Perampanel [Fycompa, 2-(2-oxo-1-phenyl-5-pyridin-2-yl-1,2-dihydropyridin-3-yl)benzonitrile hydrate 4:3; Eisai Inc., Woodcliff Lake, NJ] is an AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid) receptor antagonist used as an adjunctive treatment of partial-onset seizures. We asked whether perampanel has AMPA receptor antagonist activity in both the cerebral cortex and hippocampus associated with antiepileptic efficacy and also in the cerebellum associated with motor side effects in rodent and human brains. We also asked whether epileptic or nonepileptic human cortex is similarly responsive to AMPA receptor antagonism by perampanel. In rodent models, perampanel decreased epileptic-like activity in multiple seizure models. However, doses of perampanel that had anticonvulsant effects were within the same range as those engendering motor side effects. Perampanel inhibited native rat and human AMPA receptors from the hippocampus as well as the cerebellum that were reconstituted into Xenopus oocytes. In addition, with the same technique, we found that perampanel inhibited AMPA receptors from hippocampal tissue that had been removed from a patient who underwent surgical resection for refractory epilepsy. Perampanel inhibited AMPA receptor–mediated ion currents from all the tissues investigated with similar potency (IC50 values ranging from 2.6 to 7.0 μM). Cortical slices from the left temporal lobe derived from the same patient were studied in a 60-microelectrode array. Large field potentials were evoked on at least 45 channels of the array, and 10 μM perampanel decreased their amplitude and firing rate. Perampanel also produced a 33% reduction in the branching parameter, demonstrating the effects of perampanel at the network level. These data suggest that perampanel blocks AMPA receptors globally across the brain to account for both its antiepileptic and side-effect profile in rodents and epileptic patients.
Since the pioneering work of Hayashi (1952), it has been known that glutamate plays a prominent role in epileptic phenomena. Ion channels regulating glutamatergic and GABAergic synaptic transmission provide the dominant control over central nervous system excitability and thus play a large role in the initiation and suppression of epileptic phenomena and represent the molecular substrate for a predominant number of currently used antiepileptic drugs (AEDs). AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid) receptors regulate fast excitatory neurotransmission in the mammalian nervous system and have been a target of AED discovery and development for many years. Additionally, data from genetic studies (Lerche et al., 2013) as well as AMPA receptor upregulation in brains from epileptic patients (Palomero-Gallagher et al., 2012) have added additional rationale for targeting AMPA receptors in the control of epilepsy.
Multiple structurally distinct AMPA receptor antagonists, including competitive and noncompetitive blockers, have been shown to possess potent and efficacious activity in a range of in vitro and in vivo models for detecting AED efficacy (Rogawski, 2011, 2013; Russo et al., 2012; Rektor, 2013). A few of these molecules have been investigated in humans for their antiepileptic potential. The AMPA receptor antagonist talampanel (GYKI 53773; LY300164; 7-acetyl-5-(4-aminophenyl)-8,9-dihydro-8-methyl-7H-1,3-dioxolo(4,5H)-2,3-benzodiazepine) has been shown to produce significant reduction in seizure frequency, with dizziness (52%) and ataxia (26%) being the only adverse events reported (Chappell et al., 2002). Final proof of the concept that AMPA receptor antagonism can serve as an antiepileptic control of refractory partial seizures came with the approval of the noncompetitive antagonist perampanel [Fycompa, 2-(2-oxo-1-phenyl-5-pyridin-2-yl-1,2-dihydropyridin-3-yl)benzonitrile hydrate 4:3; Eisai Inc., Woodcliff Lake, NJ] (Fig. 1), which is the first selective AMPA receptor antagonist on the market.
Perampanel has demonstrated efficacy in multiple preclinical models, which has helped inform drug development for epilepsy (Hanada et al., 2011; Hibi et al., 2012; Rogawski and Hanada, 2013). Perampanel has been investigated in several phase III clinical studies (French et al., 2012; Krauss et al., 2012, 2013; Rektor et al., 2012). The key clinical trials were conducted on a global level as randomized, double-blind, placebo-controlled, dose-escalation studies that examined 1480 patients with partial-onset seizures. These studies demonstrated that perampanel as an adjunctive therapy significantly reduced seizure frequency in patients with partial-onset seizures with or without secondary generalized seizures (see reviews and analysis by Gao et al., 2013; Rheims and Ryvlin, 2013; Steinhoff et al., 2013; Zaccara et al., 2013). Fycompa was approved (October 2012) by the US Food and Drug Administration for use as an adjunctive treatment of partial-onset seizures with or without secondarily generalized seizures in patients with epilepsy who are age 12 years and older. However, as observed with talampanel, perampanel also engenders a range of side effects predicted from both its mechanism of action and preclinical data, including dizziness, somnolence, fatigue, irritability, falls, nausea, ataxia, balance disorder, gait disturbance, vertigo, and weight gain. Nonetheless, at efficacious dosages, perampanel is generally well tolerated.
