18F-radiolabeling of aromatic compounds using triarylsulfonium salts

Article abstract of DOI:10.1002/ejoc.201101730

A new method for the ¹⁸F-radiolabeling of aromatic compounds based on the aromatic nucleophilic substitution (S_NAr) reaction using triarylsulfonium salts has been developed. Experiments and DFT calculations indicated that sulfonium i

Full text of DOI:10.1002/ejoc.201101730

SHORT COMMUNICATION
DOI: 10.1002/ejoc.201101730
18F-Radiolabeling of Aromatic Compounds Using Triarylsulfonium Salts
Linjing Mu,[a] Cindy R. Fischer,[a] Jason P. Holland,[a] Jessica Becaud,[a]
P. August Schubiger,[a] Roger Schibli,[a] Simon M. Ametamey,*[a] Keith Graham,[b]
Timo Stellfeld,[b] Ludger M. Dinkelborg,[b] and Lutz Lehmann[b]
Keywords: Sulfonium salts / Fluorine / Isotopic labeling / Density functional calculations
A new method for the 18F-radiolabeling of aromatic com-
pounds based on the aromatic nucleophilic substitution
(SNAr) reaction using triarylsulfonium salts has been devel-
oped. Experiments and DFT calculations indicated that sulfo-
nium ions have the potential to be optimized for labeling
nonactivated and deactivated aryl rings that have Hammett
σP substituent constants greater than –0.170. This method is
applicable to a range of halogen-substituted aryl systems in-
cluding nonactivated and deactivated aryl rings. In particu-
lar, the high radiochemical yield of
benzene from precursor 1a represents a ready source of com-
pound 1b for subsequent use in palladium-catalyzed C–H ac-
tivation/C–C bond-forming reactions.
[
18F]-4-fluoroiodo-
Introduction
Results and Discussion
Traditional synthetic methods for the preparation of
[18F]-fluoroaromatic compounds are generally limited to
aromatic rings functionalized with electron-withdrawing
groups.[1] Standard activating/leaving groups used in the
SNAr reaction include trifluoromethyl (CF3), nitro (NO2),
and trimethylammonium salts.[2] Recently, the use of diaryl-
iodonium salts as precursors in nucleophilic 18F-substitu-
tion reactions was developed as an alternative route for the
synthesis of nonactivated or electron-rich [18F]-fluoroaro-
matic compounds.[3] However, the majority of 18F-radio-
chemical syntheses employing the aforementioned leaving
groups often require the use of harsh reaction conditions
with high temperatures (ca. Ͼ130 °C),[4] which frequently
leads to low radiochemical yields, precursor/product de-
composition, and the necessity to use complicated isolation
methods.
To the best of our knowledge, the published results in
the 1980s using the dimethylsulfonium ion as a leaving
group for 18F-fluoride nucleophilic substitution was only
possible with activated aromatic rings.[5] Here we present a
method for the 18F-radiolabeling of aromatic compounds
including nonactivated and deactivated aryl rings based on
the aromatic nucleophilic substitution (SNAr) reaction
using triarylsulfonium salts as the substrates.
Scheme 1 illustrates the general radiochemical reaction
employed here in the synthesis of a range of [18F]-para-fluo-
roaryl derivatives 1b–7b from their corresponding triaryl-
sulfonium salts 1a–7a. A variety of synthetic methods for
preparing triarylsulfonium salts have been reported.[6]
Compounds 1a–5a and their corresponding para-fluoroaryl
derivatives 1b–5b are commercially available, facilitating
their use as model systems in these feasibility studies.
Amide derivative precursors 6a and 7a and their corre-
sponding products 6b and 7b were synthesized by adapting
previously reported methods.[7] Full experimental details
are presented in the Supporting Information.
