Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

Within the central nervous system (CNS), -aminobutyric acid (GABA) serves as the principal inhibitory neurotransmitter. Once released from the presynaptic terminal, GABA binds to both presynaptic and postsynaptic GABA receptors, which are coupled to either a G protein-associated K⁺ channel (GABA_B receptors) or form Cl-permeable ion channels (GABA_A receptors). Alterations in GABA-ergic function have been implicated in a number of CNS disorders, including epilepsy, migraine, bipolar disorder, anxiety, and depression. As such, it is not surprising that the GABA_A receptor has been the target for numerous therapeutic entities, including the benzodiazepines, barbiturates, and neurosteroids. Collectively, these findings demonstrate the critical role played by GABA within the CNS and underscore the therapeutic impact of manipulating GABA-ergic neurotransmission within the CNS (for review and references, see Treiman, 2001).

Allosteric GABA_A receptor modulators such as the benzodiazepines and barbiturates have clearly played an important role in the treatment of epilepsy. However, the long-term utility of the benzodiazepines as anticonvulsants is limited by the rapid development of tolerance (for review and references, see Schmidt, 1995). Today, the primary use of the benzodiazepines as anticonvulsants is in the acute treatment of status epilepticus. Although effective as broad-spectrum antiseizure drugs, the clinical utility of the barbiturates is limited by the development of cognitive impairment and behavioral modifications [for reviews, see Devinsky (1995) and Mattson (1995)].

The action of neuronally released GABA is terminated by rapid uptake of GABA into surrounding neurons and astrocytes by selective GABA transporters (Schousboe and Kanner, 2002). Molecular cloning has identified four distinct GABA transporters (GAT1-4) that display regionally distinct distribution within the CNS (Gadea and Lopez-Colome, 2001). Attempts to design and synthesize specific inhibitors of GABA transport led to the successful identification of several investigational agents with demonstrated anticonvulsant activity in animal seizure models (Yunger et al., 1984; Schousboe et al., 1986; Krogsgaard-Larsen et al., 1987; White et al., 1993; Borden et al., 1994; Suzdak and Jansen, 1995; Bolvig et al., 1999). The GAT1-selective GABA uptake inhibitor tiagabine was subsequently demonstrated to possess clinical efficacy for the treatment of partial epilepsy in randomized clinical trials (Suzdak and Jansen, 1995) and was approved for the adjunctive treatment of partial epilepsy in 1997. The approval and subsequent registration of tiagabine as an antiepileptic drug clearly demonstrate the therapeutic potential associated with GABA uptake inhibitors. Tiagabine is an effective and safe drug for the management of epilepsy; however, it possesses a less than desirable pharmacokinetic profile. For example, it displays a short half-life (i.e., 7-9 h), which can be reduced even further (i.e., 2-3 h) by enzyme-inducing antiepileptic drugs, thereby necessitating three-time-a-day dosing (Genton et al., 2001). Tolerability to tiagabine seems to be limited by the presence of mild-to-moderate side effects, including dizziness, fatigue, and confusion (Genton et al., 2001). As such, the clinical utility of tiagabine is somewhat limited and the development of a better tolerated GABA uptake inhibitor with a more favorable pharmacokinetic profile could provide some distinct advantage over tiagabine.

It has been proposed that because GABA-mediated neurotransmission to a considerable extent relies on the recycling of GABA by reuptake into the presynaptic nerve ending it may be advantageous to selectively inhibit astrocytic GABA uptake to enhance GABA-ergic activity (Schousboe et al., 1983; Schousboe, 2000). This notion seems to be experimentally supported by the findings that selective inhibitors of neuronal GABA uptake are proconvulsants, whereas even marginally selective glia inhibitors are anticonvulsants and are able to increase the levels of extrasynaptic GABA (Schousboe et al., 1983; Gonsalves et al., 1989a,b; Juhasz et al., 1997). By selectively inhibiting astrocytic GABA uptake, presynaptic GABA levels are increased without impairing GABA metabolism. In contrast, selective inhibition of neuronal GABA transport leads to a disruption of GABA metabolism and ultimately a reduction in the total pool of releasable GABA (for review and references, see Schousboe et al., 1983).

