In a recent study, EF1502 [N-[4,4-bis(3-methyl-2-thienyl)-3-butenyl]-3-hydroxy-4-(methylamino)-4,5,6,7-tetrahydrobenzo [d]isoxazol-3-ol], which is an N-substituted analog of the GAT1-selective GABA uptake inhibitor exo-THPO (4-amino-4,5,6,7-tetrahydrobenzo[d]isoxazol-3-ol), was found to inhibit GABA transport mediated by both GAT1 and GAT2 in human embryonic kidney (HEK) cells expressing the mouse GABA transporters GAT1 to 4 (mGAT1–4). In the present study, EF1502 was found to possess a broad-spectrum anticonvulsant profile in animal models of generalized and partial epilepsy. When EF1502 was tested in combination with the clinically effective GAT1-selective inhibitor tiagabine [(R)-N-[4,4-bis(3-methyl-2-thienyl)-3-butenyl]nipecotic acid] or LU-32-176B [N-[4,4-bis(4-fluorophenyl)-butyl]-3-hydroxy-4-amino-4,5,6,7-tetrahydrobenzo[d]isoxazol-3-ol], another GAT1-selective N-substituted analog of exo-THPO, a synergistic rather than additive anticonvulsant interaction was observed in the Frings audiogenic seizure-susceptible mouse and the pentylenetetrazol seizure threshold test. In contrast, combination of the two mGAT1-selective inhibitors, tiagabine and LU-32-176B, resulted in only an additive anticonvulsant effect. Importantly, the combination of EF1502 and tiagabine did not result in a greater than additive effect in the rotarod behavioral impairment test. In subsequent in vitro studies conducted in HEK-293 cells expressing the cloned mouse GAT transporters mGAT1 and mGAT2, EF1502 was found to noncompetitively inhibit both mGAT1 and the betaine/GABA transporter mGAT2 (Ki of 4 and 5 μM, respectively). Furthermore, in a GABA release study conducted in neocortical neurons, EF1502 did not act as a substrate for the GABA carrier. Collectively, these findings support a functional role for mGAT2 in the control of neuronal excitability and suggest a possible utility for mGAT2-selective inhibitors in the treatment of epilepsy.
Reduction of GABA-mediated inhibitory neurotransmission is associated with seizure activity and drugs that elevate synaptic GABA levels either by inhibition of GABA degradation or inhibition of high-affinity transport have been demonstrated to possess anticonvulsant activity (see Dalby, 2003; Sarup et al., 2003). For example, the GABA-transaminase inhibitor vigabatrin and the GABA-transport inhibitor tiagabine [(R)-N-[4,4-bis(3-methyl-2-thienyl)-3-butenyl]nipecotic acid] are clinically effective antiepileptic drugs (for review and references, see Ben-Menachem, 2002; Kalviainen, 2002).
Since the advent of cloning of several GABA transporters from different species including mouse, rat, and human, interest has been focused on the development of inhibitors of the GABA transporter (GAT) 1. In support of this principle, a number of GAT1-selective inhibitors have been developed and demonstrated to possess anticonvulsant activity in animal seizure models (Swinyard et al., 1991; Pavia et al., 1992; Suzdak and Jansen, 1995; White et al., 2002; Dalby, 2003). For example, the first peripherally active GABA uptake inhibitor was SKF 89976A [(R)-N-(4,4-diphenyl-3-butenyl) nipecotic acid], an N-substituted diphenyl butenyl analog of nipecotic acid (Swinyard et al., 1991). The corresponding analogs of guvacine, SKF 100330A [N-(4,4-diphenyl-3-butenyl)guvacine] and CI-966 [N-2-bis[4-(trifluoromethylphenyl) methoxy]ethyl guvacine], supported the principle that modulation of GABA transport would result in an anticonvulsant action (Swinyard et al., 1991; Pavia et al., 1992). Unfortunately, subsequent clinical trials with CI-966 revealed the potential for serious psychotic adverse effects in certain patients (Radulovic et al., 1993).
Tiagabine is a second generation analog of nipecotic acid that contains the dithienyl ring system in place of the diphenyl rings and was developed clinically without any obvious psychotropic adverse effects. To date, tiagabine is the only clinically available GABA transport inhibitor used for the treatment of epilepsy. However, since its introduction into the world-wide market, a number of individual case reports have suggested that tiagabine, if not titrated correctly or employed in patients with pre-existing psychiatric disorders, may possess a similar adverse-effect profile (Trimble et al., 2000) as that predicted for CI-966 (Radulovic et al., 1993). Although acute psychosis has not been substantiated in more recent studies (Sackellares et al., 2002), tiagabine use has been associated with an increased incidence of depression (for review, see Kalviainen, 2002). These patient reports demonstrate the continued need for less toxic third generation GABA transport inhibitors.
