New Synthesis and Tritium Labeling of a Selective Ligand for Studying High-affinity γ-Hydroxybutyrate (GHB) Binding Sites

Abstract

3-Hydroxycyclopent-1-enecarboxylic acid (HOCPCA, 1) is a potent ligand for the high-affinity GHB binding sites in the CNS. An improved synthesis of 1 together with a very efficient synthesis of [³H]-1 is described. The radiosynthesis employs in situ generated lithium trimethoxyborotritide. Screening of 1 against different CNS targets establishes a high selectivity and we demonstrate in vivo brain penetration. In vitro characterization of [³H]-1 binding shows high specificity to the high-affinity GHB binding sites.

INTRODUCTION

γ-Hydroxybutyrate (GHB, Figure 1) is an endogenous substance that is present in micromolar concentrations in the mammalian brain. It is a metabolite of the major inhibitory neurotransmitter γ-aminobutyric acid (GABA) (Figure 1), but is also believed to be a neuromodulator on its own¹ although the specific function of GHB remains to be elucidated. GHB is administered orally for the treatment of narcolepsy (sodium oxybate)² and alcoholism.³ GHB is tasteless and odorless and combined with its central nervous system (CNS) effects (mild euphoria, sedation, respiratory depression and eventually coma in high doses) it is a hazardous drug of abuse (fantasy). In the CNS, GHB binds to both low-affinity and high-affinity binding sites.¹ The low-affinity site is believed to be identical to the GABA orthosteric site at the metabotropic GABA_B receptor as high dose effects of GHB (micromolar in the brain) is prevented by preincubating with a GABA_B receptor antagonist³ and several prominent GHB-induced effects are absent in GABA_B knockout mice.⁴ However, when comparing the effect of GHB with that of the GABA_B agonist baclofen (Figure 1) in, e.g., c-Fos expression studies⁵ and behavioral studies,⁶ it is clear that the two compounds have different pharmacological profiles. Thus, GHB effects may also be mediated by non-GABA_B receptors. The fact that GHB high-affinity binding sites are preserved in brains of GABA_B knockout mice,⁴ make them likely receptor candidates and has prompted the uncovering of their identity. Using photoaffinity labeling combined with proteomic analysis and molecular pharmacology studies, we recently reported that α₄β_1-3δ ionotropic GABA_A receptors correlate with the GHB high-affinity sites.⁷ In order to acquire a more complete understanding of the physiological relevance of GHB high-affinity binding sites in the CNS, and particularly the nanomolar effects mediated by GHB at the α₄β₁δ receptors, better tool compounds and further pharmacological investigations are needed.

Figure 1.

Chemical structures of the two neuroactive substances GHB and GABA. HOCPCA (1) and NCS-382 bind selectively to the high-affinity GHB binding site. Baclofen is a selective GABA_B agonist.

We have previously described 3-hydroxycyclopent-1-enecarboxylic acid (HOCPCA, 1, Figure 1) as a selective ligand for the high-affinity GHB binding site (27 times higher affinity than GHB itself), and, importantly, with no affinity for the GABA_B receptor. Given the close structural relationship to GHB and its relatively small size, 1 is a highly attractive compound to investigate GHB-like pharmacology. We therefore decided to explore synthetic routes for incorporation of hydrogen and carbon isotopes into 1, and at the same time provide a more feasible and preparative synthesis of 1, which is needed to facilitate in vivo studies. At present [³H](E,RS)-(6,7,8,9-tetrahydro-5-hydroxy-5H-benzocyclohept-6-ylidene)acetic acid ([³H]-NCS-382) is the standard radioligand in binding experiments for the high-affinity GHB binding site.⁸ NCS-382 (Figure 1) is a putative antagonist⁸ at the high–affinity GHB binding site with a K_i 14 times lower than that of GHB.⁹ The higher affinity of 1 compared to NCS-382,⁹ and closer structural relationship to GHB underscores the relevance of developing a better preparative and radiolabeling procedure. This will provide a better alternative to the only available radiolabeled agonist [³H]GHB, which has several caveats, both in terms of selectivity due to its affinity for the GABA_B receptor, and in terms of its susceptibility to metabolism and uptake.¹⁰

Here, we describe a highly improved preparative synthesis of 1 and the synthesis of tritium labeled 1. Furthermore, the stability of 1 and [³H]-1 are evaluated, binding competition studies using [³H]-1 are presented and the selectivity profile of 1 is investigated. Finally, the brain penetration of GHB, 1 and NCS-382 is investigated in vivo.

