Novel positive allosteric modulators of GABA_A receptors with anesthetic activity

Abstract

GABA_A receptors are the main inhibitory neurotransmitter receptors in the brain and are targets for numerous clinically important drugs such as benzodiazepines, anxiolytics and anesthetics. We previously identified novel ligands of the classical benzodiazepine binding pocket in α₁β₂γ₂ GABA_A receptors using an experiment-guided virtual screening (EGVS) method. This screen also identified novel ligands for intramembrane low affinity diazepam site(s). In the current study we have further characterized compounds 31 and 132 identified with EGVS as well as 4-O-methylhonokiol. We investigated the site of action of these compounds in α₁β₂γ₂ GABA_A receptors expressed in Xenopus laevis oocytes using voltage-clamp electrophysiology combined with a benzodiazepine site antagonist and transmembrane domain mutations. All three compounds act mainly through the two β+/α− subunit transmembrane interfaces of the GABA_A receptors. We then used concatenated receptors to dissect the involvement of individual β+/α− interfaces. We further demonstrated that these compounds have anesthetic activity in a small aquatic animal model, Xenopus laevis tadpoles. The newly identified compounds may serve as scaffolds for the development of novel anesthetics.

Introduction

The search for novel anesthetics has been triggered by the rising age of patients and increasing use of anesthesia outside the operating room^1,2. A key site of action of the potent anesthetics propofol and etomidate is the major inhibitory receptor in the mammalian central nervous system, the γ-aminobutyric acid type A (GABA_A) receptor. These receptors are composed of five homologous subunits organized around a central Cl⁻ selective channel³. Each subunit contains a large N-terminal extracellular domain (ECD), a transmembrane domain (TMD) with four alpha-helices (TM1 to TM4), and a variable-length intracellular domain (ICD) between TM3 and TM4. 19 subunits of the GABA_A receptor have been cloned (for review see: 4, 5, 6), denoting that numerous types of receptor isoforms exist⁵. The most abundant GABA_A receptor in the brain comprises α₁, β₂ and γ₂ subunits^3,4,5,7. The receptor possesses a 2α:2β:1γ subunit stoichiometry^8,9,10,11, with a subunit arrangement of γβαβα anti-clockwise as seen from the synaptic cleft^10,11,12,13. The receptor composition and arrangement influence its pharmacological properties^14,15.

Benzodiazepines modulate α₁β₂γ₂ GABA_A receptor function by binding to a high affinity site located at the α+/γ− ECD interface, homologous to the agonist binding sites at β+/α− ECD interfaces^16,17. In addition to the high affinity binding site for benzodiazepines (site 1), there are other low affinity sites. One of these, site 2, is located at the ECD α+/β− interface^18,19. Others, together designated as site 3, are located in the TMD, based on abolition of benzodiazepine effects by combined isoleucine substitutions at the homologous residues α₁S269, β₂N265, and γ₂S280²⁰.

GABA_A receptors are also targets for potent intravenous anesthetics, including barbiturates, propofol and etomidate^{21,22,23,24,25}. Interestingly, receptor sensitivity to intravenous anesthetics is affected by benzodiazepine site 3 mutations^{26,27,28,29,30}. Diverse anesthetics not only potentiate GABA-induced Cl⁻ currents, but additionally at high concentrations directly activate GABA_A receptors^31,32. Photo-affinity labeling has located allosteric sites for the intravenous anesthetics etomidate and propofol to the TM1 of α and TM2, TM3 of β subunits^23,24,33.

Previously, we reported a new method to identify ligands of the high affinity benzodiazepine pocket, experimental-guided virtual screening (EGVS), integrating experimental data with homology modeling of the GABA_A receptor³⁴. EGVS identified some ligands that only recognized site 1, others that recognized both site 1 and site 3³⁵, and another set that only recognize site 3.

Here we describe the actions of two compouds identified by EGVS, 31 and 132³⁴ and 4-O-methylhonokiol³⁶. Using mutations and concatenated receptors we determined that the three compounds act mainly through the TMD β+/α− interfaces (site 3), and particularly the γβ+/α−β site. The anesthetic action of these drugs was explored in vivo, revealing potencies similar to propofol.

Results

In an attempt to find novel ligands for the high affinity site for benzodiazepines on GABA_A receptors we screened 198 compounds for displacement of the high affinity benzodiazepine site (called site 1 previously) antagonist [³H]-Ro 15-1788 at receptors expressed in HEK-cells. Many high affinity ligands were identified³⁴. One compound, SJM3 acted as antagonist with high affinity at site 1, but allosterically potentiated receptor activation through sites in the membrane (called sites 3 previously)³⁵. Compounds 31 and 132 either did not or weakly displaced [³H]-Ro 15-1788 from site 1 but potently enhanced GABA_A receptor activation. 4-O-methylhonokiol shares these characteristics, potentiating α₁β₂γ₂ receptors with an EC₅₀ of 5.4 ± 1.8 μM, independent of the high affinity site for benzodiazepines³⁶. Here, we report mechanistic and animal studies of these three compounds.

