Enhancement of Glutamate Uptake as Novel Antiseizure Approach: Preclinical Proof of Concept

Abstract

Objective

Excitotoxicity is a common hallmark of epilepsy and other neurological diseases associated with elevated extracellular glutamate levels. Thus, here, we studied the protective effects of (R)‐AS‐1, a positive allosteric modulator (PAM) of glutamate uptake in epilepsy models.

Methods

(R)‐AS‐1 was evaluated in a range of acute and chronic seizure models, while its adverse effect profile was assessed in a panel of standard tests in rodents. The effect of (R)‐AS‐1 on glutamate uptake was assessed in COS‐7 cells expressing the transporter. WAY 213613, a selective competitive EAAT2 inhibitor, was used to probe the reversal of the enhanced glutamate uptake in the same transporter expression system. Confocal microscopy and Western blotting analyses were used to study a potential influence of (R)‐AS‐1 on GLT‐1 expression in mice.

Results

(R)‐AS‐1 showed robust protection in a panel of animal models of seizures and epilepsy, including the maximal electroshock‐ and 6 Hz‐induced seizures, corneal kindling, mesial temporal lobe epilepsy, lamotrigine‐resistant amygdala kindling, as well as seizures induced by pilocarpine or Theiler's murine encephalomyelitis virus. Importantly, (R)‐AS‐1 displayed a favorable adverse effect profile in the rotarod, the minimal motor impairment, and the Irwin tests. (R)‐AS‐1 enhanced glutamate uptake in vitro and this effect was abolished by WAY 213613, while no influence on GLT‐1 expression in vivo was observed after repeated treatment.

Interpretation

Collectively, our results show that (R)‐AS‐1 has favorable tolerability and provides robust preclinical efficacy against seizures. Thus, allosteric enhancement of EAAT2 function could offer a novel therapeutic strategy for treatment of epilepsy and potentially other neurological disorders associated with glutamate excitotoxicity. ANN NEUROL 2025;97:344–357

Allosteric activation of glutamate uptake as novel antiseizure approach. [Color figure can be viewed at www.annalsofneurology.org]

Intense research efforts led to the successful development and commercialization of many antiseizure medications (ASMs) during the past few decades. ¹ Unfortunately, more than one‐third of people with epilepsy still experience highly debilitating, refractory seizures that do not respond to available therapies. These patients suffer from drug‐resistant epilepsy (DRE), ² which is etiologically complex and remains poorly understood. ³ Consequently, DRE represents a significant unmet medical need and has become an important research topic driving the development of novel ASMs.

Currently available ASMs work through mechanisms aiming to restore the disturbed excitatory/inhibitory neuronal balance, i.e., (1) modulation of voltage‐gated ion channels (Na⁺, Ca²⁺, or K⁺); (2) augmentation of GABAergic inhibition by acting directly on GABA_A receptors or modifying GABA synthesis, degradation or re‐uptake; (3) blockade of ionotropic glutamate receptors; and (4) inhibition of glutamate signaling. ¹ , ⁴ However, mechanisms related to the impaired glutamate uptake, a well‐established driver of excitotoxicity and seizure development, are not specifically targeted by any of available ASMs. ¹ , ⁵

Synaptic glutamate clearance is mainly (approximately 90%) mediated by the glutamate EAAT2 transporter, named GLT‐1 in rodents. ⁵ , ⁶ This transporter is responsible for maintaining physiological concentrations of glutamate in the synaptic cleft in order to avoid glutamate accumulation and subsequent excitotoxicity. ⁷ , ⁸ Dysfunction of EAAT2 activity and expression is a well‐described phenomenon in epilepsy ⁵ and has also been reported in a broad range of other neurological, neurodegenerative, and psychiatric diseases or conditions such as neuropathic pain, amyotrophic lateral sclerosis, Alzheimer's disease, Parkinson's disease, Huntington's disease, ischemia, schizophrenia, anxiety, depression, addiction, autism, as well as traumatic brain injury. ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ Consequently, EAAT2 is an attractive target for the discovery of new therapeutics for a wide range of neuropsychiatric indications. Compounds that are positive allosteric modulators (PAMs) of EAAT2 enhance extracellular glutamate uptake resulting in glutamate homeostasis normalization and thus could be developed as future therapeutics. Selective EAAT2 PAM might potentially provide a more robust and safer clinical profile ¹⁸ over conventional glutamate signaling inhibition approaches targeting glutamatergic receptors blockade. Nevertheless, it remains to be established whether this novel therapeutic strategy will be well tolerated and effective clinically.

Only a few experimental tool compounds directly enhancing the EAAT2 function have been identified so far. ¹⁹ , ²⁰ , ²¹ , ²² We have recently discovered a novel, first‐in‐class drug candidate, compound (R)‐AS‐1, that is selective PAM of EAAT2/GLT‐1. ²³ To the best of our knowledge, (R)‐AS‐1 is the first molecule with drug‐like properties, as well as robust antiseizure activity in standard models used for ASM discovery (eg, the maximal electroshock [MES], the 6 Hz [32 and 44 mA] and the subcutaneous pentylenetetrazole (scPTZ)‐induced seizures; Fig 1, green panel on the left). Furthermore, a relatively large therapeutic window, defined as the ratio between side‐effect inducing and therapeutic doses, clearly distinguishes (R)‐AS‐1 from a range of currently approved ASMs. ²³

FIGURE 1.

Chemical structure of (R)‐AS‐1 together with a summary of the previously published data (*Abram et al. ²³ ) and studies reported in the current manuscript.

