GABAA receptor modulation by terpenoids from Sideritis extracts

Scope GABAA receptors are modulated by Sideritis extracts. The aim of this study was to identify single substances from Sideritis extracts responsible for GABAA receptor modulation. Methods and results Single volatile substances identified by GC have been tested in two expression systems, Xenopus oocytes and human embryonic kidney cells. Some of these substances, especially carvacrol, were highly potent on GABAA receptors composed of α1β2 and α1β2γ2 subunits. All effects measured were independent from the presence of the γ2 subunit. As Sideritis extracts contain a high amount of terpenes, 13 terpenes with similar structure elements were tested in the same way. Following a prescreening on α1β2 GABAA receptors, a high-throughput method was used for identification of the most effective terpenoid substances on GABA-affinity of α1β2γ2 receptors expressed in transfected cell lines. Isopulegol, pinocarveol, verbenol, and myrtenol were the most potent modifiers of GABAA receptor function. Conclusion Comparing the chemical structures, the action of terpenes on GABAA receptors is most probably due to the presence of hydroxyl groups and a bicyclic character of the substances tested. We propose an allosteric modulation independent from the γ2 subunit and similar to the action of alcohols and anesthetics.

or both [6,7]. Furthermore, GABA A Rs are important as they bind anticonvulsants and anesthetics. The binding site for diazepam, a highly potent anticonvulsant drug, has been studied intensively and is localized at the interface of the ␣ and the ␥2 subunits.
GABA A Rs are members of the Cys-loop receptor superfamily, which also includes nicotinic acetylcholine receptors (nAChR), glycine receptors, and the serotonin receptors (5HT 3 R). These all share a similar topological organization with a large extracellular N-terminus harboring a conserved disulfide bridge forming the eponymous Cys-loop. The N-terminus carries 10 ␤-sheets in an Ig-like structural organization and forms the orthosteric binding site, which is located between adjacent subunits of the pentameric receptor complexes [8]. Site-directed mutagenesis studies have provided much information about this binding site, as well as binding sites for the many agents that can modulate the function of these receptors [9,10].
Recently, it was shown that these receptors are also a target of volatile odorant extracts [3], and actions on GABA A Rs were described for substances from the essential oils such as geraniol or linalool [11]. A subunit-specific action of fragrant dioxane derivatives was demonstrated on ␤1-containing GABA A Rs, providing a novel tool to detect ␤1-containing neurons e.g. in neurons of the hypothalamus [12].
Odorants bind to odorant receptors at the olfactory epithelium [13], but it is not yet clear how they reach the brain. The axons of the olfactory sensory neurons could send information to second-order neurons in the olfactory bulb or these compounds could simply diffuse through lipid membranes due to their small size and high hydrophobicity [14,15]. Diazepam, for example, is able to pass the blood-brain barrier due to its hydrophobic nature [16]. An in vivo study has demonstrated that inhalation of essential oil extracted from Abies sachalinensis resulted in much higher concentrations of odorant compounds in the brain when compared to injection into the peritoneum [17]. Nevertheless, more detailed studies are required to solve this issue. Another behavioral study with the species Sideritis clandestina demonstrated the anxiolytic and antioxidant potential of plant extracts. Mice that had ad libitum access to S. clandestina tea over a period of 6 weeks showed decreased thigmotaxis time, an enhanced number of entries into a central area, and enhanced levels of reduced glutathione, a marker protein for antioxidant capacity [18].
Here, we have analyzed several odorant compounds originally identified in Sideritis species. Using GC analysis, we were able to show single components of these extracts and tested them separately on ␣1␤2␥2 or ␣1␤2 receptors expressed either in HEK293 cells or in Xenopus oocytes. Some terpenes, and also various terpene-derived compounds potentiated the GABAergic responses. A search for other terpenes with structural similarities revealed 13 additional candidates. A detailed study of the functional effects of these compounds has allowed us to identify, which structural features are important for their modulatory properties.