RESULTS AND DISCUSSION

Improved synthesis of 1

The previously⁹ reported synthesis of 1 is outlined in Scheme 1. Reduction of the starting material (2) followed by mesylation and elimination gave α,β-unsaturated ethyl ester (3a) in 53% yield.

Scheme 1. Previous synthesis route to 1 and its carbonyl precursor 9^a.

^aReagents and conditions: (a) (i) NaBH₄, EtOH, 0 °C – rt, 2.5h, (ii) CH₃SO₂Cl, Py, rt, 3h, (iii) Et₃N, CHCl₃, reflux, 23h; (b) CrO₃, Ac₂O, AcOH, rt, 30 min; (c) NaBH₄, CeCl₃, MeOH, 0°C, 30 min; (d) Na₂CO₃, 50 °C, 16h; (e) 1-propanol, H₂SO₄, reflux, 1h.

Oxidation of 3a with 3 eq. of CrO3 gave 4a in varying yields below 39%. Successful Luche reduction gave 5a and final deprotection with sodium carbonate gave the product 1 as hygroscopic crystals. The overall yield of the 6-step synthesis was 18%. However, this route posed several problems as a preparative synthesis, i.e. varying yields, tedious purification in particular in the oxidation of 3a, and a hygroscopic end product that is unstable during storage.

In light of the potential of 1 as a pharmacological tool for future in vivo and ex vivo studies, we wanted to develop a more suitable synthesis of 1 that could circumvent these obstacles and generate large quantities of 1. The new synthetic route to 1 is outlined in Scheme 2. First 1,3-cyclopentanedione was brominated with dibromo-triphenylphosphine and triethylamine.¹¹ It was crucial for a high yield to use freshly distilled triethylamine dried over LiAlH4, whereas sublimation (110 °C, 0.2 mbar) of the commercial brownish 1.3-cyclopentanedione to white crystals did not improve the yield. Luche reduction with NaBH₄ in methanol gave the alcohol that was immediately TBS-protected to give 6 in 86% overall yield.

Scheme 2. New synthesis route to 1^a.

^aReagents and conditions: (a) (i) Br₂PPh₃, benzene, Et₃N, rt, 18h, (ii) NaBH₄, CeCl₃, MeOH, 0°C – rt, 2.5h, (iii) TBS-Cl, DMAP, imidazole, DCM, rt, 16h; (b) t-BuLi, THF, −78 – −50 °C, 2h, −78 °C, CO₂(g), 20 min; (c) H₂SiF₆ (aq 23%), CH₃CN, rt, 1h.

Metal-halogen exchange of 6 using t-BuLi followed by addition of CO₂ (g) afforded the carboxylic acid 7 in 78% yield. Finally, deprotection of the TBS-protected alcohol using an aqueous solution of H₂SiF₆ as described by Pilcher and DeShong¹² gave 1 as white crystals in 75% yield upon recrystallization. The product was stable as the free acid as well as the sodium-salt. This new synthetic route is high yielding (56% overall), includes more simple purifications and produces a non-hygroscopic stable product without the use of chromic acid. The route is also applicable for late-stage isotopic carbon labeling of 1, since labeled CO₂(g) can be employed when incorporating the carboxylic acid group. Alternatively, the bromide 6 can be transformed to the corresponding pinacol boronic ester and treated with [¹¹C]CO₂(g) in the presence of CuI/crypt-222/KF as recently described by Riss et al.¹³ On the other hand, the route is not stereoselective and provide both enantiomers.