In this regard, it would seem advantageous if such transport inhibitors would exhibit an inhibition profile different from that of tiagabine, i.e., an affinity for transporters other than GAT1, such as GAT2 to 4. To this end, Dalby et al. (1997) have demonstrated that such inhibitors may be interesting as antiepileptic drugs. For example, it was reported that inhibitors of GAT3 and GAT4 (Thomsen et al., 1997) afforded protection against maximal electroshock, audiogenic, and kindled seizures (Dalby et al., 1997) Unfortunately, these inhibitors, which structurally are not GABA mimetics (Thomsen et al., 1997), were shown to have affinity for α-1 adrenergic and D-2 dopamine receptors (Dalby et al., 1997). As such, it is difficult to make any definitive conclusion regarding the mechanism underlying their anticonvulsant action.
The present study was undertaken in an effort to obtain detailed information about newly developed N-substituted analogs of exo-THPO (4-amino-4,5,6,7-tetrahydrobenzo[d]isoxazol-3-ol; see Fig. 1) that have been shown to inhibit to varying degrees GABA transporters other than GAT1 (Clausen et al., 2005). Of particular interest is the analog EF1502 [N-[4,4-bis(3-methyl-2-thienyl)-3-butenyl]-3-hydroxy-4-(methylamino)-4,5,6,7-tetrahydrobenzo[d]isoxazol-3-ol], which was found to equipotently inhibit both mGAT1 and mGAT2 and was completely devoid of activity at a multitude of neurotransmitter systems including adrenergic, GABAergic, and dopaminergic receptors (radioligand binding study of 74 different receptors and ion channels; data not shown). The role of mGAT2 within the brain is poorly understood. In addition to transporting GABA, this transporter is thought to regulate the uptake of the osmo-regulator betaine (for review and references, see Borden et al., 1995). However, it is not clear whether modulation of mGAT2 contributes to control of neuronal excitability. As such, the present investigation employed a number of animal seizure and epilepsy models to characterize the anticonvulsant profile of this particular exo-THPO analog and to elucidate a functional role for the betaine/GABA transporter mGAT2. The findings obtained from isobolographic analysis of results obtained from combination studies conducted with EF1502 and tiagabine or LU-32-176B [N-[4,4-bis(4-fluorophenyl)-butyl]-3-hydroxy-4-amino-4,5,6,7-tetrahydrobenzo-[d]isoxazol-3-ol] suggests a functional role for mGAT2 in the control of seizure activity. These findings provide the basis for this report.
Materials and Methods
Materials. Plastic tissue culture dishes were purchased from Taconic, Europe (Ry, Denmark), and fetal calf serum was purchased from SeraLab (Crawley, UK). Poly-d-lysine (mol wt. > 300.000), trypsin, soybean trypsin inhibitor, DNase, cytosine arabinoside and amino acids were obtained from Sigma-Aldrich (St. Louis, MO), insulin from NOVO-Nordisk A/S (Bagsvaerd, Denmark), and penicillin from Leo Pharmaceuticals (Ballerup, Denmark). [3H]GABA (79.0 Ci/mmol) was from DuPont-NEN (Frankfurt, Germany). Exo-THPO and its N-substituted analogs were synthesized as described previously (Clausen et al., 2005). Tiagabine was generously provided by Sanofi Synthelabo (Brøndby, Denmark). All other chemicals were of the purest grade available from regular commercial sources. The accession numbers for mouse GAT1 and GAT2 are M92378 and M97632, respectively.
Cultured Neurons. Cerebral cortical neurons were isolated and cultured from 15-day-old mouse embryos that were obtained from Bomholtgaard (Ry, Denmark). After dissociation of the tissue by trypsinization and trituration in a DNase solution containing soybean trypsin inhibitor as described by Hertz et al. (1989), the cells were plated onto NUNC 35-mm Petri dishes. After 48 h in culture, 20 μM (final concentration) cytosine arabinoside was added to the culture medium to prevent astrocytic proliferation (Larsson et al., 1985). Cells were cultured for 7 to 8 days, at which time the neurons had become functionally differentiated with properties resembling mature GABAergic neurons such as high activity of glutamate decarboxylase and pronounced vesicular GABA release (Hertz and Schousboe, 1987).
Subcloning and Expression of GABA Transporters. The cDNAs encoding the two murine GABA transporters GAT1 and GAT2 (mGAT1 and mGAT2, respectively; Liu et al., 1992, 1993; Lopez-Corcuera et al., 1992) were subcloned into the mammalian expression vector, pCis, as described previously (Bolvig et al., 1999). GAT-1pBSK was digested with XbaI, blunt-ended with Klenow enzyme, and subsequently digested with NheI. The 2.2-kbp fragment containing GAT-1 was then ligated into the 5′-ClaI (blunt-ended with Klenow enzyme) and 3′-XbaI sites of the pCis vector. The 1.9-kbp XbaI (5′)-NheI (3′) GAT-2 fragment from GAT-2pBSK was ligated into the XbaI site of the pCis vector. GATpCis cDNAs were transformed into XL1-Blues bacteria (Stratagene, La Jolla, CA), and plasmids were prepared by QIAGEN (Valencia, CA) column purification. Human embryonic kidney (HEK)-293 cells were maintained in complete growth medium (minimum Eagle's medium with Earle's salts, supplemented with 5% fetal calf serum, 1% anti-Pichia pastoris lysyl oxidase, and 1% Glutamax I, pH 7.3 under 5% CO2). GAT-pCis transfections were carried out as described previously (White et al., 2002).