Preparation of Sideritis extracts
Extracts containing the volatile fraction of an infusion of Sideritis were prepared as follows: 5 g of crushed plant material was added to 150 mL of boiled water and stirred at 70ЊC for 1 h. After filtration and cooling to room temperature, the infusion was mixed with 50 mL DCM and the volatile fraction was separated via solvent assisted flavor evaporation [19]. After separation, the organic phase was dried over anhydrous Na 2 SO 4 and concentrated using a Vigreux column and microdistillation [20]. The extract was reduced to a final volume of 200 L. Samples of Sideritis arguta, Sideritis condensata, Sideritis stricta, and Sideritis sipylea were extracted and diluted to an appropriate content with DCM, before analyses.

GC-MS
Analyses were performed on a Thermo Trace GC Ultra (Thermo Scientific, S.p.A., Rodano, Italy) coupled with an ITQ 900 IT mass spectrometer. Analytical capillaries used were DB5 and DB-FFAP (J&W Scientific, Fisons Instruments, Mainz, Germany), in the dimension of 30 m × 0.25 mm × 0.25 m. The GC temperature program started from initial 40ЊC, held for 7 min and increasing up to 250ЊC at a rate of 8ЊC/min, held for 15 min. Helium was used as carrier gas at constant flow of 0.8 mL/min and the injected sample volume was 2 L. The temperature was set to 250ЊC for the transfer line and to 200ЊC for the ion source. MS detection mode was EI with a current of 70 eV (Full scan, m/z  comparing the obtained mass spectra and retention indices in the samples to the spectra of reference substances injected in parallel [21]. Separation was performed on two different analytical gas chromatographic phases.

Whole-cell patch clamp technique
Current amplitudes were measured by the patch clamp technique in a whole-cell configuration. Current signals were amplified with an EPC-9 amplifier (HEKA, Goettingen, Germany). After transfection (24-48 h), whole-cell recordings from HEK293 cells were performed by application of ligand using a U-tube. The external buffer consisted of 137 mM NaCl, 5.4 mM KCl, 1.8 mM CaCl 2 , 1 mM MgCl 2 , 5 mM HEPES, pH adjusted to 7.2 with NaOH; the internal buffer was 120 mM CsCl, 20 mM N(Et) 4 Cl, 1 mM CaCl 2 , 2 mM MgCl 2 , 11 mM EGTA, 10 mM HEPES, pH adjusted to 7.2 with CsOH. Recording pipettes were fabricated from borosilicate capillaries with an open resistance of 3-6 M⍀. Current responses were measured at a holding potential of −60 mV. The ligand GABA was dissolved in extracellular buffer. Substances alone solved in DMSO and diluted in extracellular solution did not elicit a current response (data not shown). Similarly, untransfected cells did neither respond to GABA or to GABA plus modulator (data not shown).

Oocyte preparation
Following surgical preparation of the oocytes from the ovaries of Xenopus laevis, oocytes were isolated by enzymatic digestion at 19ЊC for 3-4 h with 600-700 U/mL type 2 collagenase from Clostridium histolyticum (CLS 2, Worthington, Lakewood, NJ, USA) dissolved in OR2 solution containing 82.5 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , and 1 mM HEPES (pH 7.4 with Tris). Defollicularized oocytes were obtained from the In-stitute of Cellular and Molecular Physiology (University of Erlangen-Nuernberg, Germany). The cDNAs of rat ␣1 and ␤2 GABA A subunits were subcloned into the pSGEM vector. Plasmids encoding for the human ␣1, ␤2, and ␥2 subunits were provided by Bjarke Ebert, Lundbeck (Valby, Denmark) and directly used for in vitro transcription. The corresponding cRNA was synthesized with the help of a mMessage mMachine T7 RNA polymerase kit (Ambion, Austin, TX, USA), following the manufacturer's protocol. To express a functional GABA A R, 1.2-2.4 ng of a 1:1 mixture of rat ␣1 and ␤2 receptor subunit cRNA was injected using a Nanoject II injector (Drummond Scientific, USA). The human ␣1␤2␥2 receptor configuration was expressed by injecting 1.56-3.12 ng of a 1:1:3 mixture of the corresponding cRNA. The oocytes were incubated in ND96 buffer (96 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 5 mM HEPES, pH adjusted to 7.4 with NaOH) at 16-18ЊC for 24 h, before use.

