Radiosynthesis and Evaluation of [11C]3-Hydroxycyclopent-1-enecarboxylic Acid as Potential PET Ligand for the High-Affinity γ-Hydroxybutyric Acid Binding Sites

Article abstract of DOI:10.1021/acschemneuro.6b00335

γ-Hydroxybutyric acid (GHB) is an endogenous neuroactive substance and proposed neurotransmitter with affinity for both low- and high-affinity binding sites. A radioligand with high and specific affinity toward the high-affinity GHB binding site would be

Full text of DOI:10.1021/acschemneuro.6b00335

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Letter
pubs.acs.org/chemneuro
Radiosynthesis and Evaluation of [11C]3-Hydroxycyclopent-1-
enecarboxylic Acid as Potential PET Ligand for the High-Affinity
γ‑Hydroxybutyric Acid Binding Sites
Claus H. Jensen,§ Hanne D. Hansen,† Tina Bay,§ Stine B. Vogensen,§ Szabolcs Lehel,‡ Louise Thiesen,§
Christoffer Bundgaard,∥ Rasmus P. Clausen,§ Gitte M. Knudsen,† Matthias M. Herth,§,†,‡
Petrine Wellendorph,§ and Bente Frølund*,§
§Department of Drug Design and Pharmacology, Faculty of Health and Medical Sciences, University of Copenhagen, 2100
Copenhagen, Denmark
†Neurobiology Research Unit and Center for Integrated Molecular Brain Imaging, Rigshospitalet and University of Copenhagen,
Blegdamsvej 9, 2100 Copenhagen, Denmark
‡PET and Cyclotron Unit, Copenhagen University Hospital, Rigshospitalet, Blegdamsvej 9, 2100 Copenhagen, Denmark
∥Discovery DMPK, H. Lundbeck A/S, Ottiliavej, 2500 Valby, Denmark
S
* Supporting Information
ABSTRACT: γ-Hydroxybutyric acid (GHB) is an endogenous neuroactive substance and
proposed neurotransmitter with affinity for both low- and high-affinity binding sites. A
radioligand with high and specific affinity toward the high-affinity GHB binding site would be a
unique tool toward a more complete understanding of this population of binding sites. With its
high specific affinity and monocarboxylate transporter (MCT1) mediated transport across the
blood-brain barrier in pharmacological doses, 3-hydroxycyclopent-1-enecarboxylic acid
(HOCPCA) seems like a suitable PET radiotracer candidate. Here, we report the 11C-labeling
and subsequent evaluation of [11C]HOCPCA in a domestic pig, as a PET-radioligand for visualization of the high-affinity GHB
binding sites in the live pig brain. To investigate the regional binding of HOCPCA in pig brain prior to in vivo PET studies, in
vitro quantitative autoradiography on sections of pig brain was performed using [3H]HOCPCA. In vivo evaluation of
[11C]HOCPCA showed no brain uptake, possibly due to a limited uptake of HOCPCA by the MCT1 transporter at tracer doses
of [11C]HOCPCA.
KEYWORDS: GHB, HOCPCA, GHB binding site distribution, PET, in vitro autoradiography
high-affinity GHB targets in vitro,11 which however remains to
γ-Hydroxybutyric acid (GHB, Figure 1) is an endogenous
be validated by in vivo studies.12 α4 knockout mice have ∼40%
high-affinity GHB binding sites of wild-type mice, which leave
neuroactive substance that is present in micromolar concen-
trations in the mammalian brain.1 The primary source of GHB
the identity of the remaining 60% of the high-affinity binding
sites elusive.5,11 So far, all these binding studies have exclusively
been performed in vitro.
Positron emission tomography (PET) is a noninvasive
technology and has been highly useful to quantify neuro-
receptor binding in vivo.13 In order to attain a more complete
understanding of the identity and physiological relevance of
high-affinity GHB binding sites in the CNS, the availability of a
suitable PET radiotracer for visualization of these sites would
be of significant value.
