Curcumol allosterically modulates GABA(A) receptors in a manner distinct from benzodiazepines

Inhibitory A type γ-aminobutyric acid receptors (GABAARs) play a pivotal role in orchestrating various brain functions and represent an important molecular target in neurological and psychiatric diseases, necessitating the need for the discovery and development of novel modulators. Here, we show that a natural compound curcumol, acts as an allosteric enhancer of GABAARs in a manner distinct from benzodiazepines. Curcumol markedly facilitated GABA-activated currents and shifted the GABA concentration-response curve to the left in cultured hippocampal neurons. When co-applied with the classical benzodiazepine diazepam, curcumol further potentiated GABA-induced currents. In contrast, in the presence of a saturating concentration of menthol, a positive modulator for GABAAR, curcumol failed to further enhance GABA-induced currents, suggesting shared mechanisms underlying these two agents on GABAARs. Moreover, the benzodiazepine antagonist flumazenil did not alter the enhancement of GABA response by curcumol and menthol, but abolished that by DZP. Finally, mutations at the β2 or γ2 subunit predominantly eliminated modulation of recombinant GABAARs by curcumol and menthol, or diazepam, respectively. Curcumol may therefore exert its actions on GABAARs at sites distinct from benzodiazepine sites. These findings shed light on the future development of new therapeutics drugs targeting GABAARs.

Natural compounds isolated from plants are a rich source of novel GABA A R ligands. Some natural flavonoids, first isolated from plants used as tranquilizers in folkloric medicine, together with their synthetic derivatives, possess selective affinity for the benzodiazepine-binding site of GABA A Rs with a broad spectrum of central nervous system effects 30 . In addition, a few natural terpenoids containing ether 31,32 or hydroxyl groups [33][34][35] have been identified as positive modulators of GABA A Rs (Fig. 1a), potentiating GABAergic transmission 33,36 and thereby suppressing aberrant excitability as seen during epileptiform activity 33,37 . Two compounds isolated from the Chinese medicinal herb Acorus gramineus, α -and β -asarone (1-propenyl-2,4,5-methoxybenzol) 31,32 , act on endogenous and recombinant GABA A Rs, activating the receptor and alleviating epileptic seizures. The widely-used cooling and flavouring agent menthol (5-methyl-2-propan-2-ylcyclohexan-1-ol, Fig. 1a), the best-known monoterpene extracted from the essential oil of the genus Mentha of the Lamiaceae family, suppresses hippocampal neuronal excitation and epileptic activity by enhancing GABAergic inhibition 37 . Menthol also enhances GABA A R-mediated currents in midbrain periaqueductal grey neurons 36 , suggesting a broader spectrum of GABA A R-related pharmacotherapy in future, using menthol and related compounds. Interestingly, menthol has an alike general anaesthetic activity and similar sites of action on the GABA A Rs to the intravenous agent propofol (2,6-di-isopropylphenol), but not to benzodiazepines, steroids or barbiturates 34 . Curcumol 38 [(3 S,5 S,6 S,8aS)-3-methyl-8-methylidene-5-(propan-2-yl)octahydro-6H-3a,6-epoxyazulen-6-ol] is a sesquiterpene compound and a major bioactive component of Rhizoma Curcumae oil. Notably, it induces minimal activation of GABA A Rs on its own, but facilitates the GABA-activated current in hippocampal neurons and cell lines, which express endogenous and recombinant GABA A Rs 33 , respectively. As a result, curcumol suppresses basal and epileptic activity in animals 33 , strengthening its pharmacological efficacy as a novel allosteric GABA A R modulator. However, the molecular mechanisms underlying curcumol modulation on GABA A Rs remain to be established. By comparing the electrophysiological effects of curcumol with other known modulators, and performing mutagenesis analysis on recombinant  33 , regraphed here in a different way to assess the effect of curcumol on GABA concentration-response curve. The EC 50 and Hill coefficient values were 2.4 ± 0.4 μ M, 2.0 ± 0.6 without curcumol and 1.7 ± 0.2 μ M, 1.9 ± 0.3 with curcumol, respectively. n = 6 each group.
Scientific RepoRts | 7:46654 | DOI: 10.1038/srep46654 GABA A Rs, here we identify that curcumol as an allosteric modulator of GABA A Rs in a manner distinct from benzodiazepines, but through sites shared with menthol.