[3H]GABA Uptake. The uptake of [3H]GABA in the recombinant cell systems was investigated essentially as described previously (Bolvig et al., 1999; White et al., 2002). The incubations, performed at 37°C in phosphate-buffered saline, were initiated by exchanging the culture medium with incubation medium leaving the cells attached to the bottom of the wells during the entire procedure and terminated (after 3 min) by rapid wash with nonradioactive incubation medium. In the kinetic experiments, the GABA concentration was varied over the range 1 to 1000 μM. The concentrations of uptake inhibitors present during incubations in the kinetic assays are stated in Table 2. After incubation, the cells were dissolved in 0.4 M KOH, and radioactivity (Schousboe et al., 1977) and protein concentrations (Lowry et al., 1951) were determined. Protein contents were related to bovine serum albumin used as the standard. The kinetic parameters, Vmax and Km, of the carrier-mediated, high-affinity GABA uptake in HEK-293 cells expressing mGAT1 and mGAT2 were calculated (Larsson et al., 1986a,b) by means of a computer program for unconstrained minimization or via the computer program Grafit v3.0 (Erithacus Software, Horley, Surrey, UK), fitting to the following equation: [V = (VmaxS)/(Km + S)] + KS. The nonsaturable influx component (KS) varied somewhat between batches but was in the range of 10-3 to 10-2 ml/min/mg. There was no systematic variation of this component depending on the presence or absence of inhibitors. In the case of noncompetitive inhibition, the Ki values were calculated from the determined Vmax values of the control and test situation using the following equation: Ki = I/[Vmax/V1max - 1], where I is the inhibitor concentration, Vmax is the control value, and V1max is the value determined in the presence of inhibitor.
GABA Release. Cerebral cortical neurons in culture were pre-loaded with [3H]GABA (1 μM, 0.1 μCi) for 30 min in the presence of 10 μM vigabatrin to irreversibly inactivate GABA-transaminase, thereby blocking GABA metabolism (Drejer et al., 1987; Gram et al., 1988). Individual cultures (35-mm Petri dishes) were subsequently placed in a superfusion apparatus (Drejer et al., 1987) at 37°C equipped with peristaltic pumps, and the cells were superfused at a flow rate of 2 ml/min. Fractions from the outlet were collected every 30 s, and at the end of the experiments, radioactivity was determined in all fractions. During the superfusion, either 200 μM nonradioactive GABA or 200 μM EF1502 was added to the superfusion medium for 2 min. Results were expressed as counts per minute per fraction collected. It should be noted that since the baseline of the GABA release during the entire superfusion period was constant no major loss of intracellular [3H]GABA occurred during this period.
Animals. Male albino CF no. 1 mice (18–25 g; Charles River Laboratories, Inc., Wilmington, MA) and male and female Frings audiogenic seizure-susceptible mice (18–25 g; Anticonvulsant Drug Development Program, University of Utah, Salt Lake City, UT) were used as experimental animals for studies conducted at the University of Utah. For these studies, all animals were allowed free access to both food (S/L Custom Lab Diet-7) and water except when they were removed from their cages for the experimental procedure. Mice were housed in a temperature and humidity controlled facility and were maintained on 12-h light/dark cycle beginning at 6:00 AM. Furthermore, all mice were housed, fed, and handled in a manner consistent with the recommendations in the Health, Education, and Welfare publication (National Institutes of Health) 86-23, Guide for the Care and Use of Laboratory Animals. Animals were euthanized in accordance with Public Health Service policies on the humane care of laboratory animals.
For those studies conducted at H. Lundbeck A/S, male Wistar rats (280–320 g at start of experiment) from M&B (Laven, Denmark) were used for amygdala kindling. For the i.v. pentylenetetrazol (PTZ) and i.p. pilocarpine tests, male NMRI mice (18–22 g; M&B) were used. All animals were allowed to adapt to the laboratory environment for at least 1 week prior to testing. Animals were housed on a 12-h light/dark cycle, beginning at 6:00 AM. All experiments were conducted between 8:00 AM and 2:00 PM. Experimental work was performed under a license from the Danish Ministry of Justice and in accordance with EC Directive 86/609/EEC and the Danish law regulating experiments on animals.
The test substances were either suspended or dissolved in 0.5% methylcellulose by trituration in a mortar and pestle and sonicating for 10 min. All substances were administered i.p. or s.c. in a volume of 0.01 ml/g body weight in mice and 0.05 ml/g body weight in rats.
Anticonvulsant Testing. The anticonvulsant profile of EF1502 was established by several different anticonvulsant tests that included audiogenic, 6-Hz psychomotor, amygdala kindled, PTZ, and pilocarpine seizures.
