This animal study (n=12) synthesized several different furan-substituted salvinorin A analogs and identified three potent variants which reduced drug-induced reinstatement of cocaine-seeking behavior, without producing the sedation observed with other κ-opioid agonists.
Introduction
The neoclerodane diterpene salvinorin A, found in the leaves of Salvia divinorum, is a potent κ-opioid receptor agonist, making it an attractive scaffold for development into a treatment for substance abuse. Although several successful semisynthetic studies have been performed to elucidate structure-activity relationships, the lack of analogues with substitutions to the furan ring of salvinorin A has prevented a thorough understanding of its role in binding to the κ-opioid receptor.
Methods
Herein we report the synthesis of several salvinorin A derivatives with modified furan rings.
Results
Evaluation of these compounds in a functional assay indicated that sterically less demanding substitutions are preferred, suggesting the furan ring is bound in a congested portion of the binding pocket.
Discussion
The most potent of the analogues successfully reduced drug-seeking behavior in an animal model of drug-relapse without producing the sedation observed with other κ-opioid agonists.
Kappa-opioid receptors (KOR) and their endogenous ligands, the dynorphins, are widely expressed in the central nervous system and have been implicated in pain, drug abuse and mood disorders. Development of KOR-targeted therapeutics has been limited by adverse effects associated with many classical KOR ligands, including dysphoria, sedation and depression, and by pharmacokinetic issues with some antagonists. Earlier work has shown that structurally distinct ligands can produce different signalling outcomes at a single receptor (functional selectivity), suggesting that novel scaffolds might provide beneficial effects with fewer side effects. Riley and colleagues focused on salvinorin A, a neoclerodane diterpene natural product that is a potent, selective KOR agonist but is structurally unlike classical opioids and lacks a basic nitrogen. Although many positions on salvinorin A have been explored by semisynthetic modification, the role of the furan ring (C-15/C-16 positions) in KOR binding and activity remained under-investigated. This study set out to synthesize a series of furan-substituted salvinorin A analogues, characterise their KOR activity in a cellular functional assay, and test selected potent analogues in an animal model of cocaine relapse to determine whether small furan modifications could retain activity while avoiding sedation.
Chemical synthesis began from salvinorin A and employed a selective C-16 bromination protocol using N-bromosuccinimide (NBS) with catalytic Br2 to give an intermediate (2) amenable to further cross-coupling. Suzuki–Miyaura couplings introduced a range of aryl, heteroaryl, alkenyl and alkyl groups at the furan, while Sonogashira couplings were used to install alkynyl substituents; Sonogashira conditions were adjusted or replaced (e.g. potassium propynyltrifluoroborate) when gaseous reagents prevented direct coupling. Other transformations included catalytic hydrogenations, selective reductions, photoredox-mediated trifluoromethylation to give C-15 or C-16 CF3 derivatives, and conversion of furan substituents to aldehydes and nitriles via oxidation to an oxime followed by dehydration. The authors note several preparative challenges: some reactions produced mixtures that resisted improved selectivity or yield, and selective functionalisation at C-15 proved more difficult because bromination favours C-16. In vitro pharmacology used CHO-K1 cells stably expressing human KOR to measure inhibition of forskolin-stimulated cAMP accumulation (DiscoveRx HitHunter cAMP assay). Test compounds were prepared in DMSO and applied in serial dilutions; data were normalised to vehicle and forskolin controls and analysed by nonlinear regression to produce EC50 values and efficacy estimates. Selectivity was assessed using MOR-expressing CHO cells. For in vivo evaluation, male Sprague–Dawley rats were trained to self-administer cocaine (0.5 mg/kg per infusion) on a fixed-ratio 5 schedule, extinguished to a criterion of less than 20 active-lever presses in a session, and then subjected to reinstatement testing after a cocaine prime (20 mg/kg, ip). Test compounds were administered intraperitoneally 10 minutes before the cocaine prime in a vehicle of DMSO:Tween 80:water (2:1:7) at doses of 0.1, 0.3 or 1.0 mg/kg as appropriate. Spontaneous locomotor activity in open-field chambers was measured to assess sedation or motor impairment. Statistical analyses were one-way repeated measures ANOVA with Dunnett’s multiple comparisons for reinstatement and Student’s t test for locomotor activity, with P ≤ 0.05 considered significant.
