Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

It is over 30 years since the introduction of the benzodiazepines into clinical practice for the relief of anxiety. Since then, the usage of these agents has diversified into anticonvulsant, hypnotic and also muscle-relaxant applications (Woods et al., 1992). During this period of clinical application, considerable evidence has indicated that the major site of action of the benzodiazepines occurs at the GABA_A receptor (Polc, 1988). Activation of this receptor accounts for the majority of inhibitory synaptic neurotransmission in the CNS and molecular cloning studies imply that the GABA_A receptor probably is composed of five polypeptide subunits that participate in forming an integral ion channel (Nayeem et al., 1994; Rabow et al., 1995; Sieghart, 1995). The subunits are members of distinct "families," designated as , ,  and , and, with the exception of the  subunit, the families comprise multiple subunit members, i.e., (1-6), (1-4), (1-4) and  (Sieghart, 1995).

Of particular interest to benzodiazepine pharmacology is the observation that the expression of different combinations of GABA_A receptor subunits can influence the pharmacological sensitivity to benzodiazepine agonists (Wisden and Seeburg, 1992). Previously, the sensitivity of GABA_A receptors to the benzodiazepines had been categorized into "type I or type II," a feature based largely on benzodiazepine agonist selectivity. Recently, the expression of recombinant GABA_A receptors has indicated that the differentiation of benzodiazepine pharmacology into type I or II, and now type III categories, depends largely on the type of  subunit present. For example, 1 subunit-containing receptors are associated with type I benzodiazepine pharmacology, whereas 2, 3 and 5 are associated with type II. The 6 subunit presently constitutes the type III category. In addition to influencing the binding and benzodiazepine pharmacology, the positive regulatory effect of the benzodiazepines on GABA_A receptor function can be removed by eliminating the gamma subunit from the receptor complex (Pritchett et al., 1989; Wisden and Seeburg, 1992).

The recent discovery of a secondary metabolite, (2aR^*, 5aR^*, 7E,11E,14aS^* , 14bS^* , 14cS^*)- 2a,5a,6,9,10,13,14,14a, 14b,14c-decahydro - 4 - hydroxy-5a,9,9,12-tetramethyl-1,5dioxacyclopenta[cd]-cycloundec[f]inden-3(2H)-one (Xenovulene A), extracted from the microorganism, Acremonium strictum, and its ability to displace high-affinity ligand binding to the benzodiazepine site on the GABA_A receptor (Ainsworth et al., 1995), was of interest, because this compound is structurally unrelated to the classical benzodiazepine ligands. We undertook the present study for two reasons: to ascertain the binding characteristics of this novel substance and a related derivative, with use of neuronal preparations noted for differences in benzodiazepine selectivity; and to analyze how these agents can functionally regulate both native neuronal and selected recombinant GABA_A receptors by binding to the allosteric "benzodiazepine site."

A preliminary account of some of these results has been published previously in abstract form (Thomas et al., 1996; Sundaram et al., 1996).

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Production of Xenovulene A (XR368). Xenovulene A was prepared from A. strictum (Accession number IMI 354451) by fermentation and down-stream processing by procedures described previously (Ainsworth et al., 1995; Blackburn et al., 1996).

Preparation of XR7009. Dibasic sodium phosphate (1.58 g, 11.13 mmol) was added to a solution of Xenovulene A (400 mg, 1.12 mmol) in anhydrous dichloromethane (20 ml). The resulting mixture was cooled to 0°C, and a solution of 3-chloroperbenzoic acid (424 mg, 1.23 mmol) in anhydrous dichloromethane (5 ml) was added dropwise with rapid stirring. After addition was complete the reaction mixture was maintained at 0°C for a further 30 min.