Tritium labeling

The previously reported synthesis of 1 outlined in Scheme 1 was a good starting point for a route to tritium labeled 1 as reduction of the α, β-unsaturated ketone (4) enabled introduction of tritium late in the synthesis. Still some optimization of this route was feasible. 3b is commercially available but volatility of both 3b and 4b compromised the yield in the first step and 3b was therefore transesterified to the propylester 3c in 70% yield. Attempts to replace the excess amount of chromic acid with KMnO₄¹⁴ or TBHP/Pd(OH)C¹⁵ were unsuccessful. The use of catalytic amounts of Mn(OAc)₂·4H₂O¹⁶ was at first glance successful (75% yield), but unfortunately an unidentified impurity not visible by NMR, affected the course of the following reduction negatively and the oxidation was therefore performed with chromic acid to afford 4c. The Luche reduction (1 eq. of NaBH₄ in MeOH in the presence of CeCl₃) gave the alcohol 5b in 98% yield (supporting information, Table S1). To prepare [³H]-1 with a maximal specific activity we decided to screen the reaction conditions using borodeuterides prepared in situ from deuterium gas instead of using commercial deuterides. In the literature a broad spectrum of in situ prepared borodeuterides^{17, 18} and tritides,^{19, 20} respectively are described. To suppress the risk of reduction of the double bond and ester group we were especially interested in less reactive borodeuterides. As demonstrated by Zippi et al.¹⁹ LiB(OMe)₃²H (8) can be readily generated from Li²H by addition of trimethoxyborate (Scheme 3). Lithium deuteride is formed by the treatment of a n-BuLi-TMEDA solution under gaseous ²H₂ atmosphere¹⁷ (Scheme 3).

Scheme 3. Synthesis of the lithiumboronic reducing reagent^a.

^aReagents and conditions: (a) TMEDA, 2h, room temperature; (b) B(OMe)₃, 30 min, room temperature.

The yield of in situ generated borodeuteride (8) was determined indirectly using 4-anisaldehyde as a trapping agent in independent reactions. The reaction was carried out under various ratios (1:1, 2:1 and 1:2) of (8) (theoretically generated amounts based on n-BuLi) and anisaldehyde and the yield of 8 was determined by HPLC and ¹H-NMR to be in the range of 40–47% in all three cases. That is in accordance with the yield of Li³H (35%) reported by Zippi et al.¹⁹ For radiation safety reasons during the ³H-labeling of 1, it was desirable to perform the reduction with borotritide and the ester hydrolysis of 4c in one-pot. Therefore, upon lyophilization of the solvent aq. NaOH was added and hydrolysis of the propylester gave [²H]-1 (Scheme 4). The lyophilization was crucial as the yield dropped from 88% to 15% if omitted. Next the reaction conditions for the reduction of 4c (Scheme 4) was investigated (See supporting information Table S1).

Scheme 4. Deuterium and tritium-labeling of 1^a.

^aReagents and conditions: (a) (i) 8 or 9, THF, room temperature, 2h, room temperature, lyophilized (ii) 1M NaOH, 1h, room temperature (iii) 1M HCl.

The protic solvent MeOH of the original experiment could not be exchanged with EtOH in the presence of CeCl₃. Actually the presence of CeCl₃ apparently resulted in an increased amount of hydrolyzed 4c, and could successfully be omitted. In the absence of CeCl₃ both DMF and even the easily lyophilizable solvent THF could be used.

Lowering of the reaction temperature to 0 °C or −78 °C did not improve the deuterated yield nor suppressed the hydrolysis of the ketone. For comparison in situ synthesized Li²H and LiB²H₄ (see supporting information) were also attempted, but resulted in lower yields and more complex reaction mixtures.

Finally, the best conditions (LiB(OMe)₃²H, THF, room temperature) was used to prepare [³H]-1 (Scheme 4). To achieve full conversion of 4c excess amounts (1.8 eq.) of LiB(OMe)₃³H was used. Reduction of 4c gave the ³H-labeled propylester in 95% yield according to HPLC (245 nm). Subsequent hydrolysis gave the desired product [³H]-1 (97%, 245 nm). We determined the amounts of prepared [³H]-1 to 637 mCi with a specific activity of 28.9 Ci/mmol (determined by MS and HPLC/LSC, which accounts for close to 1 tritium/molecule) and a radio chemical purity (RCP) and chemical purity > 99% after purification (HPLC, Figure S11).