Electrophysiological recordings
The oocytes were used after a 24 h incubation period. Activity of GABA A Rs was measured with the two-electrode voltage clamp technique. Recording pipettes (GB 150F-10, Science Products, Hofheim, Germany) were manufactured with a P97 micropipette puller (Sutter Instruments, USA) and filled with 3 M KCl to give a resistance from 0.2-2 M⍀. Oocytes were clamped at −50 mV using a Turbo Tec-03x npi amplifier (npi electronic GmbH, Tamm, Germany) and constantly superfused with ND96 during recording. Current traces were recorded at 300 Hz and filtered at 200 Hz using cell works software.
The concentration for GABA used in these experiments was 1 M corresponding to an EC 5-10 , as it has been previously shown that the use of a low GABA concentration is required to observe small effects by the modulators [2,11]. Modulators were tested using GABA plus increasing concentrations of compound and with a control (GABA only) response after every compound application. Washout time between measurements was at least 3 min and was prolonged to 6 min for the highest concentrations. To exclude nonspecific measurements, only reversible effects were taken into account. Recordings were repeated with at least three oocytes from two independent batches. All experiments were carried out at room temperature (∼22ЊC).

Cell culture and preparation
HEK293 cells were grown on 90 mm tissue culture plates containing DMEM/Glutamax medium (Invitrogen, Paisley, UK), supplemented with 10% fetal calf serum, at 37ЊC and 5% CO 2 in a humidified atmosphere. Cells were transiently transfected using polyethyleneimine (PEI, 25 kDa, linear,  Polysciences, Eppelheim, Germany). For this purpose, 5 g of human ␣1 and ␤2 were diluted together with 10 g of ␥2 subunit DNA in 1 mL of serum-free medium followed by the addition of 90 L PEI (1 mg/mL). After 10 min incubation at room temperature, the mixture was added dropwise to a 80-90% confluent plate, and incubated for 24 h. Cells were transferred to black 96-well plates (Greiner Bio-One, Stonehouse, UK), which were pretreated with 0.01% Cultrex poly-L-Lysine (Trevigen, Gaithersburg, MD, USA). After an additional 24 h, incubation cells were used for experiments.

Dye loading and compound preparation
Cells were washed twice with FLEX buffer (115 mM NaCl, 1 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 10 mM HEPES, 10 mM D-glucose, pH adjusted to 7.2 with NaOH). A total of 100 L fluorescent membrane potential (FMP) blue dye (FLIPR Membrane Potential Blue Kit, Molecular Devices, Wokingham, UK) in Flex buffer (dilution 1:1000) was then added to each well. In case of modulation experiments, the appropriate concentrations of the substances were directly added with dye to the plates. The plates containing the transfected cells and the fluorescent dye (± a distinct concentration of modulatory substance) were incubated for 30 min at 37ЊC before starting the measurement. Then, another plate containing the appropriate GABA concentrations diluted in FLEX buffer was prepared.

Fluorometric assay
Dye-loaded cells (± a distinct concentration of modulatory substance) and the plate containing various GABA concentrations were transferred to the FlexStation II Device (Molecular Devices). Fluorescence measurements were performed using Softmax Pro software. Before GABA was applied, a baseline was recorded for 20 s. Each well was read five times with high PMT adjustment. Fluorescence values were processed using a peak kinetic reduction algorithm.