The Epilepsy Therapy Screening Program (ETSP) of the National Institute of Neurological Diseases and Stroke (National Institutes of Health, Bethesda, USA), which has been successful in identifying many of the currently approved ASMs, has now implemented a number of novel animal models of DRE. ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ We thus subjected (R)‐AS‐1 to these additional models in which seizures are induced by electrical or chemical stimuli and by viral encephalitis for more in‐depth antiseizure potential assessment. Compared to the initial antiseizure activity studies described previously, ²³ the in vivo characterization was also extended herein to an additional rodent species (rats) and another mouse strain (both sexes). Furthermore, a more detailed side effect profile was evaluated in the rodent Irwin test. Thus, overall these additional data may provide a more comprehensive view on the therapeutic potential of (R)‐AS‐1. Noteworthy, for the first time (R)‐AS‐1 was also tested in status epilepticus model, which was induced by pilocarpine. We also performed in vitro glutamate uptake studies to confirm the mechanism of action of (R)‐AS‐1 alone or in combination with the specific competitive EAAT2 inhibitor WAY 213613. Finally, we have also confirmed that (R)‐AS‐1 does not induce changes in GLT‐1 expression, in contrast to previously described compounds (eg, amitriptyline or ceftriaxone). ¹⁰ , ²⁹ , ³⁰ , ³¹ Experimental data obtained with (R)‐AS‐1 previously together with the present results are summarized in Fig 1.

Materials and Methods

(R)‐AS‐1 Preparation

(R)‐N‐benzyl‐2‐(2,5‐dioxopyrrolidin‐1‐yl)propanamide, named as (R)‐AS‐1, has been synthesized in the Department of Medicinal Chemistry Jagiellonian University Medical College in Krakow (Poland), as previously reported. ²³ The purity was >99% determined by LC–MS method. During testing at the Epilepsy Therapy Screening Program (ETSP), (R)‐AS‐1 was administered as a suspension in 0.5% methylcellulose (MC) – formulation A or as solution in dimethyl sulfoxide (DMSO), PEG400, and water (10/40/50, v/v/v) – formulation B. In the remaining in vivo experiments, (R)‐AS‐1 was suspended in 1% solution of Tween 80 in water. Details about dose selection, formulation used, routes of administration, and treatment schedules are provided in the Data S1.

Study Approvals

ETSP procedures involving the use of animals were carried out according to the approvals received by the ETSP contract site units–Department of Pharmacology and Toxicology, University of Utah (Salt Lake City, UT, USA) and Synapcell (Saint Ismier, France). PILO SE and in vivo GLT‐1 qualitative expression protocols were performed at the Institute of Rural Health and approved by the Local Ethics Committee in Lublin, Poland (license no. 5/2021 and 99/2022, respectively). The GLT‐1 quantitative expression studies were approved by the Drexel University Institutional Animal Care and Use Committee (IACUC, license no. LA‐21‐607). Housing, handling, and testing were performed in agreement with the U.S. Public Health Service Policy on Humane Care and Use of Laboratory Animals and the European Community's Council Directive (2010/63/EU), where appropriate.

Evaluation of (R)‐AS‐1 by the ETSP

All seizure models as well as safety evaluation within the ETSP workflow were performed according to validated procedures. C57BL/6J mice and Sprague–Dawley rats were used in these studies. Within the identification phase of testing, (R)‐AS‐1 efficacy was evaluated in the 6 Hz (44 mA) test in mice, the MES test in rats, and in corneal kindled mice (CKM) model. In the differentiation phase of the ETSP workflow, (R)‐AS‐1 was tested in the intra‐hippocampal kainate model (IHKM) in mice, lamotrigine (LTG)‐resistant amygdala kindled (LTG‐K) rats, as well as in the Theiler's murine encephalomyelitis virus (TMEV)‐induced seizure model in mice. Additionally, the tolerability profile of (R)‐AS‐1 was assessed in the rotarod test, minimal motor impairment test, and Irwin test in mice and/or rats. A brief description of the assays used is provided below. The detailed protocols are available in the Data S1 (see section A) and at https://panache.ninds.nih.gov/Home/CurrentModels.

6 Hz (44 mA) Test in Mice

Seizures were induced by transcorneal stimulation (0.2 ms rectangular pulse at 6 Hz frequency for 3 s; current intensity of 44 mA). The 6 Hz‐induced seizures were characterized by stunned posture followed by jaw clonus, forelimb clonus, twitching of the vibrissae, and Straub tail. In an initial study, (R)‐AS‐1 (formulation B, 100 mg/kg, intraperitoneally [i.p.]) was tested in male mice at several time points to determine the time of peak effect (TPE). Next, quantification of the median effective dose (ED₅₀) (ie, the dose that confers protection in 50% of animals) was performed at the TPE of 1 h.

MES Test in Rats

Seizures were induced by transcorneal stimulation (sine‐wave pulses at 50 Hz frequency for 200 ms; current intensity of 150 mA). Tonic extension of the hindlimbs was taken as the endpoint. (R)‐AS‐1 was prepared in formulation A (for female testing) or formulation B (for male testing) and administered i.p. After initial qualitative screening for antiseizure activity, ED₅₀ values were determined at the TPE of 0.5 h.

CKM Model

Male and female mice were stimulated twice daily (3 s stimulation, 50 Hz, 1.5 mA) until at least 5 consecutive stage 5 seizures (according to the Racine scale ³² ) were reached. The ED₅₀ values were determined in fully kindled animals. (R)‐AS‐1 was prepared in formulation A (for female testing) or formulation B (for male testing) and administered i.p. at 0.25 and 1 h, respectively. Animals were reused multiple times for dose tests. A 3‐ to 4‐day washout period was permitted between each (R)‐AS‐1 dose.

