität Dresden,

T. Kniess et al. / Bioorg. Med. Chem. 20 (2012) 3410–3421
3415
F
fonyl-benzoylchloride
in
THF
to
give
2-N(4-meth-
F
ylsulfonylbenzoyl)-amino-40-(N,N-dimethylamino)-benzophenone
8 in 90% yield. Ultimately the trimethyl ammonium salt 9 was ob-
tained by reaction of 8 with methyl-trifluormethanesulfonate in
nitromethane as the desired precursor for radiofluorination.
O
O
S
CH3
3.2. Radiosynthesis of [18F]-3 via McMurry cyclization
N
NH
i)
O
H
The approach to obtain the radiolabeled COX-2 inhibitor [18F]-3
is outlined in Scheme 3. As mentioned above, the nucleophilic sub-
stitution with [18F]fluoride should be accomplished on precursor
molecule 9 to take advantage of the electron withdrawing effect
of the carbonyl group. The ring closure to the indole structure
had to be performed subsequently via McMurry cyclization of
the radiolabeled intermediate [18F]-4 with zinc and titanium
tetrachloride.
O
O
S
CH3
O
3
4
Scheme 1. Synthesis of reference compound 3 via McMurry cyclization. Reagents:
(i) TiCl4, Zn, THF.
electron withdrawing carbonyl group in para-position to the leav-
ing group for introduction of the 18F radiolabel and to perform the
McMurry cyclization as the second step. This radiolabeling
approach including a McMurry reaction has to the best of our
knowledge not been practiced so far. A detailed look in the litera-
ture revealed that the required dimethylamino-substituted benz-
ophenones from type of 8 or 9 are still unknown, what pushed
us to set up a new synthetic route (Scheme 2).
In a first number of experiments we investigated the optimal
conditions for introduction of [18F]fluoride into precursor 9 with
respect of solvent and temperature. Briefly 15 mg of precursor
were heated with the dried
[
18F]KF/Kryptofix222 complex in
1.5 mL of solvent for 15 min at temperature ranging between 90
and 140 °C. The reaction mixture was examined by analytical
radio-HPLC for the amount of radiolabeled intermediate [18F]-4
whereas the retention time of the UV-absorption from the nonra-
dioactive compound 4 served as the reference. The results are sum-
marized in Table 1, where it is recognizable that the ratio of [18F]-4
increased with the temperature. By using DMF as the solvent and
140 °C reaction temperature 31% of [18F]fluoride could be intro-
duced in 9 as a maximum yield (entries 1–5).
Beginning from entry 4 we tried to perform the McMurry cycli-
zation by addition of a complex formed from zinc and TiCl4 in THF
to the intermediate and by heating the mixture at 100 °C. At the
first go a ratio of 12% of the final product [18F]-3 was detected by
radio-HPLC. However a second radiolabeled product was observed,
what was supposed to be the free amino compound, formed by
alkaline hydrolysis of the amide intermediate [18F]-4 initiated from
the basic [18F]KF/Kryptofix222 conditions. To circumvent this situa-
tion we performed the radiolabeling under neutral pH conditions
with a Kryptofix222/C2K2O4 complex.23 Unfortunately the McMurry
reaction was inhibited under these conditions giving only a low
yield of [18F]-3; the reason for this behavior is the competitive
complex formation of titanium and oxalate (entry 5). In all our
attempts to replace the THF in the McMurry reaction by another
solvent, we observed by radio-HPLC in majority of cases a good
yield of the intermediate [18F]-4 but only less or nothing of the de-
sired final product [18F]-3 (entries 4, 6, and 10). This is conform
with findings in literature, describing that the low valent titanium
reagent is only formed by support of by weak coordinating solvents
like diethyl ether, THF or 1,2-dimethoxyethane (DME).20 In entries
7–9 DMF was applied as the solvent for the first and THF as the
In the literature Hu et al.13 described the formation of disubsti-
tuted benzophenones as multi-step procedure comprising the
reaction of tosylated anthranilic acid with PCl5 followed by a Fri-
edel–Crafts acylation with the second aromatic moiety without
work-up of the acid chloride 5. By application of this protocol with
N,N-dimethylaniline as aromatic partner at temperatures between
50 °C and 80 °C, we always obtained deep blue colored mixtures,
that were supposed to be derivatives of the triphenylmethane dyes
like Michler’s ketone.
