GABAA receptors meet all of the pharmacological requirements necessary to be considered important targets for the action of general anesthetic agents in the mammalian brain. In the following patch-clamp study, the relative modulatory effects of 2,6-dimethylcyclohexanol diastereomers were investigated on human GABAA (α1β3γ2s) receptor currents stably expressed in human embryonic kidney cells. Cis,cis-, trans,trans-, and cis,trans-isomers were isolated from commercially available 2,6-dimethylcyclohexanol and were tested for positive modulation of submaximal GABA responses. For example, the addition of 30 μM cis,cis-isomer resulted in an approximately 2- to 3-fold enhancement of the EC20 GABA current. Coapplications of 30 μM 2,6-dimethylcyclohexanol isomers produced a range of positive enhancements of control GABA responses with a rank order for positive modulation: cis,cis > trans,trans ≥ mixture of isomers > > cis,trans-isomer. In molecular modeling studies, the three cyclohexanol isomers bound with the highest binding energies to a pocket within transmembrane helices M1 and M2 of the β3 subunit through hydrogen-bonding interactions with a glutamine at the 224 position and a tyrosine at the 220 position. The energies for binding to and hydrogen-bond lengths within this pocket corresponded with the relative potencies of the agents for positive modulation of GABAA receptor currents (cis,cis > trans,trans > cis,trans-2,6-dimethylcyclohexanol). In conclusion, the stereochemical configuration within the dimethylcyclohexanols is an important molecular feature in conferring positive modulation of GABAA receptor activity and for binding to the receptor, a consideration that needs to be taken into account when designing novel anesthetics with enhanced therapeutic indices.
Intravenous sedatives and general anesthetics are some of the most common therapeutic agents used during surgery. Several of these agents (e.g., propofol and etomidate) are postulated to sedate patients and render them unconscious through positive modulation of GABAA receptor currents in the central nervous system (Franks and Lieb, 1994; Krasowski and Harrison, 1999; Olsen and Li, 2011). GABAA receptors are membrane-spanning chloride-selective ion channel complexes activated through the binding of GABA (Barnard et al., 1998) and they are the predominant ionotropic receptor type for fast inhibitory neurotransmission in the mammalian central nervous system. Their pentameric structure is composed of different subunits (α1–6, β1–4, γ1–3, δ, ε, π, and θ) with the predominant GABAA receptor combination of α1β2γ2 in mammalian neurons (McKernan and Whiting, 1996). Investigations of the action of anesthetics at GABAA receptors have revealed that for select agents, the potentiation of GABA currents correlates with anesthetic potency in vivo (Krasowski et al., 2001; Watt et al., 2008; Hall et al., 2011).
Given the interest in developing less toxic sedatives and anesthetics, several studies have explored the structure-activity relationship for agents that enhance GABA-evoked currents and provide anesthesia (e.g., Krasowski et al., 2001; Pejo et al., 2014). Previously we demonstrated the potential for cyclohexanols to act as positive modulators of GABAA receptor currents and as general anesthetics (Hall et al., 2004; Watt et al., 2008). The structure-activity relationship for a range of cyclohexanol analogs was further explored; among those tested, 2,6-dimethylcyclohexanol was determined to be the most potent for both receptor modulation and as a general anesthetic (Hall et al., 2011).
Stereoselectivity for positive modulation of GABAA receptor currents is not unprecedented, particularly in regard to enantiomers of general anesthetics (e.g., Hall et al., 1994; Tomlin et al., 1998). Likewise, cyclohexanol-based compounds (e.g., menthol) have also been shown to exhibit stereoselectivity of action for these receptors (Corvalán et al., 2009). In the following study, we used WSS-1 cells to investigate modulation of wild-type GABAA receptors (α1β3γ2s) by cis,cis-, trans,trans-, and cis,trans-diastereomers of 2,6-dimethylcyclohexanol (Fig. 1). In the most stable of the two possible chair conformations, cis,cis-dimethylcyclohexanol has both methyl groups equatorial and the hydroxyl group axial. The cis,trans-isomer has the hydroxyl group and one methyl substituent equatorial, whereas the other methyl is axial. In the trans,trans-isomer, the hydroxyl group and both methyl groups are all equatorial, making it the most stable configuration and closest to planarity. Molecular modeling studies were carried out for the cis,cis-, trans,trans-, and cis,trans-diastereomers of 2,6-dimethylcyclohexanol against the β3 subunit of the human GABAA receptor. The results indicate that stereochemical configuration within the dimethylcyclohexanols is an important molecular feature in conferring positive modulation of GABAA receptor activity and for binding to the receptor.
