Synthesis of the main metabolite in human blood of the A1 adenosine receptor ligand [18F]CPFPX

Article abstract of DOI:10.1021/ol900169f

In human blood, the PET radiotracer [¹⁸F]CPFPX (1) Is metabolized to numerous metabolites, one (M1) being the most prominent in plasma 30 min p.i. Because the mass of injected tracer is ≤5 nmol, concentrations In plasma are too low to analyze.

Full text of DOI:10.1021/ol900169f

ORGANIC
LETTERS
2009
Vol. 11, No. 19
4266-4269
Synthesis of the Main Metabolite in
Human Blood of the A1 Adenosine
Receptor Ligand [18F]CPFPX
Marcus H. Holschbach,*,† Dirk Bier,† Walter Wutz,† Sabine Willbold,‡ and
Ray A. Olsson†
Institute of Neuroscience and Medicine (INM-5) and Central DiVision of Analytical
Chemistry, Forschungszentrum Ju¨lich GmbH, D-52425 Ju¨lich, Germany
m.holschbach@fz-juelich.de
Received July 6, 2009
ABSTRACT
In human blood, the PET radiotracer [18F]CPFPX (1) is metabolized to numerous metabolites, one (M1) being the most prominent in plasma
30 min p.i. Because the mass of injected tracer is e5 nmol, concentrations in plasma are too low to analyze. Human liver microsomes
generate main metabolites having HPLC retention times identical to those in plasma. HPLC-MS tentatively identified M1 as 2. Synthesis of
2 and identical HPLC-MS spectra of 2 and M1 confirmed that assignment.
The radiofluorinated isotopologue of the potent and selective
A1 adenosine receptor antagonist (A1AR antagonist) 8-cy-
clopentyl-3-(3-fluoropropyl)-1-propylxanthine (CPFPX, 1),
namely [18F]8-cyclopentyl-3-(3-fluoropropyl)-1-propylxan-
thine ([18F]CPFPX, 1*),1 is used as a radioligand to image
the A1AR in human brain by positron emission tomography
(PET).2 Because the brain does not metabolize this ligand
and metabolites formed in the periphery do not cross the
blood-brain-barrier,3 specifically the bound ligand accounts
for a large fraction of brain tissue radioactivity. However,
such is not the case for ligand in the periphery, where it
undergoes rapid conversion to a number of more polar
metabolites in quantities that obscure specific binding.3 Using
the ligand in humans for clinical and research studies4-10
and with regard to future research focusing on the develop-
(3) Holschbach, M. H.; Fein, T.; Wutz, W.; Boy, C.; Mu¨hlensiepen,
H.; Hamacher, K.; Lewis, R.; Schwabe, U.; Mu¨ller-Ga¨rtner, H.-W.; Coenen,
H. H.; Olsson, R. A. Drug DeVelop. Res. 1998, 43, 71.
† Institute of Neuroscience and Medicine.
(4) Meyer, P. T.; Bier, D.; Holschbach, M. H.; Boy, C.; Olsson, R. A.;
Coenen, H. H.; Zilles, K.; Bauer, A. J. Cereb. Blood Flow Metab. 2004,
‡ Central Division of Analytical Chemistry.
(1) Holschbach, M. H.; Olsson, R. A.; Bier, D.; Wutz, W.; Sihver, W.;
Schu¨ller, M.; Palm, B.; Coenen, H. H. J. Med. Chem. 2002, 45, 5150–
5156.
24, 323–333
(5) Meyer, P. T.; Elmenhorst, D.; Bier, D.; Holschbach, M. H.; Matusch,
A.; Coenen, H. H.; Zilles, K.; Bauer, A. Neuroimage 2005, 24, 1192–1204
(6) Meyer, P. T.; Elmenhorst, D.; Matusch, A.; Winz, O.; Zilles, K.;
Bauer, A. Neuroimage 2006, 32, 1100–1105
.
.
(2) Bauer, A.; Holschbach, M. H.; Meyer, P. T.; Boy, C.; Herzog, H.;
Olsson, R. A.; Coenen, H. H.; Zilles, K. Neuroimage 2003, 19, 1760–1769.
.