Compounds 31 and 132 are allosteric modulators of α₁β₂γ₂ GABA_A receptors

First, we investigated if compounds 31 and 132 were able to act as agonists. Figure 1 shows their chemical structures compounds. Both compounds at 3 and 30 μM elicited only very small currents by themselves in α₁β₂γ₂ GABA_A receptors expressed in Xenopus oocytes. These compounds elicited at the concentration of 3 μM and 30 μM currents amounting to 0.1 ± 0.06% (mean ± SD, n = 3) and 0.3 ± 0.13% (mean ± SD, n = 3), respectively of the maximal current amplitude elicited by GABA in the same oocytes. Thus neither of the compounds tested acts as an appreciable agonist on α₁β₂γ₂ receptors.

Figure 1

Chemical structure of compounds 31, 132, 4-O-methylhonokiol, and the high-affinity benzodiazepine antagonist Ro 15-1788.

Both compounds strongly enhanced currents elicited by GABA. We established the concentration response curves of this positive allosteric modulation. After two applications of GABA at a concentration eliciting 0.5–1.5% of the maximal current amplitude, the same concentration of GABA was co-applied with increasing concentrations of the tested compounds. Figure 2a,b show current traces demonstrating positive modulation by different concentrations of compounds 31 and 132, respectively. Figure 2c summarizes the results of three of such experiments. Both compounds potentiate GABA elicited currents in oocytes expressing α₁β₂γ₂ receptors. For compound 31, at high concentrations apparent desensitization was observed, that could be partly due to open channel blocker effect. For both compounds no saturation at the highest concentration was obtained, because of poor solubility we could not test higher concentrations.

Figure 2: Compounds 31 and 132 stimulate GABA currents mediated by α1β₂γ₂ receptors in a concentration-dependent manner.

α₁β₂γ₂ receptors were expressed in Xenopus oocytes and electrophysiological experiments were performed. Original current traces of an experiment with compound 31 (a) and with compound 132 (b). Numbers indicate applied concentrations of the respective compounds. (c) Concentration dependence of the positive allosteric modulation by compound 31 (closed circles) and compound 132 (closed squares) in oocytes expressing α₁β₂γ₂ receptors. Mean data ± SD for both compounds is shown, n = 3.

Potentiation by compounds 31 and 132 is not affected by Ro 15-1788

From the binding data we did not expect that the two compounds are acting through the high affinity benzodiazepine binding site 1 in α₁β₂γ₂ receptors. Nevertheless, we tested whether 1 μM of the site 1 antagonist Ro 15-1788 inhibits potentiation of GABA currents by compounds 31 or 132. Either compound (3 μM) strongly potentiated currents elicited by GABA. Potentiation by either drug was not inhibited by 1 μM Ro 15-1788 (Fig. 3). Relative current amplitudes in the presence vs. absence of the antagonist were 112 ± 8% (mean ± SD, n = 4, p > 0.05, t test) for compound 31, and 115 ± 14% (mean ± SD, n = 4, p > 0.05, t test) for compound 132. This confirms that potentiation by compounds 31 and 132 does not result from action at the benzodiazepine site 1.

Figure 3: Compounds 31 and 132 do not act at the classical high affinity site for benzodiazepines.

GABA at a concentration eliciting 0.5% of the maximal current amplitude (EC_0.5, single bars) was applied until a stable response was obtained. Subsequently, the same concentration of GABA was co-applied with 3 μM of compounds 31 (a) or 132 (b), which resulted both in a large increase of current amplitude. Co-application of Ro 15-1788 with compound and GABA did not reduce the degree of modulation in both cases. Experiments were repeated 4 times, with three different batches of oocytes, with a similar outcome.

Compounds 31 and 132 and 4-O-methylhonokiol act at the low affinity benzodiazepine site 3 in α₁β₂γ₂ receptors

We next investigated whether benzodiazepine site 3 TMD mutations affect potentiation by compounds 31 and 132. Combining three homologous site 3 mutations in α₁β₂γ₂ receptors, α₁S269I, β₂N265I and γ₂S280I, eliminated the potentiation by high concentrations of diazepam²⁰. We investigated the effects of these mutations individually and combined, abbreviating them as α₁M, β₂M and γ₂M. Recently, we described the potency of GABA to activate all of these receptor subtypes expressed in Xenopus oocytes³⁷. In order to exclude general gating effects caused by these mutations we showed that potentiation by low concentrations of diazepam and by THDOC are not affected³⁷. For illustration, the localization of the mutations is shown in the crystalized homomeric β₃ receptor³⁸ where some of the β₃ subunits were renamed α₁, β₂ and γ₂ (Fig. 4a).