Stability

Both 1 and [³H]-1 was subjected to stability testing by HPLC (area%, 245 nm). The rate of decomposition of 1 was tested at both room temperature and −30±2°C in MilliQ water (2 mg/mL). pH of this solution was 3.5. At −30°C compound 1 proved completely stable (1 year and 4 months after formulation) and at room temperature only a minor decomposition of 1 was observed (100% at day 0 to 96% at day 35 and 88% at day 99). The stability of [³H]-1 was investigated in four different formulations for the protection against radiolysis (see supporting information). The best stability of [³H]-1 was found to be a solution of 1 mCi/mL in ethanol or in an aqueous solution of gentisic acid (50 mM).

Binding studies

Binding of [³H]-1 to rat cortical membranes was investigated using assay conditions previously described for [³H]NCS-382⁹ with GHB (1 mM) for determining non-specific binding (Figure 2A). IC₅₀ [pIC₅₀±SEM] values for 1, sodium salt of 1 and GHB were respectively 0.073 μM [7.14±0.02], 0.116 μM [6.93±0.01] and 1.14 μM [5.94±0.06]. This is in relative accordance with values obtained in the comparable [³H]NCS-382 binding assay (0.16 μM and 4.3 μM, respectively).⁹ As expected, [³H]-1 binding was inhibited by the GHB-selective ligand NCS-382 and the GABA_A antagonist gabazine^{7, 21} but notably not by baclofen and GABA at high concentrations.^{7, 9} These data suggest that [³H]-1 selectively labels the high-affinity GHB binding site. The selectivity was further explored in The National Institute of Mental Health’s Psychoactive Drug Screening Program (NIMH-PDSP) (Supplementary Table S2) where 1 was found to be more than 100 times selective for the [³H]-1 binding site over 45 different receptors and transporters. The high selectivity of 1 combined with a ligand efficiency²² of 1.05 kcal/mol establish 1 as a highly interesting compound for further studying the high-affinity GHB binding site.

Figure 2.

(A) Concentration-dependent inhibition of [³H]-1 binding (10 nM) to rat cerebrocortical membranes by GHB and 1. Results are expressed as mean (SD of a single representative experiment performed in triplicate). Three additional experiments gave similar results. (B) Displacement of 10 nM [³H]-1 binding to rat cerebrocortical membranes by 1 mM GHB, 100 μM NCS-382, 1 mM 1, 1 mM GABA, 100 μM gabazine and 100 μM baclofen. Results are expressed as mean CPM (SD of a single representative experiment performed in triplicate) and were repeated at least twice with similar results.

Brain penetration studies

The brain to plasma distribution ratio of GHB, 1 and NCS-382 was investigated acutely in mice 30 minutes after oral administration of 10 mg/kg. All three compounds were found to be brain penetrant reflected by their brain to plasma ratios of 0.20, 0.37 and 0.38 for GHB, 1 and NCS-382, respectively. Considering the carboxylic acid group shared by all three compounds, often associated with restricted passive brain penetration, these results may suggest an involvement of active transport across the blood-brain barrier. This hypothesis was supported by assessment of the in vitro permeability using MDCK-MDR1 cells²³ which showed that GHB, 1 and NCS-382 all exhibited very low passive permeability. Thus, after spiking of test compounds on either the apical or basal side of the cell monolayers, no compound could be detected at the respective receiver sides.