Radioligand binding. Rat brain membranes were prepared as described previously (Duggan and Stephenson, 1988). Membranes were isolated from adult rat forebrain, cerebellum and spinal cord, then thoroughly washed and freeze-thawed three times before incubation with 0.5 nM [³H]flunitrazepam or 5 nM [³H]Ro15-4513 in the presence and absence of competing drugs at 4°C for 1 h. Assays were performed in triplicate with a final volume of 1 ml. Samples were harvested by rapid filtration and washed in ice-cold phosphate buffered saline. Specific binding (calculated from the total and nonspecific binding) was defined as that inhibited by 10 µM diazepam and the results, expressed as mean ± S.E. mean, were analyzed by nonlinear least squares analysis using INPLOT (Ver 4.0; Graphpad Software). To assess the goodness-of-fit provided by single or two binding site models, the experimental results were compared with the theoretical fit by use of the Fisher (F) test.

Cell preparation: Xenopus oocytes. Oocytes were removed from anesthetized Xenopus laevis as described previously (Smart and Krishek, 1995) and placed in MBM containing (mM): 110, NaCl; 1, KCl; 2.4, NaHCO₃; 7.5, Tris-HCl; 0.33, Ca(NO₃)₂; 0.41, CaCl₂; 0.82, MgSO₄; 50 µg/ml gentamycin (pH 7.6). Suitable oocytes (stages IV and V) were centrifuged at 700 to 1100 × g for 10 min at 10°C to reveal the nucleus into which 10 to 20 nl of cDNA solution (1 mg/ml), encoding for murine GABA_A receptor subunits, was injected. Oocytes were then incubated at 19°C for 24 h and the MBM was replaced every 2 to 3 days. Murine cDNAs were cloned as EcoR1 fragments into the mammalian expression vector pGW1, as described previously (Krishek et al., 1994).

Cultured cortical neurons. Cerebral hemispheres were removed from postnatal rats (Wistar, day 1) and cut into 3 mm cubes before incubating in 0.25% (w/v) trypsin in Hanks' calcium-magnesium free balanced salt solution for 10 min at 37°C. Trypsin was inactivated by adding serum-containing growth medium. The tissue was triturated with polished Pasteur pipettes with decreasing orifice diameters. Cells were pooled and plated on poly-L-lysine coated 35 mm dishes in fresh growth medium based on: MEM (Earle's salts); 10% (v/v) horse serum, 2 mM glutamine, 5 mg/ml glucose, 200 U/ml and 200 µg/ml penicillin G and streptomycin, respectively. The cells were incubated at 37°C in 95% air/5%CO₂. Non-neuronal cell growth was controlled after 5 days by including 10 µM cytosine arabinoside in the growth medium for 24 h. Sympathetic neurons were prepared according to the methods described by Smart (1992).

Electrophysiology: Intracellular recording. Whole-cell membrane currents and conductances were recorded from Xenopus oocytes using a two-electrode voltage clamp technique. Oocytes were superfused with an amphibian Ringer's solution containing (mM): 110, NaCl; 2, KCl; 5, HEPES; 1.8, CaCl₂ (pH 7.4), at 10 to 15 ml/min (bath volume, 0.5 ml). Voltage and current microelectrodes (1-5 megohm) were filled with 3 M KCl and 0.6 M K₂SO₄, respectively. Currents were recorded using an Axoclamp 2A amplifier in conjunction with a Brush-Gould Ink pen recorder.

Whole-cell channel recording. Patch electrodes (1-5 megohm) were filled with the following solution containing (mM): 140, KCl; 2, MgCl₂; 1, CaCl₂; 10, HEPES; 11, ethyleneglycol-bis(-aminoethyl ether)-N,N,N,N-tetraacetic acid; 2, ATP, pH 7.1. The cells were continuously superfused in the culture dish with a Krebs' solution containing (mM): 140, NaCl; 4.7, KCl; 1.2, MgCl₂; 2.5, CaCl₂; 10, HEPES; 11, glucose, pH 7.4. Membrane currents were recorded using an Axopatch 1C patch amplifier. Series resistance compensation of 80% was routinely achieved and membrane currents were filtered at 10 kHz (3 dB, 6 pole Bessel filter, 36 dB/octave).