The present study was undertaken to further investigate the cell type selectivity of analogs of 3-hydroxy-4-amino-4,5,6,7-tetrahydro-1,2-benzisoxazole (exo-THPO) that were previously reported to exhibit some cell type selectivity as GABA uptake inhibitors (Falch et al., 1999). Moreover, studies were conducted to characterize their anticonvulsant potential in the audiogenic seizure-susceptible (AGS) Frings mouse and to correlate their anticonvulsant activity with their potency as GABA transport inhibitors. Results obtained with exo-THPO and its two N-alkyl analogs were compared with results obtained with the clinically effective antiseizure drug and prototype GAT1 inhibitor tiagabine.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials

Newborn mice and 15-day-old mouse embryos were obtained from Bomholtgaard (Ry, Denmark). Plastic tissue culture dishes were purchased from NUNC A/S (Roskilde, Denmark) and fetal calf serum from Sera-Lab (Sussex, UK). Poly-D-lysine (molecular weight >300,000), trypsin, soybean trypsin inhibitor, dibutyryl cyclic AMP (dBcAMP), DNase, cytosine arabinoside, and amino acids were obtained from Sigma-Aldrich (St. Louis, MO); insulin from Novo Nordisk (Bagsværd, Denmark), and penicillin from Leo (Ballerup, Denmark). [³H]GABA (79.0 Ci mmol¹) was from DuPont (Frankfurt, Germany). exo-THPO and its N-alkyl analogs were synthesized as described previously (Falch et al., 1999). Tiagabine was generously provided by Novo Nordisk. All other chemicals were of the purest grade available from regular commercial sources. The accession numbers for mouse GAT1-4 are M92378, M97632, L04663, and L04662, respectively.

Primary Cell Culture

Astrocytes were cultured essentially as described by Hertz et al. (1989b). Prefrontal cortex was taken from newborn mice and passed through Nitex nylon sieves (80-µm pore size) into a slightly modified Dulbecco's minimum essential medium as defined by Hertz et al. (1982) containing 20% (v/v) fetal calf serum and subsequently plated onto NUNC 24-well multidishes. After 14 days in culture, and during the following weeks, the serum concentration in the culture medium was reduced to 10% (v/v) and 0.25 mM dBcAMP was added. This addition of dBcAMP led to the formation of well differentiated astrocytes (Hertz et al., 1982).

Cerebral cortical neurons were isolated and cultured from 15-day-old mouse embryos. After dissociation of the tissue by trypsinization and trituration in a DNase solution containing soybean trypsin inhibitor as described by Hertz et al. (1989a), the cells were plated onto NUNC 24-well multidishes. 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 (Hertz and Schousboe, 1987).

Transient Expression of GABA Transporters

The cDNAs encoding the four murine GABA transporters (Liu et al., 1992; Lopez-Corcuera et al., 1992) were subcloned into the mammalian expression vector pCis as detailed previously (Bolvig et al., 1999) and used for transient transfections into HEK 293 cells. HEK 293 cells were maintained in complete growth medium (minimal essential medium with Earle's salts, supplemented with 5% fetal calf serum, 1% anti-pleuro-pneumonia-like organisms, and 1% Glutamax I, pH 7.3 under 5% CO₂). GATpCis transfections were carried out as described by Chen and Okayama (1987) and Pritchett et al. (1988).

