Ethosuximide reduces epileptogenesis and behavioral comorbidity in the GAERS model of genetic generalized epilepsy


  • Gabi Dezsi,

    1. Department of Medicine (Royal Melbourne Hospital), Royal Melbourne Hospital, Melbourne Brain Centre, University of Melbourne, Parkville, Victoria, Australia
    Search for more papers by this author
  • Ezgi Ozturk,

    1. Department of Medicine (Royal Melbourne Hospital), Royal Melbourne Hospital, Melbourne Brain Centre, University of Melbourne, Parkville, Victoria, Australia
    Search for more papers by this author
  • Davor Stanic,

    1. Florey Neurosciences Institutes, Parkville, Victoria, Australia
    Search for more papers by this author
  • Kim L. Powell,

    1. Department of Medicine (Royal Melbourne Hospital), Royal Melbourne Hospital, Melbourne Brain Centre, University of Melbourne, Parkville, Victoria, Australia
    Search for more papers by this author
  • Hal Blumenfeld,

    1. Departments of Neurology, Neurobiology and Neurosurgery, Yale University School of Medicine, New Haven, Connecticut, U.S.A
    Search for more papers by this author
  • Terence J. O'Brien,

    1. Department of Medicine (Royal Melbourne Hospital), Royal Melbourne Hospital, Melbourne Brain Centre, University of Melbourne, Parkville, Victoria, Australia
    Search for more papers by this author
  • Nigel C. Jones

    Corresponding author
    • Department of Medicine (Royal Melbourne Hospital), Royal Melbourne Hospital, Melbourne Brain Centre, University of Melbourne, Parkville, Victoria, Australia
    Search for more papers by this author

Address correspondence to Nigel C. Jones, Department of Medicine, Royal Melbourne Hospital, Melbourne Brain Centre, University of Melbourne, 4th Floor Clinical Sciences Building, Royal Parade, Parkville, Vic., Australia, 3052. E-mail:



Ethosuximide (ESX) is a drug of choice for the symptomatic treatment of absence seizures. Chronic treatment with ESX has been reported to have disease-modifying antiepileptogenic activity in the WAG/Rij rat model of genetic generalized epilepsy (GGE) with absence seizures. Here we examined whether chronic treatment with ESX (1) possesses antiepileptogenic effects in the genetic absence epilepsy rats from Strasbourg (GAERS) model of GGE, (2) is associated with a mitigation of behavioral comorbidities, and (3) influences gene expression in the somatosensory cortex region where seizures are thought to originate.


GAERS and nonepileptic control (NEC) rats were chronically treated with ESX (in drinking water) or control (tap water) from 3 to 22 weeks of age. Subsequently, all animals received tap water only for another 12 weeks to assess enduring effects of treatment. Seizure frequency and anxiety-like behaviors were serially assessed throughout the experimental paradigm. Treatment effects on the expression of key components of the epigenetic molecular machinery, the DNA methyltransferase enzymes, were assessed using quantitative polymerase chain reaction (qPCR).

Key Findings

ESX treatment significantly reduced seizures in GAERS during the treatment phase, and this effect was maintained during the 12-week posttreatment phase (p < 0.05). Furthermore, the anxiety-like behaviors present in GAERS were reduced by ESX treatment (p < 0.05). Molecular analysis revealed that ESX treatment was associated with increased expression of DNA methyltransferase enzyme messenger RNA (mRNA) in cortex.


Chronic ESX treatment has disease-modifying effects in the GAERS model of GGE, with antiepileptogenic effects against absence seizures and mitigation of behavioral comorbidities. The cellular mechanism for these effects may involve epigenetic modifications.

The epilepsies are a group of debilitating and progressive neurologic conditions affecting about 1% of the population (Hauser et al., 1993), and are characterized by spontaneous recurrent seizures. All current medical therapies for epilepsy symptomatically suppress seizure activity, but they are not disease-modifying, having no effect on the underlying propensity of the brain to generate seizures. Current treatments also have no significant effects on the psychiatric comorbidities associated with epilepsy, such as anxiety, depression, and psychosis, which are common in patients and significantly add to the disability burden (Tellez-Zenteno et al., 2007; Hermann et al., 2008; Vega et al., 2011). A major goal of translational epilepsy research is to identify treatments that are not merely symptomatic, but truly disease-modifying (Galanopoulou et al., 2012). There have been several recent reports of experimental successes using therapies that interfere with epileptogenesis (Blumenfeld et al., 2008; Russo et al., 2010; Wong, 2010; McClelland et al., 2011), suggesting that this goal is achievable.