Results
Characterization of curcumol on the GABA concentration-response curve in hippocampal neurons. A previous study 33 showed that curcumol (Fig. 1a), a bioactive component of Rhizoma Curcumae oil [39][40][41] , enhanced GABA response in a concentration-dependent manner. In that study 33 , we established that at the agonist (i.e. GABA) concentration of 1 μ M, curcumol facilitated the GABA-induced current with an EC 50 of 34.4 ± 2.9 μ M. To make an obvious and significant effect of curcumol on GABA A Rs, we chose 50 μ M as the effective concentration in the present study.
We assessed the effects of curcumol on GABA concentration-response curve in hippocampal neurons by re-examination of the effect of 50 μ M curcumol on the currents induced by a wide range of GABA concentrations shown in the previous study 33 . In contrast to the previous purpose to identify the operational range of GABA concentrations by curcumol 33 , here we perform data re-analysis to generate the concentration-response curves of GABA in the absence and presence of curcumol. As shown in Fig. 1b,c, the concentration-response curves to GABA were shifted to the left by curcumol. The EC 50 (the agonist concentration that induces the half-maximal response) values in the absence and presence of curcumol were 2.4 ± 0.4 μ M and 1.7 ± 0.2 μ M, respectively. Mechanistically, the 1 μ M GABA used in the following study falls an approximate EC 10 and EC 30 (the agonist concentrations that give rise to the 10 and 30% of maximal response, respectively) concentration of GABA, in the absence and presence of curcumol, respectively (Fig. 1c). Meanwhile, the Hill coefficients in the absence or presence of curcumol were 2.0 ± 0.6 and 1.9 ± 0.3, respectively. This increase of the apparent affinity to GABA implies a potentially allosteric regulation by curcumol of GABA-mediated GABA A R response; however, the precise mechanisms underlying the action of curcumol on GABA A Rs remain not fully understood.

Interplay of curcumol and diazepam on GABA-activated currents in hippocampal neurons.
To decipher the underlying mechanisms of curcumol on GABA A Rs, we sought to determine the potential interaction between curcumol and other known GABA A R modulators, such as the classical benzodiazepine, diazepam (DZP, Fig. 1a). Cultured hippocampal neurons were exposed to GABA, DZP, and curcumol, alone or combination with each other (Fig. 2). DZP (1 μ M) alone induced negligible inward currents but significantly potentiated GABA (1 μ M)-evoked currents (Fig. 2a,b), consistent with its allosteric modulatory nature 7 . Likewise, curcumol (50 μ M) produced minimal currents on its own but substantially enhanced GABA (1 μ M)-induced currents (Fig. 2a,b), consistent with the previous observation 33 . We also compared the enhancement of GABA-activated currents by DZP or curcumol (i.e. I DZP+GABA and I Curcumol+GABA , respectively) with the sum of the independent currents induced by GABA (I GABA ) and DZP (I DZP ) or curcumol (I Curcumol ), and found that the potentiation of GABA-mediated currents by DZP or curcumol was more than additive (Fig. 2c,d). This confirmed that curcumol, like DZP, allosterically potentiates the GABA A R activation in hippocampal neurons.
Interestingly, curcumol further increased the current induced by the combination of GABA and DZP (Fig. 2a,b), and the increase (I DZP+Curcumol+GABA ) was more than additive (I DZP+GABA + I Curcumol ; Fig. 2e), supporting the notion that curcumol causes an additional enhancement of the DZP-potentiated GABA A R activation. Consistent with this, DZP also led to a further increase in the current induced by the combination of GABA and curcumol (Fig. 2a,b), and the increase (I DZP+Curcumol+GABA ) was more than additive (I Curcumol+GABA + I DZP ; Fig. 2e). Thus, GABA, DZP, and curcumol act together to facilitate the GABA A R activation in hippocampal neurons. This suggests that curcumol, as a positive allosteric modulator of GABA A Rs, likely acts at a site distinct from the benzodiazepine-binding site.

Interplay between curcumol and menthol on GABA-activated currents in hippocampal neurons.