10.1021/ol900169f CCC: $40.75
Published on Web 09/02/2009
 2009 American Chemical Society

ment of metabolically more stable ligands with improved
bioavailability, information about its metabolism is required.
Because the dose of radiotracer administered to humans
is so small, typically 0.5-5 nmol, it is extremely difficult to
characterize the metabolites in plasma even by LC-MS.
However, in vitro metabolism by human liver microsomes
(HLM)12 and recombinant human CYP1A2 (hCYP1A2)
seem reasonable alternatives for generating quantities of
metabolites sufficient for analyzing their structures.
to furnish a cycloalkene which is subsequently hydroxylated
at a carbon atom adjacent to the double bond (R-oxidation)
followed by dehydrogenation to yield an enone. It is
noteworthy that metabolite M1 is an intensely fluorescent
molecule11 for which a large planar π-electron system is a
prerequisite. The conjugation of the double bonds in the
imidazole heterocycle with the cyclopentene double bond
and an additional carbonyl function in the cyclopentenyl
moiety is a structural feature common to a group of well-
known fluorescent organic polymethine dyes, the merocya-
nines.17,18
Once LC-MS tentatively identified 8-(3-oxocyclopent-
1-enyl)-3-(3-fluoropropyl)-1-propylxanthine (2) as the
likeliest structure of M1 it was necessary to synthesize it
as a proof of structure. A first synthetic approach was
based on the synthesis of the parent ligand CPFPX (1)19
(path a in Scheme 1, 4 f 1) by substituting the original
In a recent study,11 radio-TLC of extracts of plasma
from humans given [18F]CPFPX identified nine radio-
labeled metabolites. Extracts of reactions catalyzed by
HLM contained four of those metabolites and extracts of
hCYP1A2 reactions all nine. Thus, like naturally occurring
xanthines such as caffeine (1,3,7-trimethylxanthine) and
theophylline (1,3-dimethylxanthine), CPFPX undergoes
oxidation in liver microsomes mainly by CYP1A2;
however, unlike the natural xanthines, dealkylation does
not occur. That study12 used LC-MS to analyze the
metabolites generated by HLM; the m/z of the major
metabolite, M1, was 335 [M + H]+, consistent with an
enone species. Cone voltage induced in-source dissociation
analysis suggested that M1 was 3-(3-fluoropropyl)-8-(3-
oxocyclopent-1-enyl)-1-propylxanthine (2).
Scheme 1
.
Key Steps in the Synthesis of CPFPX 1 and
Unsuccessful Approaches to 2a
It is not surprising that the cyclopentyl moiety should be
an important site of metabolism. As pointed out previously,11
molecular modeling of the interaction of CPFPX with the
catalytic site of CYP1A2 places the cyclopentane moiety of
CPFPX in close proximity to the enzyme’s heme prosthetic
group.
It is well known that during metabolism a cyclopentyl
moiety can easily be hydroxylated and subsequently oxidized
to the respective ketone.13-15 To the best of our knowledge,
a double hydroxylation-dehydration process as postulated
for the metabolic transformation of the cyclopentyl moiety
of [18F]CPFPX has not yet been described in the literature.
The fact that this metabolic step is favored in the case of
[18F]CPFPX is most probably due to the generation of a very
stable conjugated double bond system resulting from the
dehydration of an initially formed ring-hydroxylated species
(7) Meyer, P. T.; Elmenhorst, D.; Boy, C.; Winz, O.; Matusch, A.; Zilles,
K.; Bauer, A. Neurobiol. Aging 2007, 28, 1914–1924.
(8) Elmenhorst, D.; Meyer, P. T.; Winz, O.; Matusch, A.; Ermert, J.;
Coenen, H. H.; Basheer, R.; Haas, H. L.; Zilles, K.; Bauer, A. J. Neurosci.
2007, 27, 2410–2415.
a For details, see the text.
(9) Basheer, R.; Bauer, A.; Elmenhorst, D.; Ramesh, V.; McCarley,
R. W. Neuroreport 2007, 18, 1895–1899.
(10) Boy, C.; Meyer, P. T.; Kircheis, G.; Holschbach, M. H.; Herzog,
H.; Elmenhorst, D.; Kaiser, H. J.; Coenen, H. H.; Haussinger, D.; Zilles,
K.; Bauer, A. Eur. J. Nucl. Med. Mol. Imaging 2008, 35, 589–597.