IHKM in Mice

Male mice were stereotaxically injected with 1 nmol kainate into the right dorsal hippocampus and implanted with a bipolar electrode for EEG recording. During a 4‐week recovery period, animals developed hippocampal paroxysmal discharges (HPDs) defined as rhythmic high‐amplitude sharp waves with a frequency of 5–10 Hz of at least 5‐s duration, with a minimum interevent interval of 1 s. At the day of testing, (R)‐AS‐1 was injected i.p. at the dose of 50 mg/kg (formulation B). EEG recordings were performed in freely moving animals for 20 min before administration of (R)‐AS‐1 (baseline) and up to 90 min post‐injection.

LTG‐K Model in Rats

Male rats were stereotaxically implanted with a bipolar electrode into the right amygdala and, after recovery, kindled with a 200 μA, 50 Hz stimulus for 2 s until 4–5 consecutive stage 4 or 5 (according to the Racine scale ³² ) seizures were reached. Lamotrigine (5 mg/kg) was administered 1 h before each stimulation. The ED₅₀ value of (R)‐AS‐1 was determined in fully kindled animals. (R)‐AS‐1 (formulation B) was given i.p. 30 min before the test. Animals were reused multiple times for dose tests, with a 3‐ to 4‐day washout period between each dose testing.

TMEV‐Induced Seizures in Mice

After brief anesthesia, animals were injected intracortically with the Daniels strain of TMEV (3 × 10⁵ plaque‐forming units). In post‐inoculation days 3–7, mice received twice daily i.p. injection of (R)‐AS‐1 (formulation B, 100 mg/kg) or vehicle and were monitored for handling‐induced seizure. Seizure severity was scored according to the Racine scale. ³² The cumulative seizure burden, defined as the sum of all seizures, was used as the primary outcome in this study.

Behavioral Tolerability Screens

The rotarod test in mice (assessing an investigational compound's possible adverse effect on motor function and coordination) and minimal motor impairment (MMI) test in rats (indicated by ataxia, which is manifested as an abnormal, uncoordinated gait), were performed in conjunction with the 6 Hz and MES seizure tests. Quantification of the TD₅₀ doses (ie, doses that induce toxicity in 50% of animals treated with the compound) was carried out in the same manner as quantification of the ED₅₀ doses. In addition, the safety of (R)‐AS‐1 (formulation B, 100–300 mg/kg, i.p.) was assessed in the modified Irwin test.

Evaluation of (R)‐AS‐1 in the Pilocarpine Induced Status Epilepticus (PILO SE) Model

Experiments were performed in albino Swiss mice. To induce SE, animals received a single injection of PILO (300 mg/kg) and were subsequently monitored to assess seizure severity and mortality. (R)‐AS‐1 (15 mg/kg) was suspended in 1% Tween 80 and injected 30 min before PILO. Following SE induction (72 h), animals were perfused and brain samples were collected for immunohistochemical staining. More details about the animals and methods used are provided in the Data S1 (see section B).

In Vitro Evaluation of (R)‐AS‐1 Influence on Glutamate Uptake Mediated by EAAT2 in Combination with WAY 213613, a Selective Competitive EAAT2 Inhibitor

In the glutamate uptake studies, COS‐7 cells were transiently transfected with CMV vectors containing either no insert (background control) or the EAAT2 gene CMV‐hEAAT2. A glutamate uptake assay was performed using [³H]‐L‐glutamic acid according to procedure described by Fontana et al. ³³ with modifications. Transfected cells were incubated with (R)‐AS‐1 alone or with WAY 213613 at specified concentrations (ranging from 10⁻⁴ M to 10⁻¹² M for (R)‐AS‐1) or a combination of (R)‐AS‐1 (from 10⁻⁴ M to 10⁻¹² M) with 0.1 μM WAY 213613. For details, see Data S1 (section C).

In Vivo Evaluation of the Influence of (R)‐AS‐1 on GLT‐1 Expression

The effect of (R)‐AS‐1 on GLT‐1 expression was assessed by using immunohistochemical staining with confocal imaging (qualitative analysis) and Western blot (quantitative analysis), as described in detail in the Data S1 (see sections D and E, respectively). For qualitative expression analysis, C57BL/6J mice were treated with (R)‐AS‐1 (100 mg/kg) twice daily for 7 days and perfused 24 h after the last injection. GLT‐1 expression was assessed in dentate gyrus (DG) and CA1‐CA3 regions of the hippocampus. Quantitative assessment of GLT‐1 expression was carried out in cortical samples obtained from C57BL/6 mice after 7‐day (twice a day) treatment with (R)‐AS‐1 (100 mg/kg). In both in vivo studies, amitriptyline (10 mg/kg, once a day) was used as a positive control.

In addition, expression of GLT‐1 was analyzed by immunoblotting in 14 DIV neuron–glia cultures after 72 h treatment with (R)‐AS‐1 (100 nM and 100 μM). Ceftriaxone (500 μM and 1 mM) was used as a positive control in this study.

Statistics

Parametric data were analyzed using unpaired Student's t‐test or one‐way analysis of variance (ANOVA) test with Tukey's or Dunnett's post hoc test, where appropriate. Fisher's exact probability test was used for comparison of the categorical nonparametric data, whereas multiple Mann–Whitney test for analysis of the cumulative seizure burden across the time‐points. Nonparametric data from the glutamate uptake experiments were analyzed with the Kruskal‐Wallis test. Differences were considered statistically significant when the p value was less than 0.05. All statistical analyses were performed with GraphPad Prism (GraphPad Software, La Jolla, CA, USA).