To understand in more depth molecular mechanisms underlying curcumol modulation of GABA A Rs, we further investigated the interplay of curcumol and menthol 34,37 , both belonging to terpenoid compounds carrying hydroxyl groups (Fig. 1a). Menthol at lower concentrations (up to 100 μ M) did not activate a tangible inward current (I Men = 0; data not shown), but significantly potentiated GABA (1 μ M)-evoked currents (Fig. 3a,b), consistent with the previous observation 37 . Similarly, in an independent set of experiments from that shown in Fig. 2, curcumol (50 μ M) significantly enhanced the GABA (1 μ M)-induced currents (Fig. 3a,b), and the compound current (I Curcumol+GABA ) was more than additive (I Curcumol + I GABA ; Fig. 3c). Interestingly, curcumol-mediated enhancement (I Curcumol+GABA ) occluded the further action of menthol (100 μ M) (I Men+Curcumol+GABA ; Fig. 3a,b), with menthol unable to improve the current (I Men+Curcumol+GABA ) to more than that induced by GABA and curcumol (I Curcumol+GABA ). Conversely, the compound current (I Men+GABA+Curcumol ) amplitude to the combination of GABA, curcumol, and menthol (100 μ M) was much higher than that of GABA and menthol (I Men+GABA ) (Fig. 3a,b) and, again, more than additive (I Men+GABA + I Curcumol ; Fig. 3d). These observations, in contrast to the non-overlapping effects between curcumol and DZP (1 μ M) (Fig. 2), raise the possibility that curcumol has a similar mechanism to menthol but not DZP, and that curcumol holds a much higher efficacy than menthol (100 μ M). Curcumol would thereby occlude further action of menthol, but would have no similar effects on the modulation by DZP at GABA A Rs (Fig. 2).
To characterize the interplay between curcumol and menthol more comprehensively, we increased the concentration of menthol up to 3 mM. Menthol (3 mM) alone activated a significant inward current (Fig. 3a, referred to as I MEN ) that was blocked by a selective GABA A R inhibitor, bicuculline methiodide (1 μ M), (data not shown) 37 , and enhanced by curcumol (Fig. 3a,b). Moreover, co-application of menthol (3 mM) and GABA enhanced GABA A R activation (Fig. 3a,b) in a more than additive manner (I MEN+GABA > I MEN + I GABA , Fig. 3e). In the simultaneous presence of curcumol and menthol (3 mM) with GABA, although the overall current (I MEN+Curcumol+GABA ) was significantly greater than that induced by GABA and curcumol (I Curcumol+GABA ), there was no difference Scientific RepoRts | 7:46654 | DOI: 10.1038/srep46654 between I MEN+Curcumol+GABA and I MEN+GABA (Fig. 3a,b). This shows that curcumol did not further increase the current induced by GABA and menthol (3 mM) together. In addition, the overall current induced by GABA, curcumol, and menthol (3 mM) did not differ from the sum of I MEN+GABA + I Curcumol , or I Curcumol+GABA + I MEN (Fig. 3f). Namely, menthol at higher concentrations saturates an allosteric site for GABA A R modulation and more likely Actions of GABA-activated currents by curcumol and menthol, but not DZP, are resistant to benzodiazepine antagonist in hippocampal neurons. To underline the differential interplay between curcumol and menthol or DZP, we then examined whether actions of the above compounds were differentially affected by flumazenil (1 μ M), a benzodiazepine antagonist. When flumazenil (1 μ M) was coapplied with curcumol and GABA (Fig. 4a), curcumol still enhanced the GABA-induced current to a comparable extent  (3 mM), curcumol, plus GABA-activated current. Data represent peak current amplitude normalized to that induced by GABA (1 μ M) alone (dashed line). n = 5 each group. * P < 0.05, ** P < 0.01, *** P < 0.001, compared with the current induced by GABA alone (dashed line); N.S., not significant, # P < 0.05, ## P < 0.01, ### P < 0.001, compared as indicated, paired Student's t-test.
