RPR 119990, a Novel α-Amino-3-hydroxy-5-methyl-4-isoxazolepropionic Acid Antagonist: Synthesis, Pharmacological Properties, and Activity in an Animal Model of Amyotrophic Lateral Sclerosis

Materials and Methods

Synthesis of RPR 119990

9-Carboxymethyl-4-oxo-5H, 10H-imidazo[1,2-a]indeno[1,2-e] pyrazin-2-phosphonic acid was prepared following an original six-step synthesis (Fig.1) starting from 3-(2-carboxymethylphenyl) propionic acid (Lambert et al., 1979). Full details have been reported elsewhere (Aloup et al., 1996). The starting material 1 was prepared in two steps from the commercially available 2-bromophenylacetic acid by condensation with acrylic acid in the presence of a catalytic amount of palladium(II) acetate (1% mol), tri-o-tolylphosphine (4% mol), and tributylamine (100°C). This led to 3-(2-carboxymethylphenyl) acrylic acid with a 64% yield. Hydrogenation of the ethylenic group in the presence of a catalytic amount of palladium on charcoal (10%) in acetic acid (pressure of H₂ = 17.6 psi at room temperature) gave1 with a 98% yield. RPR 119990 presented as a highly water-soluble (400 mg/ml), pale yellow solid with a melting point >260°C. Analytical and spectroscopic data of the final material were consistent with the targeted structure (see legend of Fig. 1 for abbreviated details).

[³H]AMPA Binding Assay

[³H]AMPA binding was determined by the method described by Honoré and Drejer (1988) with modifications. Membranes prepared from the cerebral cortex of rats (0.5 mg of protein/ml) were suspended in 10 mM phosphate buffer, pH 7.5, in the presence of 100 mM KSCN, and incubated for 30 min at 4°C with [³H]AMPA (10 nM) and various concentrations of RPR 119990 or 1 mM l-glutamate for the determination of nonspecific binding. The binding interaction was terminated by filtration through glass-fiber filters (Printed filtermat A) by using a Skatron microcell harvester (Molecular Devices France, St. Grégoire cedex, France). The radioactivity remaining on the filters was measured by liquid scintillometry. IC₅₀ values were determined by nonlinear regression according to a sigmoidal equation (Enzfitter software, Biosoft, Stapleford, UK). TheK_i was calculated from the Cheng-Prusoff equation K_i = IC₅₀/(1 + [L]/K_D). The dissociation constant (K_D = 34 nM) was obtained from a Scatchard analysis of saturation isotherms. Protein levels were measured by the method of Bradford (1976) (Bio-Rad protein assay; Bio-Rad, Hercules, CA). The specificity of RPR 119990 for AMPA receptors was assessed for a number of central nervous system receptors and transporters by CEREP (Le Bois l'Evêque, Celle-L'Evescault, France) except for the strychnine-insensitive glycine binding at NMDA receptors, which was performed at Aventis according to previously published methods (Boireau et al., 1996). Summary of assay conditions are described in Table1. Detailed protocols can be found in the CEREP product literature.

Voltage-Clamp Studies in Oocytes Expressing Human Recombinant AMPA Receptors

Oocytes were removed from ovarian lobes of anesthetizedXenopus laevis (150–300 g) and defolliculated for 1 h with collagenase (2 mg/ml) in modified Barth's medium as previously described (Debono et al., 1993). Oocytes were injected with human GluR1 + GluR2 mRNA (2:0.5 μg/μl) and maintained for at least 48 h at 16 to 19°C in modified Barth's medium with streptomycin (10 mg/ml) and penicillin (100 units/ml) before being used for voltage-clamp experiments. On the day of study, oocytes were mounted individually in a small recording chamber (300 μl) and superfused at room temperature with a Ringer's medium containing 90 mM NaCl, 1 mM KCl, 1 mM MgCl₂, 1 mM CaCl₂, and 5 mM HEPES at a flow rate of 6 ml/min. Current responses were recorded with 3 M KCl (0.8–2.5 MΩW)-filled glass microelectrodes at a holding potential of −60 mV with a two-electrode voltage-clamp amplifier. Appropriate final bath concentrations of RPR 119990 were prepared by dilution in the Ringer's recording medium. For pA₂ determination, concentration-response curves to kainate were constructed in the absence and in the presence of different concentrations of RPR 119990. The amplitudes of the evoked responses were expressed as a percentage of the maximal predrug baseline control responses. EC₅₀ values were determined by nonlinear least-squares regression procedure according to a sigmoidal equation (Prism 3.0, GraphPad Software, Inc., San Diego, CA).