Results

Evaluation of the Antiseizure Potential of (R)‐AS‐1 by ETSP

(R)‐AS‐1 showed potent protection in the 6 Hz (44 mA) seizure model at previously determined TPE of 1 h post i.p. injection in male mice (Table 1, for TPE evaluation see Table S1). Similarly, it also displayed strong antiseizure effects in both female and male rats in the MES test at the pre‐determined TPE = 0.5 h (i.p.) with lower median effective doses (ED₅₀) as compared to the 6 Hz test (Table 1; for screening data, see Table S2). (R)‐AS‐1 was also effective in the CKM model with similar potency in both male and female mice (Table 1). In this seizure model, the compound was tested at the TPE = 1 h in male mice (similar to 6 Hz [44 mA]) or TPE = 0.25 h in female mice (for time course of antiseizure activity in the CKM model, see Table S3). Furthermore, (R)‐AS‐1 in both genders significantly reduced severity of seizures (Racine's scale) in a dose dependent manner, starting from 15 mg/kg (male mice) and 30 mg/kg (female mice) (Table S4). Animals were seizure free at doses of 50 mg/kg (male mice) and 90 mg/kg (female mice). Additionally, at doses of 250 and 400 mg/kg, (R)‐AS‐1 revealed a robust protection in the MES test and was well tolerated in the MMI test when given orally to male rats (Table S5). In the IHKM of the mesial temporal lobe epilepsy (MTLE), (R)‐AS‐1 significantly decreased number of HPDs during the 20‐min period, that is, around the previously determined TPE in the 6 Hz 44 mA (see Fig 2A and associated Table S6). A total of 8 mice per treatment group were evaluated (50 mg/kg, i.p. at 1 h post‐drug administration) with the mean baseline HPD count determined prior to drug administration. Following administration of (R)‐AS‐1 HPD counts were reduced to 52.81 ± 13.45% when compared to baseline.

TABLE 1.

Antiseizure Activity and Safety Data for (R)‐AS‐1 After Single i.p. Administration in the 6 Hz, CKM, MES, and LTG‐K Models

Test a	Gender	Species (Strain)	PT for ED50 [h] b	ED50 (mg/kg)	PT for TD50 [h] a	TD50 (mg/kg) c
6 Hz (44 mA)	Male	Mice (C57BL/6J)	1.0	65.9 (45.2–93.7)	0.5	168.2 (117.4–242.0)
MES	Male	Rats (Sprague–Dawley)	0.5	34.9 (27.1–44.7)	0.5	253.8 (220.7–294.5)
MES	Female	Rats (Sprague–Dawley)	0.5	39.7 (30.6–50.6)	0.25	>500
CKM	Male	Mice (C57BL/6J)	1.0	38.2 (34.2–42.8)	0.5	168.2 (117.4–242.0)
CKM	Female	Mice (C57BL/6J)	0.25	40.5 (23.6–56.2)	ND	ND
LTG‐K	Male	Rats (Sprague–Dawley)	0.5	105.6 (65.7–173.8)	ND	ND

^a The following models were used: 6 Hz model (current intensity of 44 mA); MES, maximal electroshock seizure test; CKM, Corneal kindled mouse model; LTG‐K, lamotrigine‐resistant amygdala kindled model.

^b PT, pretreatment time. PTs have been determined based on screening data (for details see Data S1).

^c The median toxic (TD₅₀) doses were determined in the rotarod test for mice or in the minimal motor impairment (MMI) test in rats. ND, not determined. Values in parentheses are 95% confidence intervals.

FIGURE 2.

Efficacy of (R)‐AS‐1 in the IHKM and TMEV models in mice. (A) Mean number of spontaneous HPDs after single i.p. administration of (R)‐AS‐1 at dose of 50 mg/kg in the IHKM model. Data are presented as the mean ± SEM. Statistical analysis: paired t‐test; *p < 0.05 versus baseline. (B) Cumulative seizure burden analysis following inoculation with TMEV and treatment with (R)‐AS‐1. ( R)‐AS‐1 was administered at dose of 100 mg/kg i.p. BID for 5 days, starting 3 days post‐inoculation. Data are presented as the mean ± SEM. Statistical analysis: Multiple Mann–Whitney test; ****p < 0.001 versus control (vehicle).

The antiseizure efficacy of (R)‐AS‐1 was also evaluated in the LTG‐K rats. In this model of pharmacoresistant seizures, (R)‐AS‐1 at the doses of 35, 105, and 200 mg/kg reduced both the behavioral seizure score (Racine's scale) and seizure duration at 0.5 h post i.p. administration (see Table 1 for the ED₅₀ value and Table S7 for average seizure scores and seizure duration for each tested dose). Moreover, (R)‐AS‐1 was effective in the TMEV model of epilepsy in mice. As shown in Fig 2B, (R)‐AS‐1 (100 mg/kg, i.p., twice daily (BID) starting 3 days post‐TMEV inoculation) significantly reduced the average cumulative seizure burden at 3–7 days post infection as compared to the vehicle‐treated group (p < 0.0001). (R)‐AS‐1 also significantly decreased the proportion of mice with seizures (Table S8). For a distribution of seizure scores during each observation period, please see Table S9.