(210.6 ± 25.7% vs. 204.9 ± 17.4% of GABA-induced currents by curcumol in the absence and presence of flumazenil, respectively, n = 5-6 per group, P > 0.05, Fig. 4b). Likewise, the effect of menthol was also not altered by flumazenil (196.3 ± 16.5% vs. 180.8 ± 11.7% of GABA-induced currents by menthol in the absence and presence of flumazenil, respectively, n = 10-13 per group, P > 0.05, Fig. 4c,d), which was consistent the previous study performed on Xenopus oocytes expressing the α 1-β 2-γ 2 subtype of GABA A R 34 . By contrast, in the presence of flumazenil, DZP failed to enhance the GABA-induced current in hippocampal neurons (199.8 ± 27.6% vs. 103.4 ± 2.5% of GABA-induced currents by DZP in the absence and presence of flumazenil, respectively, n = 10 per group, P < 0.01, Fig. 4e,f), verifying flumazenil as a benzodiazepine antagonist. Together, these results strengthen the notion that curcumol and menthol do not share sites of action with benzodiazepines on GABA A Rs. Curcumol shares site of action with menthol, but not DZP, on the α1-β2-γ2 subtype of GABA A R. To investigate binding sites for the modulatory action of curcumol over other known modulators on the GABA A Rs (Fig. 1a), we turned to confirm the effects of curcumol, menthol, and DZP on recombinant GABA A Rs expressed in HEK-293T cells. As the α 1-β 2-γ 2 subtype constitutes the largest proportion (~60%) of GABA A Rs in the brain 4,7 and is primarily responsible for phasic GABAergic inhibition in hippocampal CA1 pyramidal neurons, we therefore used this subtype firstly to examine the actions by different modulators. Curcumol (50 μ M), or menthol (300 μ M), or DZP (1 μ M) each significantly enhanced currents induced by GABA (1 μ M) on HEK-293T cells expressing wild-type (WT) α 1-β 2-γ 2 GABA A Rs (Fig. 5a,b). This was analogous with the observation on the cultured hippocampal neurons shown above (Figs 1-4), and consistent with previous reports on the α 1-β 2-γ 2 subtype of GABA A Rs expressed in various recombinant systems 8,33,34 . These results therefore lay a foundation on which to examine the specific site(s) responsible for the modulation of GABA A Rs by curcumol and other modulators.
It has been established that a methionine residue at amino acid position 286 [transmembrane domain (TM) 3] and a tyrosine residue at position 444 (TM4) at the β 2 subunit are important for the anaesthetic actions [18][19][20][21][22][23][24][25][26] , including menthol 34 , but not benzodiazepines, on the α 1-β 2-γ 2 subtype of GABA A R. Mutations at either one of these residues to a tryptophan (i.e. M286W or Y444W) both selectively abolished menthol-mediated enhancement of GABA A R function. Given the structural similarity between curcumol and menthol (both are terpenoid compounds carrying hydroxyl groups; Fig. 1a), in addition to previous identification of the interplay between curcumol and menthol over DZP (Figs 2 and 3), we expected that these sites important for menthol would also be essential for the curcumol action. To investigate this, we exposed these modulators (Fig. 1a) to HEK-293T cells expressing mutant [α 1-β 2(M286W)-γ 2 or α 1-β 2(Y444W)-γ 2] GABA A Rs. Previous studies suggested that the GABA concentration-response relationships (i.e. the agonist concentration that induces the half-maximal response, EC 50 and Hill coefficient) for both mutant receptors are similar to those for the WT GABA A R 19,20,34 . Therefore, GABA (1 μ M) was also used to screen for modulation by curcumol (50 μ M), menthol (300 μ M), and DZP (1 μ M). We found no enhancement of either type of mutant receptor current by menthol (Fig. 5c-f), consistent with the previous study in Xenopus oocytes expressing these mutant receptors 34 . Notably, the modulation by curcumol was also abolished by inclusion of the mutations in the β 2 subunits (Fig. 5c-f). By contrast, the enhancement of mutant β 2-M286W or β 2-Y444W currents by DZP (Fig. 5c-f) was not significantly different from the WT α 1-β 2-γ 2 GABA A R (173.7 ± 13.6%, 143.8 ± 9.2%, and 162.0 ± 15.3% of GABA-induced currents by DZP on the WT, β 2-M286W, and β 2-Y444W GABA A Rs, respectively, n = 4-6 per group, P > 0.05 vs. WT). These results were comparable with the previous report studied in Xenopus oocytes 34 , which showed that flunitrazepam, another type of benzodiazepine, also reserved its allosterically modulatory effect. The lack of mutation effects on these sites to benzodiazepines 34 (Fig. 5c-f) agrees with a previous study showing that the α subunit adjacent to the γ 2 subunit determines the sensitivity to benzodiazepines in the recombinant receptors 8 . Together, these results collectively point to a notion that curcumol is an allosteric modulator for GABA A Rs in a manner distinct from benzodiazepines.