Patch-Clamp Studies in Rat Cerebellar Granule Neurons

Granule cells were enzymatically dispersed from cerebella of 7-day-old Sprague-Dawley rat pups and plated to a density of 10⁵ cell/cm² on 12-mm-coated cover glasses. Cultures were used at 4 to 7 days of age and examined under continuous superfusion of a bathing medium containing 130 mM NaCl, 1 mM CaCl₂, 10 mM HEPES, and 11 mM d-glucose. Currents were recorded using the whole-cell configuration of patch clamp with the membrane potential clamped at −60 mV. Kainate (50 μM) was applied for 5-s pulses by using a rapid microperfusion technique. Evoked current was measured at the end of the pulse. The effects of RPR 119990 were evaluated after a preincubation period of 5 min with the compound alone by application of a mixture of 50 μM kainate and a given concentration of RPR 119990. Return to control conditions was evaluated after a 3-min washout period. The IC₅₀ value was calculated by nonlinear regression according to a sigmoidal equation (Prism 3.0) from data obtained at five concentrations ranging from 0.3 to 300 nM. Three to six neurons were tested at each concentration.

Brain Slice Studies in Vitro

Transverse hippocampal slices (0.5 mm in thickness) were prepared from the hippocampi of male Sprague-Dawley rats weighing 270 to 350 g by using a McIlwain tissue chopper (Mickle Laboratory Engineering Ltd., Gomshall, Surrey, UK) and mounted in a submersion-type recording chamber under continuous superfusion (flow rate = 2.5–3 ml/min) of gassed (95% O₂, 5% CO₂) artificial cerebrospinal fluid (ACSF) containing 124 mM NaCl, 3 mM KCl, 1.3 mM MgSO₄, 2 mM CaCl₂, 26 mM NaHCO₃, 1.25 mM NaH₂PO₄, and 10 mM glucose at 32°C. Stimulation and recording electrodes were positioned in the CA1 stratum radiatum and field excitatory postsynaptic potentials (EPSP) evoked by constant voltage stimuli delivered at approximately 80% of maximal synaptic strength were recorded extracellularly using appropriate waveform processing and data analysis software. Field EPSP amplitude was monitored every 30 s for up to 180 min post-treatment. Treatments were applied via the perfusion bath by replacing ACSF containing a defined concentration of compound for normal ACSF during 90 min. Only one concentration of compound was studied per slice. Raw data were normalized with respect to a 15-min baseline preceding each treatment. The IC₅₀ value was calculated by nonlinear regression according to a sigmoidal equation (Prism 3.0).

In Vivo Electrophysiological Studies

Twenty-five male Sprague-Dawley rats weighing 300 g ± 10% were anesthetized with 1.2 to 1.5 g/kg i.p. urethane, equipped with an intrajugular catheter for compound administration, tracheotomized, and connected to a rodent ventilator for respiratory function support. Animals were placed in a standard stereotaxic frame on a thermostated heating blanket to maintain rectal temperature within physiological limits. A hole was drilled through the skull and stimulating and recording electrodes, pasted together approximately 1.4 mm apart, were inserted. The implantation site was identified according to the Paxinos and Watson (1986) atlas by moving the recording electrode 3 mm posterior to bregma and 2 mm lateral to midline, the array being oriented so that the stimulating electrode was positioned 4 mm posterior to the bregma and 3 mm lateral to the midline. Vertical electrode position was adjusted by maximizing the amplitude of the CA1 stratum radiatum EPSP. The latter were recorded extracellularly in response to constant voltage stimulation (0.1-ms duration) delivered at a voltage giving approximately 90% the maximal synaptic signal and were monitored every 30 s for up to 6 h post-treatment by using waveform-processing and data analysis software. RPR 119990 was given subcutaneously after 30 min of stable baseline recording. The half-maximal inhibitory dose (ID₅₀) was calculated by nonlinear regression according to a sigmoidal equation (Prism 3.0).

Anticonvulsant Action in Mouse Electroshock

RPR 119990 was administered by intravenous, subcutaneous, or oral route (0.3, 1, 3, 10, and 30 mg/kg s.c.) at various time intervals before electroshock in male Swiss White mice (CD1, b.wt. 19–23 g; Charles River France, St. Aubin lès Elbeuf, France). After an appropriate lapse of time in their home cage the animals were taken by the scruff of the neck and an electric shock (75 mA, 0.3 s, 50 pulses/s) was applied immediately by ocular electrodes. Each animal was then held for 30 s, or until a tonic convulsion occurred. This could be recognized by the gathering of the hind feet up under the belly, followed by their full extension behind the body. ED₅₀ values were calculated by linear regression (Prism 3.0).