Safety and Tolerability Evaluation

The rotarod (mice) and the MMI (rats) data, which are summarized in Table 1, show satisfying separation between effective and toxic doses. Consequently, the safety margin, expressed as a protective index (PI = TD₅₀/ED₅₀), was 2.5 in mice 6 Hz (44 mA) seizure model, 4.4 in CKM (male mice), and 7.3 or >12.6 in MES test in male and female rats, respectively. In the Irwin test (male rats, i.p.), (R)‐AS‐1 was very well tolerated at doses of 100, 150, and 250 mg/kg at all‐time points. At the highest dose applied of 300 mg/kg, rats displayed anesthesia (50%, 2 rats out of 4 tested at 0.25 h time point), ataxia (50%, 2/4 at 0.5 h time point and 25%, 1/4 at 1 h time point), or staggering (25%, 1/4 at 0.5 h time point and 50%, 2/4 at 1 h time point). These effects were limited in time, and no adverse events were noted at later post‐administration time points (2, 4, and 24 h) at this dose. At high doses of 250 and 300 mg/kg (i.p.) and only 0.5 h after administration, (R)‐AS‐1 also induced ataxia in the MMI test in male rats (Table S2). Notably, this effect was almost not observed 0.25 and 1 h post injection (1/8), as well as at lower doses of 100 and 200 mg/kg (all time intervals). While no symptoms of motor impairment were observed in female rats up to the dose of 300 mg/kg (i.p.), (R)‐AS‐1 at a higher dose of 500 mg/kg caused slight ataxia (25%, 2/8 at 0.25 h and 12.5% at 0.5 and 1 h), as well as tremors (75%, 6/8) and abnormal gait (37.5%, 3/8), but only at 0.25 h time point, as indicated in Table S2. However, no notable signs of adverse behavioral tolerability or significant body weight changes were observed after BID administration of (R)‐AS‐1 at 100 mg/kg (i.p.) in the TMEV‐infected mice (see Table S10).

Evaluation of Antiseizure Properties of (R)‐AS‐1 in Pilocarpine Induced Status Epilepticus (PILO SE) Model in Mice

Single i.p. administration of (R)‐AS‐1 at 15 mg/kg (dose equal to ED₅₀ value in the 6 Hz [32 mA] seizure model ²³ ) nearly completely protected mice against PILO SE, which was observed in only 1 out of 8 mice tested (12.5%), whereas 7 mice out of 8 mice (87.5%) developed SE in the vehicle‐treated group (p < 0.05; Fig 3A). Furthermore, a considerable reduction in lethality was observed in PILO SE mice treated with (R)‐AS‐1 (12.5%) compared to PILO SE mice treated with vehicle (37.5%) (Fig 3A). Furthermore, a preliminary qualitative analysis of the Fluoro‐Jade B staining revealed a substantial neurodegeneration in the CA1 and CA2 hippocampal subregions in the PILO SE control mice (Fig 3B, a). In contrast, in 4 out of 5 randomly selected PILO SE mice treated with a single dose of (R)‐AS‐1 (15 mg/kg), no visible neuronal degeneration was observed, which also was consistent with the protection against SE (Fig 3B, b–f).

FIGURE 3.

Efficacy of (R)‐AS‐1 in the PILO SE in mice. (A) Effect of (R)‐AS‐1 (15 mg/kg, i.p., 30 min before PILO) on PILO‐induced seizures or lethality. Statistical analysis: Fisher's exact probability test; *p < 0.05 versus the vehicle + PILO (300 mg/kg) group (B) Fluoro Jade B (FJB, green) and 4′,6‐diamidino‐2‐phenylindole (DAPI, blue) staining for neurodegeneration in the hippocampus after PILO SE. (a) vehicle + PILO (300 mg/kg) treated mice, (b–f) (R)‐AS‐1 + PILO (300 mg/kg) ‐treated mice. DG = dentate gyrus; CA1, CA2, CA3 = cornu ammonis areas 1, 2, 3.

Evaluation of (R)‐AS‐1 Influence on Glutamate Uptake Mediated by EAAT2 in Combination with Selective Competitive EAAT2 Inhibitor WAY 213613

The selective inhibitor of EAAT2, WAY 213613, as expected blocked the glutamate uptake in a concentration‐dependent manner with the IC₅₀ value of 0.21 μM (Fig 4A). In contrast, (R)‐AS‐1 in a concentration‐dependent manner (EC₅₀ = 0.28 nM) effectively enhanced glutamate uptake by approximately 140% of control (Fig 4A). Importantly, an ineffective concentration of WAY 213613 (0.1 μM) completely attenuated the glutamate uptake enhancement when co‐administered with an active concentration of (R)‐AS‐1 (10 nM) (Fig 4B), and this effect remained unchanged with increasing concentrations of (R)‐AS‐1 (Fig 4A).

FIGURE 4.

Evaluation of the effect of (R)‐AS‐1 on glutamate uptake mediated by EAAT2 in the presence or absence of the specific competitive EAAT2 inhibitor WAY 213613. Uptake assays conducted on COS‐7 cells transiently transfected with CMV or EAAT2 were pre‐incubated for 10 min with either a vehicle, (R)‐AS‐1 alone at concentrations ranging from 10⁻⁴ M to 10⁻¹² M or a combination of (R)‐AS‐1 (at concentrations 10⁻⁴ M to 10⁻¹² M) with 0.1 μM WAY 213613. (A) Results were normalized to a percentage of control (EAAT2) and expressed as mean ± SEM from 3 to 6 independent experiments performed in technical triplicate. Statistical analysis: Kruskal‐Wallis test with post‐hoc Dunn's multiple comparisons test; ^# p < 0.05, (R)‐AS‐1 versus (R)‐AS‐1 + WAY 213613 (0.1 μM), ^#### p < 0.0001 (R)‐AS‐1 versus WAY213613; A one‐way ANOVA with post‐hoc Dunnett's multiple comparison test was used to analyze differences between various concentrations of (R)‐AS‐1 versus. control; *p < 0.05 (0.1 nM), ****p < 0.0001 (1 nM–100 μM). (B) Sample results from a single experiment expressed in DPM (disintegrations per minute) expressed as mean ± SEM for selected concentration of (R)‐AS‐1 (10 nM) alone and combination of (R)‐AS‐1 (10 nM) with WAY 213613 (0.1 μM). Statistical analysis: Kruskal‐Wallis test with post‐hoc Dunn's multiple comparisons test; **p < 0.01, EAAT2 versus (R)‐AS‐1 (10 nM) or t‐test; ^### p < 0.001, (R)‐AS‐1 (10 nM) versus (R)‐AS‐1 (10 nM) + WAY 213613 (0.1 μM); p > 0.05 (n.s. = not significant), EAAT2 versus WAY 213613.