In a parallel approach we prepared the corresponding benzam-
ide of anthranilic acid and performed the acylation with
N,N-dimethylaniline with POCl3 at 150 °C under Vilsmeier–Haack
conditions.21 However the yield of 2-(4-methylphenylsulfonami-
do)-40-(N,N-dimethylamino)-benzophenone 6 via this route was
unsatisfying 17%.
Consequently we decided for two step approach; an isolation
and purification of the 2-(4-methylphenylsulfonamido)-benzoyl
chloride 5 as described in the literature16 and to react 5 with
N,N-dimethylaniline and AlCl3 under Friedel–Crafts conditions in
a second step; in this way 6 was obtained with 38% yield.
The cleavage of the tosyl group from 6 in concentrated sulfuric
acid at 120 °C according to Hu et al.13 did neither yield the amine
7 in sufficient yield nor quality. From that reason we chosed a
deprotection method described by Kudav et al.22 basing on a mix-
ture of HClO4/CH3COOH at 100 °C, and were finally able to isolate
the 2-amino-40-(N,N-dimethylamino)-benzophenone
yield. Subsequently 7 was successfully coupled with 4-methylsul-
7 in 59%
N+
N
N
N
TfO-
Cl
O
O
O
O
O
NH
NH
NH
NH
iii)
i)
ii)
NH2
iv)
S
O
S O
O
S O
O
O
O
O
S
CH3
O
CH3
O
7
6
8
5
9
Scheme 2. Synthesis of labeling precursor 9. Reagents: (i) N,N-dimethylaniline, AlCl3; (ii) HClO4/CH3COOH; (iii) CH3SO2C6H4COCl, TEA, THF; (iv) CF3SO3CH3, nitromethane.

3416
T. Kniess et al. / Bioorg. Med. Chem. 20 (2012) 3410–3421
N+
18F
18F
TfO-
O
O
O
S
CH3
i)
ii)
NH
N
NH
O
H
O
O
O
O
S
S
CH3
O
CH3
O
18F]4
[
18F]3
9
[
Scheme 3. Radiosynthesis of [18F]-3 via McMurry coupling. Reagents: (i) [18F]fluoride, K222/K2CO3, DMF; (ii) TiCl4, Zn, THF.
Table 1
Yields and conditions of 18F-radiolabeling and McMurry reaction starting from precursor 9, radiochemical yields (RCY) are estimated from the product ratio in radio-HPLC
18F labeling conditions
Temperature (°C)
RCY [18F]-4 (%)
McMurry conditions
Temperature (°C)
RCY [18F]-3 (%)
Solvent
Solvent
1
2
CH3CN
DMF
90
90
10
1
—
—
—
—
—
—
3
4
5
6
7
8
9
10
11
12
13
DMF
120
140
140
140
140
140
140
140
120
110
110
18
31
30
26
n.d.
n.d.
n.d.
44
1
—
—
—
12
4
DMF
CH3CN/THF
THF
100
100
100
100
100
90
100
100
90
DMFa
DMF
CH3CN
THF
1
DMF
DMF
DMF
DMSO
CH3CN
CH3CN
CH3CN
16
25
20
0
57
50
82
THFb
THF
DMSO/THF
THF
THF
10
2
THF
90
n.d. not determined.
a
Potassium oxalate/Kryptofix K222 complex.
