Abstract
Neuronal Kv7 channels are recognized as potential drug targets for treating hyperexcitability disorders such as pain, epilepsy, and mania. Hyperactivity of the amygdala has been described in clinical and preclinical studies of anxiety, and therefore, neuronal Kv7 channels may be a relevant target for this indication. In patch-clamp electrophysiology on cell lines expressing Kv7 channel subtypes, Maxipost (BMS-204352) exerted positive modulation of all neuronal Kv7 channels, whereas its R-enantiomer was a negative modulator. By contrast, at the Kv7.1 and the large conductance Ca2+-activated potassium channels, the two enantiomers showed the same effect, namely, negative and positive modulation at the two channels, respectively. At GABAA receptors (α1β2γ2s and α2β2γ2s) expressed in Xenopus oocytes, BMS-204352 was a negative modulator, and the R-enantiomer was a positive modulator. The observation that the S- and R-forms exhibited opposing effects on neuronal Kv7 channel subtypes allowed us to assess the potential role of Kv7 channels in anxiety. In vivo, BMS-204352 (3-30 mg/kg) was anxiolytic in the mouse zero maze and marble burying models of anxiety, with the effect in the burying model antagonized by the R-enantiomer (3 mg/kg). Likewise, the positive Kv7 channel modulator retigabine was anxiolytic in both models, and its effect in the burying model was blocked by the Kv7 channel inhibitor 10,10-bis-pyridin-4-ylmethyl-10H-anthracen-9-one (XE-991) (1 mg/kg). Doses at which BMS-204352 and retigabine induce anxiolysis could be dissociated from effects on sedation or memory impairment. In conclusion, these in vitro and in vivo studies provide compelling evidence that neuronal Kv7 channels are a target for developing novel anxiolytics.
Kv7 channels (formerly known as KCNQ channels), are voltage-gated K+-selective channels composed of four α-subunits, structurally organized as six transmembrane and one pore domain. Although functional channels are obtained by expression of Kv7 α-subunits alone, β-subunits belonging to the KCNE family are reported to influence the expression, pharmacology, and voltage-dependence of the α-subunits (Schroeder et al., 2000b; Wang et al., 2000; Grunnet et al., 2002). There are five known α-subunits (Kv7.1-Kv7.5), and the sequence alignment of the hydrophobic regions divides them into two groups, one comprising Kv7.1 and the other Kv7.2 to Kv7.5. Kv7.1 is found in the heart (Yang et al., 1997) and in epithelial tissues (Schroeder et al., 2000b), whereas Kv7.2 to Kv7.5 are expressed primarily in neurons and are therefore collectively referred to as neuronal (Biervert et al., 1998; Singh et al., 1998; Yang et al., 1998; Kharkovets et al., 2000). All Kv7 channels, with the exception of Kv7.5, have been linked with hereditary diseases: Kv7.1 with cardiac long QT syndrome, Kv7.2 and Kv7.3 with benign neonatal familial convulsions, and Kv7.4 with deafness (for reviews, see Robbins, 2001; Gribkoff, 2003).
The tetrameric complexes of neuronal Kv7 channels can be homo- or heteromeric with grossly comparable biophysical and pharmacological properties. The threshold for activation of Kv7 channels is around -60 mV (Yang et al., 1998), which is approximately the resting membrane potential of many neurons. A range of pharmacological tools elucidates the importance of Kv7 channels in controlling neuronal excitability. Retigabine, an atypical anticonvulsant in development for epilepsy (Rostock et al., 1996; Ferron et al., 2002), is a positive modulator of Kv7.2 to Kv7.5 channels at concentrations of 1 to 10 μM (Main et al., 2000; Schrøder et al., 2001; Tatulian et al., 2001; Dupuis et al., 2002; Passmore et al., 2003), whereas it has no effect on Kv7.1 (Tatulian et al., 2001). A structural analog of retigabine, the analgesic flupirtine (Katadolon), also enhances Kv7 channel function, although with lower potency (Ilyin et al., 2002; Korsgaard et al., 2003). The racemic mixture of Maxipost (BMS-204352) and the R-enantiomer activates Kv7.4 and Kv7.5 channels with an EC50 of 2.4 μM (Schrøder et al., 2001; Dupuis et al., 2002). BMS-204352 is a positive BK channel modulator (Gribkoff et al., 2001) that has been in development for stroke. Finally, two enhancers of neurotransmitter release (Zaczek et al., 1998), XE-991 and linopirdine, inhibit neuronal Kv7 channels with low micromolar affinity (Wang et al., 2000; Schrøder et al., 2001).
Kv7 channel activators tend to hyperpolarize or prevent depolarization of a cell and are therefore expected to affect pathological conditions underpinned by hyperexcitability. A direct correspondence between neuronal hyperexcitability and anxiety is seen in amygdala-kindled rats (Racine, 1972), which show a lowered threshold for neuronal excitability and enhanced anxiety in various tests (Rosen and Schulkin, 1998). A large body of research indicates that the amygdala is a key structure in fear/anxiety (Davis, 1992), including an induction of the neuronal marker c-fos in rats exposed to anxiety tests (Sullivan et al., 2003). Compared with most central nervous system neurons (for review, see Steriade and McCarley, 1990), the spontaneous firing rate of amygdala neurons are among the lowest in the brain, e.g., the spontaneous firing rate of some projection neurons in the amygdala are typically <1 Hz (Pare and Collins, 2000). This suggests a powerful inhibitory tone in the amygdala, preventing inappropriate fear responses (Davis, 1992; Pare and Collins, 2000). Also, provoking fear in anxiety patients increases glucose utilization, i.e., activity, in the amygdala and leads to an exaggerated fear response (Anand and Shekhar, 2003; Rauch et al., 2003).