Effect of (R)‐AS‐1 on GLT‐1 Expression in Neuron–Glia Cultures and after in Mouse Brain Repeated Administration

Previous studies indicate neuroprotective effects associated with increased expression of GLT‐1/EAAT2 induced by either β‐lactam antibiotics ¹¹ , ³⁴ , ³⁵ , ³⁶ , ³⁷ or viral gene delivery. ³⁸ We aimed to investigate whether exposure to (R)‐AS‐1 influenced the expression of GLT‐1 transporter. Neuron–glia cultures were treated for 72 h with either vehicle, 100 nM or 100 μM (R)‐AS‐1, or positive control ceftriaxone (500 μM or 1 mM), a known inducer of GLT‐1 expression. ³⁹ The cells were then harvested and processed for analysis of GLT‐1 expression via Western blotting. Fig 5A, B illustrates that incubation with 100 nM and 100 μM (R)‐AS‐1 did not alter the protein expression of GLT‐1, whereas treatment with 1 mM ceftriaxone resulted in significantly increased GLT‐1 expression. Furthermore, (R)‐AS‐1 (100 mg/kg) administered twice a day (BID) in mice, i.p. for 7 days did not alter GLT‐1 expression in the hippocampus, compared to vehicle‐treated mice, as revealed by qualitative confocal microscopy analysis (Fig 5C). In contrast, amitriptyline (positive control)‐treated mice (10 mg/kg, once a day [QD], i.p. for 7 days) showed a strong increase in GLT‐1 signal, as shown by the intense red color in Fig 5C. Consistently, quantitative Western blot analysis of prefrontal cortex mouse brain lysates also did not show any alteration of GLT‐1 protein expression after (R)‐AS‐1 treatment, while amitriptyline‐treated induced a significant increase in GLT‐1 expression (*p < 0.05) (Fig 5D, E).

FIGURE 5.

Effect of (R)‐AS‐1 on GLT‐1/EAAT2 expression. (A) Representative immunoblot of 14 DIV neuron–glia cultures stained with EAAT2 antibody (green) and β‐actin‐control (red), after 72 h treatments with vehicle (control), 100 nM and 100 μM of (R)‐AS‐1 and 500 μM and 1 mM of ceftriaxone (CEF) in neuron–glia cultures. Molecular weight ladder (kDa) on far‐left side. (B) Quantification of blot intensities of the EAAT2 monomers represented in A. Graphs are replicates of 3–4 independent groups of neuron–glia cultures combined. Data are presented as the mean ± SEM. Statistical analysis: One‐way ANOVA with Dunnett's post hoc test; **p < 0.01. (C) Qualitative assessment of GLT‐1 expression in the mice hippocampus: Control (vehicle); (R)‐AS‐1 (100 mg/kg, i.p., BID for 7 days); amitriptyline (AMI, 10 mg/kg, i.p., QD for 7 days), a positive control for GLT‐1 expression. Positive GLT‐1 signals are stained red; blue – cell nuclei. DG = dentate gyrus; CA1‐CA3 = cornu ammonis areas. (D) Western blotting for GLT‐1 expression in the mice cerebrum lysates (n = 6/group). Representative immunoblot illustrating EAAT2 (green bands) and β‐actin (red bands) expression in mice cortical samples after 7‐day BID (i.p.) administration of either vehicle, (R)‐AS‐1 (100 mg/kg, i.p.) or AMI (10 mg/kg, i.p.). (E) Quantification of immunoblots shows no significant difference in EAAT2 expression between vehicle and (R)‐AS‐1‐treated mice, and a significant difference between vehicle and AMI‐treated animals. Data are presented as the mean ± SEM. Statistical analysis: One‐way ANOVA with Dunnett's post hoc test; *p < 0.05. EAAT2 was normalized to β‐actin, and (R)‐AS‐1 and AMI were normalized to vehicle.

Discussion

The present study shows that EAAT2/GLT‐1 activation with a novel small molecule PAM compound, (R)‐AS‐1, leads to robust protection against seizures across a range of rodent models, including those characterized by a high degree of drug resistance. To the best of our knowledge, this is the first report providing in vivo proof of concept for a potential therapeutic value of EAAT2/GLT‐1 PAM drug candidate in epilepsy.

(R)‐AS‐1 displayed potent protection against seizures induced by both MES and 6 Hz stimulation irrespectively of rodent species, strain, or gender. The MES test is traditionally used as a model of generalized tonic–clonic seizures, whereas the mouse 6 Hz 44 mA test is considered a model of difficult‐to‐treat focal seizures. ²⁵ , ⁴⁰ In this study, (R)‐AS‐1 prevented the MES‐induced seizures in male and female rats and suppressed 6 Hz seizures at 44 mA in male C57BL/6J mice, which confirms its previously reported efficacy in these assays in CD‐1 mice. ²³ As shown in Tables S11 and S12, MES‐induced seizures are sensitive to numerous, but not all, ASMs with varying mechanisms of action. For example, levetiracetam and tiagabine were ineffective in this assay, whereas valproate a broad‐spectrum ASM did prevent MES‐induced seizures but with lower PI (protective index, ED₅₀/TD₅₀) values as compared to (R)‐AS‐1. In the 6 Hz test, several clinically relevant ASMs do not show protective activity (Tables S11 and S12). Moreover, some ASMs lose efficacy and/or PI value decreases when the stimulus intensity increases from 32 mA to 44 mA. This is particularly seen for levetiracetam. ⁴¹ , ⁴² Lacosamide, carbamazepine, valproate, and cannabidiol retained their activity at 44 mA with PI value between 1.4 and 1.8, whereas lamotrigine and ezogabine were active at doses inducing impairment of motor coordination. By contrast, cenobamate and tiagabine effectively suppressed 6 Hz seizures at 44 mA at doses devoid of behavioral toxicity (PI = 3.5 and 6, respectively). (R)‐AS‐1 does not lose its efficacy when the stimulus intensity increases from 32 to 44 mA, and importantly, shows activity at non‐toxic doses.