Double amount of TiCl4.
b
solvent for the second step, but even with increasing amount of
titanium tetrachloride and zinc the yield of [18F]-3 did not exceed
25%. We suspected that the DMF still present in the second step so
far, could act as an inhibitor of the McMurry cyclization. Reasoning
we decided to perform the radiolabeling in acetonitrile at 110–
120 °C, to remove the solvent in a stream of nitrogen under vac-
uum and to perform the McMurry reaction in pure THF. Under
these conditions the amount of desired radiotracer [18F]-3 was en-
hanced to 82% and the intermediate [18F]-4 was more or less com-
pletely consumed (entries 11–13). Finally the reaction steps were
transferred to an automated sequence on the synthesizer module,
covering radiolabeling, McMurry cyclization, and purification on a
semi-preparative column, solid phase extraction and formulation
of the radiotracer [18F]-3 for radio pharmacological investigation.
It should be noted that prior semi-preparative HPLC the mixture
was passed through a filter unit containing see sand to remove
any insoluble material originated from titanium. A second evapo-
ration step under vacuum was useful, because the injection of
the THF raw product solution into the semi-preparative HPLC
was responsible for a poor separation on the column. Whole pro-
cess of automated radiosynthesis of [18F]-3 is displayed in a flow
diagram at Scheme 4. In this way starting from 8000 MBq of
(0.1 M) and an equal volume of water-saturated n-octanol in a sep-
aration funnel. After vortexing for 1 min, the mixture was fixed,
and the two phases were allowed to separate. Aliquots of the sep-
arated phases were assayed for tracer activity concentration by the
gamma well counter. Samples were analyzed in fivefold and re-ex-
tracted three times to ensure stability of the logDoct7.4 value. In this
way the logDoct7.4 of [18F]-3 was determined to be 1.2 0.2.
3.4. Cell uptake studies in vitro
Cellular expression of COX proteins was detected after SDS–
PAGE by Western blotting using specific antibodies for COX-1
(68 kDa) and COX-2 (70 kDa). As internal control also b-actin
(42 kDa) was analyzed. Multiple protein bands in the COX-1 lane
between 60 kDa and 75 kDa result from different glycosylated or
splicing variants of COX-1 (Fig. 2).24,25
Different human tumor cell lines were used to study the uptake
of compound [18F]-3 in vitro. All cell lines used show a weak con-
stitutive expression of COX-1 with exceptionally low COX-1 signals
in FaDu and HT-29 cells. To differentiate between the specific con-
tribution of COX-1 and COX-2 the overall tracer uptake and intra-
cellular association tracer uptake were determined, unstimulated
human monocytes (THP-1) were used as model showing no or very
low baseline expression of COX-2.15,26,27 In contrast, TPA-stimu-
lated THP-1 cells as well as human tumor cell lines FaDu, HT-29,
and A2058 served as models for upregulated COX-2 expression
(Fig. 2).28,29
[
18F]fluoride 400–500 MBq of [18F]-3 was obtained in >98% radio-
chemical purity with a specific activity about 74–91 GBq/
lmol
(EOB) after 80 min total synthesis time.
3.3. Distribution coefficient
In the literature, the human malignant melanoma cell line A375
also has been described as COX-2 overexpressing cell line.29 In con-
trast, the present investigation showed no or only very low COX-2
The octanol/buffer distribution coefficient (logDoct7.4) at pH 7.4
was measured using radiolabeled [18F]-3 in phosphate buffer

T. Kniess et al. / Bioorg. Med. Chem. 20 (2012) 3410–3421
3417
Scheme 4. Flow diagram of automated radiosynthesis of [18F]-3 via McMurry cyclization.
Figure 2. Cellular expression of COX proteins in THP-1 and different tumor cell lines detected after SDS–PAGE by Western blotting.
protein expression in A375 cells, but an abundant synthesis of a
glycosylated or splice variant of COX-1 (Fig. 2). However, the radi-
otracer [18F]-3 uptake data obtained from all cell models used were
consistent to the published COX-1/COX-2 selectivity of 3-(4-fluo-
rophenyl)-2-(4-methylsulfonylphenyl)-1H-indole 3 (see Fig. 1)
and the observed protein expression of COXs.
macrophages, human tumor cells FaDu, A2058 and A375 and hu-
man tumor cell HT-29 monolayers, small- and large-sized spher-
oids is shown in Figure 3.