A mutation in γ2 subunit of GABA A R resistant to benzodiazepine preserves the actions of curcumol and menthol. Finally, to underpin the differential mechanisms conferring the modulatory actions of curcumol over menthol or DZP (Fig. 1a), we then examined the effects of these modulators on the mutant GABA A Rs resistant to benzodiazepine modulation. It has been established that a phenylalanine at position 77 in the γ 2 subunit is essential for the binding of benzodiazepine and the resultant regulation of GABA A Rs 6,46 . Consistent with the previous report studied in Xenopus oocytes 46 , inclusion of the F77Y mutation (Phe → Tyr) in the γ 2 subunit indeed abolished the enhancement of GABA-induced currents by DZP (1 μ M) in α 1-containing GABA A Rs (Fig. 7a,b). Similarly, the α 5-β 2-γ 2(F77Y) GABA A R also became insensitive to DZP (Fig. 7c,d). Notably, the effect of curcumol was completely preserved (α 1-containing: 155.7 ± 10.9% and 188.1 ± 18.3% of GABA-induced currents by curcumol on the WT and γ 2-F77Y GABA A Rs, respectively, n = 4-5 per group, P > 0.05, Figs 5b and 7b; α 5-containing: 170.1 ± 10.0% and 169.9 ± 11.7% of GABA-induced currents by curcumol on the WT and γ 2-F77Y GABA A Rs, respectively, n = 7-8 per group, P > 0.05 , Figs 6b and 7d). Similarly, the effect of menthol on the GABA-induced response was also largely retained (α 1-containing: 204.3 ± 25.4% and 225.5 ± 31.1% of GABA-induced currents by menthol on the WT and γ 2-F77Y GABA A Rs, respectively, n = 3-4 per group, P > 0.05, Figs 5b and 7b; α 5-containing: 190.3 ± 18.9% and 247.3 ± 28.5% of GABA-induced currents by menthol on the WT and γ 2-F77Y GABA A Rs, respectively, n = 5-8 per group, P > 0.05 , Figs 6b and 7d). Thus, the mutation in the γ 2 subunit of GABA A R resistant to benzodiazepine by no means affect the actions of curcumol and menthol. In summary, our results collectively establish the notion that curcumol exerts its facilitatory actions on GABA A Rs at sites distinct from benzodiazepine sites (Fig. 8).

Discussion
In the present study, we have shown that curcumol (Fig. 1a), a natural compound and major bioactive component of Rhizoma Curcumae oil, acts as an allosteric modulator of GABA A Rs (Fig. 1b,c) in a manner different from that of the classical benzodiazepines. Curcumol significantly potentiated the GABA A R activation in neurons in a way that did not overlap with modulation by DZP, a well-characterized benzodiazepine, but acted together with DZP to enhance receptor function (Fig. 2). By contrast, curcumol occluded the effects of menthol, another type of GABA A R modulator, at the concentration of 100 μ M, and was occluded by this compound at the concentration up to 3 mM, indicative of a shared binding site between curcumol and menthol (Fig. 3). Moreover, the benzodiazepine antagonist flumazenil had no impact on the enhancements of GABA response by curcumol and menthol, but abolished that by DZP (Fig. 4). Finally, while single mutations (M286W or Y444W) in the β 2 subunit abolished the effects of curcumol and menthol, but not DZP (Figs 5 and 6), single mutation (F77Y) in the GABA A R γ 2 subunit abolished the effects of DZP, but not curcumol nor menthol (Fig. 7). Curcumol therefore exerts its actions on GABA A Rs at sites distinct from those of benzodiazepines (Fig. 8). These findings shed more light on the modulation of GABA A Rs and could guide the development of new drugs targeting this receptor.
In line with the multifaceted physiological and pathophysiological roles of GABA A Rs in the central nervous system, the pharmacology 9,47 and the drug development 48 on these receptors have also advanced considerably in recent decades. In addition to the natural agonist GABA 5 , positive GABA A R modulators include benzodiazepines 6,7 , barbiturates 16 , steroids 17 , and anaesthetics [18][19][20][21][22][23][24][25][26] , each of which has specific binding sites on GABA A Rs. Several lines of evidence from the present study support that curcumol shares mechanisms with anaesthetics in the allosteric modulation of GABA A Rs. First, although curcumol and DZP enhanced each other's allosteric modulation (Fig. 2), curcumol and menthol reciprocally and concentration-dependently occluded each other's effects (Fig. 3), suggesting that curcumol acts on GABA A Rs via a mechanism different from that of benzodiazepines, but similar to that of menthol. Second, menthol and curcumol are both terpenoid compounds (monoterpene and sesquiterpene, respectively) with a functional hydroxyl group (Fig. 1a), a characteristic stereochemical configuration that differs from that of DZP, providing the structural basis of ligands for curcumol action independent of benzodiazepine binding sites. It is noteworthy that the structure-effect relationship of menthol indicates the importance of the hydroxyl group in these ligands 34,37 . Likewise, curdione [(3 S,6E,10 S)-6,10-dimethyl-3-propan-2-ylcyclodec-6-ene-1,4-dione], an analogue of curcumol, predominantly lacks the hydroxyl group and exhibits greatly reduced potency at the GABA A R 33 . Third, mutagenesis analysis of the GABA A R demonstrated that the TM3 and TM4 regions in the β 2 subunits are important for the potentiating effects of curcumol and menthol, but not DZP. Together with a previous study 34 showing that menthol shares general anaesthetic activity and GABA A R site of action with the intravenous agent propofol, but not with benzodiazepines, steroids or barbiturates, we determined that curcumol likely represents a new member of the anaesthetic family for allosteric modulation of GABA A Rs.