PK Monitoring in Mouse

Plasma and brain PK parameters were determined in male CD1 mice (three mice per time point) following s.c. administration of RPR 119990 in saline at dose levels of 1, 3, 10, and 30 mg/kg and times 0.5, 1, 3, and 6 h post dosing. Oral bioavailability was evaluated after administration at 3 mg/kg. Plasma samples were diluted in mobile phase and analyzed by ion pair liquid chromatography with fluorescent detection (λex = 345 nm and λem = 415 nm) [85% phosphate buffer (20 mM) containing tetrabutylammonium sulfate (10 mM)/15% acetonitrile; LiChrospher BP 5 μm, 125 × 4 mm, retention time 3.5 min].

In Vivo Metabolism

A limited investigation of the in vivo metabolism of RPR 119990 was undertaken after s.c. administration of the product to CD1 mice at 3 mg/kg. Urine was collected at 0 to 2, 2 to 4, and 4 to 7 h post administration. Urine samples were reconstituted in mobile phase and analyzed by liquid chromatographic fluorescence as described above and by liquid chromatography mass spectrometry/mass spectrometry (Finnigan LCQ positive ion electrospray, 0.05% trifluoroacetic acid, 20 mM ammonium acetate/acetonitrile; Phenomenex Luna, 5 μm, 200 × 4.6 mm).

Transgenic Mouse Studies

Transgenic mice [B6SJL-TgN(SOD1-G93A, G1H] heterozygous for the deficient SOD1 gene and wild-type litter mates were identified by polymerase chain reaction. Animals were treated starting on the 50th day of life, until their death, with RPR 119990 at 1, 3, or 2 × 3 mg/kg/day s.c. in saline solution. The following three parameters were examined.

Muscle Strength.

Animals were tested for muscle strength at intervals by using a muscle strength meter (Columbus Instruments, Columbus, OH). Briefly, each mouse was held by the tail and pulled steadily over a metal grill three times in rapid succession. The force was recorded on a force meter. The highest score for muscle strength was taken. Mice were discounted if they were considered too sick to grip the grid. In a first experiment, muscle strength in wild-type mice treated with saline or RPR 119990 was followed at 5- to 15-day intervals throughout the experiment as was strength in SOD1 transgenic mice receiving saline or RPR 119990 at 1 and 3 mg/kg s.c. In a subsequent study all animals surviving up to 150 days of age were assessed for muscle strength at this age.

Glutamate Uptake.

Animals were sacrificed at 150 days and spinal cord rapidly dissected for synaptosomal preparations and high-affinity Na⁺-dependent glutamate uptake was measured as previously described (Canton et al., 1997). Briefly, spinal cords were homogenized in 20 volumes of sucrose (0.32 M) and the homogenate centrifuged at 800g for 10 min. After centrifugation of the supernatant at 20,000g for 20 min, the pellet was resuspended and washed in 50 volumes of sucrose buffer. Transport assays were performed in Krebs-HEPES buffer. Duplicate samples, with or without sodium chloride, were incubated at 37°C for 3 min in the presence of [³H]l-glutamate after appropriate isotopic dilution. The reaction was stopped by addition of ice-cold buffer containing choline and followed by filtration and scintillation counting of the radioactivity retained on filters. High-affinity Na⁺-dependent uptake was calculated by subtracting results obtained in choline buffer from that obtained with Na⁺ buffer.

Life Expectancy.

Remaining mice were treated until day of death, which was noted.

Statistical significance of the biochemical and behavioral data was assessed using Kruskal-Wallis nonparametric analysis of variance followed by Dunn's multiple comparison test.

Results

Binding Studies

RPR 119990 is a selective ligand of rat brain AMPA receptors endowed with a submicromolar binding affinity. Tritiated AMPA displacement studies performed with rat cortex membrane preparations gave a K_i value of 107 ± 36 nM (mean ± S.E.M., n = 6). Binding studies or ion translocation assays at various receptors or transporters showed that this compound has no significant affinity for 16 neurotransmitter receptors, including other subtypes of ionotropic glutamate receptors, four ion channels, three receptor regulatory sites, five brain/gut peptide receptors, four steroid receptors, and six transporters (Table1). A single point value of 61% displacement at 10 μM on kainate binding to rat membranes suggests that the compound has a certain affinity for this closely related glutamate site, but something in the region of a 50-fold selectivity could be expected between AMPA and kainate affinity.