18F-Radiolabeling experiments were optimized by using
K[18F]F (or Cs[18F]F) and kryptofix as a source of nucleo-
philic 18F-fluoride (ca. 2–5 GBq), 2 mg of precursor com-
pound 1a–7a dissolved in 200 μL of anhydrous aprotic or-
ganic solvent [acetonitrile (MeCN), dimethyl sulfoxide
(DMSO), dimethyl formamide (DMF)] in the presence of a
base (K2CO3 or Cs2CO3) and reacted at various tempera-
tures (80–130 °C) for a total time of 15 min. Product con-
version yields and radiochemical purity (RCP) were deter-
mined from integration analysis of the radio-ultra perform-
ance liquid chromatography (radio-UPLC) chromatograms
based on the ratio of the radioactivity area of product to
the total radioactivity area. In all cases, product identity
was confirmed by co-injection of the known reference com-
pound by using UPLC. Results for the radiolabeling reac-
tions are presented in Table 1.
[a] Animal Imaging Center-PET, Center for Radiopharmaceutical
Sciences of ETH, PSI and USZ, ETH-Hoenggerberg, D-CHAB
IPW HCI H427,
Wolfgang-Pauli-Str. 10, 8093 Zurich, Switzerland
Fax: +41-44-6331367
The radiochemical experiments demonstrate that 18F-la-
beling of aromatic systems by nucleophilic substitution with
the use of sulfonium ion precursors is feasible. In MeCN,
compound 1a could be 18F-radiolabeled to give [18F]-4-
E-mail: simon.ametamey@pharma.ethz.ch
[b] Global Drug Discovery, Bayer Healthcare Pharmaceuticals,
Berlin, Germany
Supporting information for this article is available on the
WWW under http://dx.doi.org/10.1002/ejoc.201101730.
Eur. J. Org. Chem. 2012, 889–892
© 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
889

S. M. Ametamey et al.
SHORT COMMUNICATION
Table 1. Selected data for optimization of 18F-radiochemical reac-
tions by using sulfonium precursors 1a–7a (n = 2, 3).
Entry Precursor
Solvent
T
Conversion [%]
[°C]
Byproduct Product
1
2
3
4
5
6
7
8
1a
1a
1a
1a
1a
1a
1a
1a
1a
1a
2a
2a
3a
3a
4a
4a
5a
5a
6a
6a
7a
7a
7a
7a
7a
7a
7a
MeCN[a,c]
MeCN[a]
DMF[a]
80
90
80
7Ϯ1
8Ϯ2
10Ϯ2
9Ϯ1
13Ϯ2
16Ϯ3
31Ϯ2
6Ϯ1
12Ϯ2
7Ϯ1
6Ϯ2
10Ϯ1
0
91Ϯ1
90Ϯ2
81Ϯ4
88Ϯ1
76Ϯ5
62Ϯ2
42Ϯ4
49Ϯ3
61Ϯ6
48Ϯ1
36Ϯ1
23Ϯ2
48Ϯ3
70Ϯ7
1Ϯ1
2Ϯ1
0
DMF[a]
90
DMF[a]
130
130
130
130
130
90
80
90
80
90
80
90
80
90
80
90
80
90
DMF[b]
DMF[b], 10 μL H2O
DMSO[b]
DMSO[a]
DMSO[a]
MeCN[a]
DMF[a]
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
MeCN[a]
DMF[a]
0
MeCN[a]
DMF[a]
19Ϯ4
10Ϯ3
25Ϯ5
71Ϯ4
0
MeCN[a]
DMF[a]
0
MeCN[a]
DMF[a]
98Ϯ1
89Ϯ5
0
0
0
0
0
0
0
0
0
MeCN[a]
DMF[a]
0
DMF[b]
90
110
90
90
110
18Ϯ1
68Ϯ4
30Ϯ5
15Ϯ7
63Ϯ6
DMF[b]
DMSO[b]
DMSO[b], 10 μL H2O
DMSO[b]
[a] K2CO3 was used as the base. [b] Cs2CO3 was used as the base.
[c] Specific activity was 5 GBqμmol–1 with starting activity at
2.3 GBq.
Scheme 1. Radiochemical (SNAr) reaction employed in the synthe-
sis of [18F]-para-fluoroaryl derivatives and structures of 6a, 6b and
7a, 7b.