Construction of Expression Vectors for Stable Transfection of GATs

A modified version of the bicistronic expression vector pIRES (CLONTECH, Palo Alto, CA) was constructed, pIRES-BLAS-AN. A DNA fragment containing the gene from Aspergillus terreus coding for the enzyme blasticidin S deaminase was inserted into the multiple cloning site B of pIRES vector, allowing the use of blasticidin as the selection marker. The multiple cloning site A of pIRES-BLAS was expanded by digestion with NheI followed by treatment with Klenow enzyme to give blunt ends. Then the complementary primers 5'-ACCGGTGATATCTCTAGAGGCGCGCCGCTAGC-3' and 5'-GCTAGCGGCGCGCCTCTAGAGATATCACCGGT-3' were annealed and ligated into this position of pIRES-BLAS, generating pIRES-BLAS-AN and pIRES-BLAS-NA, which have opposite orientations of the added unique enzyme restriction sites. The new multiple cloning sites AN and NA were verified by automated DNA sequencing (MWG Biotech, Ebersberg, Germany).

To stably express the four mGATs, the corresponding cDNAs were subcloned into pIRES-BLAS-AN. GAT1 cDNA was subcloned from GAT1pBSK by digestion with BamHI, (blunt-ended with Klenow enzyme) and NheI and ligated into the EcoRV (5') and XbaI (3') sites of pIRES-BLAS-AN. GAT2 cDNA was subcloned from GAT2pBSK by digestion with XbaI and NheI and ligation into the XbaI site of pIRES-BLAS-AN. GAT3 cDNA was subcloned from GAT3pGEM4Z by digestion with DraIII (blunt-ended with Klenow enzyme) and EcoRI. The GAT3 insert was ligated into the (5') EcoRI and (3') MluI (blunt-ended with Klenow enzyme) sites of pIRES-BLAS-AN. GAT4 cDNA was subcloned from GAT4pBSK by digestion with XbaI and ligating the insert into the XbaI site of pIRES-BLAS-AN. All clones were identified by restriction enzyme analysis and subsequently constructs were confirmed by automated DNA sequencing (MGW Biotech).

Isolation of Stably Transfected Cells

Each of the four GATpIRES-BLAS-AN cDNAs was transfected into HEK 293 cells as described above for transient transfections. Cells were selected in the presence of 20 µg/ml blasticidin S (CAYLA, Toulouse, France) in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal bovine serum, 100 U/ml penicillin, and 100 µg/ml streptomycin. Colonies of stably transformed cells were obtained after 7 to 14 days of exposure to blasticidin S and were subsequently pooled and maintained in 2 µg/ml blasticidin S.

[³H]GABA Uptake

The uptake of [³H]GABA in cultured astrocytes and neurons as well as in recombinant cell systems was investigated essentially as described previously (Schousboe et al., 1977; Larsson et al., 1986a). The incubations were performed at 37°C in phosphate-buffered saline with an incubation time of 3 min. Incubations were initiated by exchanging the culture medium with incubation medium leaving the cells attached to the bottom of the wells during the entire procedure. Incubations were terminated by rapid wash with nonradioactive incubation medium. In the kinetic experiments, the GABA concentration varied over the range 1 to 1000 µM, whereas in IC₅₀ studies it was constant at 1 µM. The concentrations of uptake inhibitors present during incubations in the kinetic assays are stated in Tables 2 and 3. After incubation, the cells were dissolved in 0.4 M KOH, and radioactivity (Schousboe et al., 1977) and protein concentration (Lowry et al., 1951) were determined. Protein contents were related to bovine serum albumin used as the standard.

Data and Statistical Analysis

The IC₅₀ values were determined by use of SigmaPlot for Windows, version 3.02. The percentage of GABA uptake as a function of the inhibitor concentration in probit-log scale was plotted. The kinetic parameters V_max and K_m of the carrier-mediated high-affinity glial, neuronal, and GAT1 uptake were calculated (Larsson et al., 1986a,b) by means of a computer program for unconstrained minimization (Stewart, 1967; Dixon, 1972) or via the computer program of SigmaPlot for Windows, version 3.02, fitting to the following equation: V = [(V_maxS)/(K_m + S)] + kS. In neurons, astroglial cells, and HEK 293 cells expressing the GAT subtype, the nonsaturable influx component (kS) varied somewhat between batches but was in the range of 10³ to 10² ml · min¹ · mg¹. There was no systematic variation of this component, depending on the presence or absence of inhibitors. In the case of simple competitive inhibition, the K_i values were calculated from the determined K_m values of the control and test situation using the following equation: K_i = I/[(K'_m/K_m  1]; where I is the inhibitor concentration, K_m is the control value, and K'_m is the value determined in the presence of inhibitor.