Audiogenic Seizures. The ability of LU-32-176B, EF1502, and tiagabine to prevent sound-induced seizures in the Frings AGS-susceptible mouse model was assessed following i.p. administration.
For each of the analogs examined, time-response curves were constructed using a dose that produced a submaximal anticonvulsant effect in a group of four mice per time point tested (5, 15, 30, 60, and 120 min). Dose-response studies were then conducted at the subsequently determined time of peak effect of each test substance. For these studies, individual mice were placed into a Plexiglas cylinder (diameter, 15 cm; height, 18 cm) fitted with an audio transducer (Model AS-ZC; FET Research and Development, Salt Lake City, UT) and exposed to a sound stimulus of 110 decibels (11 KHz) delivered for 20 s. Sound-induced seizures are characterized by wild running followed by loss of righting reflex with fore limb and hind limb tonic extension. Mice not displaying hind limb tonic extension were considered protected.
The anticonvulsant activity for each of the test compounds was then quantitated in groups of four to eight mice per dose level. Varying doses of each substance were administered to groups of mice until at least two points were established between the limits of 0 and 100% protection.
In addition to assessing the percentage of protection following exposure to audiogenic stimulation, the effect of each analog on the audiogenic seizure severity was determined by assigning a score to the seizure phenotype as follows: 0, no response; 1, wild running for <10 s; 2, wild running for >10 s; 3, all-limb clonus; 4, fore limb extension; and 5, fore limb and hind limb extension.
The dose of drug required to produce the desired endpoint in 50% of animals tested (i.e., the ED50) and the 95% confidence interval were then calculated by Probit analysis of the dose-response data using a computer program based on the method described by Finney (1971).
EF1502 and LU-32-176B were also tested in combination with tiagabine in the Frings AGS mouse. For these studies, varying doses (administered as a fraction of their ED50 value) of one of the drugs were administered i.p. The effect of a given dose level on the efficacy of the second drug was then established by redetermining the ED50 for the second drug. The time of drug administration was optimized so that the audiogenic seizure test was conducted at the time to peak effect for each experimental drug. The results were plotted as function of their ED50, and 95% confidence intervals and isobolos were constructed in an effort to determine the nature of the observed interaction between EF1502 and tiagabine or LU32-176B (i.e., additive, antagonistic, or synergistic).
Six-Hertz Psychomotor Seizure Test. The ability of EF1502 to prevent seizures induced by 6-Hz corneal stimulation (3-s duration) was assessed at the time to peak effect. Current was applied by an apparatus similar to that originally described (Woodbury and Davenport, 1952). EF1502 was evaluated at a current sufficient to produce a seizure in 97% of the control animals tested (i.e., 22 mA). Prior to placement of the corneal electrodes, a drop of anesthetic/electrolyte solution (0.5% tetracaine hydrochloride in 0.9% saline) was applied to the eyes of each animal.
Six-Hertz seizures are characterized by a minimal clonic phase that is followed by stereotyped, automatistic behaviors described originally as being similar to the aura of human patients with partial seizures (for historical review and references, see Barton et al., 2001). Animals not displaying this behavior were considered protected.
Subcutaneous PTZ-Induced Seizures and i.v. PTZ Seizure Threshold. EF1502 was tested for its ability to prevent a minimal clonic seizure induced by the chemoconvulsant PTZ administered s.c. at a dose of 85 mg/kg. Absence of a 3-s clonic episode was used as the endpoint for protection in these three tests. PTZ was dissolved in 0.9% saline and injected in a volume of 0.01 ml/g body weight in mice. Animals were observed for at least 30 min for the presence or absence of a seizure.
Infusion of the chemoconvulsants PTZ (5 mg/ml, 0.5 ml/min) through the lateral tail vein induces clonic seizure with a loss of righting reflex in mice. The threshold for producing these seizures, calculated as milligrams per kilogram, was determined in groups of 7 to 12 mice 30 min after pretreatment (i.p.) with EF1502 (four–five doses) and the corresponding vehicle. The threshold doses for the drug treatment groups were normalized to the vehicle group and expressed as a percent increase from control. In experiments where EF1502 was coadministered with tiagabine, they were administered at the same time by i.p. injection on opposite sides of the abdomen. All data are reported as mean ± S.E.M., and statistical analysis was performed by one-way ANOVA.
Amygdala Kindling. Rats were anesthetized with Hypnorm (5 mg/ml midazolam), Dormicum (10 mg/ml fluanisone and 0.315 mg/ml fentanyl), and H2O in a ratio of 1:1:2 administered s.c. in a volume of 0.2 ml/0.1 kg prior to stereotaxic implantation of tripolar electrodes (Plastics One, Roanoke, VA) into the right basolateral amygdala (AP, -2.8 mm; L, -4.8; V, -8.7, measured from bregma; Paxinos and Watson, 1997). After surgery, the animals received the analgetic Rimadyl (5 mg/kg) and were subsequently allowed to recover for 2 weeks before the kindling protocol was initiated.