The effect of changing the substituent on the aryl ring
was also studied (Scheme 1 and Table 1). In general, bro-
fluoroiodobenzene in excellent yields (91Ϯ1%, n = 3) at mine derivative 2a gave lower conversion yields under the
80 °C in 15 min (Table 1, Entry 1). High radiochemical same conditions compared to that obtained with iodine de-
yields were also obtained by changing the solvent to DMF. rivative 1a. Notably, the conversion yield for compound 2a
For compounds 1a, 2a, 6a, and 7a, 18F-fluorination oc- in MeCN at 80 °C was only 36%, whereas for 1a 91% yield
curred preferentially on the substituted aromatic ring, with was obtained. The lower yield for 2a is consistent with the
minimal [18F]-fluorobenzene observed. In the case of non- slightly less activating Br (σP = 0.232) vs. I (σP = 0.276)
activated aromatic compound 3a (Hammett σP = 0.0) and substituent. Replacing the halogen substituent on the aryl
deactivated compounds 4a (σP = –0.17) and 5a (σP
=
ring with a hydrogen, methyl, or methoxy group (i.e., 3a–
–0.268), [18F]-fluorobenzene was the favored product, which 5a) favors the formation of [18F]-fluorobenzene. For deacti-
could be obtained in up to 70% yield (Table 1, Entries 14, vated compound 4a, although [18F]-fluorobenzene was the
15, and 18) under mild radiolabeling conditions. Purifica- preferred product, a small amount of [18F]-fluorotoluene
tion of compound 3a in acetonitrile by using a Sep-Pak C18 (1–2%) was also obtained. The fact that conversion into
light cartridge (Waters) resulted in 31Ϯ2% (n = 2) isolated [18F]-fluorobenzene occurred for compounds 3a, 4a, and
radiochemical yield (RCY, decay corrected). The measured 5a, as well as the observation of some [18F]-fluorotoluene
difference between the isolated RCY and the conversion product 4b, confirms that the use of sulfonium ions as leav-
yield of 48Ϯ3% determined by UPLC (Table 1, Entry 13) ing groups for 18F-substitutions can facilitate (albeit mod-
is in the normal range observed for fluorine-18 labeled com- est) the radiolabeling of nonactivated and even deactivated
pounds in our laboratory. All the radiolabeling steps were aryl rings.
performed in sealed reaction vials, which were cooled to
To assess the chemical scope and potential applications
room temperature 5 min prior to taking an aliquot for of sulfonium ions as precursors for 18F-radiolabeling of
UPLC analysis. Nevertheless, the loss of radioactive prod- peptides, the influence of amide bonds in the para-position
uct, and the possibility of erroneously high conversion was investigated by using model compound 6a and tetra-
yields arising from the potential volatility of [18F]-fluoro- peptide 7a (NH-Val-βAla-Phe-Gly-OH). For compound 6a,
benzene was investigated. In this regard, we found that no desired product 6b was obtained in near quantitative con-
radioactivity was lost during our procedure.
version (98%; Table 1, Entry 19) by using the optimized la-
890
www.eurjoc.org
© 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2012, 889–892

18F-Radiolabeling of Aromatic Compounds Using Triarylsulfonium Salts
beling conditions derived from compound 1a (K2CO3,
Trends calculated by DFT are fully consistent with the
MeCN at 80 °C). Encouraged by this positive result, we ap- observed experimental data on reaction conversion and
plied the same conditions to tetrapeptide 7a, however, no with the expected reactivities, as predicted from Hammett
product was observed in either MeCN or DMF. Changing σP substituent constants. Using the calculated free energy
the base to Cs2CO3 and increasing the temperature from 90 of TS2 (the transition state for the common fluorobenzene
to 110 °C did provide the product in 18 and 68%, respec- byproduct in all reactions) as an internal reference, the data
tively.
indicate that for halogenated compounds 1a, 2a, and 8a and
To understand further the influence of the substituent model amide 9a, attack of the fluoride at Cipso of the substi-
effect on radiolabeling efficiency, we used density functional tuted ring is favored by 2–5 kJmol–1 for the halogens and
theory (DFT) calculations to map the reaction potential en- by 13.5 kJmol–1 for the more-activated/electron-poor amide
ergy surface. As accurate predictions of overall reaction free ring system (Table 2). In contrast, SNAr substitution on the
energies are not accessible due to the difficulty in calculat- methoxy-substituted aryl group of compound 5a is strongly
ing the energy for the fluoride anion, we explored the na- disfavored by 14.9 kJmol–1, which is consistent with the ex-
ture of the transition state for the SNAr substitution. For perimental data showing that only the [18F]-fluorobenzene
each system, we used the common fluorobenzene byproduct byproduct was formed.