Anticonvulsant Testing

Animals. Adult male and female Frings AGS mice (20-25 g of body weight) were obtained from an in-house breeding colony at the University of Utah (Salt Lake City, UT). Mice were housed in Association for the Assessment and Accreditation of Laboratory Animal Care-approved facilities under a constant 12-h light/dark cycle with a constant temperature of 21-23°C and a relative humidity of 30 to 50%. All animals were permitted free access to standard laboratory chow (Prolab RMH 3000; PMI Nutrition International, LLC, Brentwood, MO) and water except when removed from their home cages for testing. All mice were handled in a manner consistent with the recommendations in the National Research Council Publication Guide for the Care and Use of Laboratory Animals. All test compounds were dissolved in water, and the solutions made alkaline (pH 8-10) with 1 M NaOH. Once in solution, the final pH was adjusted to a value between 7 and 8 with 1 M HCl.

Audiogenic Seizures. The ability of exo-THPO and its corresponding N-methyl-exo-THPO and N-ethyl-exo-THPO analogs to prevent sound-induced seizures in the AGS Frings mouse model was assessed after i.c.v. administration (White et al., 1992). A similar study was conducted with the prototype GABA uptake inhibitor tiagabine. Each test substance was injected i.c.v. into the lateral ventricle of conscious mice with a 10-µl Hamilton syringe. Each compound was administered i.c.v. in a constant volume of 5 µl over a 10-s period. The i.c.v. route of administration was used due to the apparent poor penetration of the blood-brain barrier (BBB) by the bicyclic isoxazoles, and although tiagabine is able to cross the BBB, it was also administered i.c.v. so that a direct comparison of the results could be made.

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

Inhibition of GABA Uptake. Previous studies on the inhibitory action of exo-THPO, N-methyl-exo-THPO, and N-ethyl-exo-THPO (Fig. 1) on neuronal and astrocytic GABA transport were extended to include the cloned mouse GABA transporters GAT1-4. Concentration-response curves for inhibition of GABA uptake into neurons, astrocytes, and HEK 293 cells transfected with mGAT1 were used to calculate the IC₅₀ values for exo-THPO, its two alkylated analogs, and tiagabine. The results from this analysis are summarized in Table 1 along with inhibition data for all four compounds in GAT2-4-transfected HEK 293 cells. It can be seen that the IC₅₀ values for exo-THPO and N-methyl-exo-THPO on mGAT1 reflect the IC₅₀ values found for inhibition of GABA uptake into neurons by the two compounds. Moreover, none of the compounds inhibited GABA transport in HEK 293 cells transfected with GAT2-4. Furthermore, kinetic analysis of the inhibitory action of exo-THPO and N-methyl-exo-THPO on GAT1 (Table 2) demonstrated that both of these GABA analogs are competitive inhibitors of GABA uptake. Similarly, exo-THPO was found to display competitive inhibition kinetics for GABA uptake into neurons and astrocytes (Table 3). The results obtained with exo-THPO, N-methyl-exo-THPO, and N-ethyl-THPO were compared with the nipecotic acid derivative tiagabine (Fig. 1), which is currently in clinical use for the management of partial seizures. As can be seen from Table 1, tiagabine inhibited GABA uptake into neurons, astrocytes, and GAT1-transfected HEK 293 cells much more potently than the inhibitors derived from 4,5,6,7-tetrahydroisoxazolo[4,5-c]pyridin-3-ol (THPO). Tiagabine, like exo-THPO and the two exo-THPO analogs, weakly or insignificantly inhibited GABA uptake into GAT2-4-transfected HEK 293 cells.