The neuronal Kv7 channels are widely expressed in the brain, including the amygdala (Saganich et al., 2001), and the electrophysiological correlate of Kv7 channels, known as the M current (Brown and Adams, 1980), has been described in the amygdala (Womble and Moises, 1992). In this study, it is demonstrated that BMS-204352 enhances, whereas the R-enantiomer inhibits, currents mediated through neuronal Kv7 channels expressed in HEK293 cells. The observation that the S- and R-forms exhibited opposing effects on neuronal Kv7 channel subtypes was surprising and proved to be a powerful in vivo tool in determining the potential role of Kv7 channels in anxiety. BMS-204352 and the R-enantiomer were compared in various anxiety tests, with the former compound showing anxiolytic effects and the latter antagonizing these effects. This led us to conclude that Kv7 channels are novel targets for treating anxiety. Additional studies reported here support this conclusion.
Materials and Methods
In Vitro
Molecular Biology. Stable cell lines for Kv7.2/3 and 7.3/4 (Schrøder et al., 2001), Kv7.4 (Søgaard et al., 2001), Kv7.5 (Dupuis et al., 2002), the large conductance Ca2+-activated potassium (BK) channel (Ahring et al., 1997), and the GABAA receptors (Jensen et al., 2002) have been described previously. In short, for stable expression of Kv7 channels the coding regions were subcloned into mammalian expression vectors belonging to the pNS series (NeuroSearch custom-made expression vectors derived from pcDNA3 from Invitrogen, Carlsbad, CA). Host cells were HEK293 (American Type Culture Collection, Manassas, VA), and transfections were carried out using 2 μg of the appropriate construct(s) and LipofectAmine (Invitrogen) according to manufacturer's instructions. Transfected HEK293 cells were cultured in Dulbecco's modified Eagle medium (Invitrogen) at 37°C in 5% CO2 and 95% air supplemented with 10% fetal bovine serum and the appropriate selection agent(s). When the selection medium resulted in single colonies, these were picked and propagated. Expression of ion channels was confirmed functionally with patch-clamp electrophysiology.
Patch-Clamp Electrophysiology. Experiments with the HEK293 cell lines were performed in the whole-cell configuration. On the day of experiment, cells were plated on cover slips (Ø = 3.5 mm) and allowed to attach. The cover slip was then placed in a custom-made 15-μl perfusion chamber (flow rate, ∼1 ml/min) mounted on an Olympus IX-70 microscope (Tokyo, Japan). The microscope was placed on a vibration-free table (TMC, Peabody, MA) in a grounded Faraday cage. All experiments were performed at room temperature (20-22°C). An EPC-9 patch-clamp amplifier (HEKA, Lambrecht, Germany) was connected to a personal computer via an ITC16 interface and controlled by Pulse software (HEKA). Data were stored directly on the hard disc and analyzed by the IGOR software (Wavemetrics, Lake Oswego, OR). The pipette capacitance (Cfast) was compensated when a giga-ohm seal was obtained, and cell capacitance (Cslow) as well as series resistance (Rs) were compensated electronically before every application of the voltage protocol. Usually, Cslow ranged from 5 to 20 pF, and Rs was in the range 2 to 5 MΩ. All experiments with drifting Rs values (more than 100%) were discarded, and calculated voltage errors due to Rs were always <5 mV. After the establishment of the whole-cell configuration, the voltage protocol was applied to the cell every 5 s from a holding potential of -90 mV. Since the holding potential was very close to the equilibrium potential for potassium (EK,20°C = -91 mV), and since the studied channels are closed at negative potentials, a brief hyperpolarizing step to -150 mV was used to assess the degree of leak current. Due to the length (>1.5 s) of the protocol, this leak assessment was performed instead of traditional leak subtraction with a number of downscaled sweeps, and it also allowed the study of drug effects on the leak currents. The experiment was discontinued if the leak current in the control situation exceeded 10% of the voltage-activated current. Since the leak was measured at a potential with approximately the same driving force on K+-ions as the activating step, it was subtracted postexperimentally when appropriate.