Next, we found that (R)‐AS‐1 dose‐dependently protects against seizures in the CKM model, which in contrast to the acute electrically‐evoked seizures in naïve animals, is characterized by chronically lowered seizure threshold, hyperexcitability of neuronal network, and neuroanatomical changes typical for human epilepsy. ⁴³ This model is helpful in identifying compounds may fail to demonstrate activity in the acute seizure assays, but still can be effective against focal seizures in humans. ⁴⁴ For example, levetiracetam shows efficacy in the CKM model, although it is ineffective in the MES and scPTZ test (Table S11). The CKM model has a good predictive validity since most of ASMs tested in this model were active at non‐toxic doses. ⁴³ However, when compared with lamotrigine and ezogabine, (R)‐AS‐1 displayed a better side effects profile (PI = 4.4 vs 1.0 and 1.5, respectively).

In the differentiation phase of the ETSP workflow for DRE, (R)‐AS‐1 was tested in models that reflect various clinical aspects of human epilepsy and thereby provide a more etiologically relevant approach to the development of new ASMs. In the IHKM model of MTLE, an initial brain insult induced by kainate results in non‐convulsive SE and subsequent spontaneous electrographic seizures. ²⁵ , ²⁸ , ⁴⁵ Acute treatment with (R)‐AS‐1 significantly decreased spontaneous HPDs to ~50% of the baseline levels, which further suggests its potential against difficult‐to‐treat focal seizures. The IHKM model is considered pharmacoresistant because only some ASMs effectively reduce HPDs at doses that do not cause motor impairment, and generally GABAergic drugs (eg, tiagabine) show the highest efficacy in this model (Table S13). ²⁴ , ²⁵ In contrast to (R)‐AS‐1, levetiracetam was effective in this assay at the dose that was not toxic, though it was ~50 times higher than doses active in the 6 Hz (32 mA) and CKM model.

In LTG‐K in rat, seizures are resistant to not only sodium channel‐blocking ASMs (eg, lacosamide) but also levetiracetam or cannabidiol (Table S12). ²⁶ , ²⁸ Importantly, (R)‐AS‐1 dose‐dependently supressed seizures in LTG‐K rats. This is a noteworthy finding as this model allows for identification of compounds effective against secondarily generalized focal seizures as well as for the differentiation of compounds that may show efficacy in DRE. ²⁶ , ²⁸ In fact, the only ASMs showing activity in this assay at doses not causing behavioral impairment, that is, with PI >1, were ezogabine, phenobarbital, tiagabine, and valproate.

Finally, (R)‐AS‐1 was tested in the mouse TMEV infection model of TLE, in which seizures arise due to viral encephalitis. ⁴⁴ This assay is used to screen compounds that may confer anti‐inflammatory and antiseizure activity. We note that (R)‐AS‐1 administered after TMEV inoculation significantly reduced the incidence of seizures and cumulative seizure burden during the acute infection period without any observable side effects. For comparison, cannabidiol (180 mg/kg) effectively reduced TMEV‐induced seizures, but simultaneously induced significant side effects (Table S14). ⁴⁶ Carbamazepine and lamotrigine were found to be ineffective in this model, whereas lacosamide, levetiracetam, tiagabine, and valproate significantly reduced seizure burden. None of the ASMs evaluated in the TMEV model increased the number of seizure‐free animals, ⁴⁷ which indicates superior activity of (R)‐AS‐1 as this compound significantly increased the number of animals without seizures (Table S14). This raises the possibility that suppression of the TMEV‐induced acute seizures may also prevent the development of spontaneous recurrent seizures, which occur weeks later in this model. A dedicated study investigating potential anti‐epileptogenic effects of (R)‐AS‐1 is warranted.

Additionally, (R)‐AS‐1 was tested in the PILO‐induced seizure model. PILO is commonly used in animal models to induce prolonged SE, which in turn leads to hippocampal injury, neuronal loss (especially in the DG and CA1–CA3 regions of the hippocampus), and circuit remodeling. A number of neurochemical and gene expression alterations, as well as significant change in glutamatergic neurotransmission, including glutamate release and uptake, have been reported in this model. ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵¹ , ⁵² We observed that treatment with (R)‐AS‐1 reduced neuronal loss in the hippocampus that is likely related to its strong antiseizure activity in this model, where neuronal loss is directly proportional to the duration of SE. ⁵³

To sum up, treatment with (R)‐AS‐1 demonstrated robust therapeutic efficacy in several epilepsy models, also those characterized by seizures that are difficult to control with current ASMs. This included seizures not only induced by acute electrical or chemical stimulation, but also seizures that arise after brain insults caused by viral infection or brain lesions related to focal injury. Tables S11–S14 summarize the potency and efficacy profile of (R)‐AS‐1 in comparison to several commonly used ASMs. Notably, (R)‐AS‐1's PI values were comparable or superior to the majority of the approved ASMs. However, it should be noted that such efficacy and tolerability comparisons between compounds tested in different models, genders, strains, and routes of administration are difficult to interpret without detailed pharmacokinetic data in each experimental condition. Furthermore, as recently discussed in the scientific literature, even highly differentiated preclinical efficacy profile does not always translate in differentiated clinical efficacy, particularly in patients with difficult to treat seizures. ⁵⁴ , ⁵⁵