View larger version (17K): [in this window] [in a new window]   Fig. 1.   Chemical structures of THPO, exo-THPO, and its N-alkyl analogs N-methyl-exo-THPO and N-ethyl-exo-THPO; the nipecotic acid analog tiagabine; and the O-pivaloyloxymethyl prodrug of N-methyl-exo-THPO.

View this table: [in this window] [in a new window]   TABLE 2 Kinetic characterization of the inhibition of GABA transport in mGAT1 transfected HEK 293 cells by exo-THPO and N-methyl-exo-THPO HEK 293 cells expressing mouse cloned GAT1 were obtained as detailed under Experimental Procedures and the kinetic analysis of GABA transport performed in the absence or presence of 500 or 1000 µM exo-THPO and 200 or 400 µM N-methyl-exo-THPO using a series of GABA concentrations, as described under Experimental Procedures. The Vmax value in the absence of the inhibitor was 2.1 to 3.8 nmol · min1 · mg1 of protein. Statistically significant differences from the control (Student's t test) are indicated by asterisks (*p < 0.01; **p < 0.001).

View this table: [in this window] [in a new window]   TABLE 3 Kinetic characterization of the inhibition of GABA transport in neurons and astrocytes by exo-THPO (EXO) Cells were cultured from mouse cerebral cortex as detailed under Experimental Procedures and the kinetic analysis of GABA transport performed in the absence or presence of 900 µM (neurons) or 400 µM (astrocytes) EXO using a series of GABA concentrations as described under Experimental Procedures. The Vmax values in the absence of the inhibitor were 3.20 ± 0.15 nmol · min1 · mg1 of protein and 0.13 ± 0.01 nmol · min1 · mg1 of protein in the neurons and astrocytes, respectively. Statistically significant differences from the control (Student's t test) are indicated by asterisks (*p < 0.01; **p < 0.001).

Inhibition of Audiogenic Seizures. Inhibition of GABA uptake by i.c.v. administration of the GAT1-selective GABA uptake inhibitors resulted in a time- and dose-dependent inhibition of audiogenic seizures in the Frings mouse. As shown in Fig. 2, peak anticonvulsant activity for all three exo-THPO analogs was observed within 15 to 30 min of i.c.v. administration. In contrast, the onset of action of the prototype anticonvulsant tiagabine was delayed relative to the three exo-THPO analogs tested. Hence, the peak anticonvulsant activity of tiagabine was observed 2 h after an i.c.v. dose of 48.5 nmol (Fig. 2).

View larger version (16K): [in this window] [in a new window]   Fig. 2.   Time-effect relationship for the anticonvulsant activity of exo-THPO (128 nmol; solid circle, solid line), N-methyl-exo-THPO (120 nmol; solid square, dotted line), N-ethyl-exo-THPO (114 nmol; solid triangle, dashed line), and tiagabine (49 nmol; solid diamond, dotted line) after i.c.v. administration to Frings AGS mice. Each test substance was injected i.c.v. into the lateral ventricle of awake mice (n = 4-8 mice/time point) in a total volume of 5 µl using a 10-µl Hamilton syringe. At various times after dosing, individual mice were challenged with a high-intensity sound stimulus. The percentage of animals at each time point not displaying a tonic extension seizure was recorded and plotted as a function of time.

View larger version (14K): [in this window] [in a new window]   Fig. 3.   Dose-response relationship for the anticonvulsant action of exo-THPO (solid circles, solid line), N-methyl-exo-THPO (solid squares, solid line), N-ethyl-exo-THPO (solid triangle, dashed line), and tiagabine (solid diamond, dashed line) after i.c.v. administration to Frings AGS mice (n = 8-13 mice/dose level). Results are expressed as a percentage of animals protected against a tonic extension seizure induced by a high-intensity sound stimulus. Anticonvulsant activity was quantitated at the time to peak effect as defined by the time-response study shown in Fig. 2.