Xenopus laevis Oocyte Electrophysiology. Selected oocytes were injected with 25 to 50 nl of cRNA mixture using a Pico Pump (WPI, Sarasota, FL). The cRNA mixture was GABAAR subunits α1/2, β2, and γ2s in the ratio of 2:2:3 and in a total concentration of 1 μg/μl. For two-electrode voltage-clamp experiments, an oocyte was placed in an RC26Z recording chamber (Warner Instruments, Hamden, CT) that was continuously perfused with 2.5 ml/min OR2. Recording electrodes were fabricated from BF150-110-10 borosilicate glass with filament (Sutter Instrument Company, Novato, CA) using a DMZ-Universal puller (Zeitz, Augsburg, Germany). Electrodes were initially filled with a small plug of 0.2% agarose made in 3 M KCl and then backfilled with 3 M KCl and stored in 3 M KCl until needed. Typical electrode resistances were 0.4 to 0.8 MΩ. Currents were amplified by a Geneclamp 500B (Axon Instruments, Inc., Union City, CA), low-pass filtered at 20 Hz, digitized at 500 Hz by a Digidata 1322A (Axon Instruments), and then recorded as well as analyzed by a personal computer (Compaq Evo) using the pClamp8 suite (Axon Instruments). Drug application was controlled by placing a capillary tube with an inner diameter of 1.5 mm (Modulohm 214813; Modulohm, Herlev, Denmark) 1 to 2 mm from the oocyte and connecting this through Teflon tubing to a Gilson 233XL autosampler (Gilson Medical Electric, Middleton, WI). A flow of 2.5 ml/min through the capillary tube ensured a rapid exchange of liquid surrounding the oocyte in the matter of few seconds. During the time interval between recordings, which was set at 5 min, the oocyte was perfused with OR2 through the capillary tube as well as the bath perfusion. Effects of compounds were evaluated by comparing the currents recorded from an EC20 concentration of GABA with or without compound.
Solutions: Patch-Clamp Experiments. The extracellular (bath) solution was 140 mM NaCl, 4 mM KCl, 2 mM CaCl2, 1 mM MgCl2, and 10 mM HEPES (pH = 7.40 with NaOH). Test compounds (see Chemistry) were dissolved in the extracellular ringer by dilution at least 1000 times from DMSO stock solutions (10 mM). The less than 1%thou DMSO did not have any effect on channel function. The intracellular (pipette) solution was 144 mM KCl, 5.37 mM CaCl2, 1.75 mM MgCl2, 4 mM NaATP, 0.4 mM NaGTP, 10 mM EGTA, and 10 mM HEPES (pH = 7.20 with KOH). By use of the program EqCal (Biosoft, Ferguson, MO), the concentration of free Ca2+ was calculated to 100 nM, the free Mg2+ was 0.1 mM, and MgATP was 1.45 mM.
Solutions: Two-Electrode Oocyte Experiments. The OR2 solution was 90 mM NaCl, 2.5 mM KCl, 2.5 mM CaCl2, 1 mM MgCl2, and 5 mM HEPES pH 7.4. GABA was dissolved in OR2 to a final concentration of 2 μM. Test compounds were initially dissolved at 10 mM in 100% DMSO and then diluted in the experimental solution (OR2 with 2 μM GABA).
Fitting Procedures. As previously described (Strøbæk et al., 2000), Ki values were calculated based on current versus time plots and the kinetics of drug-induced reduction in the current. This is fitted with the nonequilibrium version of the Michaelis-Menten equation: The current I at time t is a function of basal current I0 and concentration C of inhibitor. koff is the off-rate (s-1), kon is the on-rate (M-1 · s-1) and Ki = koff · kon-1 is the inhibitory constant.
The activation curves were fitted according to Boltzmann formalism; allowing an initial voltage-independent component (Ileak) and then a voltage-dependent rise to Imax with a slope of k and midpoint V0.5.
Chemistry. All compounds used in this study were synthesized in the Medicinal Chemistry department, NeuroSearch A/S (Ballerup, Denmark). The structures of the compounds synthesized are depicted in Fig. 1.
BMS-204352 (Maxipost) is the S-enantiomer of a the chiral compound (S)-3-(5-chloro-2-methoxyphenyl)-3-fluoro-6-(trifluoromethyl)-1, 3-dihydro-2H-indole-2-one. The compound was prepared (Pendri et al., 1998) by a seven-step synthesis with an overall yield of 23% and an enantiomeric excess of the S-enantiomer of 94.9%. BMS-204352 has a molecular weight of 359.7 g/mol and cLogP of 4.6.
Retigabine [D23129; N-(2-amino-4-(4-fluorobenzylamino)-phenyl)-carbamic acid ethyl ester] was obtained in an overall yield of 30% as the bis-hydrochloride by following the synthesis described in the literature (Dieter et al., 1996). The free base was isolated as a white crystalline compound by column chromatography. Retigabine as a free base is a stable compound with a molecular weight of 303.3 g/mol, provided it is stored in the dark. Retigabine has a cLogP of 2.0 and pKa of 10.8.
The R-enantiomer [(R)-3-(5-chloro-2-methoxyphenyl)-3-fluoro-6-(trifluoromethyl)-1,3-dihydro-2H-indole-2-one] was obtained from the mother liquor of the synthesis of BMS-204352 by a standard procedure (Pendri et al., 1998), giving an overall yield of 19% and an enantiomeric excess of 94.2%.
XE-991 (10,10-bis-pyridin-4-ylmethyl-10H-anthracen-9-one) was synthesized in a one-step procedure (Earl et al., 1989) from anthrone and 4-picolylchloride hydrochloride to give a yield of 30%. The yellowish crystalline compound has a molecular weight of 376.5 g/mol and a cLogP of 4.6.