AMPA receptors are widely expressed in the central nervous system. There are four different AMPA receptor subunits (GLKA1–4), and these subunits are broadly, albeit differentially, expressed in the brain. The subunit selectivity of perampanel among AMPA receptor isoforms is not known (Rogawski and Hanada, 2013). Because perampanel causes both beneficial and adverse effects, we presume that perampanel inhibits AMPA receptors broadly across the brain, regardless of their subunit composition. Our hypothesis is that the beneficial effects of perampanel are mediated by AMPA receptors in brain regions related to epilepsy (cerebral cortex, hippocampus, amygdala, thalamus) and that some of the side effects are caused by inhibiting AMPA receptors in other brain regions. The cerebellum is likely to play a role in some of the side effects observed with perampanel (dizziness, falls, nausea, ataxia, balance disorder, and vertigo) based upon the known physiologic regulation of these clinical signs by the cerebellum (Salman, 2002; Della-Morte and Rundek, 2012). The hypothesis was tested with novel electrophysiologic techniques to study the sensitivity to perampanel of AMPA receptors from different brain areas of both rats and humans. In addition, because epileptic brain tissue has shown changes in AMPA receptor density (Brines et al., 1997; Palomero-Gallagher et al., 2012), we also evaluated the effects of perampanel in human epileptic brains.
In particular, we adopted a recently developed technique (Miledi et al., 2006; Eusebi et al., 2009) based on the reconstitution of native human neurotransmitter receptors into oocytes of the African clawed toad (Xenopus laevis), followed by voltage-clamp recording. The strength of this technique is that it can be applied to postmortem frozen human brain samples as well as freshly excised tissue. The technique can also be used to study native ion channels from diseased brain. For example, it has been applied to study differences in native human GABAA receptors between control and epileptic human hippocampi (Palma et al., 2005, 2007). Finally, we examined the effects of perampanel directly on fresh cortical slices obtained from the same epileptic patient studied in oocytes, but this time using a 60-microelectrode recording array (Hobbs et al., 2010).
Materials and Methods
Rodent Behavioral Assays
Male Sprague Dawley rats (Harlan Sprague Dawley, Indianapolis, IN), weighing 90–110 g at the time of test, were used. Animals were housed five per cage with ad libitum food and water in a large colony room with a standard light cycle (lights on 6:00 AM, lights off 6:00 PM). F344 (Harlan Sprague Dawley) were used for the studies of intravenous pentylenetetrazole (PTZ) infusion. Male CD1 mice (Taconic Farms, Germantown, NY) weighing 23–31 g were used. Animals were maintained in the colony room for at least 3 days before testing. Animals were moved to a quiet room 1 hour before the start of the test. All studies were performed according to the guidelines set forth by the National Institutes of Health and implemented by the Animal Care and Use Committee of Eli Lilly and Company.
Mice (n = 10/dose) were used with Wahlquist Model H stimulator (Wahlquist Instrument Co., Salt Lake City, UT) with 0.2-second stimulation with corneal electrodes. Mice were observed for approximately 10 seconds after administration of the electrical stimulus (10 μA), and the types of convulsions were recorded (0 = no convulsion, 1 = clonus, 2 = tonic flexion, and 3 = tonic extension). Mice were euthanized immediately after the test. Only one mouse was used per treatment. Tonic extension was used as the primary end point for calculation of ED50 values.
Inverted Screen Test.
Before dosing rats or mice with PTZ, the animals were tested (30 minutes after drug administration) on an inverted screen. The apparatus is made of six 11 cm × 14 cm (mouse) or four 13 cm × 16 cm squares (rat) of round hole, perforated, stainless steel mesh (18 holes/square inch, 3/16-inch diameter, 1/4-inch staggered centers, 50% open area) that are mounted 15 cm apart on a metal rod, 35 cm above the tabletop.
On the day before the test, mice are placed on the screen, and the rod is rotated 180° over 2 to 3 seconds. The amount of time it takes for the test subject to climb to the top of the screen is recorded. For the test subjects that hang on the bottom for the maximum of 60 seconds, a 60-second score is recorded. A test with compound on board is studied the next day in the same manner.
For rat studies, animals were dosed orally with the test compound and returned to their home cage. Twenty-five minutes after pretreatment, the animals were tested on the inverted screen and were scored after 60 seconds as follows: 0 = climbed over, 1 = hanging on to screen, and 2 = fell off. After the inverted screen test, animals were dosed with PTZ in a volume of 1 ml/kg and placed in an observation cage (40.6 × 20.3 × 15.2 cm) with a floor containing 0.25 inches of wood-chip bedding material.
Drugs were studied for their ability to prevent or dampen seizures induced in mice or rats after subcutaneous PTZ dosing. Animals were then observed for 30 minutes after PTZ for clonus (defined as clonic seizure of forelimbs and hindlimbs during which the mouse demonstrates loss of righting) or for tonic seizures as exemplified by loss of righting accompanied with tonic hindlimb extension. The dose of PTZ used in the rat studies was 35 mg/kg s.c. and the dose in mice was 70 mg/kg i.p., based upon estimated ED90 values for PTZ in these assays.