Figure 4

(a) Model structure of the GABA_A receptor transmembrane domain. The major isoform of the GABA_A receptor is composed of two α₁, two β₂, and one γ₂ subunits. The model structure depicts the crystalized homomeric β₃ GABA_A receptor (PDB structure 4COF)³⁸. In this figure, some of the β₃ subunits were renamed α₁ (yellow), β₂ (blue) and γ₂ (red); structures are shown in ribbon representation. The mutated residues α₁S269, β₂N265, and γ₂S280 are located at the interfaces between subunits. (b) Potentiation of the GABA response by compound 31 (3 μM), compound 132 (3 μM), and 4-O-methylhonokiol (1 μM, abbreviated Mh) in wild-type α₁β₂γ₂, single mutant (α₁M, β₂M, γ₂M), and triple mutant receptors expressed in Xenopus oocytes. The bars indicate mean ± SD, n = 3.

Wild-type α₁β₂γ₂, α₁Mβ₂γ₂, α₁β₂Mγ₂, α₁β₂γ₂M and α₁Mβ₂Mγ₂M receptors were expressed in Xenopus oocytes. Using electrophysiological techniques we determined the effect of these point mutations on the potentiation by compounds 31, 132 and 4-O-methylhonokiol, normalizing to the potentiation in wild type α₁β₂γ₂ receptors.

Figure 4b summarizes the results obtained. The single mutations in the α₁ and in the γ₂ subunits did not significantly alter the degree of potentiation by 3 μM of either compound 31 or 132. Relative to wild-type α₁β₂γ₂ receptors, modulation in mutated α₁Mβ₂γ₂ receptors by compound 31 amounted to 77 ± 24% (mean ± SD, n = 3, p > 0.05, Tukey posthoc test), and by compound 132 to 118 ± 44% (mean ± SD, n = 3, p > 0.05, Tukey posthoc test). Modulation in α₁β₂γ₂M receptors by compound 31 was 87 ± 52% (mean ± SD, n = 3, p > 0.05, Tukey posthoc test), and by compound 132 was 91 ± 21% (mean ± SD, n = 3, p > 0.05, Tukey posthoc test). In contrast, modulation by both compounds was strongly impaired in α₁β₂Mγ₂ and triply mutated α₁Mβ₂Mγ₂M receptors. Potentiation by 3 μM compound 31 in α₁β₂Mγ₂ was 10.4 ± 8.5% (mean ± SD, n = 3, p = 0.007, Tukey posthoc test), and in α₁Mβ₂Mγ₂M receptors was 2.3 ± 0.8% (mean ± SD, n = 3, p = 0.0043, Tukey posthoc test). Potentiation by compound 132 relative to that in α₁β₂γ₂ was also dramatically reduced in α₁β₂Mγ₂ 7.3 ± 6.8% (mean ± SD, n = 3, p = 0.003, Tukey posthoc test) and α₁Mβ₂Mγ₂M −0.4 ± 5.1% (mean ± SD, n = 3, p = 0.0021, Tukey posthoc test), respectively.

Potentiation by 1 μM 4-O-methylhonokiol was also significantly reduced only in α₁β₂Mγ₂ and α₁Mβ₂Mγ₂M receptors, similar to compounds 31 and 132. Relative to wild-type α₁β₂γ₂ receptors, modulation in α₁Mβ₂γ₂ mutated amounted to 63 ± 6% (mean ± SD, n = 3, p > 0.05, Tukey posthoc test), and in α₁β₂γ₂M receptors to 79 ± 16% (mean ± SD, n = 3, p > 0.05, Tukey posthoc test). Relative to α₁β₂γ₂, residual potentiation in α₁β₂Mγ₂ receptors amounted to 15.6 ± 1.5% (mean ± SD, n = 3, p = 0.0022, Tukey posthoc test), and to 2.2 ± 0.2% (mean ± SD, n = 3, p = 0.0018, Tukey posthoc test) in α₁Mβ₂Mγ₂M.

The above data suggests that the modulatory site(s) for the three compounds we studied is located in one or both of the β+/α− TMD subunit interfaces on α₁β₂γ₂ GABA_A receptors.

