Late stage benzylic C-H fluorination with [18F]fluoride for PET imaging

Article abstract of DOI:10.1021/ja5039819

We describe the first late-stage ¹⁸F labeling chemistry for aliphatic C-H bonds with no-carrier-added [¹⁸F]fluoride. The method uses Mn(salen)OTs as an F-transfer catalyst and enables the facile labeling of a variety of bioactive molecules and building blocks with radiochemical yields (RCY) ranging from 20% to 72% within 10 min without the need for preactivation of the labeling precursor. Notably, the catalyst itself can directly elute [¹⁸F]fluoride from an ion exchange cartridge with over 90% efficiency. Using this feature, the conventional and laborious dry-down step prior to reaction is circumvented, greatly simplifying the mechanics of this protocol and shortening the time for automated synthesis. Eight drug molecules, including COX, ACE, MAO, and PDE inhibitors, have been successfully [ ¹⁸F]-labeled in this way.

Full text of DOI:10.1021/ja5039819

Communication
pubs.acs.org/JACS
Late Stage Benzylic C−H Fluorination with [18F]Fluoride for PET
Imaging
Xiongyi Huang,†,∥ Wei Liu,†,∥ Hong Ren,‡,§ Ramesh Neelamegam,‡ Jacob M. Hooker,*,‡,§
and John T. Groves*,†
†Department of Chemistry, Princeton University, Princeton, New Jersey 08544, United States
‡Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital and Harvard Medical School, Charlestown,
Massachusetts 02129, United States
§Division of Nuclear Medicine and Molecular Imaging, Department of Radiology, Massachusetts General Hospital, Boston,
Massachusetts 02114, United States
S
* Supporting Information
on applicable synthetic methods for radiolabeling and the
ABSTRACT: We describe the first late-stage 18F labeling
chemistry for aliphatic C−H bonds with no-carrier-added
[18F]fluoride. The method uses Mn(salen)OTs as an F-
transfer catalyst and enables the facile labeling of a variety
of bioactive molecules and building blocks with radio-
chemical yields (RCY) ranging from 20% to 72% within 10
min without the need for preactivation of the labeling
precursor. Notably, the catalyst itself can directly elute
[18F]fluoride from an ion exchange cartridge with over
90% efficiency. Using this feature, the conventional and
laborious dry-down step prior to reaction is circumvented,
greatly simplifying the mechanics of this protocol and
shortening the time for automated synthesis. Eight drug
molecules, including COX, ACE, MAO, and PDE
inhibitors, have been successfully [18F]-labeled in this way.
synthesis of precursors.7 Due to their short half-lives, PET
radioisotopes must typically be incorporated into tracer
molecules at a late stage of the overall synthesis process.
Combined with other constraints, including solvent compati-
bility, low reaction concentrations, and the need for rapid
process steps including product purification, PET radiotracer
synthesis has a very limited toolbox of chemical reactions when
compared to other synthetic organic disciplines. For 18F
labeling, the majority of the radiotracers and radiotracer
candidates are synthesized through nucleophilic 18F-substitu-
tion.8 A recent upsurge in fluorination chemistry has revealed a
number of novel 18F labeling methods, including preparation of
[18F]fluoroaromatics through aryl iodonium salts,9 Pd-catalyzed
allylic fluorination with [18F]fluoride,10 the preparation of
[18F]4-fluorophenols via oxidative fluorination,11 aromatic 18F
labeling through PdIV or NiII complexes,12 copper-catalyzed 18
F
labeling of aryltrifluoromethyl groups,13 and enantioselective
radiosynthesis of [18F]fluorohydrins.14 Some of these methods
have been scaled up and optimized for high specific activity
imaging applications,15 while others have yet to be demon-
strated in an imaging context due to practical limitations of the
methodologies.