In Vivo
Animals. Female NMRI or male C57 mice (20-25 g), and male Wistar rats (150-180 g) were used in behavioral studies (Taconic M&B Breeding Center, Germantown, NY). Animals were housed and habituated for at least 7 days before experiments in Macrolon III cages (20 × 40 × 18 cm) with eight mice or two rats per cage. Food (Altromin, Lage, Germany) and tap water were available ad libitum. Animals were housed on a 12-h light/dark cycle (lights on at 7:00 AM and off at 7:00 PM). All experiments were performed according to the guidelines of the Danish Committee for Experiments on Animals.
Drugs and Solutions. All drugs were administered i.p. dissolved in 5% Tween 80 15 min prior to a behavioral test unless otherwise stated. In mice, substances were administered in a volume of 10 ml/kg and in rats as 1 ml/kg volume.
General in Vivo Models
Mouse Exploratory Motility. Effect of test compounds on exploratory motility in mice was measured in automated cages (20 cm × 30 cm; TSE Systems, Bad Homburg, Germany) equipped with infrared sensors (6 × 2). Mice were individually placed in cages for 30 min, and interruptions of infrared sensor pairs were detected by a TEST control unit with the data recorded on a computer running ActiMot software (TSE Systems). Data shown are the mean distance ± S.E.M. animals traveled during the 30-min test.
Mouse Wire Grip Test. The wire grip test was used to determine the effect of test compounds on muscle relaxation/strength. Mice were suspended with both front paws on a horizontal steel wire (50 cm long, diameter 1.5 mm) attached to two poles and elevated 28 cm above a bench. The number of mice unable to maintain grip of the steel wire with both front paws and one hind paw in one of three trials was determined.
Mouse Rotarod Ataxia. Effect of test compounds on sensorimotor function was evaluated using the mouse Rotarod. The Rotarod consisted of a wooden rod (length, 50 cm; diameter, 4 cm) partitioned into sections by plastic discs (diameter, 17 cm), allowing up to eight mice to be tested simultaneously. Mice were pretrained on the day of testing, and only mice capable of walking on the rotating rod (6 revolutions/min) for 2 min were selected for drug testing. After drug administration, the mice falling off the Rotarod twice in a 2-min period were classified as ataxic.
Mouse Anxiety Models
Mouse Zero Maze. The potential of test compounds to affect anxiety levels was evaluated in the mouse zero maze (TSE Systems), an unconditioned model of anxiety (Shepherd et al., 1994). The zero maze had a total diameter of 53 cm and was elevated 50 cm above the ground. The platform (width, 2.7 cm) consisted of two open areas (wall height, 6 mm) and two closed areas (wall height, 11 cm). The walls were made of clear Perspex. The mouse was placed in the closed area facing the closed area and was allowed to explore the maze for 5 min. The initial latency to enter the open area with all four paws placed in an open area was measured. Thereafter, the Noldus Observer program (version 2.01; Noldus Information Technology, Wageningen, The Netherlands) was used to register the following five events: 1) time spent in the open area, 2) number of entries into the open areas (an open area entry was defined as the mouse having all four paws in an open area), 3) number of stretched-attend postures (defined as the mouse, while situated in a closed area, first exhibiting an elongated body posture resulting in both front paws and snout crossing over the closed-open divide, followed by rapid retraction of the whole body back to the original position), 4) number of rearings (rising on the hind paws) in the closed area, and 5) number of head dips in the open areas (scanning over the side of the maze). Data for each measure are presented as the mean ± S.E.M. Experiments were conducted in rooms where light levels were set at 20 lux (YF-1065; YFE, Taipei, Taiwan). The experimenter was blind to the treatments given to animals in all studies.
Mouse Marble Burying. Mice were placed for 1 h in novel cages (one mouse per cage, 20 × 30 cm) in which there were 20 glass marbles (15 mm in diameter) situated in four rows of five on top of 5 cm of sawdust. The mean number of glass marbles buried ± S.E.M. between 10 to 60 min was taken as an index of “anxiety”, i.e., the more marbles buried the more anxious the mouse-a marble was classified as buried when at least two-thirds was covered by sawdust (Broekkamp et al., 1986). The experimenter was blind to the treatments given to animals in all studies.
Rat Anxiety Models
Rat-Conditioned Emotional Response. Coulbourn operant chambers (Coulbourn Instruments, Allentown, PA) with associated Coulbourn pellet dispensers and operant levers were used. These chambers were housed in outer sound- and light-attenuating shells equipped with a ventilation fan that also helped to mask external noise. A single operant lever was positioned on the left side of the front panel of the operant chamber, approximately 2 cm above the grid floor. Formula P Noyes food pellets (45 mg; Research Diets, Inc., New Brunswick, NJ) could be delivered to the food magazine centered on the front wall of the chamber approximately 2 cm above the floor. A houselight (4.0 mA, 28 V) was positioned on the front wall of the chamber, above the food magazine, approximately 1 cm below the ceiling. The operant chambers were controlled and the number of lever presses recorded by MED Associates (St. Albans, VT) software.