Role of the individual β+/α− subunit interfaces in channel modulation by compounds 31, 132 and 4-O-methylhonokiol

Each GABA_A receptor contains two β+/α− subunit interfaces. Combined mutation at these interfaces greatly reduces the modulatory effects of compounds 31, 132 and 4-O-methylhonokiol. Using α₁-β₂-α₁ and γ₂-β₂ subunit concatemers, we studied the effects of individual mutated interfaces. We designated receptors containing the mutant in the γ₂-β₂ construct interface 1 M, and the mutation in the α₁-β₂-α₁ construct interface 2 M (Fig. 5a). The α₁-β₂M-α₁ and γ₂-β₂M constructs were built, and were co-expressed with non-mutated dual or triple subunit constructs forming α₁-β₂-α₁/γ₂-β₂M and α₁-β₂M-α₁/γ₂-β₂ receptors. Both constructs were expressed together to form the double mutant receptor α₁-β₂M-α₁/γ₂-β₂M.

Figure 5: Individual roles of the two β+/α− interfaces in channel modulation by compounds 31, 132 and 4-O-methylhonokiol.

(a) Scheme showing the four concatenated wild-type and mutant receptors. 1 and 2 refer to the two different β+/α− subunit interfaces, interface 1 and interface 2. The location of the β₂N265I mutations is indicated in red color. Concatenated receptors were prepared containing no mutation (α₁-β₂-α₁/γ₂-β₂, non M), a mutation at interface 1 (α₁-β₂-α₁/γ₂-β₂M, interface 1 M), a mutation at interface 2 (α₁-β₂M-α₁/γ₂-β₂, interface 2 M), or mutations in both sites (α₁-β₂M-α₁/γ₂-β₂M, double M). Interface 2 harbors a binding site for GABA with higher apparent affinity for channel gating than the one positioned at the interface 1⁵⁴. (b) Potentiation by compound 31 (3 μM), compound 132 (3 μM), and 4-O-methylhonokiol (1 μM), using an EC_0.5–1.5 concentration of GABA for each concatenated receptor subtype. Bars indicate mean ± SD, n = 3.

Wild type concatenated receptors α₁-β₂-α₁/γ₂-β₂ were also expressed. This receptor has an EC₅₀ for GABA of approximately 120 μM¹². Results are standardized to the potentiation observed in α₁-β₂-α₁/γ₂-β₂ concatenated receptors. As shown in Fig. 5b, in the double mutant concatemeric receptors (α₁-β₂M-α₁/γ₂-β₂M) potentiation for all three compounds was abolished. Relative to α₁-β₂-α₁/γ₂-β₂ concatenated receptors, in the double mutant α₁-β₂M-α₁/γ₂-β₂M receptor modulation by compound 31 was reduced to 1 ± 5% (mean ± SD, n = 3, p < 0.0005, Tukey posthoc test), by compound 132 to −2 ± 0.4% (mean ± SD, n = 3, p = 0.0001, Tukey posthoc test), and by 4-O-methylhonokilol to 1 ± 0.6% (mean ± SD, n = 3, p = 0.0001, Tukey posthoc test). In receptors containing only one mutation, either interface 1 or interface 2, modulation by compound 31 was reduced significantly compared to concatenated wild type receptors α₁-β₂-α₁/γ₂-β₂. With interface 1 M, residual relative potentiation was 28 ± 6% (mean ± SD, n = 3, p = 0.0005, Tukey posthoc test) and with interface 2 M it was 51 ± 3% (mean ± SD, n = 3, p = 0.0059, Tukey posthoc test) of wild-type. Potentiation of interface 1 M and interface 2 M receptors differed significantly (p = 0.0035, t test). Therefore, both TMD β+/α− sites seem to contribute differently to modulation by compound 31, interface 1 being more efficacious.

Likewise, modulation by compound 132 was sensitive to the mutations at both sites. In interface 1 M receptors relative potentiation was reduced to 8 ± 4% (mean ± SD, n = 3, p = 0.0002, Tukey posthoc test), and in interface 2 M receptors to 39 ± 19% (mean ± SD, n = 3, p = 0.0035, Tukey posthoc test). Again, interface 1 M produced a larger impact than 2 M, although the difference was at the statistical limit (p = 0.0510, t test). Modulation by 4-O-methylhonokiol in interface 1 M and 2 M receptors was reduced to 11 ± 6% (mean ± SD, n = 3, p < 0.0001, Tukey posthoc test) and 28 ± 8% (mean ± SD, n = 3, p = 0.0004, Tukey posthoc test), respectively. The two mutations produced significantly different effects (p = 0.043, t test), with the interface 1 M effect again larger.

Therefore, potentiation by all three compounds displayed similar sensitivity patterns with homologous mutations in distinct β+/α− TMD sites of α₁-β₂-α₁/γ₂-β₂ receptors. Both sites are necessary for full modulation, while interface 1 produces a larger impact than interface 2.