RPR 119990 Antagonizes AMPA Receptor-Mediated Functional Responses in Vitro

In X. laevis oocytes expressing human AMPA receptors, RPR 119990 antagonized responses induced by the nondesensitizing AMPA receptor agonist kainic acid in a concentration-dependent manner. This antagonism appeared to be competitive as shown by the parallel rightward shift of the concentration-response curve to kainate without depression of maximal responses (Fig. 2A). Schild plot analysis of this effect yielded a K_B value of 71 nM (Fig. 2B). Patch-clamp studies on rat cerebellar granule neurons in primary culture showed that RPR 119990 similarly has antagonist properties at native AMPA receptors with an IC₅₀value of 50 nM (Fig. 3, A and B).

RPR 119990 Blocks Glutamatergic Field Potentials in Rat Hippocampus in Vitro and in Vivo

The effects of RPR 119990 on the functioning of integral glutamatergic synapses was first studied on transverse slices of the hippocampus, a brain slice preparation containing well characterized synaptic circuits by using endogenous glutamate acting on AMPA receptors for generating excitatory postsynaptic potentials. Addition of RPR 119990 to the hippocampal slices in ACSF produced a concentration-dependent depression of the electrically evoked field EPSP amplitude (Fig. 4A). The threshold concentration for this effect was 30 nM. Higher concentrations lead to an IC₅₀ value of 93 nM. Partial reversal of the inhibitory effect was observed upon washout with compound-free ACSF for 90 min after the application of RPR 119990. To evaluate whether the compound is able to cross the blood-brain barrier in pharmacologically relevant amounts, we then tested its effect on the same hippocampal glutamatergic synapse in the anesthetized rat. Figure4B shows changes in field EPSP amplitude after subcutaneous injection of 2, 4, 8, or 16 mg/kg RPR 119990. The threshold dose for inhibition of the EPSP was 4 mg/kg and they remained depressed for over 5 h after injection. The effect of RPR 119990 was dose-dependent over the tested dose range with maximal inhibition occurring 5 h after injection. The ID₅₀ value by this route of administration was 6.2 ± 1.2 mg/kg.

Anticonvulsant Studies in Mouse

RPR 119990 showed potent antagonism of electroshock-induced convulsions in the male CD1 mouse by both intravenous (ED₅₀ value at 1 h = 0.86 mg/kg) and subcutaneous routes (ED₅₀ value at 1 h = 2.3 mg/kg). This suggests that the compound is readily absorbed after subcutaneous administration also in mice. No anticonvulsive activity was, however, seen after administration up to 90 mg/kg by oral route. These studies showed that RPR 119990 is endowed with a significantly longer duration of action compared with the quinoxalinedione AMPA antagonist YM-900 (Fig. 5).

PK Monitoring in Mouse

After s.c. administration, there was rapid uptake into the systemic circulation with high Cp_max values achieved at the first sampling time (0.5 h) at three of the four dose levels (means of 1615, 4513, and 14,625 ng/ml for 3-, 10-, and 30-mg/kg doses, respectively) and at 1 h in the case of the 1-mg/kg dose (534 ng/ml). The product was thereafter cleared rapidly with terminal elimination half-lives of 0.7 to 1.2 h. Quantifiable levels were consistently evident up to 6 h at the doses of 3, 10, and 30 mg/kg and 3 h in the case of 1 mg/kg. High bioavailabilities of 54.5 to 70.5% were determined by this route of administration, the highest value being at the highest dose level of 30 mg/kg. The plasma pharmacokinetics appeared to be roughly dose proportional between 1 and 30 mg/kg.

After s.c. administration at 1 mg/kg, brain levels were very low or not detectable. Levels were consistently detected up to 1.0 h post dose in the case of the 3-mg/kg dose (maximum brain concentration value of 25 ng/g at 0.5 h) and in all the animals dosed at 10 and 30 mg/kg (maximum brain concentration values of 54 and 262 ng/g between 0.5 and 1.0 h, at the respective dose levels).