ositron emission tomography (PET) is a molecular
Pimaging modality that has wide-ranging applications in
clinical oncology, cardiology, and neurology as well as basic
biomedical research.1 The characteristics that set PET apart
from other imaging techniques are its ability for high sensitivity,
noninvasive imaging and quantification of in vivo interactions at
a molecular level.1 Among all PET radioisotopes, 18F is the
most widely used and clinically relevant radionuclide.1b,2
Furthermore, fluorinated derivatives of known drugs often
show stronger binding to target sites, lower metabolic burden,
and higher bioavailability.3 By far the most prominent
radiotracer to date is [18F]fluorodeoxyglucose ([18F]FDG),
which has dominated the use of PET in oncology for over 20
years.4
Despite the great success of PET imaging in certain clinical
and research domains, the development of new radiotracers
remains a formidable challenge.5 Currently, there are only
seven approved PET tracers, three of which are simple
radionuclides.6 High throughput assessment of potential
radiotracers would be highly advantageous to increase the
rate of discovery as it is currently infeasible to predict whether a
particular radiolabeled molecule will exhibit the required in vivo
characteristics to serve as a target-specific radiotracer. One main
challenge tempering PET throughput stems from constraints
Methods currently available for 18F labeling, including the
newest advances, are dominated by a “prefunctionalization”
approach in which a highly reactive chemical functional group
(L) is preinstalled at the labeling site and substituted later by
18F (Figure 1a). In most cases, multistep synthesis is required
for the preparation of each “preactivated” precursor.1c
Consequently, potential radiotracer candidates are often
prioritized for radiolabeling based on synthetic accessibility of
the precursor, which amplifies dramatically with increasing
structural complexity. Frequently, harsh reaction conditions are
required for the 18F labeling step, which can further diminish
the functional group compatibility of reaction.2,16 The draw-
back of the “prefunctionalization” approach is most evident in
the screening of PET tracers, where labeling of a diverse range
of molecules is desired. Thus, iterative discovery through
arduous trial and error is adopted.5
Received: March 6, 2014
Published: April 27, 2014
© 2014 American Chemical Society
6842
dx.doi.org/10.1021/ja5039819 | J. Am. Chem. Soc. 2014, 136, 6842−6845

Journal of the American Chemical Society
Communication
labeling. We found that while Mn(salen)Cl gave only trace
amounts of 18F-labeled product, manganese salen complexes
with more labile triflate (OTf) or perchlorate counterions
showed substantial higher radiochemical conversions (RCC) to
18F-labeled product 2 (16% and 34%, respectively). Various
more weakly associated ligands were evaluated, and p-
toluenesulfonate (OTs) was found to afford the highest RCC
(53%) in the test reaction. Further optimizations based on the
Mn(salen)OTs catalyst achieved a maximum of 65% RCC for
the initial substrate (ibuprofen) using iodosylbenzene (PhIO)
as the oxidant (Table 1, entry 5). No labeling products were
detected in control experiments in which the manganese salen
catalyst or iodosylbenzene was omitted.
Figure 1. Approaches for labeling molecules with 18F. (a) Multistep
synthesis of preactivated labeling precursors, which is time and
resource consuming. (b) Direct C−H 18F-fluorination of parent
molecules.
We found that this procedure allowed for the efficient 18F
labeling of benzylic C−H bonds in a wide range of substrates
with RCC ranging from 20% to 68% (Figure 2). A range of
Here we describe an 18F labeling strategy that implements a
direct replacement of sp3 hydrogen with fluorine (Figure 1b).
The method avoids the need for target preactivation, enabling
high throughput radiolabeling of parent compounds and
building blocks. We demonstrate selective 18F substitution at
benzylic C−H bonds, which are common to drug and drug-like
molecules. The method shows promise to significantly increase
the efficiency of PET tracer synthesis and evaluation and to
provide ready access to labeled molecules that are difficult to
access or cannot be prepared by conventional methods.