reaction as an internal reference and calculated the vi-
Calculated frequencies for the asymmetric νas (F–Cipso–
brational modes, ν, as well as the difference in free energy, S) vibrational stretching modes of the transition states re-
ΔTSG, between the transition states for the reaction path- veal that across all compounds studied, the frequency of
ways leading to either the desired product (TS1) or the fluo- the imaginary TS2 stretch was very similar and in the range
robenzene byproduct (TS2), where ΔTSG = G[TS1] – G[TS2]. from –359 to –369 cm–1. This confirms the validity of using
A schematic diagram of the DFT calculated reaction co- the calculated structure of TS2 as an internal reference for
ordinate with optimized reactant, transition state, and the energy calculations (control compound 3a: R = H,
product structures based on compound 1a is shown in Fig- SPh3). The imaginary stretching frequencies for TS1 leading
ure 1. Calculated energies are presented in Table 2 and full to the desired product vary substantially (ranging from
details are given in the Supporting Information.
–323 cm–1 for 9a to –393 cm–1 for 5a). For systems that un-
dergo experimentally facile SNAr reactions like 1a, 2a, and
7a (here modeled by compound 9a), the vibrational fre-
quency was lower in energy than that of TS2, whereas for
nonreactive methoxy-substituted compound 5a, ν(TS1) was
calculated to be 44 cm–1 higher in energy. Changes in the
calculated free energy difference between TS1 leading to the
product and TS2 are also consistent with the experimental
data in Table 1. For activated, electron-poor systems, 1a,
2a, 8a, and 9a, ΔTSG Ͻ 0 kJmol–1 indicate that the pathway
leading to the desired product is favored. For deactivated
precursors 4a and 5a, calculated ΔTSG values Ͼ0 kJmol–1
are consistent with the higher observed radioactive conver-
sion to the [18F]-fluorobenzene byproduct. However, it is
noteworthy that for methyl-substituted compound 4a, for-
mation of product 4b is not as strongly disfavored as for
more deactivated methoxy compound 5a. Collectively, these
experimental and computational data also confirm the
theory that transition states having high frequency vi-
brational modes are generally higher in energy and are as-
sociated with a less favored reaction pathway.
Figure 1. Schematic of the DFT calculated reaction coordinates for
the SNAr substitution reaction between fluoride and sulfonium ion
precursor 1a.
Table 2. DFT calculated energies and vibrational modes.[a]
Compd.
Substituent
ΔTS
G
ΔPG
ν(TS1)
[cm–1
Hammett
constant σP
[kJmol–1
]
[kJ mol–1
]
]
Conclusions
1a
2a
3a
4a
5a
8a
9a
I
Br
H
Me
OMe
–4.3
–5.1
0
4.7
4.6
0
3.1
3.4
3.9
2.4
–348
–349
–364[b]
–378
–393
–356
–323
0.276
0.232
0.0
–0.170
–0.268
0.227
–
In conclusion, a new method for the 18F-radiolabeling of
nonactivated aromatic compounds using triarylsulfonium
salts has been successfully developed. The method is appli-
cable to a range of substituted aryl systems including
amides. In particular, the high radiochemical yield of [18F]-
4-fluoroiodobenzene from precursor 1a represents a ready
source of compound 1b for subsequent use in palladium-
catalyzed C–H activation/C–C bond-forming reac-
tions.[2,3b,8] These studies also indicate that sulfonium ions
5.4
14.9
–2.4
–13.5
Cl
C(O)NHCH3
[a] Negative values of ΔTSG indicate that the TS for the desired
product is favored. Positive values for ΔPG indicate that the byprod-
ucts are lower in energy. [b] Calculated vibrational mode for TS2
for SPh3.
Eur. J. Org. Chem. 2012, 889–892
© 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
891

S. M. Ametamey et al.
SHORT COMMUNICATION
have the potential to be optimized for labeling nonactivated
and deactivated aryl rings that have Hammett σP substitu-
ent constants greater than –0.170. In spite of the encourag-
ing preliminary results for labeling peptide-based model
systems, further work is required to fully optimize this
method as a potential tool for 18F-labeling of full, biolo-
gically relevant peptides.
