8-epi-Salvinorin B: crystal structure and affinity at the kappa opioid receptor.

There have been many reports of epimerization of salvinorins at C-8 under basic conditions, but little evidence has been presented to establish the structure of these compounds. We report here the first crystal structure of an 8-epi-salvinorin or derivative: the title compound, 2b. The lactone adopts a boat conformation with the furan equatorial. Several lines of evidence suggest that epimerization proceeds via enolization of the lactone rather than a previously proposed indirect mechanism. Consistent with the general trend in related compounds, the title compound showed lower affinity at the kappa opioid receptor than the natural epimer salvinorin B (2a). The related 8-epi-acid 4b showed no affinity.

Introduction

Salvinorin A (1a), isolated from the hallucinogenic sage Salvia divinorum,[1] is a potent and selective κ opioid receptor (KOR) agonist.[2] Because it is the first known non-nitrogenous compound to have biologically significant actions at mammalian opioid receptors, 1a enables new approaches to studies of endogenous opioid receptor systems. KOR ligands, in particular, have attracted considerable interest because of their effects on mood states.[3-6] Recently, numerous synthetic derivatives of 1a have been prepared and evaluated for activity at opioid receptors. Some potent agonists have been identified which are expected to show increased stability or solubility.[7] Others have increased affinity and potency, [8] or altered subtype selectivity.[9] As yet, however, no derivatives of 1a appear to be KOR partial agonists or antagonists, classes of agents that may have utility in the treatment of psychiatric conditions such as depression or mania.[4-510]

Salvinorins tend to isomerize under basic conditions. Valdés reported that borohydride reduction of 1a gave an unidentified stereoisomeric byproduct, which could be converted to an undetermined stereoisomer of 1a.[11] The latter compound was subsequently identified by Brown as 8-epi-salvinorin A (1b).[12] Brown also reported that deacetylation of 1a under basic conditions gave 8-epi-salvinorin B (2b), but did not characterize either compound. Several further reports of epimerization at C-8 appeared over the following decade, [13-14] but no characterization data was presented. Valdés later identified the byproduct mentioned above as 8-epi-diol 3.[15] Characterization data was given, but the basis of the structure assignment was not stated.

The first structure elucidation of one of these compounds was of 8-epi-salvinorin A (1b).[16] The trans-diaxial H-8 coupling constant found in 1a was absent in 1b, establishing an equatorial configuration. Also, irradiation of H-12 in 1b gave a strong nOe enhancement of H-8. The corresponding experiment on 1a gave instead an enhancement of H-20. These findings can be extrapolated to 2b, since acetylation gives 1b quantitatively.[9,17] Conflicting 1H NMR data for 2b itself were later reported by two groups.[8-9] The 1H NMR spectrum of 2b is reproduced in Supporting Information File 1; the corresponding amended data have been reported previously.[17] Interestingly, epimerization has also recently been reported under acidic conditions.[18]

The epimers can be readily identified by TLC: the unnatural compounds almost invariably spot above the natural compounds in EtOAc/hexanes, and give a blue rather than pink/purple colour when visualized with vanillin.[19] The unnatural epimers are also recognizable by their distinctive H-12 multiplet in 1H NMR, which resembles a broad doublet shifted upfield to ~δ 5.30 ppm. Many 8-epi-salvinorin derivatives have now been reported, although many have not been fully characterized.[7-917-1820-24] Thus, the many reports of 8-epi-salvinorins and derivatives have been based on limited data.

Results and Discussion

The crystal structure presented here (Figure 1) is the first reported for an 8-epi-salvinorin or derivative. It firmly establishes the structure of 2b, and therefore of 1b. The lactone carbonyl C-17 is axial with respect to the B ring (C6-7-8-17 torsion angle 77° versus 173° in 1a).[1] The lactone itself adopts a boat conformation with the furan equatorial (C9-11-12-13 torsion angle 179°). This is as predicted in solution, on the basis of a trans-diaxial coupling constant for H-12.[17] This is also consistent with the crystal structures of furanolactones with all other possible C8/9/12 stereochemistries (trans/anti, trans/syn and cis/syn) – the furan is equatorial in all cases.[17] The rest of the structure is very similar to the crystal structure of 1a.[1] The hydroxyl group participates in an intramolecular hydrogen bond with the ketone (O2-H2···O1, 2.12 Å). There are no intermolecular hydrogen bonds. The asymmetric unit consists of two molecules; the only substantial difference between them is in the rotation of the furan ring (C11-12-13-14 torsion angle -87° (A) versus 53° (B)). The crystals are monoclinic, space group P21 (see Figure 2). The crystallographic data can be found in Supporting Information File 2; the structure factors are in Supporting Information File 3. The crystallographic data have also been deposited with the Cambridge Crystallographic Data Centre (CCDC 626179).[25] 8-epi-Salvinorins and derivatives have a much weaker tendency to crystallize than their natural counterparts. Unsurprisingly, therefore, 2b has a lower melting point (192–196°C) than 2a (239–240°C).[17]

Figure 1

Stereoview of the molecular structure of 2b, showing 50% probability displacement ellipsoids and the atom-numbering scheme. Only one of the two molecules in the asymmetric unit is shown.