We have previously described the conformationally restricted
GHB analogue 3-hydroxycyclopent-1-enecarboxylic acid
(HOCPCA, 1, Figure 1) as a highly selective ligand for the
high-affinity GHB binding site (Ki 16 μM)10 (27 times higher
affinity than GHB itself), and, more importantly, with no
is believed to be metabolic derivation from the major inhibitory
neurotransmitter γ-aminobutyric acid (GABA) (Figure 1).2
Additionally, GHB is a recreational drug (Fantasy or liquid
ecstacy), but also a clinically prescribed drug for treatment of
alcohol dependence (Alcover)3 and narcolepsy (Xyrem, sodium
oxybate).4 In the central nervous system (CNS), GHB binds to
specific high-affinity binding sites that are abundantly expressed
and conserved through evolution.5 The functional role of these
has been a matter of thorough investigation for several years.5
Although many of the observed in vivo pharmacological effects
of GHB are mediated by the metabotropic GABAB receptors,
the high-affinity GHB binding sites are preserved in brains of
GABAB(1) knockout mice, providing evidence that the high-
affinity GHB binding sites are molecularly distinct from the
GABAB receptor complex.6 Efforts to uncover their molecular
identity are of high interest and have so far been largely driven
by the advent of nanomolar affinity and highly selective
Received: September 30, 2016
Accepted: November 14, 2016
Published: November 14, 2016
compounds and radioligands.7−10 Recently, the α4β1−3
δ
ionotropic GABAA receptors have been identified as potential
© 2016 American Chemical Society
22
DOI: 10.1021/acschemneuro.6b00335
ACS Chem. Neurosci. 2017, 8, 22−27

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
ACS Chemical Neuroscience
Letter
affinity for the GABAB receptor and 45 other neurotargets.14
Additionally, it has recently been shown that 1 penetrates into
the brain by means of the monocarboxylate transporter
MCT1.14,15 Given the close structural relationship to GHB,
its selectivity and in vivo brain penetration, 1 is a highly
attractive compound to investigate GHB pharmacology
mediated by high-affinity binding sites. Recently, we reported
the successful tritium labeling of 1 in high radioactive yield.14
Binding of [3H]1 could be displaced by GHB and 1 in a
concentration-dependent manner and not by the GABAB
receptor agonist baclofen (Figure 1), which makes [3H]1 a
promising radio-pharmacological tool to study exclusively the
high-affinity GHB binding sites.14 Additionally, we reported a
new preparative synthetic route to 1 that can be scaled up with
a high and reproducible overall yield.14 The synthetic route can
be developed further for isotopic carbon labeling.
Scheme 1. Synthesis of [11C]1 Using an Organometallic
Approach
a
a
Reagents and conditions: (a) (i) Br2PPh3, benzene, Et3N, rt, 18 h, (ii)
NaBH4, CeCl3, MeOH, 0 °C to rt, 2.5 h, (iii) TBDMS-Cl, DMAP,
imidazole, CH2Cl2, rt, 16 h; (b) iPrMgCl, iPrMgBr, or EtMgBr, THF,
Et2O, 1,4-dioxane or HMPA, 0 °C to 50 °C, 2−24 h.
exchange rate.17 However, based on the total absence of
halogen-metal exchange we decided to try out other strategies.
Recently, a method for Cu-mediated [11C]CO2 fixation for
the [11C]carboxylation of boronic acid esters was reported.16,18
Compared to organolithium and Grignard reagents, boronic
acid esters are in general less sensitive to air and moisture and
therefore attractive as precursors for 11C-carboxylation.18 In
order to use this approach, the pinacol boronic ester 6 was
synthesized as outlined in Scheme 2.
Similar to 2, 1,3-cyclopentanedione was brominated followed
by a Luche reduction. Because of the short half-life (t1/2 = 20.4
Scheme 2. Synthesis of [11C]1 and Optimization of the
a
Reaction Conditions for the Carbonylation of 6
Figure 1. Chemical structures of the two neuroactive substances GHB
and GABA. HOCPCA (1) binds selectively to the high-affinity GHB
binding site. Baclofen is a selective GABAB receptor agonist.
Here, we report the 11C-labeling of 1 using Cu-mediated
[11C]carbon dioxide fixation, examine [3H]1 in in vitro
autoradiography, and subsequently evaluate [11C]1 as a
potential PET radioligand in a domestic pig.