(11) Matusch, A.; Meyer, P. T.; Bier, D.; Holschbach, M. H.; Woitalla,
D.; Elmenhorst, D.; Winz, O.; Zilles, K.; Bauer, A. Nucl. Med. Biol. 2006,
33, 891–898.
cyclopentyl ring by a cyclopentenyl moiety. In the
published synthesis,19 a set of orthogonal protecting groups
consisting of a benzyl (Bn) and a pivaloyloxymethyl (Pom)
group was used for the regioselective incorporation of a
3-fluoropropyl substituent at N-3 of the xanthine hetero-
cycle in a late synthetic step. Since unsaturation in the
cycloalkenyl building block used in the envisioned
(12) Bier, D.; Holschbach, M. H.; Wutz, W.; Olsson, R. A.; Coenen,
H. H. Drug Metab. Dispos. 2006, 34, 570–576.
(13) Zacchei, A. G.; Wishousky, T. I.; Arison, B. H.; Hitzenberger, G.
Drug Metab. Dispos. 1978, 6, 303–312
.
(14) Krause, W.; Kuhne, G.; Jakobs, U.; Hoyer, G. A. Drug Metab.
Dispos. 1993, 21, 682–689
.
(15) Chauret, N.; Guay, D.; Li, C.; Day, S.; Silva, J.; Blouin, M.;
(17) Da¨hne, S. Z. Chem. 1965, 5, 441–451
.
Ducharme, Y.; Yergey, J. A.; Nicoll-Griffith, D. A. Bioorg. Med. Chem.
(18) Da¨hne, S.; Leupold, D. Ber. Bunsen-Ges. 1966, 70, 618–625
.
Lett. 2002, 12, 2149–2152
.
(19) Holschbach, M. H.; Fein, T.; Krummeich, C.; Lewis, R. G.; Wutz,
W.; Schwabe, U.; Unterlugauer, D.; Olsson, R. A. J. Med. Chem. 1998,
41, 555–563.
(16) Subrahmanyam, V.; Renwick, A. B.; Walters, D. G.; Price, R. J.;
Tonelli, A. P.; Lake, B. G. Xenobiotica 2002, 32, 521–534
Org. Lett., Vol. 11, No. 19, 2009
.
4267

synthesis was not compatible with hydrogenation condi-
tions necessary to remove an amide benzyl protecting
group, an alternative protecting group was needed. It is
well documented that introduction of methoxy substituents
into the aromatic benzylic core leads (poly)methoxy benzyl
derivatives with increased acid sensitivity.20 Thus, a
4-methoxybenzyl (p-methoxybenzyl, PMB) amide can be
efficiently cleaved under acidic (trifluoroacetic acid, TFA)
conditions. Starting the synthesis with 1-PMB-6-aminou-
racil instead of 1-benzyl-6-aminouracil, it became obvious
that during PMB cleavage (TFA, reflux overnight) con-
comitant hydrolysis of the Pom group on N-7 of the
xanthine imidazole ring had occurred. However, protection
with a more acid labile 2,4-dimethoxybenzyl (DMB)
group20 allowed the selective cleavage of the latter under
much milder acidic conditions (TFA, 50 °C, 5 h) while
leaving the Pom group completely unaffected. Having
established an efficient combination of protecting groups,
various ways to introduce a functionalized cyclopentane
ring into CPFPX were examined. Condensation of the key
intermediate 5,6-diamino-1-(2,4-dimethoxybenzyl)-3-pro-
pyluracil (8 in Scheme 1) with 3-oxocyclopent-1-enecar-
boxylic acid21 (path b in Scheme 1, 8 f 10) under
standard coupling reactions (DMF, EDAC, DMAP) gave
the desired amide 9 in moderate (35-40%) yield, but
subsequent ring closure either induced by aqueous base
(LiOH, NaOH, Ca(OH)2) or by heating with condensing
agents (HMDS, BTSA) decomposed the amide and
furnished only intractable mixtures. The next approach
(path c in Scheme 1, 8 f 2) envisioned the condensation
of the DMB protected diaminouracil 8 with cyclopent-1-
enecarboxylic acid, ring closure to the respective xanthine
and subsequent oxidation of the carbon atom R to the
double bond of the cyclopentene ring (allylic oxidation).