Unstimulated THP-1 cells comparatively showed a low uptake,
also did A375 cells. Interestingly, radiotracer uptake in A375 cells
was substantially higher than in unstimulated THP-1 monocytes
although COX protein expression was on a similar level. Of note,
in these cells radiotracer uptake was not or only weakly influenced
The in vitro cellular uptake and intracellular association of
[
18F]-3 in human THP-1 monocytes and differentiated THP-1

3418
T. Kniess et al. / Bioorg. Med. Chem. 20 (2012) 3410–3421
by preincubation with the nonradioactive reference compound 3. A
possible contribution of the COX-1 variant to overall tracer uptake
in A375 cells cannot be excluded. However, COX-2 overexpressing
cells showed a drastically increased uptake of [18F]-3 when com-
pared to cells showing no or only very low COX-2 synthesis. In or-
der to study a regulatory situation with hypoxia-modulated COX-2
expression, uptake of compound [18F]-3 was studied in multicellu-
lar HT-29 spheroids with or without intrinsic hypoxia as model
system.2,30,31 In this model, [18F]FMISO was used to confirm devel-
opment of intrinsic hypoxia in large-sized spheroids compared to
small-sized spheroids and monolayers, respectively.32 The
18F]FMISO uptake experiments in HT-29 monolayers (1.03
0.34%ID/mg protein) as well as individual HT-29 spheroids with a
diameter of both 300 10 m (1.17 0.29%ID/mg protein) and
600 10 m (9.86 3.4%ID/mg protein) clearly confirmed devel-
opment of an intrinsic hypoxia in large-sized spheroids that is high
enough to trap [18F]FMISO (pO2 <10 mm Hg). In consequence of
hypoxia-induced stimulation of COX-2 expression in HT-29 spher-
oids, large-sized spheroids also showed a substantially higher
uptake [18F]-3 when compared to small-sized spheroids and mon-
olayers, respectively. The specificity of the radiotracer uptake was
further tested by performing blocking experiments with all cells
[
l
l
after preincubation with a 100 lmol/L solution of the correspond-
ing nonradioactive reference compound 3 for 30 min (Fig. 3). More-
over, the possible interaction with the P-glycoprotein (P-gp) efflux
pump was examined in the presence of P-gp inhibitor cyclosporine
A and the calcium blocker verapamil, respectively.14,33,34 Therefore,
cells were incubated with cyclosporine A and verapamil for 60 min
before [18F]-3 was added to the loading buffer. Accordingly, a de-
crease in [18F]-3 cellular uptake or efflux may be an outcome of
P-gp interaction, However, both cyclosporine A and verapamil,
compared with the control, did not induce any significant effect
on [18F]-3 overall uptake in either cell line and time group (data
not shown).
Treatment of HT-29, FaDu and A375 cells with the correspond-
ing nonradioactive reference compound 3 at pharmacological
Figure 4. LC–MS/MS analysis of PGE2 in cell culture supernatants of three tumor
cell lines incubated with or without compound 3 for 25 h. Concentrations of
compound 3 were 1 lM or 10 l
M. Data are expressed as pg/106 cells (means
standard deviation), n = 6 from two independent experiments. bld, below limit of
detection. The limit of detection was ascertained at 15 pM.
0
25
50
3
3
(COX-2) IC50 = 1.2 µM
(COX-1) IC50 = 6.6 µM
75
Figure 3. In vitro study on overall cellular uptake and intracellular association of
[
18F]-3 in human THP-1 monocytes and differentiated THP-1 macrophages (A),
human tumor cells FaDu, A2058 and A375 (B), and human tumor cell HT-29
monolayers, small- and large-sized spheroids (C). Results are given as mean SD
(n = 6). TPA, stimulation by 64 nM phorbol myristate acetate (TPA) for 72 h;
100
-11
-10
-9
-8
-7
-6
-5
-4
-3
-2
Concentration [mol/L]
blocked, preincubation (for 30 min) with a 100
ing nonradioactive reference compound 3; (M), monolayer; (300), spheroid
diameter of 300 10 m; (600), spheroid diameter of 600 10 m.
lmol/L solution of the correspond-
l
l
Figure 5. Curve of inhibition of 3 for COX-1 and COX-2.