Belonging to the non-classical anaesthetic subclass of GABA A R modulators, curcumol not only shares an obvious chemical scaffold with menthol and propofol, but also contains new information about the structureactivity relationship for this particular form of GABA A R pharmacology [18][19][20][21][22][23][24][25][26] . As discussed earlier, the hydroxyl group in these compounds 33  aliphatic chain is also a prerequisite for the activity of propofol or menthol analogues, including both the allosteric modulation 19,34,49 and direct activation of GABA A Rs 50 . Accordingly, curcumol shares equivalent positioning of an isopropyl adjacent to their respective hydroxyl groups (Fig. 1a), which likely plays a major part in the interaction with GABA A Rs. Notably, curcumol preferentially enhances receptor function, which is different from propofol and menthol that hold both efficacies of allosterically enhancing and directly activating GABA A Rs. In the present study, together with the previous report 33 , curcumol at the concentrations even up to its water solubility limit (~300 μ M) 38 induced only minimal direct activation of GABA A Rs. Moreover, curcumol was more potent than menthol, but probably less than propofol 19,34 . Curcumol (50 μ M) could significantly occlude the action of menthol (100 μ M, Fig. 3). These pharmacological efficacy differences would be ascribed to the backbone structure of these compounds: propofol is a phenol (pKa 11.0, planar ring structure) and menthol is a neutral cyclohexanol (chair structure), but curcumol is an epoxy azulen (a more complex structure). A better understanding of the structurefunction relationship of curcumol interaction with GABA A Rs will aid the design of new drugs with higher efficacy and specificity for GABA A Rs.
Curcumol does not always run parallel with menthol on the modulation of GABA A Rs (Fig. 8). In the α 5-β 2 (Y444W)-γ 2 mutated GABA A Rs, while the action of curcumol was eliminated, that of menthol kept intact (Fig. 6e,f). These effects were α subunit specific, as in the α 1-β 2(Y444W)-γ 2 mutated GABA A Rs, the actions of curcumol and menthol were both abolished (Fig. 5e,f). In addition, these effects were dependent on the specific residue(s) in the β 2 subunit. In the β 2-M286W mutated GABA A Rs, both α 1- (Fig. 5c,d) and α 5-containing subtypes (Fig. 6c,d) became unresponsive to curcumol in addition to menthol. The more consensus involvement of β 2-M286 residue located at TM3 region in the GABA A R modulation implies a more direct role of this site 19,34 in conferring the anaesthetic modulation of GABA A Rs 19,34 . This is also reminiscent of an observation that GABA-induced inter-subunit conformational movements in the α 1-Τ M1-β 2-Τ M3 transmembrane subunit interface are necessary to gate the GABA A R channels 21,25 . Of note, the β 2-Y444 residue located at TM4 region is also important for anaesthetic modulation 20,34 , of which the dynamic structural arrangements 15,25 are still being actively investigated. It is definitely meaningful to further dissect these subtle variances, including the possibility that different subunit interfaces are being used in the α 1-β 2-γ 2 and α 5-β 2-γ 2 GABA A Rs for anaesthetic modulation, which would be helpful for identification of receptor subtype-selective compounds for drug development in the future. In fact, many compounds, including propofol, etomidate, avermectin, and many others have been reported to mediate their effects through the same anaesthetic site 15 . Not only, multiple propofol-binding sites [18][19][20][21][22][23][24][25][26]51 have also been identified. Nevertheless, the present identification of curcumol working in a similar way to menthol through acting at anaesthetic sites distinct from the benzodiazepine site will inspire more structural and functional studies using this novel compound.