Rats were water-deprived 24 h prior to being placed in the operant chambers and initially trained to associate lever pressing with water reward on a fixed ratio 1 reinforcement schedule. The schedule of requirement was progressed until rats had been trained to lever press for water reward on a variable interval 60-s schedule (Lattal, 1991). Training took place 5 days a week, and conditioning sessions lasted 20 min. Every training session consisted of 5-min light (L) and 2.5-min dark (D) periods in the fixed sequence L D L D L. During both L and D periods, water was delivered on the variable interval 60-s schedule. In the D periods, lever presses also elicited mild scrambled foot shocks (0.4 mA, 0.1 s, Lafayette Master shocker; Lafayette Instrument Co., Lafayette, IN) through the grid floor on an independent variable interval 20-s schedule. After approximately 10 weeks of training, lever pressing was almost completely suppressed during the D periods, and testing began.
Testing. Rats were tested with drugs twice a week on Tuesdays and Thursdays (shock off), and the other three workdays were used as baseline training sessions (shock on). Prior to every baseline session, rats were injected with physiological saline to eliminate any cues deriving from the injection. The number of lever presses during the L and D periods was determined, and an increase in response rate during the D period with no effect on response rate in the L period was indicative of an anxiolytic effect. In addition, these response rates were used to calculate a suppression ratio (SR) according to the following formula: A suppression ratio of 0 indicates that the D has evoked conditioned fear and has completely suppressed lever pressing, whereas a suppression ratio of 0.5 indicates the response rate is unchanged by the D, i.e., the complete absence of fear.
Data Analysis. All data were analyzed using SigmaStat version 2.0 (SPSS Inc., Chicago, IL). For zero maze, marble burying, and exploratory motility, data were analyzed by one-way analysis of variance followed by Dunnett's or Tukey's honest significant difference tests where appropriate. For the conditioned emotional response test in rats, separate two-way analyses of variance for the L, D, and suppression ratio measures were conducted with treatment as the between-subjects factor and trial as the within-subjects factor. Dunnett's post hoc test was used in the case of a significant main effect of treatment. In no conditioned emotional response study was a significant treatment × trial interaction found and, therefore, is not reported. Significance was set at P < 0.05.
Results
It was previously shown, that the racemic mixture of BMS-204352 and the R-enantiomer activates Kv7.4 channels expressed in HEK293 cells (Schrøder et al., 2001, 2003). In continuation of these studies and as a prerequisite to the in vivo studies, the resolved enantiomers were initially tested for their effects on Kv7 channels.
In Vitro Effects
Opposing Effects of the Enantiomers on Kv7.4 Channels.Figure 2 shows the effect of both enantiomers in comparison with retigabine on Kv7.4 channels expressed in HEK293 cells. The top panel depicts the voltage protocol. The holding potential was -90 mV, and a brief (50 ms) hyperpolarizing step to -150 mV was performed to assess an eventual leak component before the Kv7.4 channels were activated by a depolarizing step to -30 mV. Interestingly, the effects of the enantiomers were opposing (middle panel); BMS-204352 activated whereas the R-enantiomer inhibited the voltage-activated current. The bottom panel in Fig. 2 shows the time course of a representative experiment. The Kv7.4 current was measured at the end of the depolarizing step and plotted as a function of time. The Kv7.4 currents were, in general, small just after establishment of the whole-cell configuration but increased within the first 1 to 5 min due to an unknown process, and usually 5 to 10 min were allowed to pass before a stable baseline was obtained. In the experiment shown, 10 μM retigabine was first added as a control, and a fast and prominent increase in current was induced. After a washing period, 10 μM BMS-204352 was applied, and a similar effect was obtained although the washout was somewhat slower compared with retigabine. It was noted (bottom panel) that after an initial peak effect of the compounds, there was a slight desensitization to the drugs. This was due to an acceleration of the activation kinetics, which imposed a dynamic effect on the current trace (not shown). This meant that the effect at measuring point (arrow in top panel in Fig. 2) would overshoot before reaching equilibrium. In addition to the increase in steady-state current, deactivation was dramatically slower in the presence of BMS-204352, and a prominent current was induced at -150 mV (see traces in middle panel). The current induced by BMS-204352 at the step to -150 mV appeared similar to the previously described voltage-independent Kv7.4 current induced by the racemate (Schrøder et al., 2003). The following application of the R-enantiomer (10 μM) was neither similar nor inert but induced a slow reversible inhibition of the current. Note that the kinetics of the drug-channel interaction (bottom panel) differed among the enantiomers in that the activator was faster than the inhibitor. Furthermore, as described for the racemate, the drug-channel kinetics for activation induced by BMS-204352 were slower at the voltage-independent current than at the voltage-dependent current (data not shown).
Effects of the Enantiomers on the Activation Curves of Kv7.4 Channels. We then wanted to test the effects of BMS-204352 and the R-enantiomer on the activation curve of Kv7.4 measured as tail currents (Fig. 3). The channels were exposed to potentials ranging from -160 mV to 40 mV for 1 s before stepping to -120 mV (top panel). BMS-204352 lowered the threshold for activation, accelerated the rate of activation, and slowed the deactivation at -120 mV. In contrast, the R-enantiomer suppressed the current at all potentials. The peak tail current as a function of the preceding potential is shown in the bottom panel. The activation curve in the control situation has a threshold for activation of -60 mV and a half-maximal activation potential (V0.5)of -24 mV (interval: -33 to -21 mV, n = 5). The activation curve was shifted in the hyperpolarized direction by 10 μM BMS-204352, resulting in a threshold of approximately -85 mV and a V0.5 = -43 mV (interval, -54 to -36 mV, n = 4) and a maximal voltage-dependent current approximately 80% above the control situation. The R-enantiomer gave approximately the same threshold of activation as the control and an insignificantly different V0.5 of -20 mV (interval, -28 to -11 mV, n = 3, P = 0.2).