Pentylenetetrazole-Induced Seizure Threshold.
The threshold dose of PTZ to produce seizures was measured in rats with and without perampanel pretreatment during intravenous PTZ infusion. Male F-344 Harlan rats were randomly assigned to treatment groups and dosed with vehicle or test compound. After the appropriate pretreatment time as noted earlier, the rats were placed in a restrainer, and a winged infusion needle was inserted into the lateral tail vein. Intravenous infusion with 10 mg/ml of PTZ at a rate of 0.5 ml/min was then initiated until a clonic convulsion occurred, and the time to clonic convulsion was recorded in seconds or a maximum of 4 minutes, 240 seconds was recorded. After the infusion, the rats were euthanized. The dose of PTZ required to elicit a clonic convulsion was calculated using the infusion rate, concentration of PTZ, time to clonic convulsion, and animal weight.
Basolateral Amygdala Kindling.
Drugs were evaluated for their ability to impact seizure parameters in rats that were fully kindled by daily electrical stimulation of the amygdala. The electrical kindling equipment consisted of a computer connected to an amplifier (ISASI Quad Amplifier, Grass Model 15RX; Grass Instruments Division, Astro-Med, Warwick, RI). The amplifier was connected to a maximum of two telefactor units (Grass Telefactor W MOID S88; Grass Instruments Division, Astro-Med). Each telefactor unit was connected to a maximum of two battery units (photoelectric stimulator isolation unit constant current output). Each battery unit was then connected to one shock chamber. Valproic acid was used as a positive control.
A bipolar electrode for electrical stimulation and electroencephalogram recording was stereotaxically implanted into one hemisphere of the basolateral amygdala (anteroposterior −2.2, mediolateral −4.8, dorsoventral −8.5 mm, relative to bregma) of male Wistar rats. After postoperative recovery, electrical kindling begins, where a subthreshold constant current (400 µA, 1 millisecond, monophasic square-wave pulses, 60 Hz for 1 second) is given once a day Monday–Friday for about 4–6 weeks until a rat is fully kindled. A fully kindled rat has experienced 10 consecutive stage 5 seizures or 10 of its last 12 were stage 5 according to the Racine scale.
Twelve fully kindled rats were assigned to this study, and eight rats were selected and randomized to initial compound treatment groups from baseline after discharge threshold (ADT), seizure severity score, and after discharge duration (ADD). A pseudo within-subjects Latin square design was used for subsequent testing, as replacement rats were used in the event that an assigned rat did not meet the precompound testing baseline criteria or a rat lost its head cap during a seizure.
On test day, the rats were dosed 30 minutes before beginning the stimulation. After the pretreatment, rats were stimulated using an ascending staircase sequence beginning at 10 µA and increasing in log unit steps of 10, 16, 25, 40, 65, 100, 160, 250, and 400 μA. Animals were stimulated until they were assigned a seizure severity score for a visually observable seizure or they reached the 400-μA threshold limit. ADD was determined after testing. ADT was the current that induced a scorable seizure. For seizure severity, the Racine score of behavioral response to stimulation was 0 = no behavioral response; 1 = immobility, staring, and/or facial clonus; 2 = head nodding, jaw clonus, and/or tongue protrusion; 3 = unilateral forelimb clonus; 4 = bilateral forelimb clonus and/or rearing; or 5 = bilateral forelimb clonus with rearing and loss of balance. The ADD is the duration of the first after-discharge.
ADT, ADD, and seizure severity scores were recorded for each animal. Data were analyzed by analysis of variance and post-hoc Dunnett’s t-test. Statistical analyses were performed using JMP statistical software (SAS Institute Inc., Cary, NC). P ≤ 0.05 was considered statistically significant.
Perampanel (Fig. 1) was synthesized by Eli Lilly and Company. The other compounds were obtained from commercial suppliers: lacosamide (10 mg/ml) (UCB, Smyrna, GA), diazepam (Sigma-Aldrich, St. Louis, MO), valproic acid Na (Sigma-Aldrich), picrotoxin, and PTZ (Sigma-Aldrich). Perampanel was suspended in 10% acacia and 0.05% Dow antifoam in a volume of 10 ml/kg for both rat (administered via the oral route, 30 minutes before testing) and mouse (administered 30 minutes via intraperitoneal route) studies. Lacosamide was diluted from a stock solution of 10 mg/ml with saline (0.9% NaCl) and dosed in a volume of 1 or 3 ml/kg for rat (dosed via intraperitoneal route, 30 minutes before testing) and a volume of 10 ml/kg was used for mice (dosed via intraperitoneal route, 30 minutes before testing). Diazepam was dissolved in 20% 2-hydroxypropyl-β-cyclodextrin in water. Diazepam was administered in a volume of 1 ml/kg for rat (dosed via intraperitoneal route, 30 minutes before testing), and a volume of 10 ml/kg was used for mice (dosed via intraperitoneal route, 30 minutes before testing). Valproic acid was dissolved in saline and administered in a volume of 4 ml/kg to rats via intraperitoneal route, 30 minutes before testing and dosed as the acid form.