Materials and Methods
WSS-1 cells (CRL-2029; American Type Culture Collection, Manassas, VA) were used for all electrophysiology experiments. WSS-1 cells are human embryonic kidney (HEK) cells that have been stably transfected with cDNAs encoding for the rat α1 and γ2s subunits of the GABAA receptor along with expression of an endogenous human β3 subunit (Wong et al., 1992; Davies et al., 2000) and thus are a convenient cell line for generating GABA-evoked currents consistently. WSS-1 cells were grown in standard media (90% Dulbecco’s modified Eagle’s medium and 10% fetal bovine serum, with 100 U/ml penicillin and 100 μg/ml streptomycin) including 500 μg/ml geneticin (G-418) to select for cells expressing GABAA receptors (Wong et al., 1992). Cells were maintained in culture flasks in a humidified incubator with 5% CO2/95% air at 37°C and passaged on a weekly basis. During passaging, cells were either plated on poly(l-lysine) (Trevigen, Gaithersburg, MD)–coated glass coverslips for electrophysiological recording or were used to reseed another flask. Cells were used for up to 30 passages after purchase from ATCC. All culturing reagents were purchased from Life Technologies (Carlsbad, CA) unless stated otherwise.
Electrophysiological recordings were performed using a standard whole-cell patch-clamp technique at room temperature. Coverslips were transferred to a recording chamber that was continuously superfused at 3 ml/min with extracellular recording medium containing 140 mM NaCl, 5 mM KCl, 1.2 mM MgCl2, 2.5 mM CaCl2, 11 mM glucose, and 5 mM HEPES (pH 7.4 with NaOH). The electrode solution contained 140 mM KCl, 2 mM MgCl2, 11 mM EGTA, 0.1 mM Mg2+-ATP, and 10 mM HEPES (pH 7.4 with KOH). Pipettes, fabricated using a Flaming/Brown micropipette puller (Sutter Instrument Company, Novato, CA), typically had resistances in the range of 2–4 MΩ. Pipettes were maneuvered onto cells to form “gigaohm” seals using a micromanipulator (MP-225; Sutter Instrument Company). Junction potentials were zeroed in the chamber prior to all recordings, the liquid junction potential was negligible (approximately 2 mV), and cells were routinely voltage clamped at −50 mV.
Drugs were superfused onto cells using a motor-driven exchange device (Rapid Solution Changer, RSC-100; Bio-Logic Science Instruments, Claix, France) controlled via Clampex 10 acquisition software (Molecular Devices/Axon Instruments, Sunnyvale, CA). Flow of extracellular solutions onto cells was driven by a multichannel infusion pump (KD Scientific, Holliston, MA). Currents evoked by GABA with and without the modulators in extracellular solution were amplified via an Axopatch 200A (Molecular Devices/Axon Instruments), filtered at 1 kHz via a low-pass Bessel filter (Frequency Devices, Ottawa, IL), and digitized using a Digidata 1440 (Molecular Devices/Axon Instruments). All currents were measured using Clampex 10 (Molecular Devices/Axon Instruments) and were further analyzed using Origin software (OriginLab Corp., Northampton, MA). Data are expressed as the mean ± S.E.M. calculated from at least five individual cells for each data point reported (unless stated otherwise). Positive modulation of GABA-induced currents was defined as the percentage increase of the control GABA response (average of pre- and postdrug). Concentration-response data were fitted with the Hill equation (eq. 1) using Origin software (OriginLab Corp.):(1)where I is the agonist-evoked current at a given concentration, Imax is the peak current at saturating [agonist], EC50 is the concentration of agonist that elicits a half-maximal response, and nH is the Hill coefficient.
Drugs and Reagents
All reagents were purchased from Sigma-Aldrich (St. Louis, MO) unless stated otherwise. During experiments, GABA and the cyclohexanol isomers were co-applied to assess the level of current modulation. Stock solutions (10–100 mM) of the 2,6-dimethylcyclohexanol isomers in dimethylsulfoxide were diluted daily to the required concentrations in extracellular medium with dimethylsulfoxide concentrations in final solutions never exceeding 0.1% (a concentration that had no effect on GABA-activated currents).
Isolation of 2,6-Dimethylcyclohexanol Isomers
2,6-Dimethylcyclohexanol from Acros (Geel, Belgium) was a mixture of 45% cis,cis-isomers, 32% trans,trans-isomers, and 23% cis,trans-isomers (Fig. 1). Thin-layer chromatography (TLC) was performed as follows: eluant, 10% diethyl ether in petroleum ether; Rf = 0.1, a mixture of trans,trans- and cis,trans-diastereomers; Rf = 0.2, cis,cis-diastereomer; both spots were visualized with vanillin.
2,6-Dimethylcyclohexanone was obtained from Sigma-Aldrich as a mixture of 80% cis-isomers and 20% trans-isomers. TLC was performed as follows: eluant, 5% diethyl ether in petroleum ether; Rf = 0.23 (trans), 0.42 (cis), visualized with KMnO4.