Subtype-Dependent Effects Induced by the Enantiomers. To test whether the opposing effects of this pair of enantiomers were an isolated phenomenon on Kv7.4 channels, similar experiments were performed on other α-subunits of the Kv7 family (Fig. 4A). For the neuronal Kv7 channel subtypes expressed as homo- or heteromers, the effect of the enantiomers were qualitatively like that on Kv7.4 with respect to the voltage-dependent current: an activation by BMS-204352 and an inhibition by the R-enantiomer. In Table 1, the activating effect of BMS-204352 has been calculated as percentage of baseline current, and Ki values for the inhibition by the R-enantiomer have been calculated according to eq. 1. Across the subunit combinations, Kv7.5 and Kv7.3/5 were the most sensitive (605 and 931% of baseline, respectively), Kv7.4 and Kv7.3/4 exhibited medium sensitivity (230-245%), whereas Kv7.2/3 appeared to be least sensitive (125% of baseline at 10 μM). The potentiation of Kv7.5 containing channels showed rather large variations. The inhibition by the R-enantiomer was very similar for channels containing Kv7.4 or Kv7.5, whereas the inhibition was weaker for Kv7.2/3 channels (Ki = 24 μM compared with 3.4 to 5.4 μM for the other subunit combinations tested). When the enantiomers were applied to Kv7.1, the effect was surprisingly an inhibition of the current for both BMS-204352 and the R-enantiomer, amounting to Ki values of 3.7 and 4.3 μM, respectively.
Effects of Enantiomers on BK and GABA Receptors. BMS-204352 is a positive modulator of the large conductance calcium-activated K+ channel (BK), (Gribkoff et al., 2001). The enantiomers were tested on BK channels expressed in HEK293 cells, and it was found that both enhanced the BK-mediated currents at 0 mV by approximately 3-fold (Fig. 4B) with identical kinetics (not shown). The compounds were also tested on GABAA receptors using two different subunit combinations: α1β2γ2 and a2β2γ2 expressed in Xenopus oocytes (data not shown). In combination with an EC20 concentration of GABA, BMS-204352 (10 μM) reduced the current by 20% (n = 3-5) on both receptor subtypes, whereas the R-enantiomer (10 μM) on average enhanced the current by 13% (n = 2-7). These effects were concentration-dependent, reaching maximum inhibition of 46% (n = 5) on α1 (100 μM BMS-204352) and a maximum potentiation of 46% (n = 2) on α1 (100 μM R-enantiomer). For comparison, retigabine was tested on the same subunit combinations and showed concentration-dependent potentiations up to 100% (n = 2, 30 μM retigabine).
In Vivo Effects
Ancillary Models (Mouse). To aid in the interpretation of drug effects in the anxiety models described further below, we initially tested the drugs for any nonspecific sensorimotor effects using standard motility cages, a Rotarod, and a simple index of muscle strength.
Exploratory Motility (See Fig. 5). BMS-204352 (3-30 mg/kg) had no effect on exploratory activity [F(3,24) = 1.7], whereas the R-enantiomer (3-30 mg/kg) significantly reduced distance traveled by the mice [F(3,24) = 20.8], with a modest reduction compared with the vehicle group at 10 to 30 mg/kg (P < 0.05). Retigabine had a significant effect on exploratory activity [F(3,24) = 30.2], with a significant reduction in distance traveled at 30 mg/kg compared with vehicle (P < 0.05).
Grip Test (Muscle Strength) and Rotarod Ataxia. Neither retigabine (3-30 mg/kg) nor BMS-204352 (3-30 mg/kg) affected performance of mice in the grip test (data not shown). Likewise, BMS-204352 (3-30 mg/kg) had no effect on the ability of mice to maintain balance on the Rotarod, whereas 80% of mice were unable to perform this task when administered 30 mg/kg retigabine (data not shown).
Anxiolytic Models (Mouse)
Zero Maze (See Fig. 6 and Table 2). In this unconditioned animal model of anxiety, both positive modulators of neuronal Kv7 channels, BMS-204352 and retigabine, induced a dose-dependent anxiolytic effect. BMS-204352 significantly and dose dependently increased time spent in the open areas [F(3,27) = 18.2], reduced latency to enter an open area [F(3,27) = 6.9], and increased the number of entries in to the open areas [F(3,27) = 5.0], with a significant difference at doses of 10 to 30 mg/kg on these measures relative to the vehicle control group (P < 0.05). Furthermore, at 30 mg/kg, BMS-204352 also significantly reduced stretch-attend postures [F(3,27) = 5.7] and increased head dips [F(3,27) = 3.9], with no effect on rearing activity [F(3,27) = 0.8].
Retigabine dose dependently and significantly increased time spent in the open areas [F(3,25) = 4.9], reduced latency to enter an open area [F(3,25) = 5.5], and increased the number of head dips [F(3,25) = 4.3], at doses between 10 to 30 mg/kg relative to the vehicle control group (P < 0.05). There was no significant effect on the number of entries to the open areas [F(3,25) = 2.7] or stretched-attend postures [F(3,25) = 2.3]. However, there was a significant main effect of treatment on the number of rearings [F(3,25) = 3.0], although Dunnett's tests showed that no drug treatment group differed from the vehicle group.