Analysis of ligand-modulated membrane conductances. The ligand-induced membrane conductance change (G) was calculated by subtraction of the resting conductance. Conductances were ascertained by the application of brief hyperpolarizing voltage command steps (1 s duration, 10 mV amplitude, 0.2 Hz) which were superimposed on the holding potential (30 to 50 mV) in the absence and presence of ligand as described previously (Krishek et al., 1994). These data were used to construct equilibrium concentration-response relationships for GABA. To pool dose-conductance data from more than one oocyte, all the conductances were normalized (G_N) to the conductance change produced by 10 µM GABA. These data were used to construct equilibrium concentration-response curves and fitted according to the following equation: G/G_max = 1/(1 + (EC₅₀/A)ⁿ), where G and G_max represent the conductance increases induced at a given concentration (A) and a saturating concentration of ligand, respectively. EC₅₀ defines the ligand concentration producing a half-maximal response, and n_H represents the Hill coefficient.

Compounds. For electrophysiology, stock solutions of Xenovulene A and XR7009 were made up in ethanol and diluted before use in Ringer's solution. Final ethanol concentrations did not exceed 0.05% (v/v) and did not affect the GABA-activated responses. For radioligand binding, to avoid the precipitation of XR368 and XR7009 in the assay buffer, 10% (v/v) ethanol was also incorporated. This concentration of ethanol did not significantly affect [³H]flunitrazepam binding which was determined as 93 ± 4% of the controls in the absence of ethanol (n = 3, mean ± S.E.M.).

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Structures of Xenovulene A and XR7009. Xenovulene A and XR7009 bear no apparent resemblance to the structure of classical benzodiazepines (see fig. 1). The structure of Xenovulene A comprises an 11-membered humulene ring fused to a very unusual tricyclic ring system, the structure of which has been described previously (Ainsworth et al., 1995). The humulene ring is conformationally restricted by the presence of two trans double bonds and the rigid tricycle, which contains five chiral centers present at the ring junctions. The relative stereochemistries of these chiral centers leads to an overall concave conformation for the Xenovulene A molecule. XR7009 occurs as a minor co-metabolite of Xenovulene A and was first detected on the scale-up of the fermentation of Xenovulene A. However, XR7009, which differs from the parent compound by the selective epoxidation of one face of the trisubstituted C₁₁-C₁₂ double bond in the humulene ring, was routinely and more conveniently prepared as a semisynthetic derivative of Xenovulene A. 

View larger version (10K): [in this window] [in a new window]   Fig. 1.   Structural comparison of diazepam with Xenovulene A and its synthetic derivative, XR7009. Diazepam (A) bears little structural resemblance to the compounds isolated in this study. Most significantly, the metabolite Xenovulene A (B) and its derivative XR7009 (C) uniquely possess an 11-membered humulene ring which appears to be significant in determining affinity for the "benzodiazepine receptor." It should be noted that bonds to hydrogen atoms are not represented, except in cases where their stereochemistry has particular importance.

Radioligand binding studies on GABA_A receptors. The first indication that Xenovulene A could specifically interact with the "benzodiazepine binding site" on the mammalian GABA_A receptor was evident from the ability of Xenovulene A to inhibit [³H]flunitrazepam binding to bovine brain membranes with an IC₅₀ of 40 nM (Ainsworth et al., 1995). To further analyze the radioligand binding profile, three neuronal preparations containing GABA_A receptors exhibiting differential sensitivities to some benzodiazepines (classified as types I to III) were selected from the rat CNS. These included membranes prepared from the forebrain (types I and II), cerebellum (predominantly type I and diazepam-insensitive type III) and spinal cord (type II > type I).