Brain levels were low in relation to the doses administered with mean penetration in terms of AUC_0–∞ always <4% of the equivalent systemic exposure. Terminal elimination half-lives at the two higher doses suggested that at these doses at least, the product was more persistent in the brain than in the plasma (3.1 and 2.2 h at 10 and 30 mg/kg, respectively). Unlike systemic exposure, the increase in brain exposure with increasing dose did not appear to be proportional in terms of AUC_0–∞, i.e., for a 10 times increase in dose between 3 and 30 mg/kg there was a 30 times increase in AUC_0–∞, although it should be noted that systemic levels were proportional in terms of Cp_max and AUC_0–∞ between these doses.

After oral administration at 3 mg/kg, very low plasma levels (consistently below the limit of accurate quantification) were detected, giving rise to a mean Cp_max of only 18 ng/ml at 1 h. Due to the low and variable drug levels only just above the calculated limit of detection of 14 ng/ml, an accurate determination of the AUC_0–∞ and terminal elimination half-life was not possible but rough estimates suggest that absolute bioavailability post oral dose is extremely low (ca. 1%).

In Vivo Metabolism

A limited investigation of the in vivo metabolism of RPR 119990 was undertaken after s.c. administration of the compound to mice at 3 mg/kg. Preliminary analysis of urine revealed the presence of RPR 119990, however, no metabolites could be detected. In addition, the fraction eliminated in the urine was determined to be 0.7, which appears to indicate that renal elimination is a major route of clearance for this compound and that metabolism may play a more minor role.

Transgenic Mouse Studies

Treatments with 1, 3, or 2 × 3 mg/kg RPR 119990 per day were well tolerated in mice.

Muscle Strength.

A group of 10 wild-type mice received treatment with RPR 119990 at 3 mg/kg/day s.c. up to 200 days of age; this treatment had no effect on muscle strength and no toxic symptoms or mortality was observed. A first experiment after the evolution of muscle strength throughout the animals' lives showed that wild-type mice showed a progressive increase in muscle strength up to 70 days of age (Fig. 6), whereupon the increase in muscle strength plateaued. Familial ALS transgenic mice reached a more rapid plateau with muscle strength of about 75% of the wild-type controls. Although muscle strength varied from time point to time point, it was only from 100 days of age that a consistent decrease in strength started to be noticed. Treatments with RPR 119990 at 1 mg/kg s.c. induced no significant effects but a dose of 3 mg/kg s.c. caused a significantly greater muscle strength from about 75 days of age, and this remained consistently better than for control animals up to 135 days of age (Fig. 6). After this time point deaths of the weaker mice reduced group size in the vehicle and 1-mg/kg-treated groups. All mice surviving up to 150 days of age were assessed for muscle strength and treatment at 3 and 2 × 3 mg/kg/day s.c. significantly increased muscle strength in these survivors (Table2).

Glutamate Uptake.

In a subgroup of animals (150 days of age), glutamate uptake was examined in synaptosomal preparations from spinal cords. K_m remained unchanged between wild-type, vehicle-treated, or compound-treated transgenic mice, indicating that there appeared to be no change in the affinity of the uptake site for glutamate (Table 2). However,V_max was significantly lower in the group of vehicle-treated transgenic ALS mice compared with wild-type controls (45% reduction, p < 0.001). When mice received chronically either 3 or (2 × 3) mg/kg RPR 119990 a day, this compound almost totally prevented the decrease inV_max in ALS mice compared with wild-type controls. A clear sparing of the number of uptake sites was observed in animals treated with the AMPA antagonist.

Life Expectancy.

Mean survival times show that treatment at 3 and 2 × 3 mg/kg RPR 119990 significantly increased survival, in a dose-dependent manner (Table 2), as is illustrated by the survival curve for these animals (Fig. 7).

Discussion

The novel AMPA antagonist RPR 119990 has a potent affinity for the rat AMPA receptor in membrane-binding studies that compares favorably with results for other AMPA receptor antagonists described in the literature (Takahashi et al., 1998; Turski et al., 1998). The compound was selective with respect to other ionotropic glutamate receptors, although a certain affinity for the kainate-binding site at around 50-fold higher concentrations was noted. The compound shows low or negligible affinity for 37 other binding or uptake sites, suggesting strong specificity of action. The activity on the closely related kainite site is expected, because cross-reactivity has already been reported (Bleakman and Lodge, 1998) and may account for some of the compound's anticonvulsant and neuroprotective actions.