2,6-Dimethylcyclohexanone (1.50 g, 11.9 mmol) was applied to a column of silica gel (150 g) and eluted with 5% ethyl ether in petroleum ether to yield cis-2,6-dimethylcyclohexanone (1.01 g, 8.00 mmol), a mixture of both isomers consisting of 20% cis-2,6-dimethylcyclohexanone and 80% trans-2,6-dimethylcyclohexanone (115 mg, 0.910 mmol), and trans-2,6-dimethylcyclohexanone (72.1 mg, 0.570 mmol).
For cis-2,6-dimethylcyclohexanone, TLC was as follows: eluant, 5% diethyl ether in petroleum ether; and Rf = 0.42, visualized with KMnO4. 1H nuclear magnetic resonance (NMR) (300 MHz, CDCl3) was as follows: 1.0 (d, 6 H), 1.2–1.4 (m, 2 H), 1.7–1.9 (m, 2 H), 2.0–2.2 (m, 2 H), and 2.3–2.5 (m, 2 H)
For trans-2,6-dimethylcyclohexanone, TLC was as follows: eluant, 5% diethyl ether in petroleum ether; and Rf = 0.23, visualized with KMnO4. 1H NMR (300 MHz, CDCl3) was as follows: 1.1 (d, 6 H), 1.5–1.6 (m, 2 H), 1.7–1.8 (s, 2 H), 1.9–2.0 (m, 2 H), and 2.5–2.6 (m, 2 H).
Reduction of cis-2,6-Dimethylcyclohexanone with Lithium Aluminum Hydride.
An oven-dried 100-ml three-neck flask, equipped with magnetic stirrer, rubber septum, and gas inlet was filled with N2. Tetrahydrofuran (anhydrous, 20 ml) was added followed by a 1 M solution of lithium aluminum hydride in tetrahydrofuran (13.1 ml, 13.1 mmol, 1.10 Eq). cis-2,6-Dimethylcyclohexanone (1.46 g, 11.5 mmol) was then added dropwise via a syringe and the reaction was stirred at room temperature for 1 hour.
The reaction was quenched with 2 ml water added dropwise, then diluted with 5 ml 15% NaOH solution, followed by another 2 ml water. The mixture was filtered through silica gel under a vacuum and rinsed with diethyl ether (approximately 50 ml). The filtrate was transferred to a 250-ml separatory funnel. The aqueous layer was extracted with ethyl ether (3 × 50 ml), and the combined ether layers were dried over anhydrous MgSO4. 1H NMR on the crude product (1.65 g, 100%) showed a mixture of cis,cis-2,6-dimethylcyclohexanol (54%) and trans,trans-2,6-dimethylcyclohexanol (46%). Flash column chromatography of the crude product using 240 g silica gel and 10:1 petroleum ether to ethyl ether as eluant gave the cis,cis-isomer (646 mg, 42%), a combination of both isomers (41 mg, 3%), and the trans,trans-isomer (551 mg, 36%). The total yield was 1.24 g (9.65 mmol, 81%).
cis,cis-2,6-Dimethylcyclohexanol (Isomer 1).
TLC was as follows: Rf = 0.2, eluant 10% diethyl ether in petroleum ether, visualized with vanillin. 1H NMR (300 MHz, CDCl3) was as follows: 0.98 (d, 6 H), 1.16 (d, 1 H), 1.28–1.38 (m, 5 H), 1.45–1.59 (m, 2 H), 1.65–1.74 (m, 1 H), and 3.51–3.56 (m, 1 H)
trans,trans-2,6-Dimethylcyclohexanol (Isomer 2).
TLC was as follows: Rf = 0.1, eluant 10% diethyl ether in petroleum ether, visualized with vanillin. 1H NMR (300 MHz, CDCl3) was as follows: 1.07 (d, 6 H), 1.20–1.29 (m, 2 H), 1.30–1.42 (m, 2 H), 1.47 (d, 1 H), 1.56–1.65 (m, 2 H), 1.66–1.76 (m, 2 H), and 2.72 (t of d, 1 H).
Reduction of trans-2,6-Dimethylcyclohexanone with Lithium Aluminum Hydride.
It should be noted that 3 and 4 shown in eq. 2 are enantiomers and may (or may not) have equal activity dependent on whether the site is chiral (not the same activity) or achiral (same activity).
An oven-dried 25-ml two-neck flask, equipped with magnetic stirrer, rubber septum, and gas inlet, was filled with dry N2. Tetrahydrofuran (anhydrous, 1.5 ml) was added together with a 1 M anhydrous solution of lithium aluminum hydride in tetrahydrofuran (1.10 ml, 1.10 mmol). A solution of trans-2,6-dimethyl cyclohexanone (128.5 mg, 1.01 mmol) in anhydrous tetrahydrofuran (1 ml) was transferred by cannula from a pear-shaped flask and the mixture was stirred at room temperature for 1 hour.