In contrast to the two positive modulators of neuronal Kv7 channels, the R-enantiomer, a negative modulator of these channels, was inactive up to 30 mg/kg on any measure in the zero maze [F(3,25), range of 1.0-2.2 for all measures]. There was a tendency for the R-enantiomer to reduce time spent in the open areas and increase latency to enter an open area, i.e., effects opposite to those seen with BMS-204352 on these parameters.
Marble Burying (See Figs.7and8). In this unconditioned model of anxiety in mice, in which animals bury harmless glass marbles placed on top of sawdust in an observation cage, both BMS-204352 and retigabine reduced burying behavior, indicating an anxiolytic effect. BMS-204352 significantly reduced marble burying behavior in a dose-dependent manner [F(3,28) = 15.2], with a significant difference (P < 0.05) from vehicle at the highest dose of 30 mg/kg. Likewise, retigabine dose dependently and significantly [F(3,28) = 25.8] reduced burying behavior, with a significant reduction compared with vehicle at 3 to 10 mg/kg (P < 0.05). From the data in the section on ancillary models described above, it is clear that at the doses effective in the marble burying test, neither retigabine (1-10 mg/kg) nor BMS-204352 (3-30 mg/kg) affected exploratory activity in mice, suggesting that the reduction in marble burying seen with these two compounds at these doses cannot easily be attributed to nonspecific motor deficits (compare Figs. 5 and 7).
The two Kv7 inhibitors, XE-991 and the R-enantiomer, were also tested in marble burying (Fig. 7). The R-enantiomer tended to reduce marble burying at 30 mg/kg, although there was no significant main effect of treatment [F(3,28) = 2.5]. In exploratory activity experiments described above, the R-enantiomer modestly reduced activity levels in mice at 10 to 30 mg/kg but not at 3 mg/kg (Fig. 5C). XE-991 significantly [F(3,28) = 16.3] reduced burying behavior, with a significant reduction compared with vehicle-treated animals at all doses tested (P < 0.05). However, it should be stressed that after 3 mg/kg XE-991, mice became very still and quiet, and were hypersensitive to any movement or noise in the laboratory, as noted in an earlier a preliminary Irwin screen (Irwin, 1969).
In the following studies where the Kv7 blockers were used to antagonize the effects of the positive modulators of neuronal Kv7 channels, we selected a dose of 1 mg/kg XE-991 and 3 mg/kg of the R-enantiomer so as to avoid any adverse effects of the blockers themselves as indicated in the exploratory motility and/or Irwin studies described above. Furthermore, since XE-991 has a short t1/2 (Earl et al., 1998), we administered this compound 5 min before the marble burying test and 10 min after administration of 10 mg/kg retigabine. In Fig. 8A, it is clear that XE-991 at 1 mg/kg completely reversed the reduction in marble burying induced by retigabine (10 mg/kg). There was a main effect of treatment [F(2,21) = 8.4], with the post hoc multiple comparison test showing significant differences between the vehicle and retigabine alone treated groups, as well as between the XE-991 + retigabine and retigabine alone groups (P < 0.05). Likewise, Fig. 8B shows that the R-enantiomer (3 mg/kg) antagonized the reduction in marble burying induced by BMS-204352 (30 mg/kg). There was a main effect of treatment [F(2,21) = 5.8], with the post hoc multiple comparison test showing a significant difference between the control and BMS-204352 alone groups, as well as between the BMS-204352 alone and BMS-204352 + R-enantiomer groups.
Anxiolytic Models (Rat)
Conditioned Emotional Response (See Table 3). In this shock-based conditioned model of anxiety in rats, BMS-204352 engendered an anxiolytic profile, although this was not seen with retigabine. During baseline responding in the light period, retigabine completely disrupted lever pressing at 10 mg/kg, and therefore, this dose was excluded from further analysis. Neither retigabine (1-3 mg/kg) nor BMS-204352 (3-60 mg/kg) significantly affected response rate in the dark (anxiety) period of the test [main effect of treatment: F(2,25) = 1.5 and F(4,41) = 1.8, respectively] or during baseline responding in the light period of the test [main effect of treatment: F(2,25) = 0.8 and F(4,41) = 1.0, respectively]. By contrast, whereas retigabine had no effect on the suppression ratio [F(2,25) = 0.1], BMS-204352 significantly affected the ratio between light and dark responses [F(4,41) = 2.6, P < 0.05], indicative of an anxiolytic profile.