View larger version (19K): [in this window] [in a new window]   Fig. 4.   Correlation between the ED50 for protection against sound-induced seizures and the IC50 for GABA uptake into astrocytes (A) and neurons (B) for exo-THPO, N-methyl-exo-THPO, N-ethyl-exo-THPO, and tiagabine. ED50 values for tiagabine, exo-THPO, and its analogs were calculated by probit analysis of the dose-response data plotted in Fig. 3 and plotted against the IC50 values obtained from GABA inhibition studies conducted with neuronal and glial cells grown and maintained in tissue culture.

View larger version (13K): [in this window] [in a new window]   Fig. 5.   Time- and dose-effect relationship for the anticonvulsant activity of the O-pivaloyloxymethyl prodrug of N-methyl-exo-THPO after i.p. administration to Frings AGS mice. Results obtained from the time-response study are plotted in the inset as a percentage of mice not displaying a tonic extension seizure at the various times (n = 4 mice/time point) tested. For the dose-response study, groups of eight mice were treated with varying doses of the O-pivaloyloxymethyl prodrug and tested at the time of peak anticonvulsant effect (i.e., 30 min) for protection against sound-induced tonic extension seizures.

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

The results of the present investigation confirm previous findings that the bicyclic isoxazole exo-THPO and its N-methylated and N-ethylated analogs are potent inhibitors of neuronal and glial GABA uptake (Falch et al., 1999). In addition, the IC₅₀ values reported herein for exo-THPO, N-methyl-exo-THPO, N-ethyl-exo-THPO, and tiagabine were found to be essentially identical to previously reported values (Braestrup et al., 1990; Falch et al., 1999). The present study extends these previous findings by demonstrating that exo-THPO and its two N-alkylated analogs are highly selective inhibitors of the cloned GAT1 transporter being inactive at GAT2-4. Furthermore, all three isoxazoles were found to exert their inhibitory effect through a competitive interaction with the GABA transporter. These findings are similar to those reported for the bicyclic isoxazole THPO, which is isomeric with exo-THPO (Fig. 1; Bolvig et al., 1999). Importantly, the present study further confirms the marked glial selectivity of N-methyl-exo-THPO. This is particularly interesting in light of the finding that this compound was found to be highly selective for GAT1. The fact that both neurons and glial cells express GAT1 is not sufficient to explain the cellular selectivity of N-methyl-exo-THPO. At the present time, no exact explanation is obvious. However, it may be hypothesized that cooperative interactions may exist among the different GABA transporters, several of which may be expressed on the same cell (Gadea and Lopez-Colome, 2001; Schousboe and Kanner, 2002). Thus, it is conceivable that such an interaction may alter the pharmacological properties of the transporters, thereby explaining the unexpectedly high potency of N-methyl-exo-THPO as an inhibitor of GABA uptake in astrocytes. It can, however, not be excluded that transporters so far not cloned and with unknown pharmacological properties may exist.

Both exo-THPO and its two N-alkylated analogs were found to possess potent anticonvulsant activity when administered i.c.v. to Frings AGS mice. Moreover, for all three compounds tested, the anticonvulsant activity was observed at i.c.v. doses devoid of behavioral impairment. Based on previous findings in the isoniazid seizure model wherein it was demonstrated that the anticonvulsant activity of N-methyl-exo-THPO was restricted to the R-enantiomer (Falch et al., 1999), it is likely that the anticonvulsant activity of N-methyl-exo-THPO in the Frings mouse is also stereoselective. This conclusion is supported by the previous demonstration that inhibition of GABA uptake is more pronounced by the R-enantiomer rather than the S-enantiomer (Falch et al., 1999). Interestingly, N-methyl-exo-THPO was the most potent of the three analogs tested followed by exo-THPO and N-ethyl-exo-THPO. This rank order of potency was found to correlate very well with the rank order of inhibition of GABA uptake found in astrocytes (r² = 0.988). In contrast, a weak correlation was observed between inhibition of neuronal GABA uptake and anticonvulsant activity of exo-THPO, its two alkylated analogs, and tiagabine (r² = 0.534). Although preliminary in nature, these findings provide theoretical support for the potential therapeutic utility of a highly selective glial GABA uptake inhibitor. Furthermore, as mentioned above pure neuronal selective GABA uptake inhibitors display proconvulsant activity, presumably as a result of their ability to deplete the transmitter GABA pool (Schousboe et al., 1983; Gonsalves et al., 1989b). In contrast, glial selective inhibitors increase both the extrasynaptic and synaptosomal levels of GABA without disrupting GABA metabolism (Wood et al., 1983; Juhasz et al., 1997).