Analysis of data
Fluorescence data were normalized and the dose-response curves were generated using Prism 4 (GraphPad Software) with a sigmoidal dose response (variable slope) equation. Statistical analysis was performed using GraphPad InStat 3.0 (Graphpad Software). Comparisons with control application were made with one-way ANOVA method and Dunnett post hoc test. Within a particular concentration, the Tukey post hoc test was used to determine substance specific effects. p-values of significance were indicated in the appropriate figures and within the text.
These substances were tested for their potentiating effect on GABA-evoked currents in injected Xenopus oocytes and transfected HEK293 cells expressing ␣1␤2 or ␣1␤2␥2 GABA A Rs (Fig. 2). The presence of the ␥2 subunit was tested using a coapplication of Zn 2+ , which only inhibits ␣1␤2 receptors (data not shown) [22], although recordings from both cell types yielded similar results arguing for an effect independent from the ␥2 subunit. Application of compounds in the absence of GABA did not elicit any currents (data not shown). Pinenes had little or no modulatory effects; the current values using high concentrations (1mM) of either ␣-pinene or ␤-pinene showed only a small increase in the GABA-evoked responses of 0.35 ± 0.08 A (compared to 0.25 ± 0.08 A without ␣-pinene) and 0.35 ± 0.08 A (compared to 0.24 ± 0.06 A without ␤-pinene), in Xenopus oocytes and 1.37 ± 0.66 nA for ␣-pinene compared to 1.18 ±. 0.58 nA without modulator, and 0.52 ± 0.16 nA for ␤-pinene compared to 0.29 ± 0.07 nA for GABA alone in transfected HEK293 cells (Table 1; see relative I-values of 1 M GABA). Similarly, substances such as sabinene, ␣-phellandrene, and ␤-caryophyllene did not enhance responses (Fig. 2B). A significant potentiation, however, was observed for 1-octen-3-ol, linalool, and carvacrol (Fig. 2B). Therefore, carvacrol was tested in a dose-dependent manner using concentrations from 30 M to 1 mM (Fig. 2C). The relative current amplitudes compared to GABA alone showed that carvacrol at concentrations of 100 M (1.8 fold potentiation) or greater significantly enhanced GABAergic currents (up to 4.5 fold, Fig. 2D). Thus, carvacrol was among the substances identified from Sideritis, the most potent modulator on GABA A Rs comparable to linalool and 1-octen-3-ol (Fig. 2B) [11].
To determine, if there are structural motives within the terpene family that may account for the observed modulatory effects on GABA A Rs, we probed the effects of 13 structurally related terpenes ( Table 2). Positive modulation was observed in almost all cases when the substances contained hydroxyl groups, and was enhanced when these substances showed a mono-or bicyclic character (Table 1). GABA-evoked responses were significantly potentiated up to 417% by 600 M isopulegol with I abs values of 1.25 ± 0.43 A compared to 0.3 ± 0.04 A in the absence of the volatile compound (Fig. 3, Table 1). Less effective were cedrol, theaspirane, myrtenal, 2-methyl-3-buten-2-ol, carveol, and pinocarveol, which resulted in 2-to 3.8-fold potentiation (Fig. 3, see Table 1 for absolute current values). Substances that did not potentiate responses included ␣-thujone, nerolidol, nootkatone, ␤-citronellol, terpinolene, and ␣-santonin, most