Curcumol preferentially enhances GABA-induced GABA A R activation, its prominent feature over other known anaesthetic modulators (i.e. propofol and menthol). However, it is unlikely to open the chloride channel considerably in the absence of GABA, which gives this compound its intriguing potential to be an ideal candidate GABA A R drug. This self-limiting property of curcumol for GABA A R modulation is also reminiscent of the widely-prescribed benzodiazepines in current therapeutic use. In contrast to barbiturates, benzodiazepines 6,7 do not directly activate GABA A Rs in the absence of GABA (Fig. 2). Nevertheless, the clinical use of benzodiazepines is currently limited because their various pharmacological effects are not clearly separable by dosing. For instance, although the anxiolytic actions of benzodiazepines are observed at lower doses than their sedative actions, sedation is still a problem if benzodiazepines are used as daytime anxiolytics. Benzodiazepines also have addictive properties and are liable to be abused 13,14 , which limits their long-term use, and physical dependence and tolerance are areas of concern 7 . Considering this, curcumol holds a potential promise for the future development of novel GABA A R drugs. Importantly, curcumol not only potentiates GABA-induced GABA A R activation, but also amplifies the modulation of GABA A Rs in the presence of benzodiazepines (i.e. DZP) (Fig. 2). Therefore, as a non-classical anaesthetic modulator, curcumol and its derivatives might represent an alternative or supplementary strategy to alleviate or remove the side-effects that limit long-term and high-dose administration of benzodiazepines. However, the assumption remains under-developed yet, which needs to be carefully investigated in the future.
Curcumol is a natural compound isolated from Rhizoma Curcumae oil. Used alone or mixed in a specific type of traditional Chinese medicine, knowledge of its pharmacological effects on the central nervous system is increasing. Rhizoma Curcumae (rhizome of Curcuma; Ezhu) has been used as a condiment and home remedy in China for thousands of years, illustrating its lack of prominent toxicity in human. Rhizoma Curcumae oil has been suggested to possess pharmacological efficacy in a number of domains, including neuroprotection 39 , cognitive enhancement 40 , and anti-seizure efficacy 41 . Of the three main ingredients in Rhizoma Curcumae oil (curcumol, curcumin, and curdione), curcumol is the most potent GABA A R modulator, and probably confers, at least in part, the pharmacological effects reported above. Moreover, like most naturally derived substances, curcumol is lipophilic and readily crosses the blood-brain barrier 52 , with the maximal concentration of curcumol after intravenous injection of Rhizoma curcuma oil up to 108.85 ± 65.91, 92.38 ± 17.63 μ g/g in the liver and brain, equivalent to 458.43 ± 278.87 and 390.86 ± 74.59 μ M (both the densities of liver and brain tissue were assumed to be 1.0 g/ml), respectively. Using the radioactive [ 3 H]-curcumol, a previous study 53 demonstrated that curcumol can be rapidly and completely absorbed orally in rats; it emerged in the blood at 5 min and peaked at 15 min, respectively, after the oral administration. In addition, tissue distribution (including the penetration into the brain), drug stability and metabolism, expressing as the area under concentration time curve of curcumol, under oral administration all were comparable with that by intravenous injection 53 , supporting a more easily administration way for using this drug. Based on the pharmacokinetics of curcumol, together with the pharmacological effects on GABA A Rs, it is not surprising that curcumol is capable of targeting against the central nervous system to treat neurological diseases. Indeed, curcumol alone decreased basal locomotor activity and chemically induced seizure activity in mice 33 , confirming its effectiveness as a GABA A R modulator to target the central function. However, despite curcumol belonging to the anaesthetics class of GABA A R modulators, its anaesthetic effects remain unexplored. Of note, whether the long-term use of curcumol would produce dependence or tolerance, as with benzodiazepines, remains to be determined in the future studies. Nevertheless, the present study has contained new information about the pharmacological nature of curcumol on the central nervous system, and provides a primary basis for further in-depth studies regarding the pharmacological development of curcumol and its related drugs.
In summary, we have identified the natural compound curcumol as an allosteric modulator of GABA A Rs. Curcumol possesses an intriguing self-limiting efficacy at GABA A Rs, in addition to its mechanisms being similar to anaesthetics but independent on benzodiazepine binding sites. This work therefore suggests a novel approach to the development of drugs targeting GABA A Rs. Cell culture. Primary cultures of mouse hippocampal neurons were prepared according to previously described techniques 33 . In brief, 15-day-old embryonic C57BL/6 J mice were anesthetized with halothane. Brains were removed rapidly and placed in ice-cold Ca 2+ -and Mg 2+ -free phosphate buffered solution. Tissues were dissected and incubated with 0.05% trypsin-EDTA for 10 min at 37 °C, followed by trituration with fire-polished glass pipettes, and plated on poly-D-lysine-coated 35 mm culture dishes at a density of 1 × 10 6 cells per dish. Neurons were cultured with Neurobasal medium (Invitrogen) supplemented with B27 (Invitrogen) and maintained at 37 °C in a humidified 5% CO 2 atmosphere incubator. Cultures were fed twice a week and used for electrophysiological recording 10-20 days after plating. For neuron cultures, glial growth was suppressed by addition of 5-fluoro-2-deoxyuridine (20 μ g/ml; Sigma-Aldrich) and uridine (20 μ g/ml; Sigma-Aldrich).