T. Kniess et al. / Bioorg. Med. Chem. 20 (2012) 3410–3421
3419
Figure 6. Radio-HPLC chromatogram of [18F]-3 in rat blood plasma after 1 min and 2 h in vitro incubation at 37 °C (left) and of rat arterial plasma samples immediately after
intravenous injection and 60 min p.i. (right), metabolites are marked as a and b.
Figure 7. Representative small animal PET images of a HT-29 tumor bearing mouse (maximum intensity projections) at 1, 5, and 60 min after single intravenous injection of
[
18F]-3; (circle—position of the subcutaneously xenotransplanted tumor).
concentrations resulted in substantial inhibition of PGE2 synthesis
in all cell lines (by 52–89% at 1 M; by 79–100% at 10 M; 100%
inhibition equals to below limit of determination (bld)) independent
compound was applied to an enzymatic competitive inhibition as-
say with celecoxib as reference compound. By using this assay in
our laboratories the inhibitory potency for 3 (IC50) was determined
l
l
on their individual COX expression levels (Fig. 4).
to be 6.6
coxib IC50 values of 115
l
M for COX-1 and 1.2
l
M for COX-2 (Fig. 5) while for cele-
M (COX-2) were
PGE2 is one of the major metabolites of arachidonic acid synthe-
sized by both cyclooxygenase isoforms. The determination of PGE2
level in cell culture supernatants is therefore a direct measure of
the COX activity. The cellular inhibition assay results demonstrated
the compound 3 to be a potent cyclooxygenase inhibitor, however,
also point to an only low COX-1/COX-2 selectivity of compound 3.
In order to further evaluate the actual selectivity of 3, the
l
M (COX-1) and 0.06 l
found. Of note, the data measured for compound 3 result in a
low COX-1/COX-2 selectivity ratio of 5.5 and are in a strong contra-
diction to the primary presumed selectivity ratio of 500 for com-
pound 3 that has been published in the literature.13 Nevertheless,
the overall results demonstrated that [18F]-3 is a suitable indicator
of increased COX-2 expression in conditions associated with

3420
T. Kniess et al. / Bioorg. Med. Chem. 20 (2012) 3410–3421
Figure 8. Time activity curves. Left: 18F-activity in the blood (vein, ROI over the abdominal vena cava and aorta), brain, liver and kidney measured with PET. Data are means
SEM of 9 mice. Right: external measured arterial blood 18F-activity (% of maximum activity) and the relative (% of activity) distribution of the 18F-activity between the blood
components (blood cells, plasma proteins, plasma water). Data are means SEM of blood samples of two rats.
inflammatory, tumorigenic or hypoxia-related stimuli in vitro.
Thus, the results showed promise for evaluating [18F]-3 as a posi-
tron-emitting probe for imaging of distribution or functional
expression of COX-2 by PET in HT-29 tumor xenotransplanted
mice.
the elimination phase was, similar to the situation in mice, very
short with 8.2 min calculated using a two-phase elimination mod-
el. These data indicate that [18F]-3 is not suitable for functional
imaging of COX-2 in rodent models in vivo.
The reasons for that behavior are manifold; mainly the affinity
of 3 for COX-2 is with 1.2 lM not specific enough to address the
3.5. Stability studies in vitro and in vivo
target enzyme, for that purpose an affinity in the nanomolar range
is required. On the other hand, the low COX-1/COX-2 selectivity
ratio may result in a substantial binding of the radiotracer in the
multitude of cells constitutively expressing COX-1, this also could
contribute to the observed lower overall availability of [18F]-3 in
the HT-29 tumor. Furthermore, it can be hypothesized that the per-
fusion of the tumors in the model used was too low for significant
intracellular tracer accumulation and that the lipophilic profile of
The in vitro stability of [18F]-3 was determined by incubation in
rat whole blood and plasma for 1 min and 2 h at 37 °C, respectively.