The protected xanthine heterocycle 13 was obtained in a
straightforward manner using the classical Traube purine
synthesis22 followed by Pom-protection of N-7 to give
12, hydrolysis of the DMB moiety, and subsequent
fluoropropylation of N-3 to yield 13. Allylic oxidation of
the cyclopentene ring was tried with a variety of oxidants
and catalysts according to published methods23-25 for the
allylic oxidation of cyclopentene esters. Among the oxidizing
systems investigated were CrO3-acetic anhydride-acetic
acid,23 Pd(OH)2-tert-butyl-hydroperoxide,24 and Rh2-
(cap)-tert-butyl hydroperoxide-K2CO3,25 but all efforts to
isolate even traces of the desired enone were unsuccessful.
Scheme 2. Synthesis of Compound 2
synthesis required several modifications to the approach used
to synthesize CPFPX.19 1-(2,4-Dimethoxybenzyl)uracil 15
was prepared by condensation of 2,4-dimethoxybenzylurea
3 with cyanoacetic acid22 followed by sodium ethanolate
induced ring closure. The 6-amino-1-(2,4-dimethoxybenzyl)-
3-propyluracil 18 was synthesized according to a modifica-
tion of a known procedure27 for the regioselective alkylation
of the uracil N-3 position via protection of the amino group
at the 6-position as the N-[(dimethylamino)methylene]
derivative. Thus, reaction of 15 with dimethylformamide
dimethyl acetal (DMF/DMA) in dimethyl formamide (DMF)
at 40 °C yielded the 6-N-[(dimethylamino)methylene] de-
rivative 16 in excellent yield. Phase-transfer-catalyzed (PTC)
alkylation28 of 16 with 1-bromopropane, CsCO3, and tet-
rabutylammonium bromide as a catalyst in a mixed solvent
system of DMF and acetonitrile gave the 3-n-propyl deriva-
tive 17. Alkaline hydrolysis of the N-[(dimethylamino)-
methylene] group using KOH in methanol furnished 6-amino-
1-(2,4-dimethoxybenzyl)-3-propyluracil 18.
Finally, a completely different synthetic strategy to
compound 2 was followed (Scheme 2), namely a Pd-
catalyzed Stille cross-coupling reaction26 of a brominated
xanthine with tri-n-butylstannyl-2-cyclopentene-1-one. This
Subsequent nitrosation of 18 at the C-5 position with
NaNO2 in aqueous acetic acid afforded the 5-nitroso deriva-
tive 19. Reduction of the nitroso group with Na2S2O4 in dilute
aqueous ammonium hydroxide solution19 gave 5,6-diamino-
1-(2,4-dimethoxybenzyl)-3-propyluracil (8), and 3-(2,4-
dimethoxy)benzyl-1-propylxanthine (21) was obtained by
(20) Weygand, F.; Steglich, W.; Bjarnason, J.; Akhtar, R.; Chytil, N.
Chem. Ber. 1968, 101, 3623–3641.
(21) Herzog, H.; Koch, H.; Scharf, H.-D.; Runsink, J. Tetrahedron 1986,
42, 3547–3558.
(26) For comprehensive reviews on organometal-mediated C-C bond-
forming reactions, see: Hocek, M. Eur. J. Org. Chem. 2003, 24, 5–254.
Schro¨ter, S.; Stock, C.; Bach, T. Tetrahedron 2005, 61, 2245–2267.
(27) Priego, E. M.; Camarasa, M. J.; Perez-Perez, M. J. Synthesis 2001,
3, 478–482.
(22) Traube, W. Ber. 1900, 33, 3035–3056.
(23) Lange, G. L.; Decicco, C. P.; Willson, J.; Strickland, L. A. J. Org.
Chem. 1989, 54, 1805–1810
(24) Yu, J.-Q.; Corey, E. J. J. Am. Chem. Soc. 2003, 125, 3232–3233
(25) Catino, A. J.; Forslund, R. E.; Doyle, M. P. J. Am. Chem. Soc.
.
.