Effect of the β₂N265S mutation

The β₂N265 residue is important for allosteric modulation of GABA_A receptors by many compounds acting through the TMD. This residue was initially described as a determinant for the modulatory action of loreclezole, where the β₂N265S mutation created a receptor unresponsive to this compound³⁹. Mutations in this residue also abolish potentiation by the anesthetics etomidate and propofol^28,40. As shown in Fig. 6, the β₂N265S mutation in α₁β₂γ₂ receptors significantly reduced potentiation by 3 μM compound 31, from 485 ± 230% in wild-type receptors (mean ± SD, n = 11), to 125 ± 25% in the mutated receptor (mean ± SD, n = 4, p = 0.009, Tukey posthoc test). In contrast, this mutation did not significantly reduce potentiation by 3 μM compound 132: 415 ± 126% in wild-type receptors (mean ± SD n = 6) versus 280 ± 36% in the mutated receptor (mean ± SD, n = 4, p > 0.05, Tukey posthoc test). We have shown earlier that potentiation by 4-O-methylhonokiol in receptors carrying the β₂N265S mutation was greatly reduced to about 40%³⁶.

Figure 6: Effect of the β₂N265S mutation on modulation by compounds 31 and 132.

Wild-type α₁β₂γ₂ and mutated α₁β₂N265Sγ₂ receptors were expressed in Xenopus oocytes and studied. Potentiation of GABA currents was determined using 3 μM of compound 31 or 132. Bars indicate mean ± SD, n = 4–11.

Subunit specificity of compounds 31 and 132

From the above experiments we inferred that the β+/α− TMD subunit interfaces mediate potentiation of compounds 31 and 132. We also wanted to know if potentiation by these compounds depends on subunit isoforms. First we replaced the α₁ subunit by different α subunit isoforms: α₁β₂γ₂, α₂β₂γ₂, α₃β₂γ₂, α₄β₂γ₂, α₅β₂γ₂ and α₆β₂γ₂ (Fig. 7). Compound 31 displayed a similar degree of potentiation in α₁β₂γ₂ receptors 485 ± 230% (mean ± SD, n = 11), α₂β₂γ₂ receptors, 407 ± 172% (mean ± SD, n = 4, p > 0.05, Tukey posthoc test), α₃β₂γ₂ receptors, 735 ± 234%, (mean ± SD, n = 5, p > 0.05, Tukey posthoc test), and α₄β₂γ₂ receptors, 412 ± 175% (mean ± SD, n = 4, p > 0.05, Tukey posthoc test). The α₅β₂γ₂ receptor showed a significant decrease in potentiation compared to α₁β₂γ₂ receptors to 253 ± 92% (mean ± SD, n = 8; p = 0.023, Tukey posthoc test), and the α₆β₂γ₂ receptor an increase in potentiation, amounting to 780 ± 236% (mean ± SD, n = 4; p = 0.048, Tukey posthoc test). This discrepancy in modulation of α₅β₂γ₂ and α₆β₂γ₂ receptors maybe explained by the fact that 3 of the first 7 residues of M1 located at the minus side of the α subunit are different (Fig. 8a). Compound 132 produced similar potentiation in receptors with all α subunits tested. Amounting to 415 ± 126% (mean ± SD, n = 6) in α₁β₂γ₂ receptors, 694 ± 293% (mean ± SD, n = 4, p > 0.05, Tukey posthoc test) in α₂β₂γ₂ receptors, 652 ± 215% (mean ± SD, n = 4, p > 0.05, Tukey posthoc test) in α₃β₂γ₂ receptors, 476 ± 152% (mean ± SD, n = 4, p > 0.05, Tukey posthoc test) in α₄β₂γ₂ receptors, 399 ± 147% (mean ± SD, n = 4, p > 0.05, Tukey posthoc test) in α₅β₂γ₂ receptors, and 542 ± 175% (mean ± SD, n = 4, p > 0.05, Tukey posthoc test) in α₂β₂γ₂ receptor type. These results indicate that although the type of α subunit has differential effects on potentiation between compounds 31 and 132, these compounds modulate all receptor subtypes studied.

Figure 7: Subunit specificity of current potentiation in different GABA_A receptors.

Different subunit combinations were expressed in Xenopus oocytes. Potentiation of GABA currents at a GABA concentration eliciting 0.5–1.5% of the maximal current amplitude was determined using 3 μM of compound 31 (a) or 132 (b). Bars indicate mean ± SD, n = 4–11.

Figure 8: Alignment of the rat amino acid residue sequences of different α (a) and β (b) subunit isoforms.

(a) Sequences preceding M1 and the first part of M1 in α subunits are shown. (b) Sequences preceding M3 and the first part of M3 in β subunits are shown.