RESULTS AND DISCUSSION
■
A high yielding and reproducible synthesis of 1 has recently
been reported by Vogensen et al.,14 where the carboxylic group
was incorporated by halogen-metal exchange followed by
CO2(g). This synthetic strategy seems applicable for late-stage
isotopic carbon labeling of 1, since [11C]CO2(g) can be
employed. Carboxylation of Grignard or organolithium
reagents with [11C]CO2(g) represents a direct route to 11C-
labeled carboxylic acids. However, applications of this approach
have several caveats due to the high reactivity of the
organometallic reagents. Rigorous exclusion of atmospheric
moisture and CO2 during storage and manipulations is
necessary to obtain radiolabeled products with consistently
high radiochemical yields and specific activities.16 Initially, the
more stable Grignard reagent was pursued rather than the
lithium species using the previously reported strategy.14
Bromination of 1,3-cyclopentanedione followed by Luche
reduction provided the alcohol, which was immediately
protected as the tert-butyldimethylsilyl ether (TBDMS) to
give 2 (Scheme 1).
a
(a) Reagents and conditions: (a) (i) Br2PPh3, benzene, Et3N, rt, 18 h,
(ii) NaBH4, CeCl3, MeOH, 0 °C to rt, 2.5 h; (b) BOM-Cl, DIPEA,
CH2Cl2, rt, 16 h; (c) t-BuLi, THF, −78 to −50 °C, 2 h, − 78 °C,
isopropyl pinacol boronate, 1 h; (d) conc. HCl/EtOH 1:1, rt, 3 min.
(b) Unless otherwise noted, [11C]CO2 was trapped in the reaction
mixture at −10 °C. (c) Determined by radioHPLC integration of
peaks from product, byproducts, and unreacted [11C]CO2 complex.
(d) [11C]CO2 was collected in the reaction mixture at rt.
Abbreviations: RCY, radiochemical yield; CuTc, copper(I)-thio-
phene-2-carboxylate; TBAT, tetrabutylammonium difluorotriphenylsi-
licate; TMEDA, tetramethylethylenediamine; NMP, N-methyl-2-
pyrrolidone; M, molar; equiv, equivalent.
Compound 2 was treated with several commercial Grignard
reagents. No halogen−metal exchange was observed in any of
the cases. Neither higher temperature, change in solvent, nor
prolongation of the reaction time afforded any halogen−metal
exchange either. The addition of lithium chloride to Grignard
reagents has been reported to increase halogen−metal
23
DOI: 10.1021/acschemneuro.6b00335
ACS Chem. Neurosci. 2017, 8, 22−27

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
ACS Chemical Neuroscience
Letter
min) of carbon-11, an easily removable protecting group for the
alcohol moiety was needed. Benzyloxymethyl acetal (BOM),
which readily cleaves under acidic conditions,19 was considered
as a suitable protection group for this purpose. Halogen-metal
exchange using tert-BuLi followed by addition of isopropyl
pinacol boronate provided 6.
Due to the high polarity of 1, purification of [11C]1 was
troublesome. Solid phase extraction with neither an ion-
exchange light Sep-Pak nor a normal phase Sep-Pak could
retain the byproduct in sufficient amount. Only by using two
solid-phase light C-18 Sep-Pak extraction column in a series,
purification of [11C]1 was achieved but half of [11C]1 was lost.
The pH was adjusted to 7 with 5 M NaOH. Average specific
activity was around 1.5 GBq/μmol (range 1−2 GBq/μmol) (n
= 2) with a radiochemical purity above 97%. Typically, an
amount of 831−1012 MBq could be isolated by using a
cyclotron beam time of 40 min.
The synthesis of [11C]1 was initially attempted using the
conditions reported by Riss et al.18 Cyclotron generated
[11C]CO2 was bubbled into a reaction vial containing the
pinacol boronic ester precursor 6, N,N-dimethylformamide
(DMF), copper(I) iodide (CuI), potassium fluoride (KF),
4,7,13,16,21,24-hexaoxa-1,10-diazobicyclo[8.8.8]hexacosane
(Crypt-222), and N,N,N′,N′-tetramethylethylenediamine
(TMEDA). Unfortunately, all attempts to produce the desired
[11C]1 using this approach were unsuccessful, and only
unlabeled 6 and unreacted [11C]CO2 were observed.
To investigate the regional binding of 1 in different brain
regions in the pig prior to in vivo PET studies, in vitro
autoradiography was performed using [3H]1 on pig brain
sections at both pH 6.0 and pH 7.4 (Table 1 and Figure 2).
Second, we attempted 11C-labeling of 1 using the conditions
reported by Rotstein et al.16 (Scheme 2, entry 1). By using the
more soluble copper(I) thiophene-2-carboxylate (CuTC) and
replacing the KF/Crypt-222 with the bench-stable tetrabuty-
lammonium difluorotriphenylsilicate (TBAT), formation of the
radiolabeled carboxylic acid 7 was observed. Furthermore, to
improve the catalyst and precursor solubility, the solvent was
changed from DMF to N-methylpyrrolidinone (NMP).