Methods
Human embryonic kidney (HEK)-293T cells were cultured at 37 °C in a humidified atmosphere of 5% CO 2 and 95% air. The cells were maintained in Dulbecco's modified Eagle's medium supplemented with 1 mM L-glutamine, 10% foetal bovine serum, 50 units/ml penicillin, and 50 μ g/ml streptomycin (all from Invitrogen).
Site-directed mutagenesis. Mutations of receptor cDNA were generated with the QuikChange ® mutagenesis kit (Stratagene, La Jolla, CA) in accordance with the manufacturer's protocol using high-pressure-liquid-chromatography-purified or PAGE-purified oligonucleotide primers (Sigma-Genosys, The Woodlands, TX). All mutants were verified by DNA sequence analysis.
Functional expression of the recombinant GABA A Rs. The rat α 1, β 2, and γ 2 subunit cDNA of GABA A R were obtained from Dr. Yu Τ ian Wang (University of British Columbia, Vancouver, BC, Canada). The rat α 5 subunit cDNA was kindly provided by Dr. David H. Farb (Boston University School of Medicine, Boston, Massachusetts, USA). Transient transfection of HEK-293T cells was carried out using HilyMax liposome transfection reagent (Dojindo Laboratories). Cotransfection with a green fluorescent protein expression vector, pEGFP-C3, was used to enable identification of transfected cells for patch clamp recording by monitoring the fluorescence of green fluorescent protein. Electrophysiological measurements were performed 24-48 h after transfection.
Electrophysiology. Whole-cell recordings were made using an Axon 700A patch-clamp amplifier (Axon Instruments, Foster City, CA, USA). Membrane currents were sampled and analysed using a Digidata 1440 interface and a personal computer running Clampex and Clampfit software (Version 10, Axon Instruments). In voltage clamp mode, the membrane potential was held at −60 mV for whole-cell current recording. All electrophysiological experiments were carried out at room temperature (23 ± 2 °C).
Chemicals and drugs. The chemicals used in the present study curcumol [(3 S,5 S,6 S,8aS)-3-methyl-8-methylidene-5-(propan-2-yl)octahydro-6H-3a,6-epoxyazulen-6-ol], menthol [5-methyl-2-propan-2-ylcyclohexan-1-ol], and diazepam (DZP) [7-chloro-1-methyl-5-phenyl-3H-1,4-benzodiazepin-2-one] were purchased from Sigma-Aldrich (St. Louis, MO). Curcumol, menthol, and DZP were initially dissolved as concentrated stock solutions in dimethyl sulfoxide and subsequently diluted to the desired concentration in the standard external solution. The final concentration of dimethyl sulfoxide was lower than 0.1% and was confirmed to be ineffective alone at the same concentration in control experiments (data not shown). Other drugs were either first dissolved in deionized water and then diluted to a final concentration in standard external solution just before use or dissolved directly in the standard external solution. Drugs were applied using a rapid application technique termed the "Y-tube" method as described previously [54][55][56] . The tip of the drug tube was positioned 50-100 μ M away from the patched cells. This system allows a complete exchange of external solution surrounding a cell within 20 ms. Throughout the experiment, the bath was superfused continuously with the standard external solution.

Data analysis.
Values are expressed as the mean ± S.E.M. Groups are compared using Student's t test.
P < 0.05 was considered to be statistically significant. P and n represent the value of significance and the number of neurons or cells, respectively. Clampfit 10.5 (Molecular Devices) was used for data analysis. The smooth concentration-response curves of curcumol on facilitation of the GABA response in hippocampal neurons were drawn according to a modified Michaelis-Menten equation by the method of least squares (the Newton-Raphson method) after normalizing to the maximal GABA response: I = I max × C h /(C h + EC 50 h ), where I is the normalized value of the current, I max is the maximal response, C is the drug concentration, EC 50 is the concentration which induces the half-maximal response and h is the apparent Hill coefficient.