Xenopus Oocyte Studies: Effects on Human and Rat Hippocampal and Cerebellar AMPA Receptors Reconstituted into Xenopus Oocytes
Rat hippocampal and cerebellar brain samples were taken from 2- and 3-month-old male Sprague Dawley rats (Charles River, Margate, UK). Frozen fragments of human hippocampal and cerebellar brain samples were obtained from Asterand (Detroit, MI) and kept frozen at −80°C. Transport of the human tissue samples from the United States to the United Kingdom took place on dry ice. Both human brain samples were from the same donor, a 67-year-old Caucasian male whose cause of death was cardiac arrest. During his life, the donor’s height was 172 cm, weight was 74 kg, body mass index was 25.0, and he was clinically diagnosed with hypertension, chronic pulmonary disease, and hyperlipidemia. He was on the following medications: unspecified inhalers, unspecified statins, and Vicodin (Abbvie, North Chicago, IL). The excision of the tissues took place in 2010, and the postmortem intervals were 723 minutes (hippocampal tissue) and 758 minutes (cerebellar tissue). Membrane preparations of the frozen human tissues were made on the August 4, 2011.
Membrane preparations were also made from hippocampus obtained from a 13-year-old male patient undergoing surgical resection for refractory epilepsy. This tissue was resected, prepared, and used in accordance with an institutional review board (0612, Indiana University School of Medicine, Indianapolis, IN), and approval from the parent of the patient was obtained before surgery. The patient voluntarily elected to be in the study. Cortical tissue was harvested from left temporal lobe in an area with the most active epileptogenic profile of the patient’s seizure focus as identified by preoperative video electroencephalography and intraoperative electrocorticography. A 2 × 2 mm piece of cortex within the seizure focus was removed immediately after the intraoperative electrocorticography; removal took less than 30 seconds. Cerebral vasculature was left intact as the tissue was resected and immediately placed for 3 minutes in ice-cold artificial cerebrospinal fluid (ACSF) containing in mM: sucrose 125, KCl 3, NaH2PO4*H2O 1.25, NaHCO3 26, MgSO4*7H2O 2, CaCl2*2H2O 2, d-glucose 10, saturated with 95% O2/5%CO2.
Membrane preparations from these tissues were prepared according to the method developed and described by Miledi et al. (2006) and Eusebi et al. (2009). In short, 0.1–0.5 g of fresh rat tissue or of frozen human tissue was homogenized in ice-cold glycine buffer (in mM: 200 glycine, 150 NaCl, 50 EGTA, 50 EDTA, 300 sucrose) to which 10 μl of protease inhibitor cocktail (Sigma-Aldrich) was added per milliliter of glycine buffer. The homogenate was centrifuged at 4°C for 15 minutes at 9500g. The supernatant was subsequently centrifuged at 4°C for 2 hours at 100,000g with an ultracentrifuge, and the pellet was resuspended in ice-cold assay buffer (5 mM glycine). The protein concentration of the membrane preparations were measured using the Pierce BCA protein assay kit (Thermo Scientific, Rockford, IL) and were between 2.5 and 7 mg/ml. Aliquots of the suspensions were kept at −80°C and were thawed just before injection into Xenopus oocytes.
Xenopus oocytes (stages V–VI) were removed from schedule I sacrificed frogs and defolliculated after treatment with collagenase type I (5 mg/ml calcium-free Barth’s solution) for 1.5 hours at room temperature. We injected 60 nl of membrane suspension per oocyte using a Drummond (Broomall, PA) variable volume microinjector. After injection, oocytes were incubated at 18°C in a modified Barth’s solution containing 88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO3, 0.3 mM Ca(NO3)2, 0.41 mM CaCl2, 0.82 mM MgSO4, 15 mM HEPES, and 50 mg/l neomycin (pH 7.6 with NaOH; osmolarity 235 mOsm). Experiments were performed on oocytes after 2–5 days of incubation.
Oocytes were placed in a recording chamber (internal diameter 3 mm), which was continuously perfused with a saline solution (115 mM NaCl, 2.5 mM KCl, 1.8 mM CaCl2, 1 mM MgCl2, 10 mM HEPES, pH 7.3 with NaOH, 235 mOsm) at a rate of approximately 10 ml/min. Dilutions of drugs in external saline were prepared immediately before the experiments and applied by switching between control and drug-containing saline using a BPS-8 solution exchange system (ALA Scientific Inc., Westbury, NY). Between responses, the oocytes were washed for 2 minutes. Oocytes were impaled by two microelectrodes filled with 3 M KCl (0.5–2.5 MΩ) and voltage-clamped using a Geneclamp 500B amplifier (Axon Instruments, Union City, CA) according to the methods described by Stühmer (1998). The external saline was clamped at ground potential by means of a virtual ground circuit using an Ag/AgCl reference electrode and a Pt/Ir current-passing electrode. The membrane potential was held at −60 mV. The current needed to keep the oocyte’s membrane at the holding potential was measured. Membrane currents were low-pass filtered (four-pole low-pass Bessel filter, −3 dB at 10 Hz), digitized (50 Hz), and stored on disc for offline computer analysis. Data are expressed as mean ± S.E.M. All experiments were performed at room temperature.