Next, we examined the role of the β subunit, replacing the β₂ by β₁ or β₃. For compounds 31 and 132, α₁β₃γ₂ receptors showed a similar potentiation as α₁β₂γ₂ receptors. Amounting to 381 ± 115% for compound 31 and 555 ± 169% for compound 132 (mean ± SD, n = 4, p > 0.05, Tukey posthoc test). In the case of α₁β₁γ₂ receptors, potentiation by both compounds was significantly reduced compared to that in α₁β₂γ₂, 41 ± 13% for compound 31 and 37 ± 7% for compound 132 (n = 4; p = 0.0023, Tukey posthoc test for compound 31; n = 4; p = 0.0004, Tukey posthoc test, for compound 132). These results indicate that the type of β subunit is important for the potentiation by both compounds. It is interesting to note in this context that β₁ and β₂/β₃ differ not only in the residue 265, but also in the fourth residue of M3 predicted to be close to the latter residue (Fig. 8b).

When the γ₂ subunit was omitted, no statistical difference was observed for either compound tested. Compound 31 displayed a similar degree of potentiation between α₁β₂γ₂ receptors and α₁β₂ receptors, 339 ± 73% (mean ± SD, n = 4, p > 0.05, Tukey posthoc test). Compound 132 also showed a similar potentiation between both receptors, potentiation in α₁β₂ receptors amounting to 625 ± 219% (mean ± SD, n = 4, p > 0.05, Tukey posthoc test). When the γ₂ subunit was replaced by a δ subunit, in the case of α₁β₂δ receptors, potentiation was affected only in the case of compound 31, where a significant reduction was observed relative to α₁β₂γ₂ receptors. Where potentiation amounted to 168 ± 48% (mean ± SD, n = 4, p = 0.0026, Tukey posthoc test) for compound 31, and 640 ± 275% (mean ± SD, n = 4, p > 0.05, Tukey posthoc test) for compound 132. In α₄β₃δ receptors, potentiation by both compounds was statistically reduced relative to α₄β₃γ₂ receptors. For compound 31 potentiation amounted to 345 ± 77% (mean ± SD, n = 4) in α₄β₃γ₂ receptors, and was reduced to 125 ± 22% (mean ± SD, n = 4, p = 0.0015, Tukey posthoc test) in α₄β₃δ receptors. Potentiation of compound 132 in α₄β₃γ₂ receptors was 728 ± 203% (mean ± SD, n = 4), and was reduced to 214 ± 71% (mean ± SD, n = 4, p = 0.0031 Tukey posthoc test, for compound 132) in α₄β₃δ receptors.

Previous work³⁶ showed that potentiation by 4-O-methylhonokiol was dependent on the α subunit in a similar fashion as compound 31, since modulation was reduced by receptors containing α₅ and α₆ subunits. The type of β subunit was also important, as the presence of the β₁ subunit strongly reduced potentiation by this compound. On the contrary, the presence of a γ or a δ subunit did not affect potentiation.

Anesthetic activity in tadpoles

The anesthetic activity for compounds 31, 132 and 4-O-methylhonokiol was determined as loss of righting reflex (LoRR) in Xenopus tadpoles, Fig. 9 shows the concentration dependence curve for each. Compound 31 yielded an EC₅₀ of 2.7 μM (95% confidence interval = 2.0 to 3.7 μM), while the EC₅₀ compound 132 was 1.2 μM (95% confidence interval = 0.73 to 2.0 μM), and 4-O-methylhonokiol EC₅₀ = 1.0 μM (95% confidence interval = 0.46 to 2.2 μM). For comparison, the EC₅₀ for propofol-induced LoRR in tadpoles is 1.3 μM⁴¹. Anesthesia was fully reversible for compound 31; for animals tested with compound 132, recovery was minimal at concentrations above 3 μM. For 4-O-methylhonokiol, animals tested at a concentration of 10 μM did not recover, whereas recovery was complete at 3 μM and lower concentrations.

Figure 9: Concentration-response curves for loss of righting reflexes (LORR) in tadpoles for compounds 31 (closed circle), 132 (closed square), and 4-O-methylhonokiol (closed triangle).

The percent of animals anesthetized is plotted against aqueous anesthetic concentration, overlaid with logistic fits. Each point represents data from ten animals. Data were fitted to a Hill equation.

Discussion

Here we functionally characterized compounds 31, 132 and further investigated the properties of 4-O-methylhonokiol. All three compounds are potent allosteric potentiators of α₁β₂γ₂ GABA_A receptors that do not act at site 1. We suspected that they instead act through the low affinity TMD site(s) for benzodiazepines (site 3). Indeed, potentiation by compounds 31, 132 and 4-O-methylhonokiol was abolished in the triple mutant receptor α₁S269Iβ₂N265Iγ₂S280I as well as by the single mutation β₂N265I, but unaffected by the homologous mutations α₁S269I and γ₂S280I. Assuming that these TM2 mutations alter drug actions through local steric effects in adjacent TMD interfacial sites, our results indicate that of the five such sites, only the two β+/α− interfaces 1 and 2 mediate the potentiating effects of these three compounds.