A diverse set of furan-substituted salvinorin A analogues was prepared using the selective bromination/cross-coupling strategy and complementary transformations; the Conclusions state that 47 novel KOR agonists were synthesised. Many substitutions were tolerated synthetically, though some reactions yielded mixtures or proved difficult to make selective. Notably, the C-2 acetate of salvinorin A remained intact under some aqueous Suzuki conditions, an observation highlighted by the authors. In the KOR functional assay most furan-substituted derivatives retained full agonist efficacy but many showed reduced potency compared with salvinorin A (1). Bromination at C-16 produced little change in potency (EC50 0.030 ± 0.004 nM for 1 versus 0.040 ± 0.010 nM for 2). Conversion of that bromide to a methyl (compound 24) caused an ~11-fold loss in potency (EC50 = 0.41 ± 0.15 nM). Introducing an ethynyl group at C-16 (compound 36) yielded a probe marginally more potent than 1 (EC50 = 0.019 ± 0.004 nM). Larger or branched substituents appended to the terminal position of the alkyne produced large reductions in potency (for example, compounds 28 and 33 showed EC50 values of 26.0 ± 10.0 nM and 9250 ± 260 nM, respectively), although adding a terminal alcohol to an n-propyl terminus produced an approximately 10-fold improvement relative to the nonpolar analogue. Comparison of hybridisation states showed that reduction of an alkyne to an alkene or alkane tended to decrease potency (e.g. vinyl 29, EC50 = 0.96 ± 0.24 nM; reduced 37, EC50 = 2.9 ± 0.6 nM), an effect attributed to increased steric bulk after reduction. Small electron-withdrawing modifications at the furan (aldehyde 49 and nitrile 50) substantially reduced potency relative to geometrically similar alkene/alkyne analogues: 50 was reported as more than 500-fold less potent than 36, and 49 was about 7-fold less potent than 29; 49 and 50 were noted to be nearly equipotent with each other. Replacement of the furan with a phenyl (compound 3) decreased potency versus 1 but produced a still-potent agonist (EC50 = 1.3 ± 0.4 nM). Exploration of substituted phenyl rings and extended π-systems revealed no substitutions that improved potency over 3; overall the steric bulk of substituents appeared to be the dominant factor influencing potency. Within the fluoro- and trifluoromethyl-substituted phenyl series there was a pattern of potency by substitution position (ortho > meta > para). The extracted text does not clearly report complete results for substitutions at C-15 due to regioselectivity limitations in the bromination step. Selectivity profiling indicated that the lead analogues were highly selective for KOR with no measurable activity at MOR in the assays described. In behavioural assays, compounds 2 and 36 significantly attenuated cocaine-prime-induced reinstatement of drug-seeking in male Sprague–Dawley rats; a larger dose of 2 (0.3 mg/kg) was required compared with 36 (0.1 mg/kg). Compound 24 showed a trend toward reduced responding with increasing dose but did not reach statistical significance even at 1.0 mg/kg. Locomotor testing at doses that reduced reinstatement (2 at 1.0 mg/kg, 24 at 1.0 mg/kg, 36 at 0.3 mg/kg) showed no significant reduction in spontaneous activity compared with vehicle, and inactive lever responding during self-administration, extinction and reinstatement was unchanged. These findings indicate that decreased responding in the reinstatement paradigm was not due to overt sedation or motor impairment.
The investigators interpret their data to indicate that the furan ring of salvinorin A occupies a sterically congested region of the KOR binding pocket, since small substituents were generally better tolerated than bulkier groups. The superior potency of small linear substituents such as an ethynyl group and the marked loss of potency seen with larger terminal groups or with substitutions that increase steric bulk after reduction are cited as supporting evidence. Electronic effects were also considered: introduction of electron-withdrawing groups (aldehyde, nitrile) reduced potency relative to geometrically similar alkene/alkyne analogues, suggesting that either electron density of the furan contributes to binding or that other subtle structural changes (for example, terminal hydrogen atoms) account for activity differences. Riley and colleagues note that several furan-modified analogues (notably 2, 24 and 36) retained subnanomolar agonist potency and that 2 and 36 reduced cocaine-primed reinstatement without causing measurable sedation in rats, supporting further investigation of these compounds as leads for drug-abuse therapeutics. The authors acknowledge synthetic limitations encountered, including regioselectivity favouring C-16 bromination that hampered more extensive C-15 functionalisation and several reactions that produced low selectivity or mixtures. They also indicate ongoing studies to further characterise the effects of novel KOR agonists on self-administration, reward-related measures and other potential side effects, but detailed results from those follow-up studies are not presented in the extracted text.
Using selective furan substitution methods, the study produced a library of salvinorin A derivatives (47 novel KOR agonists reported). Functional cAMP assays showed that only relatively small substitutions to the furan are well tolerated, consistent with binding in a sterically congested portion of the KOR pocket. The most potent analogues (2, 24 and 36) attenuated cocaine-induced reinstatement in a rat model at doses that did not produce sedation, identifying them as promising leads for further investigation in drug-abuse therapy.