The concept of direct hydrogen substitution with 18F was
first demonstrated by Firnau et al. in the 1980s,17 but there has
been very limited development of 18F labeling methods based
on C−H fluorination since this pioneering work. Recently,
several of us reported a series of manganese-catalyzed aliphatic
C−H fluorination reactions that exhibit promising features for
18F labeling applications.18 These reactions utilize nucleophilic
fluoride (F−) as the fluorine source, in contrast to methods that
require reactive, electrophilic fluorinating agents.19 Although
there are several C−H fluorination methods based on
nucleophilic fluoride sources,20 application to 18F labeling has
yet to be demonstrated.
The pivotal factor for adapting fluoromanganese(IV) fluorine
transfer reactions to 18F labeling is the formation of reactive
18F-containing intermediates using substoichiometric, low-
concentration [18F]fluoride. Under catalytic 19F reaction
conditions, excess fluoride serves as the axial ligand for the
oxomanganese(V) species in the hydrogen abstraction and
subsequently as the fluorine transfer agent.18a−c
To address this challenge in the context of radiochemistry,
we employed the exploratory conditions shown in Table 1 to
evaluate the efficacy of various manganese salen catalysts for 18F
Figure 2. Direct 18F labeling of aliphatic C−H bonds of substrates and
decay-corrected radiochemical conversions (RCCs).
Table 1. Aliphatic C−H 18F-Fluorination of Ibuprofen Ester
functional groups were well tolerated, including esters, amides,
imides, ketones, alkynes, ethers, cyanides, heterocycles,
carbamates, and aryl and aliphatic halides. The labeling was
generally more efficient for substrates bearing electron-donating
groups, presumably due to the electrophilic nature of the
hydrogen-abstracting oxomanganese(V) intermediate. The
method can be used to prepare 18F-labeled synthons (in
addition to direct labeling for radiotracer evaluation). For
example, dibenzosuberone (10), the chemical precursor of a
series of tricyclic antidepressant drugs (TCAs) including
amitriptyline and nortriptyline, was readily labeled with 18F in
50% RCC. The tolerance of reactive functional groups such as
halogens and alkynes enables the rapid incorporation of 18F-
catalyst
solvent
T
RCC
1
2
3
4
5
6
7
Mn(salen)Cl
CH3CN
CH3CN
CH3CN
CH3CN
acetone
acetone
acetone
50
50
50
50
50
25
90
trace
16%
34%
53%
65%
45%
50%
Mn(salen)OTf
Mn(salen)ClO4
Mn(salen)OTs
Mn(salen)OTs
Mn(salen)OTs
Mn(salen)OTs
6843
dx.doi.org/10.1021/ja5039819 | J. Am. Chem. Soc. 2014, 136, 6842−6845

Journal of the American Chemical Society
Communication
labeled motifs into complex structures through well-established
methods such as nucleophilic substitutions or “click”
reactions.21 Notably, the 18F labeling reaction can be performed
under air and without rigorous exclusion of water, greatly
simplifying the protocol and facilitating scale-up.
chemistry, we found that no drying procedure was required.
[18F]Fluoride deposited on an anion exchange cartridge (AEC)
could be eluted using an organic solution of the catalyst,
Mn(salen)OTs, with over 90% recovery of the radiolabel from
the column and no erosion of the RCC in the subsequent
fluorination reaction. For example, 18F-labeled celestolide was
obtained with 10% non-decay corrected RCY with a specific
activity of 2.68 Ci/μmol (end of bombardment) (Figure 4).
These results demonstrate the significant potential of the
present method for PET imaging applications.