Animals were stimulated daily with monophasic square wave pulses (2 s, 50 Hz) using an Ellegaard Systems (Faaborg, Denmark) stimulator. The intensity of the stimulus for kindling was determined by increasing the stimulus in 25-μA steps until an afterdischarge was observed (electroencephalogram recorded using Spike 2 software); this current afterdischarge threshold was then used for kindling. Stimulations were then delivered daily until five consecutive grade 5 seizures were observed; at this point, the animals were considered to be fully kindled.
Assessment of Drug Effects on Afterdischarge Threshold. The threshold for induction of an afterdischarge was determined in kindled rats using an ascending stairstep procedure; using 10 μA as the initial stimulus, steps of 10 μA were applied at 1-min intervals until an afterdischarge was observed. Afterdischarges were defined as being electroencephalogram spikes at least twice the amplitude of the prestimulus recording, with a frequency greater than 1/s, for a minimum duration of 3 s.
The effect of EF1502 on afterdischarge threshold was determined 24 h later, 30 min after administration of EF1502 (0.3–2.5 mg/kg s.c.). The afterdischarge threshold was determined using 10-μA steps, with an initial stimulus 20 μA less than that previously measured for each rat. In addition to observing the afterdischarge threshold, the duration of the afterdischarge, the severity of the seizure according to the Racine scale (Racine, 1972), and the number of rats displaying fully generalized seizures (Racine scale ≥ 3) were also recorded. The ED50 and 95% confidence interval for protection against the fully kindled secondarily generalized seizure (Racine scale ≥ 3 was then calculated by Probit analysis; Finney, 1971).
Pilocarpine Seizure Test. Groups of eight mice were injected with pilocarpine (250 mg/kg i.p.) 30 min after s.c. administration of EF1502 or vehicle; the dose of pilocarpine was chosen from a previously constructed dose-response curve to induce convulsions in 97% of control animals (data not shown). The percentage of animals per group that exhibited clonic convulsions within 30 min was recorded, and results were compared by Fisher's exact probability test.
Behavioral Impairment. For those studies conducted at the University of Utah, minimal motor impairment was identified in mice by the rotarod procedure as described previously (Dunham and Miya, 1957). Inability of the mouse to maintain its equilibrium for one min in each of three trials on this rotating rod (6 rpm) was used as an indication of such impairment. The effect of EF1502 on rotarod impairment was then quantitated in groups of four to eight mice per dose level. Varying doses of each substance were administered to groups of mice until at least two points were established between the limits of 0 and 100% impairment. The dose of drug required to produce the desired endpoint in 50% of animals tested (i.e., the TD50), and the 95% confidence interval was then calculated by a computer program based on the method described by Finney (1971).
In combination studies conducted at H. Lundbeck A/S designed to assess whether EF1502 enhanced the behavioral toxicity of tiagabine, a modified rotarod procedure was employed. For this test, groups of eight untrained mice were tested for six runs (30 s) on a 75-mm-diameter rotarod rotating at 17 rpm (Rotamex 4/8; Columbus Instruments, OH) 30 min after i.p. administration of drug or vehicle. For each mouse, the total time spent on the rotarod was recorded, and mean ± S.E.M. was calculated for each group. Significant differences between drug- and vehicle-treated groups were determined by one-way ANOVA.
Inhibition of GABA Uptake. The inhibition profiles of tiagabine and the exo-THPO analogs (see Fig. 1) on GABA uptake in neurons and astrocytes as well as HEK cells transfected with the cloned mouse GABA transporters mGAT1 to 4 obtained in a recently conducted study (Clausen et al., 2005) are summarized in Table 1. It can be seen that EF1502 (the dithienyl-butenyl analog of N-methyl-exo-THPO), in contrast to tiagabine (the dithienyl-butenyl analog of nipecotic acid), exhibited pronounced inhibitory activity toward mGAT2, whereas mGAT3 and mGAT4 were only weakly inhibited. It should be noted that EF1500 as well as LU-32-176B, both of which are lipophilic analogs of exo-THPO and not of N-methyl-exo-THPO, did not display any affinity for mGAT2–4 but only inhibited GABA transport mediated by mGAT1. All of the compounds potently inhibited GABA uptake in neurons and astrocytes. Of the drugs tested, tiagabine was approximately 10-fold more potent than the exo-THPO analogs. No neuronal or glial selectively was observed for any of the compounds tested. The kinetics of the inhibition was investigated for EF1502, an analog of N-methyl-exo-THPO, with regard to GABA uptake in HEK cells expressing mGAT1 or mGAT2. As shown in Table 2, EF1502 was found to act as a noncompetitive inhibitor of both mGAT1 and mGAT2.
Based on this information, cultured neocortical neurons preloaded with [3H]GABA were used to investigate whether GABA release could be stimulated by EF1502. The cells were superfused, and the addition of a high concentration of non-radioactive GABA (i.e., 200 μM) to the superfusion medium led to a pronounced stimulation of release of radioactive GABA caused by homoexchange (Fig. 2). When 200 μM EF1502 was used in place of GABA, no increase in the release of [3H]GABA was seen (Fig. 2). These results demonstrate a lack of heteroexchange between [3H]GABA and EF1502.