In all analyses only one more polar metabolite (a) in very small
amounts (lower 5% of activity) was observed after 2 h (Fig. 6).
A similar metabolic behavior was observed in vivo. Metabolite
analysis of arterial blood samples with radio-HPLC after intrave-
nous injection of the radiotracer into rats indicated that a minor
part of [18F]-3 was rapidly converted immediately after application
into the more polar metabolite (a), amounting to about 5% of the
applied activity at 5 min after injection. After 60 min a second
more hydrophilic radioactive metabolite (b) could be detected
and the original compound [18F]-3 in rat plasma still amounted
to 75% with the two metabolites a (12.3%) and b (13.7%) compris-
ing 25% of the applied (decay corrected) activity (Fig. 6).
[
18F]-3 is not optimal for crossing the cell membrane and is respon-
sible for high unspecific binding in nontargeted tissues such as the
gastrointestinal tract, liver, and kidneys. On the other hand, an
important criterion for a ‘successful’ radiotracer is a high specific
radioactivity. The specific radioactivity should be as high as possi-
ble to minimize a binding competition between radiolabeled and
non-labeled inhibitor. In case of [18F]-3 a specific radioactivity
ranging from 74 to 91 GBq/lmol at the end of synthesis was
achieved. This results in an administered dose of 3 of 0.1
lg
3.6. Small animal PET studies
(0.28 nM) that should be adequate low not to compete with the
radiotracer.
To further assess the feasibility of compound [18F]-3 as PET
radiotracer for imaging COX-2, particularly, COX-2 overexpressing
tumors, small animal PET imaging was performed in HT-29 tumor-
bearing mice. In the small animal PET experiments, 20 MBq of
4. Summary and concluding remarks
The aim of the present work was to develop an 18F-radiolabeled
cyclooxygenase inhibitor with the scaffold of a diarylsubstituted
indole exhibiting high affinity and selectivity towards COX-2. Be-
cause a direct nucleophilic substitution with 18F-fluoride on the
lead structure was not very promising, we used a specially de-
signed benzophenone 9 with a trimethylammonium leaving group
as precursor molecule for introduction of the radiolabel. The radi-
otracer [18F]-3 was synthesized by cyclization of the intermediate
[
18F]-3 was administered intravenously into the tail vein (corre-
sponding to 0.1 lg (0.28 nM) of 3). In contrast to the in vitro exper-
iments, the small animal PET studies showed no substantial
accumulation of [18F]-3 in human colorectal adenocarcinoma
(HT-29) xenotransplanted in nude mice. Dynamic PET studies
revealed that the main activity was fast accumulated in the liver
and was eliminated by the hepatobiliar route into the intestine
(Fig. 7). A substantial amount of activity was accumulated in the
Harderian glands and the brown adipose tissue. The kidneys
showed a temporary accumulation of minor part of the activity.
A very low amount of activity temporarily could be detected in
the brain.
The clearance of the activity in mice from the blood (ROI over
the abdominal part of the vena cava and aorta), liver, kidneys and
brain were similar (Fig. 8). The main activity was eliminated into
the intestine. For comparison, in rats the arterial blood clearance
was externally measured as shown in Figure 6. The half-life of
[
18F]-4 with low valent titanium and zinc by application of the
McMurry protocol from organic synthesis onto radiochemistry. In
this manner [18F]-3 was synthesized via a two-step procedure in
a remote controlled synthesizer unit covering purification, solid
phase extraction and formulation for radio pharmacological inves-
tigation in 10% total decay corrected yield with high radiochemical
purity and a specific activity of 74–91 GBq/lmol at EOB. Consider-
ation of in vitro properties, including affinity and distribution
between the blood cells, plasma proteins and water, showed
promise for in vivo imaging application of the radiotracer [18F]-3.