The reaction was quenched with a few drops of water and then treated with 9% NaOH solution added dropwise. The mixture was filtered through silica gel under a vacuum, rinsed with diethyl ether (approximately 15 ml), and the filtrate was transferred to a 50-ml separatory funnel. The aqueous layer was extracted with ethyl ether (3 × 10 ml). and the ether layers were combined and dried over MgSO4. 1H NMR on the crude product (42.9 mg, 33%) revealed a mixture of cis,trans-2,6-dimethylcyclohexanol (98%) and trans,trans-2,6-dimethylcyclohexanol (2%).
cis,trans- and trans,cis-2,6-Dimethylcyclohexanol (Enantiomers 3 and 4).
TLC was as follows: Rf = 0.1, eluant 10% diethyl ether in petroleum ether, visualized with vanillin. 1H NMR (300 MHz, CDCl3) was as follows: 0.97 (t, 6 H), 1.36–1.55 (m, 6 H), 1.63–1.80 (m, 2 H), 1.90–2.04 (m, 1 H), and 3.28–3.35 (m, 1 H)
A sample of the cis,cis-isomer was also isolated directly from the commercially available mixture via column chromatography of the mixture (5 g) on silica gel (300 g) using hexane/ethyl acetate (20:1) to yield 1.1 g pure cis,cis-isomer (by 1H NMR) as a colorless liquid. Samples of the cis,cis-isomer from both isolation procedures produced similar modulation of GABA currents.
Molecular docking studies were carried out to define the mode of interaction between the GABAA receptor propofol and each diastereomer of 2,6-dimethylcyclohexanol. Given previous literature highlighting the role of β subunits in the binding of propofol to GABAA receptors (Yip et al., 2013) and the availability of a crystal structure [Protein Data Bank (PDB) identifier 4COF; Miller and Aricescu, 2014], the β3 subunit of the human GABAA receptor was the considered target. This target was prepared for the docking process by protonating, minimizing, and examining the missing side chain residues in the protein using Chimera Software (University of California, San Francisco) (Pettersen et al., 2004; Goddard et al., 2005). The prepared target was uploaded to the ProBiS server (http://probis.nih.gov/; National Institutes of Health, Bethesda, MD) (Carl et al., 2010) for the detection of binding sites using protein binding site structure similarities. The ProBiS program aligns and superimposes protein binding sites, and it enables pairwise alignments and fast database searches for similar binding sites.
We focused on propofol binding sites highlighted in previous studies (Nury et al., 2011; Yip et al., 2013; Chiara et al., 2014) and sites based on the structural similarities of the following proteins (shown by PDB identifiers): 2M6B (structure of transmembrane domains of human glycine receptor α1 subunit; Mowrey et al., 2013), 4X5T (α1 glycine receptor transmembrane structure fused to the extracellular domain of gloeobacter ligand-gated ion channel(GLIC); Moraga-Cid et al., 2015), and 3P50 (structure of propofol bound to GLIC; Nury et al., 2011). Based on these previous studies, we explored intrasubunit binding sites within chain A of a single subunit of 4COF (Miller and Aricescu, 2014). The best binding site with a Z score of 4.21 included the key amino acid residues Tyr220, Phe221, Gln224, His267, and Thr271. A second binding site with key amino acid residues Tyr143, Thr225, Pro228, Ile264, and Leu268 gave a Z score of 4.01. The high scoring binding sites were then merged because there was considerable spatial overlap between the key amino acids. Although it is recognized that intersubunit sites have been proposed for propofol binding to GABAA receptors (e.g., Bali and Akabas, 2004), sites between subunits (e.g., chains A and B) were not considered because steric clashes were encountered when modeling multiple subunits at these sites.
The protein was loaded to MGL-AutoDock Tools (Scripps Research Institute, La Jolla, CA) to define the custom binding site grid box for the docking process. In brief, polar hydrogen atoms and Kollman charges were assigned to the protein by converting to the PDBQT format of AutoDock (Morris et al., 2009). The program AutoGrid was used to generate grid maps for the custom binding site on the protein and the grid was generated based on selected residues from binding site analysis. To generate grid maps for different types of ligands (with possible hydrogen bonding), the grid parameter file was modified to include O-H bond types from ligands. To generate the custom grid maps, we defined grid points as x = 40, y = 40, and z = 50 and the grid center as x = 10.70, y = −13.28, and z = 157.35 with a spacing of 0.38 in the protein three-dimensional structure. The binding site for 2,6-dimethylcyclohexanols included Tyr143, Tyr220, Phe221, Gln224, Thr225, Pro228, Ile264, His267, Leu268, and Thr271 amino acids from the M1 and M2 domains of chain A (Fig. 2).