However, radiochemical yield remained suboptimal (Scheme
2, entry 1), so we turned to alternative conditions to improve
the 11C-labeling of 6. Higher temperature (Scheme 2, entry 2)
was unable to afford greater yield as the release of [11C]CO2
presumably increased. Lowering the temperature (Scheme 2,
entry 4) only markedly lowered the yield. Significant
improvements were realized by increasing the concentration
of the precursor (Scheme 2, entry 5) in the reaction mixture
from 0.1 to 0.2 M, which led to a moderate conversion to the
labeled product 7. Furthermore, an increased precursor
concentration combined with an increase in the addition of
catalyst (CuTC) and additive (TBAT) (Scheme 2, entry 6)
provided an analytical radiochemical yield of 28% determined
by radioHPLC.
Table 1. Quantifications of Specific Binding (fmol/mg) of
[3H]1 (4.5 nM) As Determined from Autoradiography in the
Pig Brain at pH 6.0 and 7.4
region
pH 6.0
pH 7.4
cortical regions
prefrontal cortex
frontal cortex
temporal cortex
cingulate cortex
occipital cortex
striatum
165.1
224.3
217.7
176.2
191.5
99.6
44.94
50.11
59.44
28.32
22.52
lateral septal nucleus
thalamus
322.7
54.27
128.6
12.47
42.95
central amygdala
hippocampal regions
CA1
284.9
233.4
254.8
318.5
383.9
14.3
53.61
32.72
42.17
52.05
70.06
6.89
CA2/3
dentate gyrus
parahippocampal gyrus
subiculum
cerebellum
Based on the optimized conditions, 7 was synthesized and
isolated. Briefly, after bubbling [11C]CO2 into the reaction
mixture, the vial was sealed and the mixture was heated to 100
°C for 5 min. The reaction was quenched with 0.1% aqueous
phosphoric acid, and radiolabeled 7 was isolated by semi-
preparative HPLC followed by trapping on a solid-phase light
C-18 Sep-Pak extraction column. Radiolabeled 7 was eluted
from the C-18 Sep-Pak using EtOH, which afforded pure 7.
Preliminary experiments revealed that diluted aqueous
solutions of TFA or HCl were sufficiently effective for the
removal of the BOM-protecting group. However, under
radioactive conditions, TFA appeared insufficient for the
deprotection even with the use of concentrated TFA at 80
°C. The use of HCl proved to be more desirable as fast
conversion of 7 was observed. Treating 7 with concentrated
HCl (12.4 M) at 80 °C for 3 min provided a 4:1 mixture of 7
and byproduct, presumably the eliminated product. Using a 3
M solution of HCl markedly increased the formation of [11C]1,
and by lowering the temperature to 60 °C, the formation of
byproduct was slightly lowered. However, we concluded that
the optimal conditions for cleavage of the BOM-protection
group were concentrated HCl at room temperature for 3 min at
atmospheric pressure which provided a 9:1:0 ratio of [11C]1,
byproduct and 7, respectively.
As shown in Figure 2, a high degree of specific binding was
obtained with [3H]1, which permitted elucidation of binding
levels in various brain regions. From this experiment, the
distribution of [3H]1 binding sites in pig brain in vitro indicates
that the frontal cortex, septal nucleus, and hippocampus contain
a high density of high-affinity GHB binding sites. The thalamic
regions, amygdala, and caudate putamen have intermediate
levels of binding sites (Table 1 and Figure 2), whereas
cerebellum and hypothalamic regions display low levels of
binding sites. The regional distribution pattern is very similar to
that reported for rodents, using selective high-affinity radio-
ligands [3H]NCS-382 or [125I]BnOPh-GHB as well as for [3H]
1 itself.7,11,20,21 Thus, the distribution of high-affinity GHB
binding in vitro proves to be highly conserved across species.
Interestingly, the absolute binding of [3H]1 to pig brain was 4−
5 times higher at pH 6.0 compared with pH 7.4. The binding of
GHB to the high-affinity GHB binding sites has been reported
to be pH-dependent, with an optimum around pH 6.0.22
However, both the physiological role of the binding optimum at
pH 6.0 and the molecular explanation for this is yet unknown.
It could be either a consequence of local changes in the binding
pocket or allosteric changes in the protein due to protonation
of specific amino acid residues.