We further dissected the contribution of the individual β+/α- subunit interfaces using concatenated subunit assemblies. The γβ+/α−β interface (interface 1) and αβ+/α−γ interface (2) participated differently in modulation by the three compounds studied. For all compounds the contribution of the interface 1 to drug modulation is apparently greater than that of the interface 2.

We and others have shown that the intravenous anesthetics etomidate, propofol and pentobarbital also act via TMD interfacial sites^{26,27,28,29,37}. In α₁β₂γ₂ receptors, β₂N265I reduced potentiation by all compounds, α₁S269I reduced potentiation exclusively by pentobarbital, and the γ₂S280I mutation increased potentiation by etomidate, while reducing potentiation by propofol and pentobarbital³⁷. Different sets of residues located at subunit interfaces have been photo-labeled by etomidate, barbiturate, and propofol analogs, revealing that some anesthetics selectively bind within different TMD interfaces^{23,24,33,42,43}. Additionally, mutations of residues at the β+/α− subunit interface affecting anesthetic action have been shown to affect modulation by valerenic acid. This suggests that the binding pocket for this compound is also at or near the anesthetics binding site⁴⁴. Other subunit interfaces were not investigated.

Work by our group using receptor concatenation determined that both β+/α− subunit interfaces participated equally in modulation by propofol. In contrast, modulation by etomidate was found to be more affected by the γβ+/α−β interface site (interface 1) than the αβ+/α−γ site (interface 2)³⁷. Interestingly, studies using another mutation (β₂M286W) and different concatenated subunit assemblies suggest that etomidate interactions are equivalent in the two β+/α− sites of α₁β₂γ₂ receptors⁴⁵.

The homologs of β₂N265 in β₁ and β₃ are serine and asparagine, respectively, and this single residue dramatically influences sensitivity to loreclezole³⁹, etomidate and propofol^{26,28,29,39,40}. Potentiation by compound 31 was strongly affected by this mutation while that by compound 132 was affected less. On the other hand, potentiation by 132 was severely reduced in α₁β₁γ₂ receptors, compared to α₁β₂γ₂ receptors. Potentiation by 4-O-methylhonokiol is also reduced by the β₂N265S mutation or substitution of β₁ for β₂³⁶.

In subunit specificity studies, compounds 31 and 132 potentiated receptors containing the α₄ and α₆ subunits, contrasting with benzodiazepine site 1 agonists. Both compounds also potentiated receptors carrying the δ subunit, although to different degrees. Thus, these compounds not only acted at receptors shown to be located synaptically as α₁β_2–3γ₂, α₂β₂γ₂ and α₃β₂γ₂⁴⁶, but also at α₅β₂γ₂ receptors and receptors containing the δ subunit, which are all located extra-synaptically^5,46,47. Similarly, SJM-3 modulates both synaptic and extrasynaptic receptors³⁵.

The similarities between the three compounds we studied and the clinical anesthetics propofol, etomidate and pentobarbital suggested their possible use as sedative-hypnotics in animals. Indeed, all three compounds induced reversible loss of righting reflexes (LoRR) in Xenopus laevis tadpoles with EC₅₀s comparable to the anesthetics propofol⁴¹, and etomidate⁴⁸. However, LoRR was not reversible with high concentrations of compound 132 and 4-O-methylhonokiol.

In summary, the newly identified compounds 31 and 132 modulate both synaptic and extrasynaptic GABA_A receptors at molecular sites different from the classical benzodiazepine pocket. These compounds, together with 4-O-methylhonokiol, act through β+/α− TMD interfaces, with strongest effects through the interface 1. These compounds potently produced LoRR in aquatic animals and thus may be useful lead compounds in the search for novel anesthetic, sedative-hypnotic or anxiolytic drugs.

Methods

Construction of mutated receptor subunits

The point mutations α₁S269I, β₂N265I, β₂N265S and γ₂S280I were prepared using the QuikChange^TM mutagenesis kit (Stratagene, Agilent Technologies, Basel, Switzerland).

Construction of concatenated subunits

Construction of tandem and triple subunit cDNAs. The tandem construct γ₂-β₂, and triple construct α₁-β₂-α₁ has been described previously¹². Site-directed mutagenesis of β₂N265 to I was done in the tandem construct and the triple construct using the QuikChange^TM mutagenesis kit (Stratagene, Agilent Technologies, Basel, Switzerland).