View this table: [in this window] [in a new window]   TABLE 1 Inhibition constants for a range of ligands displacing [3H]flunitrazepam binding from rat brain membranes Specific [3H]flunitrazepam binding was determined in the presence and absence of individual ligands at 0.5 nM radioligand concentration. Competition curves for Xenovulene A, XR7009, zolpidem and -CCE, determined on forebrain membranes, were best described by a two-binding site model with the values for Khigh(a) and Klow(b) as shown. Values in parentheses correspond to the relative proportions (%) of high and low affinity sites. For cerebellar and spinal cord preparations, a single-binding site model sufficed. Independent of the membrane preparation, the data for the benzodiazepines, diazepam, lorazepam, flurazepam and clobazepam, fitted best to a single-binding site model. Results are expressed as mean ± S.E.M. for three to seven experiments.

View larger version (18K): [in this window] [in a new window]   Fig. 2.   Inhibition of [3H]flunitrazepam binding by Xenovulene A and XR7009 in rat brain membranes. Results are expressed as percentages (mean ± S.E.M. of triplicate determinations) of specific [3H]flunitrazepam inhibition by Xenovulene A (X-A; A) and XR7009 (B) in rat forebrain (), cerebellar () and spinal cord () membranes. Curves were fitted to data by use of a nonlinear least squares method, and the goodness-of-fit to theoretical single or two-binding site models was assessed with the Fisher (F) test. Similar results were obtained for displacements by Xenovulene A and XR7009 in three to seven separate experiments, the inhibition constants for which are summarized in table 1.

Regulation of GABA_A receptors: Electrophysiological studies on native and recombinant receptors. The investigation of the functional effects of Xenovulene A, and the derivative XR7009, was conducted initially on recombinant GABA_A receptors expressed in X. laevis oocytes. Both Xenovulene A and XR7009 (0.1-1 µM) were devoid of any intrinsic agonist activity when applied to oocytes expressing 112S subunit-containing GABA_A receptors; however, 10 µM GABA-activated responses were significantly enhanced by these agents in a reversible manner (fig. 3A). The equilibrium-concentration response curve for GABA was shifted to the left by Xenovulene A, and more so by XR7009 (1 µM), without increasing the maximum response to GABA (fig. 3B). The EC₅₀ for GABA was 10.9 ± 0.4 µM (n = 7) in control Ringer's solution, decreasing to 5.1 ± 0.2 µM (n = 4) in the presence of 1 µM Xenovulene A and 2.7 ± 0.1 µM (n = 4) in 1 µM XR7009. This type of positive modulation at the GABA_A receptor is similar to that observed with benzodiazepines, but different from that expected for the barbiturates which generally increase the maximum response to GABA. Similarly, 1 µM flurazepam displaced the concentration-response curve to the left reducing the EC₅₀ to 3.2 ± 0.16 µM (n = 3) without affecting the maximum response; whereas, 50 µM pentobarbitone displaced the curve leftward, reducing the EC₅₀ to 1.06 ± 0.1 µM (n = 3), but increasing the maximum response to GABA (data not shown).

View larger version (24K): [in this window] [in a new window]   Fig. 3.   Positive modulation of the recombinant murine 112S GABAA receptor by Xenovulene A and XR7009. (A) Traces represent whole-cell membrane currents and conductance changes recorded from Xenopus oocytes after application of 10 µM GABA or 1 µM Xenovulene A (X-A), either alone or in combination. In this, and subsequent traces, membrane conductance was monitored continuously by application of brief hyperpolarizing voltage steps (1 s, 10 mV, 0.2Hz) superimposed on the holding potential (30 to 50 mV). Solid lines represent the period of drug application, typically 30 s. Xenovulene A was observed to have no direct effect, yet was able to positively modulate the GABA-activated response in a reversible manner. (B) Equilibrium concentration-response curves for GABA (), and in the presence of 1 µM Xenovulene A () or XR7009 (). The data are normalized to the response evoked by 10 µM GABA. Points represent mean ± S.E.M. of data gathered from four to seven experiments with 18 cells. These and subsequent curves were fitted to the logistic model described under "Materials and Methods."