of which lack a hydroxyl group (Fig.3, Tables 1 and 2). In addition to the 13 candidates resulting from a comparison of terpene structures, pinene metabolites were also investigated as pinenes were identified in all Sideritis extracts analyzed. Previous studies have shown that terpene substances may undergo major biotransformatory processes in vivo so that the active compounds might be represented by the formed derivatives [23]. Interestingly, pinene metabolites verbenol and myrtenol at 100 M were able to enhance GABA-evoked current responses up to 493 ± 82% (verbenol) and 541 ± 94% (myrtenol) of the current magnitudes of the pinene metabolite used compared to GABA application alone (Fig. 3). Significant increases of the GABA-mediated currents were also observed at lower concentrations, e.g. 10 M for myrtenol and 30 M for verbenol ( Fig. 4A-C). Potent substances were tested in a high-throughput assay using a Flexstation device. Here, transfected HEK293 cells expressing the ␣1␤2␥2 configuration of the human GABA A Rs were used. GABA EC 50 s were 1.4 M for ␣1␤2 and 20.8 ± 7.0 M for ␣1␤2␥2 GABA A Rs, comparable to previously published data (e.g. Boileau et al.) (Fig. 5A) [24]. A coapplication of diazepam (10 M) led to a 8.3-fold decrease in GABA EC 50 (Fig. 5B, Table 3). Coapplication of terpenes (300 M) myrtenol, verbenol, pinocarveol, and isopulegol together with increasing GABA concentrations caused a decrease in GABA EC 50 (Fig. 5). Myrtenol was identified as the volatile compound with the most pronounced effect on  the GABA-dose-response curve (7.4 fold decrease in GABA EC 50 ) , not dissimilar to the effect of diazepam (Table 3), while the decrease was ∼ fourfold for carveol and linalool (Fig. 5D,  E, and H).
We also demonstrated that the effects of the terpenes and terpene-derived substances are similar in rat and human GABA A R subunits. Human ␣1␤2␥2 GABA A Rs were expressed in oocytes and successful incorporation of the ␥2 subunit demonstrated using Zn 2+ : no block was obtained, in contrast to Zn 2+ inhibition for the ␣1␤2 configuration ( Fig. 6A-D) [22]. Linalool and 1-octen-3-ol (1000 M) enhanced GABA-evoked currents of ␣1␤2 GABA A Rs (Fig. 2B). The coexpression of the ␥2 subunit did not lead to significant changes in the effectiveness of GABA-current modulation by either linalool or 1-octen-3ol (300 M) (Fig. 6E). These data indicate that the action of the volatile substances was independent of the ␥2 subunit. The effectiveness of other substances and pinene metabolites on human GABA A Rs in an ␣1␤2␥2 configuration was similar to the effects observed on rat ␣1␤2 GABA A Rs in oocytes and Flexstation data using human ␣1␤2␥2 in HEK293 cells. Modulation by pinocarveol, myrtenol, and verbenol achieved significance when used at concentrations >10 M whereas for carvacrol, isopulegol, and carveol concentrations of at least 30 M were required (Fig. 6F and  G). To determine the modulatory potency of these substances, however, concentration-response curves are required. The potentiating effect of structurally similar volatile substances is, therefore, independent from the origin of the GABA A R subunits. Thus, these substances most probably interact with conserved sequence motifs of the ␣1 and/or ␤2 GABA A R subunits.