Propofol, cis,cis-, trans,trans-, and cis,trans-diastereomers of 2,6-dimethylcyclohexanol were all considered as ligands. The geometries were drawn in MarvinSketch (Chemaxon, Cambridge, MA) and converted to three-dimensional structures using the MM2 force field. The diastereomeric conformers were frozen for further quantum chemical optimization using the DFT/B3LYP level of theory and the 6-31G basis set in HyperChem software (Hypercube, Inc., Gainesville, FL) (see Table 1 for quantum chemical properties). The lowest energy conformer was uploaded to MGL Tools to assign Gasteiger partial charges and for the detection of torsions to rotate the bonds during the docking procedure. For all ligands, random initial positions, fixed conformers, and torsions were parameterized. The number of active torsions and the number of torsional degrees of freedom were set to default values indicated in AutoDock. A Lamarckian genetic algorithm was used for minimization using optimum parameters (from initial docking evaluations) to generate all possible energies to rank the conformers (Morris et al., 1998). For energy evaluations, we used the following docking parameters: 250,000 evaluations, 250 genetic algorithm iterations, and a population size of 150 to generate 250 docked conformers. For reliable docking results, the root mean square deviation of the lowest energy conformer and the root mean square deviation to one another were analyzed to group families of similar conformations using clustering.
cis,cis- and trans,trans-2,6-Dimethylcyclohexanols Isomers Are Positive Modulators of GABAA Receptor Currents
We investigated the modulation of submaximal GABA currents by the three isolated cis,cis-, cis,trans-, and trans,trans-2,6-dimethylcyclohexanol diastereomers along with the mixture of the isomers. WSS-1 cells (HEK cells stably expressing α1β3γ2s GABAA receptors) were routinely exposed to applications of 10 μM GABA that evoked approximate EC20 currents (effective concentration that evoked 20% of maximal current). Coapplications of 30 μM 2,6-dimethylcyclohexanols produced potentiations of the GABA responses (Fig. 3). For example, the addition of 30 μM cis,cis-isomer resulted in approximately 2- to 3-fold enhancement of the EC20 GABA current (Fig. 3B). Pre-exposure to cyclohexanols did not affect the extent of current modulation upon subsequent coapplication and no direct activation of GABAA receptor currents was observed by the cyclohexanols even at the highest concentrations (300 μM) of the isomers tested (data not shown).
Coapplications of 30 μM 2,6-dimethylcyclohexanol isomers produced a range of positive enhancements of control GABA responses with the rank order for positive modulation of GABA EC20 currents: cis,cis > trans,trans ≥ mixture of isomers > > cis,trans-isomer (Fig. 3). We confirmed the relative extent of the potentiations of the GABA receptor activity in the presence of increasing concentrations of the isomers (1–300 μM, Fig. 4). For instance, on average with the addition of 30 μM cis,cis-isomer, the positive modulation of the current (above control) was 165% ± 27% (n = 6), whereas the equivalent for the trans,trans-isomer was 92% ± 24% (n = 5). Current modulation by the cis,trans-isomer was negligible even at the highest concentration tested (at 300 μM, 5% ± 5%, n = 6). All of the cyclohexanol effects were fully reversible upon washout (as previously reported; Hall et al., 2011).
The relative potencies for positive modulation of GABA currents were further supported by recording and plotting the leftward shifts in the GABA concentration-response curves in the presence of 30 μM 2,6-dimethylcyclohexanol isomers (Fig. 5). For instance, the EC50 for the control GABA currents (approximately 21 μM) was shifted to approximately 10 μM by the trans,trans-isomer and to approximately 7 μM in the presence of the cis,cis-isomer. The cis,trans-isomer produced a negligible shift in the concentration-response relationship (Fig. 5).
Modeling the Binding of the Isomers to the β3 Subunit of the GABAA Receptor
Molecular modeling focused on regions of the β3 subunit of the GABAA receptor (4COF; Miller and Aricescu, 2014) that, in previous studies, were implicated in propofol binding to GABAA receptor β-subunits or GLIC channels (Nury et al., 2011; Yip et al., 2013; Chiara et al., 2014). The binding site prediction server ProBiS (Carl et al., 2010) gave the best binding site score with key amino acid residues Tyr143, Tyr220, Phe221, Gln224, Thr225, Pro228, Ile264, His267, Thr271, and Leu268.