For antagonist concentration-inhibition curves, ion currents were evoked by switching from control solution to a solution containing 100 μM AMPA and 100 μM cyclothiazide (CTZ). After 30 seconds of AMPA/CTZ perfusion, the solution was switched to AMPA/CTZ plus various concentrations of perampanel for 20 seconds. The amount of inhibition was calculated from the current amplitude at the end of perampanel application and the current amplitude of the AMPA/CTZ response just before perampanel application. Concentration-inhibition curves were fitted according to the Hill equation. Curve fitting was performed using GraphPad Prism 3.01 software (GraphPad Software, San Diego, CA).
Human Epileptic Cerebral Cortex Electrophysiology
Tissue Preparation and Recording.
The block of tissue from the patient described earlier was then sliced into coronal sections with a thickness of 250 µm using a Vibratome tissue slicer (Leica, Wetzlar, Germany). Coronal sections were cut using white matter as landmark. After cutting, slices were bathed for ∼1 hour at room temperature in ACSF with the same ingredients as previously listed but with 125 mM NaCl substituted for 125 mM sucrose to restore Na+ and allow cells to fire action potentials again.
In preparation for recording, slices were adhered to microelectrode arrays (Multichannel Systems, Reutlingen, Germany) with a solution of 0.1% polyethylenimine that had been previously applied and let dry for 2 hours (Wirth and Luscher, 2004). We attempted to place the tissue so that neocortical layers I–V covered the array. Slices were maintained thermostatically at 37°C and were perfused at 1.0 ml/min first with normal ACSF for 1 hour to see whether the tissue was spontaneously active. If spontaneous activity did not develop after 1 hour, the tissue was bathed in excitable ACSF (EACSF) solution containing 5 mM KCl and 0 mM Mg+2. These external ionic concentrations are known to produce robust local field potential (LFP) activity in cortical brain slices (Schiff et al., 1994; Wu et al., 1999; and verified in cortical tissue from humans by Hobbs et al., 2010). Tissue was then recorded in EACSF with a concentration of 10 μM picrotoxin for more than 1 hour. Subsequently, tissue was recorded in EACSF with a concentration of 10 μM picrotoxin with perampanel and recorded for 1 hour. After the perampanel addition, tissue was recorded in a washout solution of EACSF and 10 μM picrotoxin for 1 hour.
Recordings were performed on microelectrode arrays purchased from Multichannel Systems (Reutlingen, Germany). Each array had 60 electrodes, and each electrode was 30 µm in diameter and 30 µm high. Electrodes were arranged in a square grid with 200 µm spacing between electrodes.
Local Field Potential Detection.
Extracellular activity from slices was recorded in the same manner as previously reported (Hobbs et al., 2010). Activity was sampled from all 60 electrodes at 1 kHz and amplified before being stored to disk for offline analysis. LFPs that showed sharp negative peaks below a threshold set at 3 S.D. of the signal were marked, and the time of the maximum excursion was recorded as the time of that LFP. Time points were binned at 4-millisecond resolution, as this was previously shown to match the average time between successive LFP events across electrodes (Beggs and Plenz, 2003).
In characterizing network activity, we closely followed the methods first described by Beggs and Plenz (2003, 2004). An avalanche is a sequence of consecutively active frames that is preceded by a blank frame and terminated by a blank frame. The length of an avalanche is given by the total number of active frames, and the size of an avalanche is given by the total number of electrodes activated during the avalanche.
Measuring the Branching Parameter.
The branching parameter, symbolized by σ, is the average number of descendant electrodes produced by a single ancestor electrode. By ancestor electrode, we mean an electrode on the array that experienced a suprathreshold LFP signal at a given time step. By descendant electrode, we mean an electrode on the array that experienced a suprathreshold LFP signal one time step after the activity of the ancestor electrode. The branching parameter is straightforward to approximate from experimental data, and can be obtained by taking the parameterwhere L is the length of each avalanche, N2:L is the total number of electrodes activated in frames 2 to L, N1:(L − 1) is the total number of electrodes activated in frames 1 to L − 1, and the angled brackets indicate averaging over all avalanches.
From this definition, it is clear that the branching parameter is a measure of collective excitability in the network. Intuitively, when σ < 1, activity is damped and will quickly die out. When σ = 1, the network is in the critical state, and long chains of neural activity can occur without unstable expansion. When σ > 1, activity is amplified every time step, eventually leading to excessive network activity. Excessive network activity can be induced by washing the slice in 10 μM picrotoxin. This increase in network activity can mimic a seizure state in the slice. Once an excited state is induced, the effects that perampanel has on network activity can be measured by the branching parameter as just described.
In Vivo Studies
Both perampanel and diazepam potently and fully prevented the tonic extension induced by electroshock in mice after intraperitoneal dosing (Table 1).