RPR 119990 acts as a competitive antagonist at the recombinant human AMPA receptor/channel expressed in X. laevis oocytes. The compound shows a profile compatible with a competitive single site antagonism of the AMPA receptor with an equilibrium constant that is in the same order of magnitude to the apparent constant observed in the binding studies on rat brain membranes. This AMPA receptor antagonist is equally active on ion currents in rat cerebella granule neurons in culture, where it shows a potent and reversible antagonism of the ion flux elicited by a submaximal concentration of kainic acid, an agonist that generates nondesensitizing responses at AMPA receptors.

Studies in the more complex system of the rat hippocampal slice maintained in vitro confirm the ability of RPR 119990 to reduce the amplitude of field potentials, which are known to be evoked by endogenous glutamate release. Furthermore, a similar activity could be observed when the same area of the hippocampus was stimulated in vivo, after administration of RPR 119990 by subcutaneous route. This indicates that the compound is absorbed after parenteral administration and penetrates the blood-brain barrier in concentrations relevant to inhibit an AMPA receptor-mediated functional response within the brain of intact, anesthetized rats.

Likewise, RPR 119990 has potent anticonvulsant activity in the CD1 mouse, showing that the compound is also effective in the mouse and has central activity over a prolonged period after subcutaneous administration. PK studies suggest that the compound passes rapidly into the plasma after subcutaneous administration and, although passage into the healthy brain was low, its half-life in the brain was quite long. AMPA antagonist compounds appear frequently to show poor solubility, short half-lives, and poor in vivo availability. The present compound appears to have a more acceptable profile, making it suitable for pharmacological testing in disease models, in vivo. From these data the doses of 1, 3, and due to the compound's relatively short half-life, 2 × 3 mg/kg/day were chosen for testing in the transgenic mouse model of ALS.

Animals treated with RPR 119990 from day 50 of age up to their death showed significant improvements in muscle strength. Glutamate uptake transport is defective in 70% of human patients with the sporadic form of ALS, with a 30 to 95% loss of the excitatory amino acid transporter 2 astroglial glutamate transporter, giving a reduction in the maximal velocity of glutamate uptake transport in patient spinal cord (Rothstein et al., 1992), thus increasing glutamate in the synaptic cleft. A fall in V_max of the glutamate transporter has been observed in the spinal cord of familial ALS mice (Canton et al., 1997). Treatment of familial ALS mice with RPR 119990 protected glutamate uptake significantly at the end of their lives. Finally, treatment with RPR 119990 prolonged life expectancy significantly in familial ALS mice.

A number of possible modes of neuroprotective action may be relevant to the use of AMPA antagonists in ALS. Immunoglobulins from ALS patients have been shown to increase glutamate release from neurons in culture (Anjus et al., 1997), whereas cerebrospinal fluid samples from ALS patients have been shown to be toxic to neurons in primary cell culture and this toxic effect can be blocked by 6-cyano-2,3-dihydroxy-7-nitroquinoxaline, an AMPA antagonist (Couratier et al., 1993). These observations suggest that there may be an excitotoxic molecule acting on the AMPA receptor/channel. Moreover, the AMPA receptors of human spinal cord appear to have lower expression of the GluR2 subunit, making them permeable to calcium (Williams et al., 1997; Weiss and Sensi, 2000). This would be expected to increase calcium influx into cells for a given quantity of glutamate in the synaptic cleft. Inappropriate glutamate activity mediates neurotoxicity by raising intracellular calcium concentrations and saturating the mitochondria, which can trigger harmful enzyme and free radical cascades within neurons (Mayer et al., 1990). Lower motor neurons also have been shown to have reduced levels of calcium-buffering proteins (Alexianu et al., 1994). Free radicals may stimulate glutamate release while glutamate uptake sites are particularly sensitive to free radical damage.

The antiexcitotoxic neuroprotectant riluzole has demonstrated efficacy in prolonging life span in patients suffering from this devastating disease (Bensimon et al., 1994). In familial ALS mice, riluzole has already been shown to prolong survival in this mouse strain (Gurney et al., 1996, 1998). Riluzole acts on glutamate release, Ca²⁺ flux through the NMDA glutamate receptor, and inactivates sodium channels (Doble, 1997), so its modes of action do not overlap, but complement those of RPR 119990.

In conclusion, RPR 119990 is a member of a new generation of AMPA antagonists with high solubility and better durations of action than the quinoxalinedione series. Nevertheless, it is only active by parenteral routes, which would complicate daily administration in humans. It is to be hoped that compounds from future series may show oral activity and have potential interest in the treatment of ALS or other neurodegenerative diseases in which an excitotoxic component is probably involved.