The major benefit of this mild C−H 18F-fluorination reaction
is its application to late-stage radiolabeling. To demonstrate this
potential, we examined a variety of well-known biologically
active molecules. The selected compounds encompass
inhibitors of important biological and pharmacological targets
including cyclooxygenase (COX), monoamine oxidase B
(MAO-B), phosphodiesterase 10A (PDE10A), and angioten-
sin-converting enzyme (ACE), as well as biomessenger
molecules such as the neurotransmitter dopamine, and the
immuno-modulating drug, fingolimod. Subjecting these mole-
cules (or protected analogues) to Mn-catalyzed fluorination led
to successful 18F labeling specifically at benzylic positions
(Figure 3). The RCC ranged from 22% to 72% at 50 °C within
Figure 4. Dry-down free procedure for the synthesis of 18F-celestolide.
It is of interest to compare and contrast the 18F fluorinations
described here using limiting fluoride ion to the 19F reactions
we have previously described that use a large excess of
fluoride.18 We suggest the mechanism shown in Figure 5a for
Figure 5. (a) Proposed mechanism for 18F labeling of benzylic C−H
bonds catalyzed by a manganese salen catalyst. (b) Detection of
enantioselectivity of labeling products of celestolide by chiral radio-
HPLC analysis. (c) Energy landscape of fluorine transfer from F−
MnIV−OH intermediate (34) to a benzyl radical.
Figure 3. Direct C−H 18F labeling of bioactive molecules. Reported
radiochemical conversions (RCCs) are decay-corrected and averaged
over (n) experiments. *0.35 equiv of oxidant, PhIO, was used.
10 min. In the case of fingolimod, 29, high regioselectivity was
observed for the protected amino diol side chain (see
Supporting Information). Notably, the 18F labeling reaction
showed a much broader substrate scope than its 19F
counterpart. For example, fluorinating C−H bonds β to
electron-withdrawing groups (e.g., a ketone or Boc-protected
amine) was very challenging on a preparative scale in the 19F
reaction, but this position was readily labeled under 18F reaction
conditions with an RCC of 41% and 51%, respectively, for 24
and 30. This seemingly counterintuitive phenomenon results
from the very low concentration of [18F]fluoride and the large
excess of the manganese catalyst. Apparently, the low
concentration of manganese [18F]fluoride species present
under 18F labeling conditions is sufficient to capture the
incipient substrate radicals with high efficiency.
this 18F labeling reaction. In 19F chemistry, a trans-
difluoromanganese(IV) complex was shown to be the reactive
fluorine transfer intermediate. This compound was isolated and
structurally characterized.18c However, due to the limiting
amount of [18F]fluoride in the labeling conditions, the
formation of the [18F]trans-difluoromanganese(IV) intermedi-
ate is not feasible. Therefore, the [18F]fluorine transfer is more
likely to proceed directly through a 18F−MnIV−OH inter-
mediate (33) even in the presence of a large excess of a
manganese catalyst that has no fluoride ligand.
The involvement of a manganese salen-bound 18F
intermediate in the fluorine transfer step was demonstrated
experimentally by analyzing the enantioselectivity of the
resulting 18F-labeled product. Using celestolide as the
diagnostic substrate, we measured a 25% ee in the fluorinated
product using chiral HPLC analysis and radio-detection (Figure
5b). Moreover, when the catalyst was changed from (R,R)-
Mn(salen)OTs to (S,S)-Mn(salen)OTs, the same 25% ee was
Having demonstrated the enabling power of this new 18F
fluorination protocol, we performed initial process optimization
to facilitate scale-up for PET imaging. While the method is
compatible with typical “dry-down” procedures used in 18F
6844
dx.doi.org/10.1021/ja5039819 | J. Am. Chem. Soc. 2014, 136, 6842−6845

Journal of the American Chemical Society
Communication
(3) (a) Purser, S.; Moore, P. R.; Swallow, S.; Gouverneur, V. Chem.
Soc. Rev. 2008, 37, 320. (b) Muller, K.; Faeh, C.; Diederich, F. Science
2007, 317, 1881.
(4) Wood, K. A.; Hoskin, P. J.; Saunders, M. I. Clin. Oncol. 2007, 19,
237.