Inverted Screen Test.
The potencies of perampanel and lacosamide to disrupt behavior of mice or rats on the inverted screen test are shown in Table 1. Both compounds produced motoric side effects as measured by this assay.
Both perampanel and lacosamide were fully effective in attenuating PTZ-induced clonic seizures in both mice and rats. The potencies of these drugs under these assays are shown in Table 1.
Pentylenetetrazole-Induced Seizure Threshold.
Perampanel was not effective in increasing PTZ seizure latency when dosed orally. At the highest dose tested, the dose of PTZ inducing seizures was 41.7 ± 4.2 mg/kg compared with 35.1 ± 1.0 mg/kg after vehicle dosing. In contrast, the positive control in this study, valproate (300 mg/kg i.p.), increased the threshold dose of PTZ to 86.8 ± 9.4 mg/kg. Diazepam had minimal effective dose of 0.3 mg/kg i.p. At this dose, the threshold dose of PTZ required for seizures was 43.4 ± 0.3 mg/kg (vehicle = 37.8 ± 1.2 mg/kg PTZ).
In amygdala kindled rats, perampanel and lacosamide were studied after oral doses of 3–30 mg/kg. Under these conditions, perampanel was efficacious in dampening all measured parameters of seizure kindling at 30 mg/kg except the ADD, which was also unaffected by lacosamide (Table 1). Although lacosamide was 1/2-log dose more potent in protecting against ADT and seizure severity than perampanel, it was less potent in altering convulsions driven by 400 µA (Table 1).
When comparing potencies of perampanel on a measure of motor deficit (inverted screen) to its anticonvulsant potencies, a protective index (PI) can be calculated as Potency inverted screen/Potency anticonvulsant. PI values >1 indicate a margin between efficacy and side-effect doses; PI values ≤1 indicate that side effects and efficacy are not dissociable. For PTZ clonus in rats and for the ADT measure of amygdala kindling, the PI = 0.36 and 0.13, respectively. For lacosamide, estimates of the PI for these same anticonvulsant end points were ∼0.15 and 0.3, respectively. In mice, perampanel had a PI of 0.44 in the PTZ clonus measure and 3.5 in the electroshock measure. Lacosamide had a PI estimate of ∼1 for PTZ clonus in mice.
Effects on Native Human and Rat Hippocampal and Cerebellar AMPA Receptors Reconstituted into Xenopus Oocytes
Oocytes injected with membranes from rat or human hippocampal or cerebellar membranes did not respond to 100 μM AMPA with measurable currents (not shown), but when 100 μM AMPA was coapplied with 100 μM AMPA receptor potentiator CTZ, AMPA receptor–mediated ion currents were clearly observed, which were almost completely inhibited when 30 μM perampanel was applied during the AMPA/CTZ response (Fig. 2A). Perampanel induced concentration-dependent inhibition of AMPA/CTZ (Fig. 2B). The IC50 and Hill slope values are shown in Table 2. The data demonstrate that perampanel inhibited native rat and human hippocampal and cerebellar AMPA receptors with similar potency and to the same extent.
Membranes were also prepared from the hippocampal tissue obtained from the epileptic patient and further injected into oocytes. In this case as well, perampanel produced a concentration-dependent blockade of ion currents evoked by application of AMPA/CTZ (Fig. 3) with an IC50 of 6.7 μM, demonstrating that perampanel is indeed blocking AMPA receptors in the diseased brain.
Human Epileptic Cerebral Cortex Electrophysiology
Characterizing Multielectrode Activity.
In characterizing network activity, we closely followed the methods first described by Beggs and Plenz (2003, 2004) and illustrated in Fig. 4. The configuration of active electrodes during one time step is called a frame.
An avalanche is a sequence of consecutively active frames that is preceded by a blank frame and terminated by a blank frame. The length of an avalanche is given by the total number of active frames, and the size of an avalanche is given by the total number of electrodes activated during the avalanche.
Spontaneous Activity in Human Slices.
As reported previously, human slices are rarely active in normal ACSF (Hobbs et al., 2010). This was also true in this experiment, and to test the effects of perampanel in this experiment, the slice was bathed in 10 μM picrotoxin, elevated K+, and reduced Mg2+. This concentration was found to evoke large field potentials on at least 45 channels of the array.
Event Size Distributions.
We commonly observed LFPs that exceeded the threshold on multiple electrodes in the same time bin. We also observed cascades of consecutively active time bins (Fig. 4). After cascades were recorded over 45 minutes or more, it was possible to plot their size distribution. We previously reported that in data collected from rats this size distribution had a very long tail that approached a power law (Hobbs et al., 2010). This distribution was not expected by chance (Beggs, 2008), but was similar to that produced by computational models of sand pile avalanches (Bak, 1996).
Perampanel Effects on Epileptic Human Brain Tissue.