Ethosuximide (ESX) is a first-line clinical symptomatic treatment for absence seizures. Recent research has demonstrated that chronic treatment with ESX, when initiated prior to the onset of the epilepsy, has antiepileptogenic effects in the WAG/Rij rat model of genetic generalized epilepsy (GGE) with absence seizures (Blumenfeld et al., 2008; Russo et al., 2010; Sarkisova et al., 2010; Russo et al., 2011). This treatment regimen also reduces the depression-like behaviors that are present in WAG/Rij rats (Sarkisova et al., 2010). Without corroboration of the antiepileptogenic potential of ESX in other models, it is unclear whether this property is limited to an isolated cause of GGE (thereby limiting the translatability of this finding), or whether it is effective in other GGE models, thereby establishing a broad applicability of this drug across animal models, and potentially extending to human cases. In this study, we investigated whether chronic treatment with ESX has disease-modifying effects against epilepsy and behavioral comorbidities in a different model of GGE with absence seizures, Genetic absence epilepsy rats from Strasbourg (GAERS).

GAERS are a well-validated polygenic model of GGE (Danober et al., 1998), with a genetic causation that is different from that of WAG/Rij rats (Gauguier et al., 2004; Rudolf et al., 2004; Powell et al., 2009). The epilepsy develops in GAERS at around 8–9 weeks of age in our colony (Jones et al., 2008), manifesting as bilateral spike-and-wave discharges on electroencephalography (EEG), which are responsive to drugs used clinically to treated absence seizures, including ESX (Tringham et al., 2012). In addition, we have shown previously that GAERS from our colonies in Melbourne exhibit an anxiety-like behavioral phenotype using the elevated plus maze and open field tests of anxiety (Jones et al., 2008). Because anxiety disorders are prevalent in pediatric generalized epilepsy patients (Caplan et al., 1998; Ott et al., 2001, 2003; Caplan et al., 2005; Jones et al., 2007; Caplan et al., 2008; Vega et al., 2011), this high-anxiety phenotype makes GAERS an appropriate model to study the influence of disease-modifying drugs on behavioral outcomes (Jones & O'Brien, 2012).

We also investigated whether ESX treatment alters expression of key components of the epigenetic molecular machinery, which may implicate an epigenetic mechanism in any long-term effects on the epilepsy and behavioral phenotype. Specifically we examined the expression levels of the DNA methyltransferase (DNMT) family of enzymes in the somatosensory cortex—a region shown previously to play a key role in spike-wave seizure expression in rodent models (Meeren et al., 2002; Polack et al., 2007; David et al., 2008; Zheng et al., 2012) following both chronic and acute ESX treatment. DNMTs catalyze the covalent attachment of methyl (CH3) groups to specific cytosine residues in the DNA sequence—an epigenetic modification termed DNA methylation, which is strongly associated with transcriptional repression of genes (Attwood et al., 2002). Any change in DNMT enzyme expression would be expected to induce enduring alterations of the expression patterns of key genes, an effect that may influence epileptogenesis.



This study used male genetic absence epilepsy rats from Strasbourg (GAERS; n = 16) and nonepileptic control rats (NEC; n = 16) bred in the Department of Zoology, University of Melbourne Biological Research Facility (BRF). These animals were derived from the original AERS/NEC colony from Strasbourg in 2007, and inbred at the University of Melbourne for seven generations prior to use in this study. Rats were weaned and transferred at 3 weeks of age to the Department of Medicine (Royal Melbourne Hospital), University of Melbourne BRF, where they were housed until the completion of the study. Both facilities were kept on a 12 h light/dark cycle, with lights on at 7 a.m. at a constant temperature of 22°C. For chronic treatment experiments, animals were group-housed up until the surgical procedure, and from then on, they were housed individually. For acute treatment studies, animals were group-housed until 14 weeks of age, when the treatments and postmortem experiments took place. All procedures were approved by the University of Melbourne Animal Ethics Committee.