This binding site is located between the transmembrane M1 and M2 helices of the β3 subunit (Fig. 2) and includes a phenylalanine at the 221 position (M1) and a histidine at the 267 position (M2) that were previously photolabeled with an ortho-propofol diazirine derivative (Yip et al., 2013). In the same study, replacement of the phenylalanine at 221 by a tryptophan residue was shown to attenuate propofol’s potentiation of GABA-evoked currents. Interestingly, the three cyclohexanol isomers were found to bind within this pocket through hydrogen-bonding interactions with a glutamine at the 224 position and a tyrosine at the 220 position plus hydrophobic interactions with the leucine at position 268, phenylalanine at position 221, and tyrosine at position 220 (Fig. 6; Tables 1–3). By comparison, propofol’s hydrogen bonding was modeled only through the glutamine residue and hydrophobic interactions with the leucine at 268, tyrosine at 220, and threonine at 225 positions (Fig. 7). The ligand efficiency, van der Waals energy, and electrostatic and H-bond energy were all considered in the calculation of binding energies and intermolecular energies for this intrasubunit site. Within the site the binding energies for the interactions (Table 2) had a rank order of propofol > cis, cis > trans,trans > cis,trans-2,6-dimethylcyclohexanol corresponding with the rank order of potencies for positive modulation of GABAA receptor currents recorded electrophysiologically [compare with Figs. 3 and 7 from Davies et al. (2000) and Hall et al. (2011), respectively].
cis-,trans-2,6-Dimethylcyclohexanol as an Inhibitor of Propofol’s Modulatory Action at GABAA Receptors
Finally, given the lack of modulation observed by cis,trans-2,6-dimethylcyclohexanol and the modeling of its binding to the receptor at a site similar to that of propofol’s, we explored the possibility that this isomer might competitively inhibit propofol's modulatory action at the receptor. As expected, submaximal GABA (3 μM) responses were potently enhanced by propofol (10 μM; Fig. 8). However, this positive modulation was only moderately attenuated by the coapplication of 100 μM cis,trans-2,6-dimethylcyclohexanol (Fig. 8). In summary, positive modulation by propofol was attenuated by 14.1% ± 1.3% (n = 4) and by 13.4% ± 2.5% (n = 5) by 100 and 300 μM cis,trans-2,6-dimethylcyclohexanol, respectively.
In the search for novel anesthetic and sedative compounds, this study investigated the potential of 2,6-dimethylcyclohexanol stereoisomers as positive modulators of GABAA receptor– mediated currents. In this study, the GABA EC50 in HEK cells expressing α1β3γ2s receptors was determined to be approximately 20 μM with a Hill coefficient of 1.6, which is reasonably consistent with other reports for similar receptor combinations using the same expression system (e.g., Ueno et al., 1996). The cyclohexanols in this study all have their aliphatic groups in the 2,6-position relative to the hydroxyl group equivalent to the commonly used intravenous anesthetic, propofol (2,6-di-isopropylphenol). Following on from studies that demonstrated the efficacy of 2,6-dimethylcyclohexanol for positive modulation of GABAA receptor currents and for inducing anesthesia (Hall et al., 2011), the major electrophysiological findings of our study are as follows. First, the mixture of 2,6-dimethylcyclohexanol isomers (3–300 µM) enhanced GABA currents evoked in HEK cells expressing α1β3γ2s receptors. Second, the cis,cis- and trans,trans-isomers of 2,6-dimethylcyclohexanol positively modulated GABA currents, with the former being marginally more potent. Finally, cis,trans-2,6-dimethylcyclohexanol had minimal effects on GABA currents.
The modulatory effects of cyclohexanols on GABA receptors have been previously investigated (Hall et al., 2011); however, in this instance, they were studied in oocytes expressing α1β2γ2s receptors (the most prevalent combination found in the mammalian brain). By comparison, the potentiation of GABA currents by 30 μM 2,6-dimethylcyclohexanol was greater in the oocyte studies (approximately 4-fold enhancement; Hall et al., 2011) than in HEK cells expressing α1β3γ2s receptor composition (approximately 1.5-fold enhancement, this study). Subunit composition, differences in expression systems and speed of agonist/modulator application probably contribute to the discrepancies in the extent of the modulations reported.