We first recorded the slice activity for 1 hour in 10 μM picrotoxin to establish a base line. The average amplitude was −23.34 ± 7.07 μV, and the average firing rate (Fig. 5A) was 0.14 ± 0.02 Hz during this recording. In the second hour of recording, 10 μM perampanel was added along with the 10 μM picrotoxin. The amplitude showed a trend toward a decrease as shown in Fig. 6A to −21.22 ± 5.24 μV, and the average firing rate decreased to 0.06 ± 0.01 Hz (Fig. 5A). Both of these decreases were statistically significant (P < 0.05, paired test). This decrease in activity was maintained after washout with 10 μM picrotoxin: −19.68 ± 8.57 μV and 0.06 ± 0.01 Hz (N3 reference Table 1). Statistically significant decreases in both amplitude and firing rate establish that perampanel decreases activity in the slice from a human epileptic patient.
Epilepsy is a network phenomenon—groups of neurons exhibiting abnormal electrical activity. Understanding epilepsy thus requires scrutiny of the activation of groups of neurons at the network level. To establish the effects of perampanel at the network level, we measured the branching ratio.
As previously mentioned in Materials and Methods, 10 μM picrotoxin was washed into the slice bath to induce excitability in the slice network. After recording this picrotoxin-induced excitability for 1 hour, we washed 10 μM perampanel into the slice bath while maintaining the 10 μM picrotoxin solution. The perampanel at 10 μM significantly decreased the branching ratio in the network (as data are non-normally distributed, we use the Kolmogorov-Smirnov paired t-test at P < 0.05). In Fig. 6, the tissue in 10 μM picrotoxin had an average branching parameter of 0.36. The addition of 10 μM perampanel decreased the average branching parameter to 0.24, which was statistically significant at P < 0.05. The branching parameter continued to decrease after washout in 10 μM picrotoxin to 0.16.
In our present study, we document that in rodent models seizures in both rats and mice are well controlled by perampanel. Previous preclinical reports have documented efficacy for perampanel against seizures in rodents (Hanada et al., 2011; Rogawski and Hanada, 2013). In those studies, perampanel attenuated audiogenic, electroshock-induced, PTZ-induced seizures, 6 Hz–driven seizures, and amygdala-kindled seizures in rodents (Hanada et al., 2011). Our present findings provide the first systematic replication of the preclinical findings from Eisai Corporation (Hanada et al., 2011), which markets perampanel. The translational value of these data are documented in the correspondence in the efficacy in rodents and the antiepileptic efficacy in epileptic patients with partial-onset seizures with or without secondary generalized seizures (see reviews and analysis by Gao et al., 2013; Rheims and Ryvlin, 2013; Steinhoff et al., 2013; Zaccara et al., 2013).
In addition to anticonvulsant protection in rodents, perampanel induced motor deficits on the inverted screen test in mice and rats. The potencies for inducing motor impairments were greater than that for efficacy as an anticonvulsant with the exception of mouse electroshock seizures, where a protective index of ≥1 was found (3.5). Low-dose margins of efficacy over motor effects have been reported before in rodents (Hanada et al., 2011). Although it is well tolerated, perampanel has shown motor impairments in human studies and in phase III data (French et al., 2012; Krauss et al., 2012, 2013; Rektor et al., 2012). Thus, although protective indices in rodents are not always fully translatable into human motor side-effect liabilities (Löscher and Nolting, 1991), the data with perampanel in the current study and in Hanada et al. (2011) have uncovered the dose-dependent motor consequences of perampanel in humans.
As a first step in understanding the molecular basis of perampanel efficacy as well as its motoric effects, we explored perampanel pharmacologic effects in native rodent and human preparations. With the aid of the innovative technique of reconstituting native neurotransmitter receptors into Xenopus oocytes, we were able to demonstrate that perampanel inhibits native rat and native human AMPA receptors from both the hippocampus and the cerebellum. In addition, direct measurements from human epileptic cortical brain slices showed that perampanel decreased amplitudes and firing rates of large evoked field potentials. These findings are important because the inhibition of cortical and hippocampal AMPA receptors in human tissues provides an explanation for the beneficial effects of perampanel in epileptic patients.
Our results showing that perampanel also inhibits native hippocampal AMPA receptors from a refractory epileptic patient (both in oocytes and in fresh slices) show, for the first time, that perampanel acts on AMPA receptors in epileptic human brain tissue and that the AMPA receptors in the diseased state are still sensitive to the drug, a finding consistent with the antiepileptic efficacy of perampanel.
Some aspects of the fresh brain slice experiments are worth considering in more detail. Our data extend previous findings obtained with perampanel in cultured cortical neurons (Hanada et al., 2011) and studying synaptic transmission in hippocampal slice preparations of rat brain (Ceolin et al., 2012). As previously reported, tissue from humans is rarely spontaneously active in vitro (Hobbs et al., 2010). Picrotoxin, which acts as a noncompetitive antagonist for GABAA receptor chloride channels and is well known for increasing excitability in neural tissue (Mackenzie et al., 2002), was used to investigate the effects of perampanel. The results from these in vitro studies, consistent with the phase III efficacy data, show that perampanel provides seizure protection in cortical slices from the active region of the left temporal lobe from a medication-unresolved epilepsy patient. Perampanel significantly dampened local population activity and network activity in human tissue from an epileptic patient. This conclusion was derived from three specific measures: firing rate, amplitude, and the branching parameter, all of which decreased when 10 μM perampanel was washed into the slice bath.