Anticonvulsant Profile. EF1502 was evaluated in a battery of anticonvulsant models to assess its ability to block both generalized and partial seizures in mice and rats (Table 3). EF1502 was effective in preventing sound-induced seizures in the Frings audiogenic seizure-susceptible mouse model of reflex epilepsy. EF1502 was found to block both tonic extensor and clonic seizure components of the audiogenic seizure. Although less potent (ED50 4.4 versus 0.55 mg/kg i.p. for EF1502 and tiagabine, respectively), EF1502 was equally effective to tiagabine in this reflex seizure model. As shown in Table 3, the anticonvulsant activity of EF1502 was associated with the R-isomer; i.e., the S-isomer was found to be inactive at doses up to 20 mg/kg. In a separate model, EF1502 was found to afford protection against pilocarpine-induced seizures with 100% of the animals being protected at 10 mg/kg. In the same test, tiagabine was slightly more potent offering complete protection against clonic seizures at 2.5 mg/kg.
EF1502 displayed dose-dependent protection against clonic seizures induced by s.c. PTZ and limbic seizures induced by 6-Hz corneal stimulation. In the s.c. PTZ test, EF1502 prevented clonic seizures (ED50 18.6 mg/kg). In addition, EF1502, at a dose as low as 2.5 mg/kg i.p., produced a 17% increase in the i.v. PTZ threshold. In this particular test, EF1502 was equipotent to tiagabine (results not shown).
EF1502 was also found to prevent limbic seizures in two different rodent models of partial epilepsy (i.e., the 6-Hz psychomotor seizure test and the amygdala kindled rat model). Compared with results obtained in the Frings audiogenic seizure model, EF1502 was only slightly less potent in the 6-Hz psychomotor seizure test than in the Frings audiogenic seizure model (ED50 4.4 versus 10.4 mg/kg i.p., respectively).
In the amygdala kindled rat model of partial epilepsy, EF1502 produced a dose-dependent reduction in both the electrographic afterdischarge duration and behavioral seizure score without affecting the afterdischarge threshold (Table 4). At the highest dose tested (2.5 mg/kg), EF1502 significantly decreased the afterdischarge duration by 50% and reduced the seizure score from 5 ± 0 to 2.6 ± 0.6.
Combination Studies. In an effort to assess whether the mGAT2 inhibitory properties of EF1502 contributed to its anticonvulsant activity, an isobologram study was conducted in the Frings audiogenic seizure-susceptible mouse. As shown in Fig. 3A, when the two mGAT1-selective inhibitors (LU-32-176B and tiagabine) were combined, an additive anticonvulsant effect was observed (i.e., experimental results fell along the predicted line of additivity). In contrast, when the mixed mGAT1 and mGAT2 inhibitor EF1502 was combined with tiagabine or LU-32-176B, a synergistic interaction was observed (Fig. 3, B and C; i.e., experimental results fell to the left of the line of additivity). In a subsequent study, the combination of minimally active doses of EF1502 and tiagabine were found to produce a highly significant synergistic anticonvulsant action in the i.v. PTZ threshold test (Fig. 4). Collectively, these results support the hypothesis that mGAT2 plays a role in regulating brain excitability and that an inhibitor of mGAT2 possesses anticonvulsant activity.
Behavioral Impairment. The effect of EF1502 on motor performance was evaluated in two different variants of the rotarod impairment model. In the first study, the ability of a mouse to maintain balance on a 2.5-cm-diameter knurled rod rotating at 6 rpm was evaluated. The dose producing impairment in 50% of the mice tested (i.e., TD50) was found to be 57 mg/kg (Table 3). In contrast, the TD50 for tiagabine in this test was 1.3 mg/kg i.p. In the second model, the total time spent on a 7.5-cm-diameter rod rotating at 17 rpm was assessed over six separate sessions, each lasting 30 s. As shown in Fig. 5, doses of EF1502 as high as 40 mg/kg did not negatively impair performance in this test. In contrast, tiagabine produced marked impairment at doses of 5 mg/kg and higher (Fig. 5). Interestingly, when EF1502 was tested in combination with tiagabine, the degree of impairment was not greater than that observed with tiagabine alone (Fig. 5).
In contrast to the parent compounds THPO and N-methyl-THPO, of which the latter has been shown to be a glial-selective GABA uptake inhibitor (Falch et al., 1999; White et al., 2002), the lipophilic analogs of these compounds described here did not exhibit cell-type selectivity (Clausen et al., 2005). However, EF1502, the lipophilic analog of N-methyl-exo-THPO, was found to display significant affinity for mGAT2 as well as mGAT1. This was not the case for EF1500, its non-N-methylated counterpart (Clausen et al., 2005). That this N-methyl group (see Fig. 1) is an important determinant of GAT2 inhibitory activity is underlined by the finding that exchanging the lipophilic moiety of the molecule (i.e., EF1500 and LU-32-176B) did not by itself affect the affinity of the compound toward mGAT1 and mGAT2.