24
DOI: 10.1021/acschemneuro.6b00335
ACS Chem. Neurosci. 2017, 8, 22−27

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
ACS Chemical Neuroscience
Letter
Figure 3. (A) Time−activity curves for [11C]1 in the indicated regions
of the pig brain. (B) Summed PET image (0−90 min, sagittal view) of
the pig brain. (C) Corresponding MR-image of the pig brain. SUV:
Standardized uptake value (g/mL).
both brain and plasma.15 Given the conserved role for MCT1
in transporting L-lactate in pigs and the localization at the BBB,
it is anticipated that 1 is also actively transported in pig cells in
a similar fashion. Under this assumption, a primary concern is
the Km of 1 for MCT1 (16.3 mM at pH 7.4 at recombinant
MCT1 expressed in Xenopus oocytes).15 Carrier-mediated
transport across the BBB follows Michaelis−Menten kinetics
plus a nonsaturable component that increases linearly with
concentration. Assuming Michaelis−Menten kinetics alone and
an estimated plasma concentration of [11C]1 of 240 nM, it
means that only a relative transporter velocity of 0.002% was
reached in the current experiment. Thus, the most obvious
approach to try to obtain better brain penetrance would be to
increase transporter activity. This can be attained either by
increasing the dosage of [11C]1 or by decreasing the
extracellular pH (Km for 1 at MCT1 decreases at more acidic
pH values).15 Increasing the dosage of [11C]1 is not an option
since that would violate tracer kinetics. Acidification is routinely
done in vitro, but has also been attempted in vivo by inducing
hypercapnia in humans to study the pH dependency of lactate
BBB permeability. Here, results showed increased permeability,
though this was not significant.25 Despite active transport into
the brain, the total brain exposure of 1 could be adversely
affected by transporter-mediated efflux. Various transporters
have been found on the BBB that acts as efflux transporters, but
no reports have yet addressed this issue for 1. If transporter-
mediated efflux of 1 is involved, inhibition of the relevant efflux
transporter could be an alternative approach to enhance brain
exposure. Indeed, this approach has been studied for diclofenac
and mefenamic acid, where brain penetration seems to be
enhanced by inhibition of the organic anion efflux transporter
3.26 Here, the potential efflux of 1 mediated by organic acid
transport carriers was investigated in mice by coadministering 1
with increasing doses of the inhibitor probenecid. As shown in
Figure 4, the extent of brain penetration of 1 was not influenced
by this inhibitor. Thus, if this translates to pigs, this indicates
Figure 2. Representative autoradiograms of [3H]1 (4.5 nM) binding
to pig brain sections (20 μm) at pH 6.0 (top). Nonspecific binding
was established in the presence of 1 mM GHB (bottom). Coronal
sections showing: prefrontal cortex (Pfx), frontal cortex (Fcx),
temporal cortex (Tcx), occipital cortex (Occ), cingulate cortex
(Cin), putamen (Put), caudate (Cau), thalamus (Tha), amygdala
(Amy), and hippocampus (Hip); and sagittal section of cerebellum
(Cb) as guided by the stereotaxic atlas of Fel
́
ix et al.23
Using the chemistry delineated in Scheme 2, [11C]1 was
prepared for evaluation in in vivo PET imaging studies in a
domestic pig using a high resolution research tomography
(HRRT) PET scanner. The summed PET images (Figure 3B)
and time−activity curves (Figure 3A) showed that the
radioligand [11C]1 did not enter the pig brain. A tendency to
slow rising activity curves can be seen in Figure 3A, which could
indicate radiometabolites. However, no information on radio-
tracer metabolism is available in this specific case.
The lack of detectable brain penetrance of [11C]1 in the live
pig was at first sight rather surprising given the known brain
penetrance of 1 into the mouse brain at an oral dose of 10 mg/
kg.14 The first obstacle in terms of getting brain exposure is the
blood-brain barrier (BBB). Being 99% charged at physiological
pH, little passive diffusion of 1 is expected. This is supported by
studies using MDCK-MDR1 cells showing low passive
permeability toward 1.14 Indeed, it has been reported that
both GHB and 1 enter the brain via active transport by MCT1
at the BBB, at least in rodents.15,24 In the same study, 1 showed
a rapid absorption, a B/P ratio of 0.4 and a first order
elimination profile with a half-life of approximately 20 min for
25
DOI: 10.1021/acschemneuro.6b00335
ACS Chem. Neurosci. 2017, 8, 22−27