Although the 2,6-dimethylcyclohexanol mixture of stereoisomers enhanced GABA currents at concentrations of 3–300 μM, the mixture is composed of isomers that may have different modulatory effects on the receptors. Stereoselective action has been previously observed for GABA receptor modulation by anesthetic agents. For instance, the S(+)-isoflurane isomer was observed to be more effective in potentiating GABA-induced currents than the R(−)-isoflurane (Hall et al., 1994) with steroselectivity also observed for (+)- and (−)-pentobarbital (Tomlin et al., 1998). It seemed plausible, therefore, that individual isomers of 2,6-di-methylcyclohexanol would exhibit a range of potencies due to differing chemical configurations. 2,6-Di-isopropylphenol (propofol), a potent positive modulator of GABA receptor currents, consists of a hydroxyl group and two ortho isopropyl groups in the plane of the benzene ring. By contrast, all three 2,6-di-methylcyclohexanols are chair shaped and are not planar molecules. The trans,trans configuration of 2,6-di-methylcyclohexanol is the closest to planarity with the hydroxyl and two methyl groups in equatorial positions in the most stable conformer. Thus, we expected that the trans,trans-isomer would be the most potent modulator since it might fit most effectively into an equivalent propofol binding pocket within the GABAA receptor. Indeed, previous studies reported that cyclohexanols and propofol may share similar sites of action on GABAA receptors (Watt et al., 2008). The cis,cis configuration of 2,6-dimethylcyclohexanol was postulated to be the next most potent since the most stable conformation would have the hydroxyl group axial and the methyl groups equatorial. Finally, the cis,trans configuration of 2,6-dimethycyclohexanol was expected to be the least potent since one methyl must necessarily be axial thereby reducing planarity. Contrary to expectations, although the trans,trans-isomer was still an effective modulator, the cis,cis-isomer was the most potent diastereomer for positive modulation of GABA responses. As anticipated, the cis,trans-isomer had minimal impact on the modulation of GABA receptor currents, suggesting that the 23% of this isomer present in the commercially available mixture acts as pharmacological ballast with regard to GABAA receptor modulation.
In previous studies, a range of 2,6-substituted cyclohexanols with varying sizes of the aliphatic chains were assessed for anesthetic potency and for loss of righting reflex in a tadpole assay (Hall et al., 2011). 2,6-Dimethylcyclohexanol was one of the most potent in this regard with an EC50 of approximately 13 μM, with 2,6-di-isopropylcyclohexanol demonstrating equivalent potency (EC50 of approximately 14 μM). To date, yields of isolated isomers from mixtures of both 2,6-dimethylcyclohexanol and 2,6-di-isopropylcyclohexanol have been insufficient to enable similar analyses of the relative anesthetic potencies of the cis,cis- versus trans,trans-isomers in in vivo tadpole assays.
Although our electrophysiological studies were conducted on a GABAA receptor consisting of α1, β3, and γ2s subunits, the closest approximating crystal structure for the molecular modeling studies was for a homopentamer of human β3 subunits (4COF; Miller and Aricescu, 2014). The binding pocket for the cyclohexanol isomers, revealed through our modeling studies of a GABAA receptor β3 subunit, was composed primarily of residues from M1 and M2 transmembrane helices. This intrasubunit binding site was predicted by reference to previous propofol binding studies, by similarity in the templates used for the homology modeling and by identifying key amino acid residues within two sites that produced the highest Z scores. The site included a phenylalanine at the 221 position and a histidine at the 267 position (Fig. 7) that have already been implicated in propofol’s positive modulation of receptor currents and in propofol’s binding (Yip et al., 2013). Both active isomers, cis,cis and trans,trans, were also observed to bind to the receptor subunit via the histidine 267 and through hydrophobic interactions with the phenylalanine 221 (Fig. 6, A and B). It should be noted that Yip et al. (2013) observed interactions of an equivalent binding site with the main chain of the neighboring subunit. In our modeling studies, intersubunit sites of action were not explored because of steric clashes between subunits (e.g., chains A and B) at these sites that were severe even after extensive protein preparation (i.e., minimization to refine protein structure). Because this questions the availability of the proposed site in an assembled receptor, future studies will require a thorough protein minimization to refine the structure and molecular dynamics simulations to address other potential intra- and intersubunit binding sites.
Our modeling studies revealed several important points regarding the binding of the 2,6-dimethylcyclohexanols at the chosen site. First, the binding energies (Table 2) vary from −4.42 through −4.38 to −4.21 kcal/mol for the cis,cis-, trans,trans-, and cis,trans-isomers, respectively. Such small differences, although corresponding to the rank order of potency for receptor modulation, are unlikely to explain changes in current modulation from approximately 300% (300 μM cis,cis) to 0% (300 μM cis,trans). Binding energy comprises a combination of hydrophobic interactions, van der Waal’s forces, and hydrogen bonding. Figure 6 shows that the H-bond length to glutamine 224 (Gln224) increases from 1.898 Å (cis,cis) through 2.014 Ǻ (trans,trans) to 2.152 Å (cis,trans). Interestingly, these H-bond lengths are inversely related to the extent of current enhancements derived electrophysiologically, suggesting that H bonding may be an important factor in determining the extent of receptor-positive modulation (Table 3).