Expression of GABA_A receptors in Xenopus oocytes

Capped cRNAs were synthesized (Ambion, Austin, TX, USA) from the linearized plasmids with a cytomegalovirus promotor (pCMV vectors) containing the different subunits, respectively. A poly-A tail of about 400 residues was added to each transcript using yeast poly-A polymerase (United States Biologicals, Cleveland, OH, USA). The concentration of the cRNA was quantified on a formaldehyde gel using Radiant Red stain (Bio-Rad) for visualization of the RNA. Known concentrations of RNA ladder (Invitrogen) were loaded as standard on the same gel. cRNAs were precipitated in ethanol/isoamylalcohol 19:1, the dried pellet dissolved in water and stored at −80 °C. cRNA mixtures were prepared from these stock solutions and stored at −80 °C.

Animal experiments were carried out in strict accordance to the Swiss ethical guidelines, and have been approved by the local committee of the Canton Bern Kantonstierarzt, Kantonaler Veterinärdienst Bern (BE85/15). Surgery was done under anesthesia, and all efforts were made to diminish animal suffering. Xenopus laevis oocytes were prepared, injected and defolliculated as described previously^49,50. Oocytes were injected with 50 nL of the cRNA solution containing wild type or mutated rat α₁, β₂ and γ₂ subunits of the GABA_A receptors at a concentration of 10 nM:10 nM:50 nM⁵¹. For concatenated tandem and triple constructs, cRNA combinations ratios of 25: 25 nM were used. Injected oocytes were incubated in modified Barth’s solution at 18 °C for at least 24 h before the measurements.

Functional characterization of GABA_A receptors

Currents were measured using a modified two-electrode voltage clamp amplifier Oocyte clamp OC-725 (Warner Instruments) in combination with a XY-recorder (90% response time 0.1 s) or digitized at 100 Hz using a PowerLab 2/20 (AD Instruments) using the computer programs Chart (ADInstruments GmbH, Spechbach, Germany). Tests with a model oocyte were performed to ensure linearity in the larger current range. The response was linear up to 15 μA. The holding potential was −80 mV. The perfusion medium contained 90 mM NaCl, 1 mM KCl, 1 mM MgCl₂, 1 mM CaCl₂, and 5 mM Na-HEPES (pH 7.4). Concentration response curves for the compounds were fitted with the equation I(c) = I_max/[1 + (EC₅₀/c)ⁿ], where c is the concentration of the compound, EC₅₀ the concentration eliciting half-maximal current amplitude, I_max is the maximal current amplitude, I the current amplitude, and n is the Hill coefficient. Maximal current amplitudes (I_max) were obtained from the fits of the concentration-response curves. For all receptors studied, modulation was measured at a GABA concentration eliciting 0.5–1.5% of the maximal GABA current amplitude. GABA was applied twice alone for 20–60 s, and then in combination with the different compounds for 45 s or 1 min. The duration of washout periods was 4 min in between agonist or agonist/drug aplications to prevent receptor desentization. At the beginning of the experiments, GABA applications were repeated when the elicited current amplitude altered by >5%. Potentiation was calculated by the following equation: (I_{Modulator + GABA}/I_GABA − 1) * 100%. The perfusion solution was applied through a glass capillary with an inner diameter of 1.35 mm, the mouth of which was placed about 0.4 mm from the surface of the oocyte. This allowed fast changes in agonist concentration around the oocyte. The rate of change was estimated 70% in less than 0.5 s¹⁴. The perfusion system was cleaned between drug applications by washing with DMSO to avoid contamination. All media contained a final concentration of 0.5% DMSO (v/v) to ensure drug solubility.

All data are from at least two different batches of oocytes. Data represent mean ± SD. An unpaired t test was used to compare two means. One-way analysis of variance (ANOVA) was used for multiple comparisons followed by a Tukey post hoc test. *p < 0.05; **p < 0.01; ***p < 0.001.

Loss of righting reflex assay in Xenopus tadpole

Animals were used and experiments were carried out with approval and according to the guidelines of the MGH Institutional Animal Care and Use Committee. General anesthetic potency was assessed in Xenopus laevis tadpoles as previously described^41,52,53. In brief, groups of 10 tadpoles were placed in aqueous solutions containing compound 31, 132, or 4-O-methylhonokiol, and tested every five minutes for loss of righting reflexes (LoRR). Each animal was assigned a score of either awake or LoRR, and the percent of animals anesthetized was plotted against the concentration of the compound tested. Concentration-response data was fitted by non-linear least squares to logistic functions of the form Y = 1/(1 +10^((LogEC₅₀ − Log[Drug])*HillSlope)) using Graphpad Prism 6. Results are reported as EC_50s and 95% confidence intervals.