(R)‐AS‐1 has a clear‐cut PAM effect on glutamate uptake in COS‐7 cells transfected with EAAT2. This observation is in line with the initially published results, ²³ and now can be contrasted with the opposite effect of a selective inhibitor of EAAT2, WAY 213613, validating the experimental glutamate uptake model. Even very low concentrations (nM) of (R)‐AS‐1 exert the activating effect on EAAT2 function, and this effect reaches about 140% of control. Importantly, an ineffective concentration of WAY 213613 attenuates the PAM effect of (R)‐AS‐1 further confirming the specificity of the EAAT2 interaction. It is worth reminding that (R)‐AS‐1 is highly selective for EAAT2 and does not interact with EAAT1 and EAAT3, as previously reported. ²³ Structural biology modeling studies suggest that (R)‐AS‐1 interacts with EAAT2 at an allosteric site, ²³ that is different from the orthosteric site postulated for WAY 213613, which acts as a competitive inhibitor. ⁵⁶ Nevertheless, more detailed structural biology and cystalography studies are needed to fully elucidate the exact binding mode of (R)‐AS‐1 to EAAT2.

It was also important to rule out a potential effect of (R)‐AS‐1 on the transporter expression, since such observations were reported for several other compounds. ³¹ , ⁵⁷ Qualitative confocal image analysis in mice showed that (R)‐AS‐1 did not affect the expression of GLT‐1 compared to control, whereas a clear enhancement of the GLT‐1 signal was observed after amitriptyline treatment, a known activator of GLT‐1 expression. ³¹ These results were confirmed additionally by quantitative Western blot assessment of GLT‐1 expression in neuronal‐glia primary cultures and mouse cerebral cortex. In contrast, ceftriaxone, a reference transcriptional up‐regulator of GLT‐1⁵⁷ also increased the transporter expression in the present study. Overall, these results suggest that the antiseizure efficacy of (R)‐AS‐1 is likely driven by the enhancement of glutamate uptake due to direct interaction with the EAAT2/GLT‐1 transporter and not by increasing its expression.

Compounds positively modulating glutamate uptake by EAAT2/GLT‐1 reported thus far can be divided into 2 main classes, ¹⁰ namely (1) transcriptional or translational enhancers (eg, amitriptyline, ³¹ riluzole, ³⁰ ceftriaxone ⁵⁸ ) or (2) allosteric modulators directly interacting with the transporter (eg, (R)‐AS‐1, ²³ GT949, ²⁰ , ²² GT951, ¹⁹ etc.). Due to these mechanistic differences, such compounds are likely to have different safety and tolerability profiles. For example, increased expression of EAAT2/GLT‐1 could have more broad impact on glutamatergic neurotransmission throughout the brain, irrespectively of the transporter activity, which can lead to adverse events. ⁵⁸ Additionally, transcriptional/translational modulators may potentially have non‐specific off‐target effect on other genes/proteins, which is often the case for such classes of compounds. ¹⁰ Finally, the pharmacokinetic/pharmacodynamic (PK/PD) effects of these compounds might be difficult to predict due to differences in the duration of transcription/translation processes or protein turnover, which might lead to a delayed therapeutic effect. ¹⁰ In contrast, positive allosteric EAAT2/GLT‐1 modulators, such as (R)‐AS‐1 reported herein, have fast onset of action with more predictable PK/PD and selectivity profiles, which may provide therapeutic advantages versus EAAT2/GLT‐1 expression enhancers. ⁵⁹ , ⁶⁰ However, we do recognize that EAAT2 PAMs might have some safety liabilities and might not be fully devoid of side effects due to interference with glutamate signaling, an essential excitatory neurotransmission system in the central nervous system, that is involved in learning and memory, ⁶¹ mood and anxiety. ⁶² Therefore, future studies are necessary to assess the potential effects of (R)‐AS‐1 in models of cognition, drug abuse, anxiety, and depression.

Conclusions

The data reported herein strongly highlight a potentially broad therapeutic utility of (R)‐AS‐1, a PAM of EAAT2 in seizure therapy. Dynamic changes in expression and function of this glutamate transporter during epileptogenesis, ⁵ indicate that compounds of this class may also have disease‐modifying or antiepileptogenic potential. Furthermore, many neurological and psychiatric diseases share common mechanisms related to abnormal excitation/inhibition balance, ⁶³ , ⁶⁴ , ⁶⁵ where EAAT2 activation may play a potential therapeutic role. In short, our current results and the availability of this mechanistically novel drug‐like molecule could open several therapeutic avenues for the treatment of a broad range of neuropsychiatric conditions.

Author Contributions

K.K., M.A.M., and A.C.K.F. contributed to the conception and design of the study; K.K., K.S., K.L.R., R.T., M.Z., M.K., A.F.G., M.M., A.S., M.A.M., M.Q.H., A.C.K.F., A.R., and R.M.K. contributed to the acquisition and analysis of data; K.K., K.S., M.A., M.J., K.L.R., R.T., M.Z., A.F.G., M.A.M., M.Q.H., A.C.K.F., A.R., and R.M.K. contributed to drafting the text or preparing the figures.

Potential Conflicts of Interest

Prof. Krzysztof Kamiński is the CSO in iQure Pharma, Princeton, NJ, US. The company is developing (R)‐AS‐1, currently in IND enabling studies.

Supporting information

Data S1. Supporting Information.

Data Availability

All data generated in this study are provided within the article and the Supporting Information file.