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Relative potencies of Xenovulene A and XR7009 in modulating the control GABA response in oocytes expressing 112S GABAA receptors. Concentration-response relationships for Xenovulene A () and XR7009 (), relative to the GABA modulators diazepam (), flurazepam () and pentobarbitone () in enhancing the control response to 10 µM GABA (defined as 0% enhancement). Values represent the mean ± S.E.M. of data gathered from 18 cells.

View larger version (18K): [in this window] [in a new window]   Fig. 5.   Pharmacological profile of Xenovulene A. (A) Potentiation of the 10 µM GABA-activated response by 1 µM Xenovulene A (X-A) in oocytes expressing 112S GABAA receptors was significantly compromised in the presence of the benzodiazepine antagonist flumazenil (FLZ: 5 µM). (B) Exclusion of the 2S subunit by the expression of 11 constructs resulted in a loss of activity for Xenovulene A (1 µM).

View larger version (21K): [in this window] [in a new window]   Fig. 6.   GABAA receptor subtype selectivity exhibited by Xenovulene A and XR7009. Control equilibrium concentration-response curves for GABA (), and in the presence of either 1 µM Xenovulene A () or XR7009 () derived from cells expressing 312S or 612S GABAA receptors. A leftward displacement of the control curve was observed for these ligands with 3 subunit-containing receptor constructs. The GABA concentration-response relationship for 6 subunit-containing, diazepam-insensitive, GABAA receptors was not enhanced by Xenovulene A. Curves are normalized with respect to the response evoked by 100 µM (312S) or 1 µM (612S) GABA. Points represent means ± S.E.M. of data from three to six cells.

View this table: [in this window] [in a new window]   TABLE 2 Potentiation of the GABA EC20 response for alpha-1 subunit- and alpha-3 subunit-containing GABAA receptors in the presence of Xenovulene A and XR7009 Measurements taken directly from the equilibrium concentration-response curves in figures 3 and 6 for the enhancement of the GABA EC20 value by Xenovulene A (1 µM) or XR7009 (1 µM) reveal the selectivity exhibited for the 112S GABAA receptor.

View larger version (8K): [in this window] [in a new window]   Fig. 7.   Regulation of GABA-activated responses of cortical neurons by Xenovulene A. Traces represent whole-cell membrane currents recorded from cultured cortical neurons after application of 5 µM GABA in the presence and absence of 1 µM Xenovulene A (X-A). The positive modulatory effect of co-applied Xenovulene A was fully reversible.

View larger version (14K): [in this window] [in a new window]   Fig. 8.   Steady-state enhancement of the GABA-induced response by Xenovulene A. (A) Traces represent whole-cell membrane currents recorded from a single human embryonic kidney cell transiently expressing 112L GABAA receptors after the rapid application of 10 µM GABA in the absence, or presence (after preincubation for 5, 15 or 30 s) of Xenovulene A (X-A: 1 µM). The positive modulatory effect of Xenovulene A was fully reversible. (B) The percentage enhancement of the 10 µM GABA-activated response (IGABA) after various pre-exposure times with 1 µM Xenovulene A was adequately described by a single exponential curve. The time constant was 5.8 ± 0.8 s (n = 3). A steady state enhancement was attained in approximately 15 to 20 s.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

Benzodiazepines and the GABA_A receptor. The usefulness of the benzodiazepines as therapeutic agents, particularly in treating anxiety, has been compromised by several well-known side effects ranging from ataxia, sedation and amnesia to the chronic problems associated with benzodiazepine addiction. Two approaches have been adopted to obviate some of the side effects and to improve the therapeutic potential of anxiolytic benzodiazepines. The first has targeted the synthesis of numerous agents, derived with the purpose of improving therapeutic selectivity. Secondly, the realization of potential GABA_A receptor heterogeneity, following studies with recombinant GABA_A receptors, offers the prospect that novel benzodiazepines, targeting receptors comprising specific subunit constructs, may possess unique therapeutic profiles perhaps devoid of many of the deleterious side effects. In particular, agents that can distinguish Type II over Type I benzodiazepine receptors would be of considerable interest, particularly because the behavioral profile of such ligands is currently unresolved. An alternative approach to these strategies, which we have adopted, involves the isolation of novel compounds from natural sources, unrelated to the benzodiazepine class of ligands, that are identified by radioligand binding to be potential ligands for the benzodiazepine binding site(s).