(5) Agdeppa, E. D.; Spilker, M. E. AAPS J. 2009, 11, 286.
(6) (a) Vallabhajosula, S.; Solnes, L.; Vallabhajosula, B. Semin. Nucl.
Med. 2011, 41, 246. (b) Koo, J.; Byun, Y. Arch. Pharm. Res. 2013, 36,
1178.
observed in the labeling product but with reversed
enantioselectivity. The fluorine transfer reactivity of the 18F−
MnIV−OH complex was further supported by density func-
tional theory (DFT) computations (Figure 5c). The activation
barrier of fluorine transfer from the F−MnIV−OH complex
(34) to the benzyl radical was only 9.6 kcal/mol in an acetone
solvent continuum. The molecular orbitals involved in the C−F
2
bond formation are the σ*(dz ) orbital of 34 and the benzyl
radical SOMO. The overall fluorine transfer process is
thermodynamically favored with a calculated free energy
change of −38.0 kcal/mol.
(7) Mach, R. H.; Schwarz, S. W. PET Clinics 2010, 5, 131.
(8) Le Bars, D. J. Fluorine Chem. 2006, 127, 1488.
(9) (a) Pike, V. W.; Aigbirhio, F. I. J. Chem. Soc., Chem. Commun.
1995, 2215. (b) Ross, T. L.; Ermert, J.; Hocke, C.; Coenen, H. H. J.
Am. Chem. Soc. 2007, 129, 8018.
(10) Hollingworth, C.; Hazari, A.; Hopkinson, M. N.; Tredwell, M.;
Benedetto, E.; Huiban, M.; Gee, A. D.; Brown, J. M.; Gouverneur, V.
Angew. Chem., Int. Ed. 2011, 50, 2613.
(11) Gao, Z.; Lim, Y. H.; Tredwell, M.; Li, L.; Verhoog, S.;
Hopkinson, M.; Kaluza, W.; Collier, T. L.; Passchier, J.; Huiban, M.;
Gouverneur, V. Angew. Chem., Int. Ed. 2012, 51, 6733.
(12) (a) Lee, E.; Hooker, J. M.; Ritter, T. J. Am. Chem. Soc. 2012,
134, 17456. (b) Lee, E.; Kamlet, A. S.; Powers, D. C.; Neumann, C.
N.; Boursalian, G. B.; Furuya, T.; Choi, D. C.; Hooker, J. M.; Ritter, T.
Science 2011, 334, 639.
(13) Huiban, M.; Tredwell, M.; Mizuta, S.; Wan, Z.; Zhang, X.;
Collier, T. L.; Gouverneur, V.; Passchier, J. Nat. Chem. 2013, 5, 941.
(14) Graham, T. J. A.; Lambert, R. F.; Ploessl, K.; Kung, H. F.; Doyle,
A. G. J. Am. Chem. Soc. 2014, 136, 5291.
(15) Kamlet, A. S.; Neumann, C. N.; Lee, E.; Carlin, S. M.; Moseley,
C. K.; Stephenson, N.; Hooker, J. M.; Ritter, T. PLoS One 2013, 8, 10.
(16) Hollingworth, C.; Gouverneur, V. Chem. Commun. 2012, 48,
2929.
(17) (a) Firnau, G.; Chirakal, R.; Garnett, E. S. J. Nucl. Med. 1984, 25,
1228. (b) Chirakal, R.; Firnau, G.; Couse, J.; Garnett, E. S. Int. J. Appl.
Radiat. Isot. 1984, 35, 651.
(18) (a) Liu, W.; Huang, X.; Groves, J. T. Nat. Protoc. 2013, 8, 2348.
(b) Liu, W.; Groves, J. T. Angew. Chem., Int. Ed. 2013, 52, 6024.
(c) Liu, W.; Huang, X.; Cheng, M.-J.; Nielsen, R. J.; Goddard, W. A.,
III; Groves, J. T. Science 2012, 337, 1322. (d) Liu, W.; Groves, J. T. J.