Drug regimens

For the chronic treatment studies, wherever possible, animals from a single litter were divided and allocated as either ESX-treated or control (tap water)–treated animals in a paired fashion. Ethosuximide syrup was purchased from the Royal Melbourne Hospital pharmacy (Zarontin, Pfizer Pharmaceuticals Group, New York, NY, U.S.A.). At 3 weeks of age, ESX treatment was initiated in GAERS (n = 5) and NEC (n = 7) rats, in an attempt to achieve a palatable dose of 300 mg/kg/day in the drinking water. Volumes consumed and animal weights were measured daily for the first week, and then weekly thereafter, and the drug dosage received (mg/kg/day) was calculated. The concentration of drug in the water bottles was updated weekly to maintain the appropriate dosage in an iterative fashion. All animals were able to drink independently when the study began. Control-treated rats received tap water ad libitum for the duration of the study. Due to the light-sensitive nature of ESX, drinking bottles were wrapped with black tape. At 22 weeks, the drug treatment ceased, and all animals reverted to tap water (see Fig. 1A for study timeline) until 34 weeks of age.

Figure 1.

Study timeline and ethosuximide dosing. (A) Littermates were separated to receive ESX or control treatment at 3 weeks of age, which persisted until 22 weeks. EEG recordings were made at 20 and 34 weeks of age, as well as 1 day prior to, and following, treatment cessation in GAERS. Behavioral assessments were conducted at 7 and 34 weeks in all rats. (B) Over the course of the study, the concentration of ESX in the drinking water presented to GAERS and NEC rats was altered depending on their drinking habits. This stabilized for each rat at around 10 weeks of age, and thereafter, all animals consistently received 300 mg/kg/day. (C) GAERS weigh significantly less than NEC rats across the study, and ESX treatment also results in lower weights. (D) Fluid intake was equal between GAERS and NEC, but reduced in ESX-treated rats. Once treatment was removed (i.e., 22 weeks), drinking volumes returned to control levels, and actually rebounded above control levels in GAERS.

For the acute treatment studies, GAERS were administered ESX (Sigma, St Louis, U.S.A.: 200 mg/kg i.p., n = 5) or vehicle (0.9% saline, n = 5), and culled 90 min later, a time when the acute seizure-suppressing effects of ESX are present (Tringham et al., 2012). By comparing acute and chronic effects of drug treatment, we could discern whether any observed genomic effects were related to acute antiseizure effects of ESX, or whether they were specifically associated with chronic treatment, and therefore more closely linked to any disease-modifying effects.

Surgery and EEG

At 11 weeks, animals underwent surgical implantation of EEG recording electrodes, as described previously (Hakami et al., 2009), with minor modifications. Briefly, under isoflurane anaesthesia, a midline incision was made on the scalp and the connective tissue removed. Six small burr holes were made in the skull, and electrodes were implanted (Plastics One, Bioscientific, Australia) to facilitate recording of the EEG. The electrodes were held in place with dental cement (Henry Schein Halas, Melbourne, Australia) and applied to the skull, and animals were left to recover for 1 week in individual cages.

To assess the effects of ESX treatment on seizures, electrical cables (Plastics One, Bioscientific) were connected to the implanted electrodes, and the EEG recorded continuously for 24 h, beginning at midday, using COMPUMEDICS software (Melbourne, Australia). These recordings occurred four times: at week 20, 1 day prior to drug cessation, 1 day following drug cessation, and then again at week 34. Recordings were performed in the home cages in a quiet room under standard lighting conditions of the BRF. Analysis of the seizure frequency was performed by a reviewer blinded to treatment using Spike-and-Wave seizure detection program (Electronic Research Group, Radboud University, Nijmegen, The Netherlands) over the entire 24 h recording period. The Spike-and-Wave detection program program automatically identifies spike-wave episodes, with each episode then visually verified. Inclusion criteria for seizures was a Spike-Wave discharge of amplitude more than two times baseline, a frequency of 7 to 12 Hz, and a duration of longer than 1 sec (see Fig. 2A).

Figure 2.

Ethosuximide reduces seizures during the treatment period, and this effect endures following treatment cessation. (A) An example of a recorded SWD from the EEG. GAERS exposed to ESX (n = 5) show significantly reduced % time in seizure (B) and the number of seizures recorded (C) during the 24 h recording period both during and after the treatment period, compared to control-treated GAERS (n = 6). No differences were observed in seizure duration throughout the study period (D). *Indicates significant (p < 0.05) post hoc comparison with control-treated GAERS. Data represent group mean + SEM.