(28) Khalafi-Nezhad, A.; Zare, A.; Parhami, A.; Soltani Rad, M. N.
ARKIVOC 2006, xii, 161–172.
2004, 126, 13622–13623
.
4268
Org. Lett., Vol. 11, No. 19, 2009

3-ethoxy-2-cyclopenten-1-one,30 were unsuccessful. We as-
sumed that this was due to interferences of the organostan-
nane with the free imidazole NH function of the xanthine
core. A similar observation has been described31 during the
synthesis of 8-alkynyladenine analogues by the cross-
coupling of 8-bromoadenine with alkynes.31 These authors
concluded that preparation of the 8-alkynyladenine analogues
was difficult by direct coupling and suggested that efficient
coupling requires protecting N-9. Accordingly, N-7 of
xanthine 26 was reprotected with a Pom group, giving key
intermediate 27. Finally Pd-catalyzed Stille coupling of 27
with tri-n-butylstannyl-2-cyclopenten-1-one gave 28, which,
upon deprotection in methanolic ammonia, supplied com-
pound 2.
On HPLC the retention times and LC-MS fragmentation
spectra of M1 and 2 coincided exactly (parts A and B of
Figure 1, respectively).
Figure 1.
(A) HPLC analysis of the metabolites of [18F]CPFPX in
Experiments examining the metabolism of M1 by HLM
showed little or no consumption of this substrate after 4 h.
This was demonstrated in control experiments (n ) 3) by
comparison of the areas of the respective M1-UV signals at
t ) 0 with those obtained at t ) 4 h (using Student’s t test
the null hypothesis is significant with p ) 0.01). Under
identical conditions, the consumption of CPFPX used as a
control was ∼90%. Thus, M1 appears to be at best a very
poor substrate for further metabolism by HLM. This resis-
tance of M1 to the most important enzymes of phase I
metabolism, together with its relatively high lipophilicity
(LogPcalc ) 1.45, calculated with “Marvin” from http://
www.chemaxon.com), which would slow its elimination
from plasma more than those of more polar metabolites (e.g.,
log Pcalc values for monohydroxylated cyclopentyl derivatives
range from 0.4 to 0.6)), might account for why M1 is the
major metabolite in plasma.
human plasma 30 min postinjection (bottom trace) and of synthetic
compound 2 (top trace). (B) ESI-MS “in-source” fragmentation
spectra of M1 generated by HLM (bottom trace) and of synthetic
compound 2 (top trace).
condensing the diaminouracil 8 with formic acid29 to afford
carboxamide 20, which was not isolated but directly sub-
jected to alkaline ring closure. Alkylation with pivaloy-
loxymethyl chloride (Pom-Cl) protected N-7 of xanthine 21
to give 22, which then underwent debenzylation by treatment
with neat trifluoroacetic acid to form 23. Subsequent alky-
lation of N-3 by 1-bromo 3-fluoropropane gave 24. Attempts
to brominate compound 24 at the 8-position under a variety
of conditions were unsuccessful, most probably due to the
sterically hindering and electron-withdrawing Pom group that
prevented electrophilic bromination of C-8. Thus, the Pom
group was removed by alkaline hydrolysis yielding 3-(3-
fluoropropyl)-1-propylxanthine 25, which was brominated
with bromine and sodium acetate in acetic acid to generate
bromoxanthine 26 in excellent yield. All attempts to perform
a Pd-catalyzed Stille cross-coupling between compound 26
and tri-n-butylstannyl-2-cyclopenten-1-one, prepared from
Supporting Information Available: Detailed descriptions
of experimental procedures and spectroscopic data for
compounds 2, 3, 8, and 14-28.This material is available free
of charge via the Internet at http://pubs.acs.org.
OL900169F
(30) Laborde, E.; Leshaski, L. E.; Kiely, J. S. Tetrahedron Lett. 1990,
31, 1837–1840.
(29) Kramer, G. L.; Garst, J. E.; Mitchell, S. S.; Wells, J. N. Biochemistry
1977, 16, 3316–3321.
(31) Koyama, S.; Kumazawa, Z.; Kashimura, N. Nucleic Acids Symp.
Ser. 1982, 11, 41–44.
Org. Lett., Vol. 11, No. 19, 2009
4269