Am. Chem. Soc. 2010, 132, 12847.
In conclusion, we have developed a facile, no-carrier-added,
18F labeling method that allows efficient late-stage labeling of a
variety of organic molecules and known drugs. The reaction is
operationally simple, requiring no dry-down operations, and is
tolerant of both moisture and air. This protocol can be
immediately adapted in any laboratory site with a basic PET
chemistry infrastructure. We are working to expand the concept
of 18F labeling via direct C−H activation demonstrated in this
study.
ASSOCIATED CONTENT
■
S
* Supporting Information
Detailed experimental procedures, spectroscopic data for all
new compounds, and details for DFT calculation. This material
is available free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Authors
hooker@nmr.mgh.harvard.edu
jtgroves@princeton.edu
Author Contributions
∥X.H. and W.L. contributed equally.
Notes
The authors declare no competing financial interest.
(19) (a) Hull, K. L.; Anani, W. Q.; Sanford, M. S. J. Am. Chem. Soc.
2006, 128, 7134. (b) Wang, X.; Mei, T.-S.; Yu, J.-Q. J. Am. Chem. Soc.
2009, 131, 7520. (c) Bloom, S.; Pitts, C. R.; Woltornist, R.; Griswold,
A.; Holl, M. G.; Lectka, T. Org. Lett. 2013, 15, 1722. (d) Amaoka, Y.;
Nagatomo, M.; Inoue, M. Org. Lett. 2013, 15, 2160. (e) Xia, J.-B.; Zhu,
C.; Chen, C. J. Am. Chem. Soc. 2013, 135, 17494. (f) Bloom, S.; Pitts,
C. R.; Miller, D. C.; Haselton, N.; Holl, M. G.; Urheim, E.; Lectka, T.
Angew. Chem., Int. Ed. 2012, 51, 10580. (g) Fier, P. S.; Hartwig, J. F.
Science 2013, 342, 956.
ACKNOWLEDGMENTS
■
This research was supported by the Center for Catalytic
Hydrocarbon Functionalization, an Energy Frontier Research
Center, U.S. Department of Energy, Office of Science, Basic
Energy Sciences, under Award No. DE SC0001298 (J.T.G.).
Fluorination of biomolecules was supported by the US National
Science Foundation award CHE-1148597 (J.T.G.). A portion
of this research was carried out at Martinos Center for
Biomedical Imaging using resources provided by the Center for
Functional Neuroimaging Technologies, P41EB015896, and
shared instrumentation grants S10RR017208 and
S10RR023452. H.R. was supported by a US Dept. of Energy
radiochemistry training grant DE-SC0008430 (J.M.H.). X.H.
thanks the Howard Hughes Medical Institute for fellowship
support. W.L. thanks Merck, Inc. for fellowship support. The
authors also thank Prof. A. G. Doyle and T. Graham for helpful
discussions.
(20) (a) McMurtrey, K. B.; Racowski, J. M.; Sanford, M. S. Org. Lett.
2012, 14, 4094. (b) Braun, M.-G.; Doyle, A. G. J. Am. Chem. Soc. 2013,
135, 12990.
(21) Marik, J.; Sutcliffe, J. L. Tetrahedron Lett. 2006, 47, 6681.
REFERENCES
■
(1) (a) Phelps, M. E. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 9226.
(b) Miller, P. W.; Long, N. J.; Vilar, R.; Gee, A. D. Angew. Chem., Int.
Ed. 2008, 47, 8998. (c) Ametamey, S. M.; Honer, M.; Schubiger, P. A.
Chem. Rev. 2008, 108, 1501.
(2) Tredwell, M.; Gouverneur, V. Angew. Chem., Int. Ed. 2012, 51,
11426.
6845
dx.doi.org/10.1021/ja5039819 | J. Am. Chem. Soc. 2014, 136, 6842−6845