Allosteric modulation of the GABA_A receptor by Xenovulene A: Type I vs. Type II pharmacology. The displacement of [³H]flunitrazepam binding from rat brain membranes coupled with the electrophysiological observations of enhanced GABA-mediated currents, all indicated that the metabolite, Xenovulene A, and the synthetic derivative, XR7009, bind to the GABA_A receptor. The ability of flumazenil to inhibit the enhancement of the GABA-induced current by Xenovulene A and XR7009, and the dependence of the potentiation by these agents on the presence of the 2 subunit in recombinant GABA_A receptors, is in accordance with Xenovlene A and XR7009 binding to the benzodiazepine recognition site. In comparison with classical benzodiazepines (e.g., diazepam and flurazepam), XR7009 was more potent in enhancing GABA-mediated responses, whilst the parent molecule, Xenovulene A, was less potent than diazepam; however, both compounds behaved as full agonists at the benzodiazepine receptor, achieving similar maximal responses.

Comparison with other novel ligands at the benzodiazepine binding site. Apart from the benzodiazepines, there are other ligands that appear to fulfill the structural requirements necessary to interact at the allosteric benzodiazepine binding site on the GABA_A receptor. The most notable of these is -lumicolchicine, an inactive analog of the microtubule-disrupting agent, colchicine, which like our compounds also bears little structural resemblance to the benzodiazepines. -Lumicolchicine enhanced muscimol-mediated Cl uptake in cortical microsacs and inhibited the binding of [³H]flunitrazepam binding (Mihic et al., 1994). With human recombinant GABA_A receptors comprising 122S subunits, -lumicolchicine enhanced GABA-activated currents. An interaction at the benzodiazepine binding site was indicated by the inhibition of modulation with flumazenil and the inability of -lumicolchicine to enhance responses mediated by 12 subunits (Mihic et al., 1994). Interestingly, as for Xenovulene A, -lumicolchicine was less active on Type II GABA_A receptors (222S), which suggests some degree of subtype selectivity.

Structure-function relationship for Xenovulene A. The structures of the A. strictum metabolites have little resemblance to the structures of the classical benzodiazepines (fig. 1). To understand the details of the benzodiazepine pharmacophore, we attempted some preliminary structural modifications to Xenovulene A to identify those parts of the structures that are important for their functional properties at the GABA_A receptor, and to identify those which bestow subtype-selective characteristics. A range of semisynthetic derivatives were prepared and characterized. Modifications included: the removal of one or both of the humulene ring double bonds, blocking the enolized -dicarbonyl group, complete reduction of the molecules to the corresponding diols and the selective reduction of the tricycle with retention of the humulene ring double bonds. With the exception of XR7009 (produced by epoxidation of the C₁₁-C₁₂ double bond in the humulene ring of Xenovulene A), removal or chemically blocking any of these structural features resulted in a substantial loss of activity in radioligand displacement binding and functional electrophysiological assays. In particular, the rigidity of the humulene ring appears to be essential for activity at the GABA_A-benzodiazepine receptor. Interestingly, the diol analogs exhibited increased stability, a feature absent from the parent compounds. Despite these substantial structural modifications, we were unable to change the selectivity of the molecule such that it became a Type II preferring benzodiazepine receptor ligand. Xenovulene A and XR7009 therefore remain novel and fascinating templates for the design of new potentially subtype-selective ligands for the benzodiazepine receptor.