Discussion
In this study we show that 1) BMS-204352 is a positive modulator of all neuronal Kv7 channel subtypes expressed in HEK293 cells, whereas the R-enantiomer is a negative modulator; 2) both enantiomers are negative modulators of Kv7.1 channels; 3) BMS-204352 and the R-enantiomer are positive modulators of BK channels; 4) whereas BMS-204352 is a negative modulator of GABAA receptors, the R-enantiomer is a positive modulator; 5) BMS-204352 as well as retigabine show clear dose-related anxiolytic efficacy in the mouse zero maze; 6) in another unconditioned model of anxiety, the marble burying test (Broekkamp et al., 1986), BMS-204352 and retigabine reduced burying behavior at doses not affecting general activity levels; and 7) these effects on burying behavior can be antagonized by a Kv7 channel blocker, either XE-991 or the R-enantiomer of BMS-204352. These behavioral and electrophysiological data suggest that the anxiolytic efficacy of BMS-204352 can be ascribed to positive modulation of neuronal Kv7 channels, indicating that this is a novel target for developing anxiolytics. Finally, neither BMS-204352 (3-30 mg/kg) nor retigabine (1-10 mg/kg) impaired the performance of rats in a delayed nonmatching to position task (Dunnett et al., 1989), a test for memory (data not shown).
The racemic mixture of BMS-204352 and its R-enantiomer was previously shown to activate voltage-dependent and voltage-independent Kv7.4 currents (Schrøder et al., 2003). In this study, we investigated whether these two effects were separately induced by one enantiomer. We found that BMS-204352 induced activation of the voltage-dependent as well as the voltage-independent current and that the R-enantiomer inhibited both Kv7 current components. This profile holds true for all the neuronal Kv7 channel subtypes, whereas for the Kv7.1 subtype both enantiomers block the current. In light of the presence of Kv7.2/7.3 (Saganich et al., 2001) and Kv7.5 (Schroeder et al., 2000a) in the amygdala, and the absence of Kv7.1 in the brain, this implies positive modulators of these channels have a potential as anxiolytics.
To ascertain if an effect at Kv7 channels specifically mediates the anxiolytic effects of BMS-204352 and retigabine observed in this study, we additionally tested the compounds on BK and GABAA channels. The BK channel is found in the basolateral amygdala (Meis and Pape, 1997) but is unlikely to mediate the anxiolytic effects of BMS-204352, since both this and the R-enantiomer are positive modulators of the BK channel. GABAA receptors are the targets of traditional anxiolytics belonging to the benzodiazepine class. Retigabine has been reported to affect GABAA currents (Rundfeldt and Netzer, 2000), which in part could account for its antiepileptic effects. We also describe a positive modulatory effect in vitro of retigabine on GABAA receptors, but in line with others (Rundfeldt and Netzer, 2000), we found that the concentration required was at least one order of magnitude above that of the Kv7 effect. Moreover, we have seen that retigabine (0.3-10 mg/kg) shows no generalization to the chlordiazepoxide discriminative cue (data not shown), a cue based on the ability of benzodiazepines to allosterically enhance the effects of GABA. The negative modulatory effect of BMS-204352 and the positive modulatory effect of the R-enantiomer at GABAA receptors, respectively, suggest that if an effect at GABAA receptors is relevant to our observations in the anxiety tests then the R-enantiomer rather than BMS-204352 would have been predicted to be anxiolytic.
In the zero maze unconditioned anxiety model we saw dose-dependent effects of both BMS-204352 and retigabine. In addition, we saw a tendency for the R-enantiomer to show the opposite effect of BMS-204352. In the marble burying model, BMS-204352 had an anxiolytic effect, which could be fully antagonized by the R-enantiomer, which argues in favor of a neuronal Kv7 channel-mediated mechanism. Additionally, with the Kv7 channel blocker, XE-991 (1 mg/kg), we were able to reverse the reduction in marble burying behavior induced by retigabine, although this dose of XE-991 in itself reduced marble burying. In contrast to these effects in unconditioned models of anxiety in mice, we saw no affect of retigabine in the conditioned emotional response model of anxiety in rats. Shock-based models like the conditioned emotional response are highly sensitive to benzodiazepines (Green, 1991), and either this factor or a species difference between mouse and rat may explain the lack of efficacy with retigabine. By contrast, BMS-204352 engendered an anxiolytic response at 60 mg/kg in the conditioned emotional response test. Looking closely at the data (Table 3), it appears that the basis for the increase in suppression ratio after BMS-204352 was an increase in response rate in the dark period rather than a decrease in light period response rate, suggesting a true anxiolytic profile. Nonetheless, the effect is not equivalent to the efficacy attained with a benzodiazepine in this model (Stanhope and Dourish, 1996).
The in vitro profiles of BMS-204352 and the R-enantiomer provide powerful tools with which to delineate behavioral effects mediated by neuronal Kv7 channels. Here we have demonstrated that the positive modulator of neuronal Kv7 channels, BMS-204352, is anxiolytic in two unconditioned models of anxiety in mice and that the R-enantiomer antagonizes the effect in one model. Moreover, BMS-204352 is also active in a rat-conditioned model of anxiety, although the effect is not of the magnitude seen with benzodiazepines in this test. Further supporting data with retigabine and XE-991 demonstrate that this channel family may be a target for the development of novel anxiolytics. Clearly, further studies are needed to address potential tolerance, abuse, or other side effect liabilities that may be associated with this target. Thus far, our data suggest that separation between sedative/cognitive side effects and anxiolytic efficacy can be claimed for drugs acting at this target.
Footnotes
-
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
-
doi:10.1124/jpet.105.083923.
-
ABBREVIATIONS: BK, large conductance Ca2+-activated potassium; DMSO, dimethyl sulfoxide.
- Received January 21, 2005.
- Accepted March 21, 2005.
- The American Society for Pharmacology and Experimental Therapeutics