Although these results are statistically significant, caution must be exercised when extrapolating these findings to human tissue in vivo. Experiments with slices are limited by several factors. Long-range neuronal connections are severed, and severing these long-range connections alters network properties; activity in the network failed to recover and return to baseline after perampanel was washed out. As obtaining human tissue is rare, this study relies on a sample of one, so replication is warranted. Finally, in a previous publication (Hobbs et al., 2010), we showed that large, network-wide events occurred simultaneously for extended periods of time in some tissue from human patients with epilepsy, and that the branching parameter was positively correlated with firing rate in these recordings. We also showed that this seizure-like activation can drive down the branching parameter (Hobbs et al., 2010), as the descendant of an extended network-wide event can at most have the same size or more likely fewer branches than its ancestor. To confirm that the effects on the branching parameter were not derived from network-wide activation, we searched for sustained network-wide events on time scales of up to 50s and determined that this type of activation did not decrease the branching parameter.
Finally, understanding the source of perampanel’s side effects is crucial to developing AEDs with better margins of safety. Given the motor impairments observed in rodents (Hanada et al., 2011; Rogawski and Hanada, 2013; present study) and the ataxia and dizziness reported in epileptic patients at higher doses of perampanel (French et al., 2012; Krauss et al., 2012, 2013; Rektor et al., 2012), it is suggested that cerebellar AMPA receptor blockade is a primary mediator of the main side effects of perampanel. Our results showing that both rodent and human cerebellar AMPA receptors are equally inhibited by perampanel further corroborate this hypothesis.
We hypothesize that the optimal AMPA receptor antagonist to be used as an AED would block AMPA receptors in affected brain regions (cerebral cortex, hippocampus, amygdala, thalamus) while leaving AMPA receptors in other brain regions (e.g., cerebellum) unaffected. It is conceivable that the subunit composition of AMPA receptors in the cerebellum differ from AMPA receptors in other brain regions. If that is the case, the possibility to develop subunit-selective AMPA receptor antagonists can be explored. AMPA receptor structure and function are further affected by proteins associated with AMPA receptors such as the transmembrane AMPA receptor–associated proteins (TARPs). TARPs are known to influence the pharmacologic properties of AMPA receptors (Menuz et al., 2007; Cokic and Stein, 2008; Kato et al., 2010). It is currently not known whether the activity of perampanel is influenced by TARPs. TARPγ2 is highly expressed in the cerebellum (Tomita et al., 2003). Thus, an AMPA receptor antagonist that has reduced potency and/or efficacy at AMPA receptors associated with TARPγ2 relative to other AMPA receptor isoforms might provide antiepileptic control without the cerebellar-mediated motor side effects.
In conclusion, our experimental findings confirm our hypothesis that perampanel antagonizes AMPA currents in both brain areas responsible for epileptic discharge and brain areas controlling motoric side effects. These experimental data in rodent and human tissue are consistent with the antiepileptic and motoric profile of perampanel in patients with refractory partial seizures. These data add to the growing body of data implicating that the dampening of excitatory neurotransmission through AMPA receptors is a safe and effective means of seizure control. The introduction of perampanel (Fycompa) to the clinic will enable the opportunity for the first time to evaluate this mechanism compared with others in epilepsy therapeutics.
The authors thank the patient and parents for consent and contribution of cortical and hippocampal tissues to this research—the authors are grateful for their courage and generosity. The authors also thank Randy Hunter, Mario Sgro, Deah Modlin, Brain Spegal, and Elizabeth Eberle of Covance for excellent work in performing the studies with intravenous PTZ, mouse electroshock, and rat amygdala kindling.
Participated in research design: Witkin, Ping, Zwart, Sher, Gleason, Hobbs, Smith.
Conducted experiments: Witkin, Valli, Ping, Jin, Zwart, Sher, Gleason, Hobbs, Smith.
Contributed new reagents or analytic tools: Hahn, Gardinier, Gernert, Smith.
Performed data analysis: Witkin, Ping, Zwart, Sher, Gleason, Hahn, Gardinier, Gernert, Hobbs.
Wrote or contributed to the writing of the manuscript: Witkin, Ping, Jin, Zwart, Sher, Sims, Chappell, Gleason, Hahn, Gardinier, Gernert, Hobbs, Smith.
- Received January 3, 2014.
- Accepted July 14, 2014.
This work was supported by a Pence Foundation grant (to J.L.S.).
- artificial cerebral spinal fluid
- after discharge duration
- after discharge threshold
- antiepileptic drug
- α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid
- excitable artificial cerebral spinal fluid
- GYKI 53773/LY300164/talampanel
- local field potential
- protective index
- transmembrane AMPA receptor–associated protein
- Copyright © 2014 by The American Society for Pharmacology and Experimental Therapeutics