The finding that EF1502 was unable to stimulate the release of preloaded GABA in cultured neocortical neurons strongly suggests that it does not act as a substrate for the carrier. This is in line with the finding that it acted as a noncompetitive inhibitor of GABA uptake in HEK cells expressing mGAT1 or mGAT2. It also agrees with the previous observation that analogous lipophilic analogs of nipecotic acid, which in itself is a substrate for the carrier (Larsson et al., 1983), do not act as substrates (Larsson et al., 1988; Braestrup et al., 1990). It is of interest to note that N-methyl-THPO was previously found to be a competitive inhibitor of mGAT1 (White et al., 2002), whereas in the present study, its lipophilic analog, EF1502, was found to be a noncompetitive inhibitor of mGAT1. This strongly suggests that the addition of the lipophilic side chain induces a molecular restraint that interferes with the interaction between the GABA mimetic part of the molecule and the GABA binding site of the carrier protein. Since such an alternative binding site would not have any affinity for GABA, this kinetic analysis would indicate a mechanism of action of EF1502 that is noncompetitive. Such a model would be compatible with the finding that tiagabine, having the same lipophilic side chain (Fig. 1), binds to a site overlapping but not identical to the GABA recognition site (Braestrup et al., 1990).
Given the unique properties of EF1502 as an mGAT1 and mGAT2 GABA uptake inhibitor, it was of interest to assess whether it possessed any unique attributes as an anticonvulsant. Results obtained from a series of studies in well established anticonvulsant models demonstrated that EF1502 possessed a broad-spectrum anticonvulsant profile consistent with that of tiagabine. Furthermore, the results from these studies suggest that EF1502, like tiagabine, exerts its anti-convulsant action by elevating seizure threshold.
Further examination found EF1502 to be active in two models of human partial epilepsy; i.e., 6-Hz psychomotor seizure test and the amygdala kindled rat. The former seizure test possesses a unique pharmacological profile in that the characteristic seizure induced by low frequency stimulation is refractory to phenytoin, carbamazepine, lamotrigine, and topiramate but not to valproic acid and the novel anti-epileptic drug levetiracetam (Barton et al., 2001). Another model where EF1502 showed marked anticonvulsant activity was the pilocarpine model of acute generalized seizures. Carbamazepine, lamotrigine, and phenytoin are ineffective in this model, even at doses causing marked ataxic effects (Watson, 2001). To characterize EF1502's potential for the treatment of partial epilepsy, it was evaluated in the traditional amygdala kindled rat model of partial epilepsy. In this model, it was demonstrated to decrease electrographic seizure duration and behavioral seizure severity. These results suggest that EF1502 would be effective against primary partial seizures and partial seizures secondarily generalized. Finally, it should be noted that the anticonvulsant activity of EF1502 was restricted to its R-stereoisomer. This finding is consistent with previous results, wherein inhibition of GABA uptake was also found to be stereoselective (Clausen et al., 2005). Interestingly, EF1502 seems to possess a tolerability profile that is superior to that of tiagabine. For example, in the rotarod performance tests conducted, doses substantially exceeding anticonvulsant effective doses did not seem to produce any notable impairment. In contrast, at a dose that was only slightly higher than its anticonvulsant dose in the i.v. PTZ threshold test, tiagabine produced a significant dose-dependent impairment in the rotarod test (Fig. 5). Thus, based on these initial findings, it would seem that EF1502 offers some beneficial improvement over the GAT1-selective inhibitor tiagabine.
At the present time, the role of mGAT2 within the brain is not understood. This transporter is thought to regulate the uptake of the osmo-regulator betaine (for references, see Borden et al., 1995). In an effort to determine whether inhibition of mGAT2 might contribute the anticonvulsant action of EF1502, three separate combination studies with tiagabine were undertaken. The first of these, conducted in the Frings audiogenic seizure-susceptible mouse, clearly demonstrated that the combination of the GAT1 and GAT2 inhibitor EF1502 with the GAT1-selective inhibitors tiagabine or LU-32-176B results in a synergistic anticonvulsant action (Fig. 3, B and C). This conclusion is supported by the negative finding wherein the combination of two GAT1-selective inhibitors (i.e., tiagabine and LU-32-176) in the same model resulted in only an additive effect. That this synergism is indeed related to the difference in the affinity for mGAT1 and mGAT2 and not to differences in chemical structures of the GABA-mimetic moiety is supported by the fact that the GAT1 inhibitors tiagabine and LU-32-176B, which are structurally different from each other, both with regard to their lipophilic side chain and their GABA mimetic moiety (Fig. 1), did not exhibit synergism. This is further substantiated by the finding that synergism could be demonstrated with EF1502 in combination with either of the structurally different mGAT1 inhibitors tiagabine and LU-32-176B. Additional support for this conclusion is provided by the combination study conducted in the i.v. PTZ seizure threshold test. In this study, minimally effective doses of EF1502 and tiagabine, when administered together, resulted in a greater than additive effect. Collectively, these results suggest that inhibition of mGAT2 by EF1502 contributes significantly to its overall anticonvulsant activity. It is worth noting that the synergy between EF1502 and tiagabine or LU-32-176B could be related to pharmacokinetic interaction between these compounds. Such an interaction could theoretically lead to an increase in the intracerebral concentration and an apparent supra-additive effect. However, the fact that synergism was observed after combination of EF1502 with either of the two mGAT1 inhibitors, but was not seen with the combination of tiagabine and LU-32-176B, the structures of which are totally different (Fig. 1), makes this highly unlikely. In addition, rotarod impairment was not increased when EF1502 was combined with tiagabine. Such an effect would have been predicted by a purely pharmacokinetic interaction.