[1] a) Z. B. Li, P. S. Conti, Adv. Drug Delivery Rev. 2010, 62, 1031–
1051; b) H. H. Coenen in PET Chemistry: The Driving Force
in Molecular Imaging (Eds.: P. A. Schubiger, L. Lehmann, M.
Friebe), Springer, Berlin. 2007, vol. 62, pp. 15–50; c) J. Becaud,
L. Mu, M. Karramkam, P. A. Schubiger, S. M. Ametamey, K.
Graham, T. Stellfeld, L. Lehmann, S. Borkowski, D. Berndorff,
L. Dinkelborg, A. Srinivasan, R. Smits, B. Koksch, Bioconjug-
ate Chem. 2009, 20, 2254–2261.
[2] P. W. Miller, N. J. Long, R. Vilar, A. D. Gee, Angew. Chem. Int.
Ed. 2008, 47, 8998–9033.
[3] a) T. L. Ross, J. Ermert, C. Hocke, H. H. Coenen, J. Am. Chem.
Soc. 2007, 129, 8018–8025; b) F. R. Wust, T. Kniess, J. Labelled
Compd. Radiopharm. 2004, 47, 457–468; c) A. Shah, V. W. Pike,
D. A. Widdowson, J. Chem. Soc. Perkin Trans. 1 1998, 2043–
2046; d) B. C. Lee, C. S. Dence, H. B. Zhou, E. E. Parent, M. J.
Welch, J. A. Katzenellenbogen, Nucl. Med. Biol. 2009, 36, 147–
153; e) S. G. DiMagno, B. J. Wang, L. L. Qin, K. D. Neumann,
S. Uppaluri, R. L. Cerny, Org. Lett. 2010, 12, 3352–3355.
[4] J. Ermert, C. Hocke, T. Ludwig, R. Gail, H. H. Coenen, J.
Labelled Compd. Radiopharm. 2004, 47, 429–441.
Experimental Section
The syntheses and chemical characterization of compounds 6a, 6b,
7a, and 7b have been described in the Supporting Information. No-
carrier-added 18F-fluoride was produced by the 18O(p,n)18F nuclear
reaction by irradiation of isotopically enriched 18O-water in a fixed-
energy Cyclone 18/9 cyclotron (IBA). Dried 18F-fluoride–cryptate
complex was prepared by using a standard separation and azeo-
tropic drying procedure in the presence of Kryptofix 2.2.2 and po-
tassium carbonate.
[5] a) M. Maeda, T. Fukumura, M. Kojima, Chem. Pharm. Bull.
1985, 33, 1301–1304; b) M. Maeda, T. Fukumura, M. Kojima,
Appl. Radiat. Isot. 1987, 38, 307–310.
Supporting Information (see footnote on the first page of this arti-
cle): All synthetic and radiochemical details, HPLC chromato-
grams, and full computational methods.
[6] a) Y. Endo, K. Shudo, T. Okamoto, Chem. Pharm. Bull. 1981,
29, 3753–3755; b) R. D. Miller, A. F. Renaldo, H. Ito, J. Org.
Chem. 1988, 53, 5571–5573; c) S. Imazeki, M. Sumino, K. Fu-
kasawa, M. Ishihara, T. Akiyama, Synthesis 2004, 1648–1654.
[7] Z. J. Kaminski, B. Kolesinska, J. Kolesinska, G. Sabatino, M.
Chelli, P. Rovero, M. Blaszczyk, M. L. Glowka, A. M. Papini,
J. Am. Chem. Soc. 2005, 127, 16912–16920.
Acknowledgments
[8] F. R. Wust, T. Kniess, J. Labelled Compd. Radiopharm. 2005,
Financial support was provided in part by the Eidgenössische Tech-
nische Hochschule (ETH) Fellowship Program (FEL-01 10–1;
J. P. H.).
48, 31–43.
Received: December 1, 2011
Published Online: January 2, 2012
892
www.eurjoc.org
© 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2012, 889–892