Tiagabine was also found to be a potent anticonvulsant in the Frings AGS mouse. This finding is in keeping with its previously demonstrated anticonvulsant activity (Suzdak and Jansen, 1995). Interestingly, tiagabine was less potent as an anticonvulsant than might be predicted by its ability to inhibit GABA uptake. For example, compared with exo-THPO and its N-alkylated analogs, tiagabine was between 100- and 1000-fold more potent as an inhibitor of GABA uptake; however, it was only 2.5- to 7-fold more potent as an anticonvulsant. At the present time, there is no obvious explanation for this interesting dichotomy. However, it should be noted that the actual concentration of tiagabine at the site of action could be lower than predicted from the amount injected because the onset of action of tiagabine was delayed considerably from the time of injection, a phenomenon not seen for the exo-THPO family of compounds.

Previous investigations have shown the O-pivaloyloxymethyl prodrug of N-methyl-exo-THPO to be active against isoniazid-induced seizures (Falch et al., 1999). The present study found this prodrug to be active against audiogenic seizures after i.p. administration to Frings mice. After i.p. administration, this prodrug displayed a time- and dose-dependent inhibition of audiogenic seizures that is thought to result from the cleavage of the pivaloyloxymethyl from the parent compound upon entrance into the CNS. As noted in the time course study (Fig. 5, inset) a dramatic decrease in anticonvulsant activity was observed 60 min after i.p. administration. This differs from the rather long duration of action observed after i.c.v. administration of the parent compound (i.e., more than 2 h). This may be explained in part by a relatively rapid peripheral metabolism of the prodrug before penetration of the BBB. By decreasing the total amount available, less of the prodrug is therefore available for cleavage of the pivaloyloxymethyl group upon entrance into the CNS. Interestingly, the direct i.c.v. injection of the prodrug into the ventricular compartment of the brain failed to exert an anticonvulsant action. This apparent discrepancy is explained in part by the presence of marked toxicity after i.c.v. administration of the prodrug that was not observed after i.c.v. administration of the parent compound. Nonetheless, the results of the present study when coupled with previous findings in the isoniazid seizure model (Falch et al., 1999) confirm the validity of the prodrug approach and may represent a viable formulation for the systemic administration of N-methyl-exo-THPO and other bicyclic isoxazole derivatives. Alternatively, more lipophilic analogs could be developed to enhance penetration of the BBB while at the same time preserving the glial selectivity of N-methyl-exo-THPO.

Surprisingly, N-methyl-exo-THPO was found to be only 2-fold less potent than tiagabine as an anticonvulsant. This observation, taken together with the finding that N-methyl-exo-THPO-treated animals perform better than tiagabine-treated mice in the rotarod toxicity test, suggests that this new class of GABA transport inhibitors may represent a better tolerated alternative to tiagabine as antiseizure agents. Clearly, testing of these compounds is at a very early stage of development and other seizure models must be evaluated before conclusions can be reached regarding the overall therapeutic potential of this class of compounds. Nonetheless, the results of the present investigation support the continued evaluation of this class of compounds as anticonvulsants and development of more lipophilic analogs.