It is instructive to compare the modeling data for the cyclohexanols with those for propofol, which shows a binding energy of −4.96 kcal/mol. If binding energy was the determining factor, this only corresponds to an estimated modulation enhancement over cis,cis-2,6-dimethylcyclohexanol of approximately 3-fold, whereas the observed enhancements using propofol are considerably greater (e.g., compare Fig. 3B with Fig. 8). According to our modeling, propofol also forms a H bond with Gln224 with a bond length of 1.874 Ǻ only slightly shorter than that with cis,cis-2,6-dimethylcyclohexanol. There is, however, a fundamental difference. The H bond is described between the hydrogen of the OH group and the basic amide oxygen of Gln224, possibly due to the acidity of propofol (pKa of approximately 9). By contrast, the H bond between the cyclohexanols and Gln224 is between the NH2 group (in CONH2) of the glutamine and the oxygen atom of the OH group. Therefore, one may conclude that it is the nature and lengths of the hydrogen bonds that determines the degree of receptor-positive modulation, although additional contributions from the hydrophobic and van der Waal’s interactions must also contribute to the binding energy.
Although we focused on a site selected from predocking studies with the highest Z scores, many other sites had the potential to accommodate the isomers, albeit with lesser scores. Indeed, many other sites have been proposed for propofol’s interactions and modulation of GABAA receptor activity, including other sites in M1 (Chang et al., 2003), M2 (Siegwart et al., 2002), and also M3 and M4 helices (Krasowski et al., 1998; Siegwart et al., 2002; Richardson et al., 2007). For example, in previous studies, a tyrosine in the 444 position of β2 subunits proved important for modulation of the GABA currents by propofol (Richardson et al., 2007) and by menthol, a monoterpenoid with a neutral cyclohexanol chair structure (Watt et al., 2008).
The drugs investigated in this study were all positive modulators and, therefore, likely stabilize the “open” state of the ligand-gated ion channel through an allosteric mechanism. By contrast, the 4COF structure used for our modeling studies is described by Miller and Aricescu (2014) as representative of a “desensitized” state. This state is evidently distinct from that predicted to be stabilized by positive modulators. Moreover, the 4COF structure represents a modified β3 subunit of the GABAA receptor with extensive intracellular loop domains removed. Therefore, although 4COF provides an approximation of the relevant structure for binding studies, interpretation of the molecular modeling data must be viewed with caution. Furthermore, the structure of 4COF is that of a homopentameric receptor, whereas our electrophysiological recordings were derived from receptors with two additional subunits (α1 and γ2s), which could contribute either directly to drug binding or indirectly to binding site conformations. Given these caveats, other candidate binding sites may be equally or more relevant to producing the positive modulations of the receptor activity observed electrophysiologically.
Finally, we performed competition experiments to determine whether cis,trans-2,6-dimethylcyclohexanol could act as a competitive inhibitor of propofol’s action at the receptor (Fig. 8). In these recordings, we observed only modest inhibition of the positive modulation induced by propofol. This result suggests that although cis,trans-cyclohexanol produced no modulation of GABA responses, its ability to compete for a propofol binding site is limited, presenting a further caveat for overinterpretation of the modeling data.
In conclusion, the enhanced activity of cis,cis-2,6-dimethylcyclohexanol presents an interesting lead in the development of novel anesthetics, given that cyclohexanols in general are typically well tolerated (Thorup et al., 1983). Further refinement of such agents through isolation of individual isomers may lead to novel anesthetics with improved therapeutic indices.
The authors thank Harrison Hunter and Salma Bargach for helping to set up the patch-clamp electrophysiology. They also thank Dr. Andrew Jenkins (Department of Anesthesiology, Emory University, Atlanta, GA) for reading of and comments on the manuscript. G.G. Pillai thanks the University of Tartu Graduate School for Functional Materials and Technologies.
Participated in research design: Chowdhury, Croft, Goel, Zaman, Tai, Walch, Smith, Page, Shea, C.D. Hall, Jishkariani, Pillai, A.C. Hall.
Conducted experiments: Chowdhury, Croft, Goel, Zaman, Tai, Walch, Smith, Page, Shea, C.D. Hall, Jishkariani, Pillai, A.C. Hall.
Contributed new reagents or analytic tools: Smith, Page, Shea, C.D. Hall, Jishkariani.
Performed data analysis: Chowdhury, Croft, Goel, Zaman, Tai, Walch, Smith, Page, Shea, C.D. Hall, Jishkariani, Pillai, A.C. Hall.
Wrote or contributed to the writing of the manuscript: Chowdhury, Croft, Goel, Zaman, Tai, Walch, Smith, Page, Shea, C.D. Hall, Jishkariani, Pillai, A.C. Hall.
- Received September 9, 2015.
- Accepted March 22, 2016.
This work was supported by the Howard Hughes Medical Institute [Grant 52007557 (to C.J.C. and S.G.)], a Smith College Tomlinson Award [(to L.C.)], the Blakeslee Foundation [(to A.C.H.)], and the European Social Fund [Project 1.2.0401.09-0079, University of Tartu (to G.G.P.)].
- gloeobacter ligand-gated ion channel
- human embryonic kidney
- nuclear magnetic resonance
- Protein Data Bank
- thin-layer chromatography
- Copyright © 2016 by The American Society for Pharmacology and Experimental Therapeutics