Figure 2

Stereoview of the packing of 2b. H atoms are not shown.

Configuration at C-8 is biologically significant. The affinity and potency of 8-epi-salvinorin A (1b) at the KOR are dramatically lower than those of 1a.[16] This finding has been replicated several times.[8-920] The same trend is evident with many salvinorin derivatives: epimerization of active compounds at C-8 reduces affinity and potency.[8-920,23-24] Very few exceptions to this trend have been reported to date.[8,23] These include 8-epi-salvinorin B (2b) itself, whose binding affinity (Ki = 43 nM) was reportedly greater than that of the natural epimer 2a (111 nM).[8] To explore this anomaly, we submitted a new sample of 2b for in vitro testing at the KOR. Binding affinity, potency and efficacy were determined as previously described (Table 1).[26]

Table 1

Affinities (Ki), potencies (EC50), and efficacies at the KOR.

Compound	Ki ± SEM a,b	EC50 ± SEMb,c	Emax ± SEMd
	nM	nM	%
1a	2.4 ± 0.4	1.8 ± 0.5	98 ± 3
2b	304 ± 46	214 ± 33	90 ± 2
4a	>10,000	-	-
4b	>10,000	-	-
U50,488H	2.2 ± 0.3	1.4 ± 0.3	100

aInhibition of [3H]diprenorphine binding to membranes of Chinese hamster ovary cells stably transfected with the human KOR (CHO-hKOR). bMean ± SEM of three independent experiments performed in duplicate. cEnhancement of [35S]GTPγS binding to CHO-hKOR membranes. dRelative to that of U50,488H control.

The binding affinity of 2b (Ki = 304 nM) was lower than those previously reported for salvinorin B (2a) under the same conditions (66, 111 or 155 nM).[7-827] An early report that 2a was inactive employed a different radiolabeled ligand, [3H]bremazocine.[28] Subsequent testing with [3H]diprenorphine by the same group gave concordant values for the relative affinity of 2a.[17] Thus, our data suggest that 2b in fact has a lower affinity than 2a, consistent with the general trend mentioned above. We also reexamined the epimeric acids 4.[16] In a previous report, 4a was found to be inactive (Ki > 1,000 nM), but the 8-epimer 4b showed high affinity at the KOR (49 nM).[23] In contrast, our current samples of both 4a and 4b showed no affinity at the KOR (Table 1).

Given the very high binding affinity of 1a, contamination of an inactive or weakly active compound with even traces of 1a will cause large errors. Flash chromatography in EtOAc/hexanes effectively separates 2b from 2a, but not from 1a. To overcome this, we re-chromatographed our sample in acetone/CH2Cl2, which resolves 2b from 1a, and verified purity by 1H NMR [Supporting Information File 1]. No methoxy peak corresponding to 1a (δ 3.72) was apparent above baseline noise. We separated 4a and 4b with difficulty by repeated chromatography in EtOAc/hexanes. The sample of 4a contained traces of an inseparable impurity, which if active might artificially elevate the apparent binding affinity. Since the sample showed no affinity, however, this problem does not arise. The 1H NMR spectra are reproduced in Supporting Information File 1.

There is no consensus on the mechanism of base-catalyzed epimerization at C-8. Koreeda and coworkers proposed a complex mechanism, initiated by ketone enolate formation. The configuration of H-8 is inverted indirectly, without exchange, by cleavage of the C-8/9 bond (see Scheme 1). [11-13] The simpler mechanism of enolization of the lactone itself has also been proposed.[16] A detailed case for this mechanism has been presented, giving evidence that H-8 exchanges under mildly basic conditions, and that similar furanolactones lacking the ketone also undergo epimerization.[17] Other workers remain undecided.[8,18]

Scheme 1

Koreeda et al's proposed mechanism for the epimerization.

Experimental details; statement of author contributions; 1H NMR spectra of 2b, 4a and 4b (Portable Document Format).

Crystal structure of 2b (Crystallographic Information File).

Structure factors for 2b (Crystallographic Information File).