Discussion
Tea preparations from various Sideritis species are used as evening tea in Mediterranean countries. Previously, it was shown that mice with ad libitum access to Sideritis tea, improved in anxiolytic tests as well as showed enhanced antioxidative capacity in various brain structures. Various nonvolatile compounds were identified that may account for the observed effects [18]. Moreover, it was shown that the volatile fraction of Sideritis extracts modulates ␣1␤2 and ␣1␤2␥2 GABA A R subtypes [3,18]. Therefore, the action of the VOCs present in Sideritis extracts on GABA A R may be a distinct part of the underlying mechanism for a sedative action of such tea preparations in addition to effects of nonvolatile compounds. The volatile extracts potentiated GABAergic responses similar to Lavender extracts, where the main component linalool modulates GABA A R as well as other receptors such as voltage-dependent calcium channels [11,[25][26][27].
In the present study, single components identified within various Sideritis extracts have been analyzed for their ability to modulate different GABA A R subtypes. The extracts are rich in various terpenoids or terpene-derived structures including carvacrol and pinenes. Carvacrol was identified to  significantly enhance ␣1␤2 GABA A Rs with an effectiveness comparable to 1-octen-3-ol and linalool, which have been previously shown to modulate GABA A Rs [28]. Earlier studies on ␣-pinene and its modulatory effect on GABA A Rs expressed in Xenopus oocytes have demonstrated an increase in the GABAevoked currents at millimolar concentrations [29]. Our results show that the single substances ␣-pinene, ␤-pinene, and ␤-caryophyllene led to a moderate enhancement of GABA-gated responses on GABA A ␣1␤2 receptors similar to potentiation of the whole Sideritis extracts [3]. Other substances such as sabinene, ␣-phellandrene, and caryophellene oxide had no influence on GABAergic currents. However, a combinatorial effect of substances in the extract was not investigated and can, therefore, not be excluded.  Pinenes and other monoterpenes are easily metabolized, e.g. by allylic oxidation in drosophila, rodents, and in humans. The effects of such secondary metabolites on the human central nervous system might be due to molecular and biochemical similarities [30,31]. We observed that the substances, which harbored modulatory potential at GABA A Rs shared structural similarities such as their cyclic character (mono-or bicyclic), and the presence of a hydroxyl group. Accordingly, 13 similar terpenoid structures have been analyzed as individual candidates. Seven of these showed a strong modulatory effect on GABA A Rs of the ␣1␤2 subtype. The same substances acted similarly on GABA-induced currents on ␣1␤2␥2 GABA A Rs. A structural comparison revealed that almost all substances that lack modulatory potential did not contain a cyclic structure or they lacked a hydroxyl group. Recently, it was hypothesized that the adjacency of an isopropyl group to the hydroxyl group on a ring structure as present in propofol and menthol might share similar modulatory sites at GABA A Rs [32]. Similar to the data obtained from Sideritis extracts, data on single substances indicate that the modulation is independent from the presence of the ␥2 subunit [3]. The ␥2 subunit is important for the formation of the benzodiazepine-binding site between the ␣1 and the ␥2 subunit. Benzodiazepines such as diazepam are anticonvul-sant drugs used to treat patients suffering from various forms of epilepsy [33]. Diazepam is, however, structurally dissimilar from terpene structures. Therefore, a modulatory action independent of the ␥2 subunit, i.e. via a different binding interface than diazepam, is not surprising [34,35]. The Xray structure of a GABA A R homolog from prokaryotes ELIC provided evidence for two distinct binding sites for benzodiazepines an intrasubunit and an intersubunit binding site corresponding to a low-affinity and a high-affinity binding site of benzodiazepines [35]. Thus, distinct modulatory binding sites for benzodiazepines and for VOCs may also exist on the GABA A Rs.
Hydrophobicity, however, is a common feature of both substance classes. A hydrophobic nature is a requirement for the VOCs and would enable them to reach their targets in the brain. Only a few reports are available that demonstrate and discuss the route of volatile odorants in vivo besides their binding to the olfactory receptors in the nasal epithelium [36][37][38]. Odorant receptors are G-protein-coupled receptors that lead to an enhancement of cAMP in the cell, which triggers downstream signaling cascades [15,39].
These modulatory sites could have similar physiological consequences. Carvacrol, for example, was identified to harbor anxiolytic properties using an animal model of anxiety, the elevated plus maze test. In contrast, no effect on locomotor activity was observed for carvacrol. Mice pretreated with carvacrol displayed an increase in latency for the development of convulsions, however, only when used at high concentrations and less effective than (-)-borneol or citral [36,37]. Nevertheless, the concentrations of volatile odorants that indeed are present at CNS neurons are still unknown and need further studies for clarification of this issue.
Alcohols and anesthetics also modulate GABA A Rs and other members of the Cys-loop receptor family [40,41]. In vitro mutagenesis as well as X-ray crystallography of a bacterial homolog of the GABA A Rs GLIC, isolated from Gloeobacter violaceus, implied that the motion of residues in the transmembrane regions 2, 3, and 4 enable a transient communication between the inter-and intrasubunit cavities for propofol and other anesthetics [41,42]. The crystallization of GLIC in the absence and presence of ethanol intriguingly illustrated residues localized in the transmembrane cavities involved in ethanol binding. F238, for example, participates in the binding of ethanol and similar modulators, thereby stabilizing the open-state conformation [41]. Terpene structures that were able to modulate GABA A Rs are characterized by a cyclic structure and the presence of a hydroxyl group, which seem to be required for modulation. Isopulegol, verbenol, and myrtenol were identified to harbor the highest potential of modulation on ␣1␤2 as well as ␣1␤2␥2 GABA A Rs. Therefore, we postulate that they may have a similar mechanism to the action of various alcohols or anesthetics via an interaction at the GABA A Rs transmembrane domains. The enhancement of GABAergic currents mediated by VOC modulation at GABA A Rs might be explained by conformational changes following modulator interaction resulting in stabilization of the open channel.
In summary, we have identified some distinct volatiles, specifically terpenes, as modulators of the GABA A Rs providing an explanation of sedative effects. It was demonstrated by Löw et al. that in contrast to ␣1-containing receptors ␣2and ␣3-containing receptors mediate anxiolytic effects [6]. If terpenes act in a subunit-specific manner or act nonspecifically, it still needs further investigation. Recordings from GABA A Rs expressing neurons or brain slices might also prove useful to investigate the role of these substances in the target neuronal tissue. Thus, terpenes with distinct structural properties may mediate sedative or anxiolytic mechanisms involving GABA A Rs.