T. Kniess et al. / Bioorg. Med. Chem. 20 (2012) 3410–3421
3421
8. Prabhakaran, J.; Underwood, M. D.; Parsey, R. V.; Arango, V.; Majo, V. J.;
Simpson, N. R.; van Heertum, R.; Mann, J. J.; Kumar, J. S. D. Bioorg. Med. Chem.
1802, 2007, 15.
9. De Vries, E. F.; Doorduin, J.; Dierckx, R. A.; van Waarde, A. Nucl. Med. Biol. 2008,
35, 35.
10. Uddin, J.; Crews, B. C.; Ghebreselaise, K.; Huda, I.; Kingskey, P. J.; Sib Ansan, M.;
Tantawy, M. N.; Reese, J.; Marnett, L. J. Cancer Prev. Res. 2011, 4, 1536.
11. Wuest, F. R.; Hoehne, A.; Metz, P. Org. Biomol. Chem. 2005, 3, 503.
12. Wuest, F.; Kniess, T.; Bergmann, R.; Pietzsch, J. Bioorg. Med. Chem. 2008, 16,
7662.
However, although showing micromolar affinity for COX-2 and
acceptable high metabolic blood stability in vivo, [18F]-3 turned
out to be not suitable for in vivo imaging of COX-2 in mice with
xenotransplanted HT-29 tumors. It should be pointed out that
the high specificity and selectivity data of 3 published in the liter-
ature13 prompted us in the development of the radiotracer [18F]-3
as a lead compound. However in the present study two indepen-
dent COX-2 activity assays demonstrated only a low selectivity of
3 for COX-2 at pharmacological concentrations.
13. Hu, W.; Guo, Z.; Chu, F.; Bai, A.; Yi, X.; Cheng, G.; Li, J. Bioorg. Med. Chem. 2003,
11, 1153.
Nevertheless, from a radiochemical point of view it should be
underlined that the utilization of McMurry cyclization gives access
to 18F-labeled diarylsubstituted indoles, a class of compounds that
is by conservative labeling techniques not straightforward avail-
able so far. In our ongoing work a sequence of further 2,3-diaryl-
substituted indoles and similar five and six-membered nitrogen
containing heterocycles is under preparation and will be evaluated
as new inhibitors of COX-2. The compounds showing the best
affinity, selectivity, and most promising bioavailability will be sub-
jected 18F-radiolabeling to further in vivo functional characteriza-
tion of cyclooxygenases.
14. Tanaka, M.; Fujisaki, Y.; Kawamura, K.; Ishiwata, K.; Qinggeletu; Yamamoto, F.;
Mukai, T.; Maeda, M. Biol. Pharm. Bull. 2006, 29, 2087.
15. Buu-Hoi, L. Bull. Soc. Chim. Fr. 1946, 139.
16. Coombs, R. V.; Danna, R. P.; Denzer, M.; Hardtmann, G. E.; Huegi, B.; Koletar, G.;
Koletar, J.; Ott, H.; Jukniewicz, E. J. Med. Chem. 1973, 16, 1237.
17. Tang, G.; Wang, M.; Tang, X.; Gan, L.; Luo, L. Nucl. Med. Biol. 2005, 32, 553.
18. Cao, H.; Xiao, L.; Park, G.; Wang, X.; Azim, A. C.; Christman, J. W.; van Breemen,
R. B. Anal. Biochem. 2008, 372, 41.
19. Schütze, C.; Bergmann, R.; Yaromina, A.; Hessel, F.; Kotzerke, J.; Steinbach, J.;
Baumann, M.; Beuthien-Baumann, B. Radiother. Oncol. 2007, 83, 311.