Based on Northern blot studies in human brain and in situ hybridization studies in mouse brain, the betaine transporter (i.e., mGAT2) seems to be primarily localized to the extra-synaptic component (Borden et al., 1995). However, ultrastructural immunocytochemical studies have yet to add more details to these findings. On the other hand, when microinjected into cultured hippocampal neurons, mGAT2 was reported to colocalize with MAP2 but not with synapsin (Ahn et al., 1996). Moreover, after transfection of mGAT2 into Madin Darby canine kidney cells, mGAT2 is sorted to the basolateral surface of the cells (Ahn et al., 1996). These findings support a somatodendritic localization of mGAT2, and it seems reasonable to hypothesize that mGAT2 may play a role in regulating extrasynaptic GABA levels. Thus, inhibiting mGAT2 would be expected to increase extrasynaptic GABA that could then activate nonsynaptic GABAA and GABAB receptors.
Interestingly, EF1502 was found to possess an anticonvulsant profile similar to that of the nonselective GAT2–4 inhibitors NNC 05-2045 and NNC 05-2090 (Dalby et al., 1997). In this study, the anticonvulsant activity of these compounds was attributed to inhibition of mGAT3 and mGAT4 despite the fact that these two inhibitors were somewhat more potent inhibitors of mGAT2 (Thomsen et al., 1997) Hence, given that EF1502 lacks affinity for mGAT3 and mGAT4, an alternative conclusion would be that inhibition of mGAT2 did indeed contribute to the anticonvulsant efficacy of these two inhibitors of GABA transport. Clearly, the development of a purely selective, peripherally active, mGAT2 inhibitor could resolve this issue.
It is of particular interest that synergy was obtained in the anticonvulsant studies but not in the rotarod impairment study conducted. Lack of a synergistic effect in this study might suggest that a selective mGAT2 inhibitor would offer some distinct advantage over a nonselective mGAT1/mGAT2 inhibitor when it comes to tolerability. However, this hypothesis will only be substantiated or refuted at the time that a peripherally active mGAT2-selective inhibitor becomes available for testing. Overall, the results of the present investigation suggest that EF1502 possesses a favorable anticonvulsant and tolerability profile and support the continued development of EF1502 and its analogs.
We thank Lone Rosenquist, Kirsten Thuesen, and Laura Webb for expert technical assistance, Irene Kamerath for editorial assistance, and Nathan Nelson (Tel Aviv University, Israel) for providing the mGAT1 and mGAT2 clones used in this study.
- Received April 7, 2004.
- Accepted June 9, 2004.
This work was supported by Danish State Medical Research Council Grant 20-00-1011, the Lundbeck Foundation, and National Institute of Neurological Disorders and Stroke Grant NO1-NS-9-2313.
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
ABBREVIATIONS: GAT, GABA transporter; SKF 89976A, (R)-N-(4,4-diphenyl-3-butenyl) nipecotic acid; SKF 100330A, N-(4,4-diphenyl-3-butenyl)guvacine; CI-966, N-2-bis[4-(trifluoromethylphenyl) methoxy]ethyl guvacine; exo-THPO, 4-amino-4,5,6,7-tetrahydrobenzo[d]isoxazol-3-ol; EF1502, N-[4,4-bis(3-methyl-2-thienyl)-3-butenyl]-3-hydroxy-4-(methylamino)-4,5,6,7-tetrahydrobenzo[d]isoxazol-3-ol; LU-32-176B, N-[4,4-bis(4-fluorophenyl)-butyl]-3-hydroxy-4-amino-4,5,6,7-tetrahydrobenzo[d]isoxazol-3-ol; HEK, human embryonic kidney; PTZ, pentylenetetrazol; ANOVA, analysis of variance; EF1500, N-[4,4-bis (3-methyl-2-thienyl)-3-butenyl]-3-hydroxy-4-amino-4,5,6,7-tetrahydrobenzo[d]isoxazol-3-ol; NNC 05-2045, 1-(3-9H-carbazol-9-yl)-1-propyl)-4-(4-methoxyphenyl)-4-piperidinol; NNC 05-2090, 1-(3-(9H-carbazol-9-yl)-1-propyl)-4-(2-methoxyphenyl)-4-piperidinol.
- The American Society for Pharmacology and Experimental Therapeutics