CONCLUSION

In conclusion we have developed a new preparative synthetic route to 1 that can be scaled up with a high and reproducible overall yield. The new synthesis can readily be developed further for carbon labeling. Compound 1 is stable during storage and has a favorable selectivity profile when tested against other relevant targets. Furthermore, compound 1 was successfully tritium labeled using in situ generated lithium trimethoxy-borotritide. The incorporation of one tritium atom proceeded in high radioactive yield, with close to theoretical specific activity and with high radiochemical purity. Binding of [³H]-1 could be displaced by GHB and 1 in a concentration-dependent manner and not by the GABA_B receptor agonist baclofen. This new radioligand will be useful for further understanding the molecular mechanism of GHB at its high-affinity binding site and α4βδ GABA_A receptors and for elucidating the pharmacology of the GHB system in the mammalian CNS, given the brain permeability of 1, for instance through ex vivo or in vivo labeling of the binding sites with positron-emitting or other isotopes.

EXPERIMENTAL SECTION

[³H]-3-Hydroxycyclopent-1-enecarboxylic acid ([³H]-1)

Career free ³H₂ (13.8 Ci) trapped on a uranium bed (as uranium tritide) was released by heating (500 °C) and directed in a tritium manifold into a dried reaction vial (1 mL 2 necked round bottom flask). N,N,N′,N′-Tetramethylethylenediamine (50 μL) was added. n-Butyl lithium (100 μL, 160 μmol, 1.6M in hexane) was added drop wise and within 10 min a white precipitate of Li³H was formed. After 2 h of vigorous stirring trimethyl borate (18 μl, 160 μmoles) was added and the reaction was stirred for 30 min to generate LiB(OMe)₃³H. A solution of propyl 3-oxocyclopent-1-en carboxylate (4c) (15 mg, 89 μmol) in THF (300 μL) was added drop wise and the reaction was stirred for 2 h. Solvents were lyophilized. Water (500 μL) was added and a sample showed predominantly [³H]-5b by HPLC analysis (95%, 245 nm). Aq. NaOH (2 M, 100 μl) was added and the reaction was stirred for 1 h. The reaction mixture was neutralized with 1N HCl to reach pH 6. Water was lyophilized and the solid residue dissolved in accurately amounts of water. 1/10 of the crude mixture was used for qualitative and quantitative analysis (see SI). Preparative HPLC gave [³H]-1 ((97%, 245 nm), 637 mCi, 28.9 Ci/mmol) as white solid. ³H NMR (320 MHz, DMSO-D₆/H₂O 10:1) δ 4.71 (s). MS (ESI): 128.9 (100, M-1).

3-Hydroxycyclopent-1-enecarboxylic acid (1)

3-((tert-butyldimethylsilyl)oxy)cyclopent-1-enecarboxylic acid (3.17 g, 13.08 mol) was dissolved in acetonitrile (130 mL) in a polypropylene vial and H₂SiF₆ (aq, 20–25 %wt, 2.3 mL, 3.9 mmol) was added. The solution was stirred for 1h at room temperature. Sat. Na₂CO₃ (100 mL) and water (50 mL) were added and the aqueous phase was washed with diethyl ether (2 × 20 mL). The aqueous phase was made acidic by the addition of HCl (aq, 4M) and extracted with ethyl acetate (4 × 150 mL). The combined organic phase was dried (MgSO₄) and the solvent removed in vacuo to give a white solid. CC (2:1 heptane/ethyl acetate + 2% AcOH) gave a white solid. Recrystallization (acetonitrile) gave the product as white crystals (1.26 g, 75%). Mp 134–136°C. ¹H NMR (CD₃OD): δ 1.71 (m, 1H), 2.30–2.48 (m, 2H), 2.62–2.70 (m, 1H), 4.86–4.91 (m, 1H), 6.63–6.65 (m, 1H). ¹³C NMR (CD₃OD): δ 30.81, 34.23, 77.65, 139.93, 144.72, 168.62. Anal. Calcd for C₆H₈O₃: C, 56.24; H, 6.29. Found: C, 56.34; H, 6.17. LC-MS (system B): (M-H₂O)H⁺: 111.1.

ABBREVIATIONS

GHB

γ-hydroxybutyrate

HOCPCA

3-Hydroxycyclopent-1-enecarboxylic acid

NCS-382

(2E)-(5-hydroxy-5,7,8,9-tetrahydro-6H-benzo[a][7]annulen)-6-ylidene ethanoic acid