20. McMurry, J. E. Chem. Rev. 1989, 89, 1513.
21. Lee, B. C.; Lee, K. C.; Lee, H.; Mach, R. H.; Katzenellenbogen, J. A. Bioconjugate
Chem. 2007, 18, 507.
22. Kudav, D. P.; Samant, S. P.; Hosangadi, B. D. Synth. Commun. 1987, 17, 1185.
23. Hamacher, K.; Hamkens, W. Appl. Radiat. Isot. 1995, 46, 911.
24. Chandrasekharan, N. V.; Dai, H.; Roos, K. L.; Evanson, N. K.; Tomsik, J.; Elton, T.
S.; Simmons, D. L. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 13926.
25. Schumacher, K.; Castrop, H.; Strehl, R.; de Vries, U.; Minuth, W. W. Cell. Physiol.
Biochem. 2002, 12, 63.
26. Bagga, D. W.; Wang, L.; Farias-Eisner, R.; Glaspy, J. A.; Reddy, S. T. Proc. Natl.
Acad. Sci. U.S.A. 2003, 100, 1751.
27. Goppelt-Struebe, M.; Schaefer, D.; Habenicht, A. J. Br. J. Pharmacol. 1997, 122,
619.
Acknowledgements
The authors wish to thank S. Preusche for radioisotope produc-
tion and M. Barth, C. Heinig, R. Herrlich, S. Lehnert, A. Suhr and P.
Wecke for expert technical assistance in the chemical and radio-
pharmacological investigations.
28. Lev-Ari, S.; Strier, L.; Kazanov, D.; Madar-Shapiro, L.; Dvory-Sobol, H.; Pinchuk,
I.; Marian, B.; Lichtenberg, D.; Arber, N. Clin. Cancer Res. 2005, 11, 6738.
29. Denkert, C.; Köbel, M.; Berger, S.; Siegert, A.; Leclere, A.; Trefzer, U.;
Hauptmann, S. Cancer Res. 2001, 61, 303.
References and notes
30. Lee, J. J.; Natsuizaka, M.; Ohashi, S.; Wong, G. S.; Takaoka, M.; Michaylira, C. Z.;
Budo, D.; Tobias, J. W.; Kanai, M.; Shirakawa, Y.; Naomoto, Y.; Klein-Szanto, A.
J.; Haase, V. H.; Nakagawa, H. Carcinogenesis 2010, 31, 427.
31. Kaidi, A.; Qualtrough, D.; Williams, A. C.; Paraskeva, C. Cancer Res. 2006, 66,
6683.
32. Rasey, J. S.; Nelson, N. L.; Chin, L.; Evans, M. L.; Grunbaum, Z. Radiat. Res. 1990,
122, 301.
33. Herzog, C. E.; Trepel, J. B.; Mickley, L. A.; Bates, S. E.; Fojo, A. T. J. Natl. Cancer
Inst. 1992, 84, 711.
1. Blombaum, A. L.; Marnett, L. J. J. Med. Chem. 2007, 50, 1425.
2. Greenhough, A.; Smartt, H. J.; Moore, A. E.; Roberts, H. R.; Williams, A. C.;
Paraskeva, C.; Kaidi, A. Carcinogenesis 2009, 30, 377.
3. Wang, D.; BuBois, R. N. Nat. Rev. Cancer 2010, 10, 181.
4. Rizzo, M. T. Clin. Chim. Acta 2011, 412, 671.
5. Chakraborti, A. K.; Garg, S. K.; Kumar, R.; Motiwala, H. F.; Jadhavar, P. S. Curr.
Med. Chem. 2010, 17, 1563.
6. Kao, J.; Genden, E. M.; Chen, C. T.; Rivera, M.; Tong, C. C.; Misiukiewicz, K.;
Gupta, V.; Gurudutt, V.; Teng, M.; Packer, S. H. Cancer 2011, 117, 3173.
7. De Vries, E. F. Curr. Pharm. Des. 2006, 12, 3847.
34. Beck, W. T.; Qian, X. D. Biochem. Pharmacol. 1992, 43, 89.