Anxiety-like behavior

To determine the effects of ESX treatment on behavioral comorbidity, at 7 and 34 weeks of age, all rats underwent assessment of anxiety-like behavior in an open field. This was performed in a closed, quiet, light-controlled room in the Behavioural Testing Facility at the Department of Medicine, Royal Melbourne Hospital, University of Melbourne. The testing protocol was as previously described (Jones et al., 2008). Briefly, animals were individually placed in the center of a 1 m diameter circular arena with an inner circle (diameter 66 cm) separating the inner area from the outer, and allowed to explore the arena for 10 min. The lighting level within the maze was set at approximately 90 lux. Quantification of the total distance travelled, and the time spent and number of entries made into the inner area of the maze was objectively assessed using ETHOVISION TRACKING Software (Noldus, Wageningen, The Netherlands). Our previous data have demonstrated robust differences in open field behavior between GAERS and NEC, particularly the extent of exploration of the environment (Jones et al., 2008, 2010; Bouilleret et al., 2009).

Postmortem analysis

At the conclusion of the chronic study, animals were cardiac-perfused with 4% paraformaldehyde, and the brains extracted and sectioned on a cryostat at 40-μm thickness. The somatosensory cortex region was microdissected from five to six sections, and RNA was extracted using the RNeasy FFPE kit (Qiagen, Melbourne, Australia). One microgram of RNA was reverse transcribed to cDNA using the Omniscript Reverse Transcription kit (Qiagen, Australia) in the presence of an RNase inhibitor. Taqman quantitative polymerase chain reaction (PCR) was performed on 50-ng complementary DNA (cDNA) (Powell et al., 2008) using custom-designed taqman gene expression assays (Applied Biosystems, Melbourne, Australia) for Dnmt1 (Assay ID Rn00709664_m1), Dnmt3a (Assay ID Rn01469994_g1), Dnmt3b (Assay ID Rn01536414_g1). Relative quantification of messenger RNA (mRNA) levels were normalized to the reference genes, GAPDH (Assay ID Rn99999916_s1) using the ΔΔCT method (Schmittgen & Livak, 2008).

For the acute study, rats were culled 90 min after ESX or vehicle treatment, and the somatosensory cortex region dissected and frozen on dry ice. RNA was extracted from these samples using RNeasy Mini Kit (Qiagen), and gene expression analysis performed as above.

Statistical comparisons

All EEG parameters were compared using two-way ANOVA with repeated measures, separating the periods during and following the treatment period into different analyses. The between-subject factor was treatment (either ESX or control), and the within-subject factor was age of recording. In addition, we performed ANOVA to assess whether EEG parameters were altered following removal of treatment, using the same parameters as above. Bonferroni's post hoc analysis was used when appropriate. Behavior was assessed at 7 and 34 weeks separately, and used two-way ANOVA. Molecular data were analyzed using two-way ANOVA, or Student's t-tests. All comparisons were conducted using GRAPHPAD PRISM v5.04 (La Jolla, CA, U.S.A.).


Drug dosage and animal weights

Through management of the concentration of ESX in the drinking bottles, we were able to achieve stable doses of 300 mg/kg/day for chronically treated animals (Fig. 1B). Over the course of the study, GAERS weighed significantly less than NEC rats (F3,22 = 80.04; p < 0.0001: Fig. 1C), and weights were also significantly reduced by ESX treatment (F3,22 = 4.94; p = 0.039). This drug effect may be partially explained by the slight, but significant, reduction in fluid intake of ESX-treated rats (F3,22 = 14.79; p = 0.001: Fig. 1D). After cessation of drug treatment, fluid consumption returned to control levels, and actually increased in treated GAERS to well above control. Over the treatment period, fluid intake did not differ between GAERS and NEC rats (F3,22 = 0.002; p = 0.96).

Ethosuximide limits disease severity in GAERS

As expected, the percentage of time spent in seizure was significantly reduced during ESX treatment, compared to water-treated GAERS (F1,11 = 21.70; p = 0.0012). Following cessation of ESX treatment, this seizure-suppressing effect was maintained for the 3-month follow-up period (F1,11 = 11.92; p = 0.0072: Fig. 2B), indicative of a disease-modifying effect of the ESX treatment. Likewise, the number of seizures experienced was significantly reduced in treated animals both during treatment (F1,11 = 31.13; p = 0.0003), and this persisted after treatment had stopped (F1,11 = 5.36; p = 0.046: Fig. 2C). When examining treated GAERS, it should also be noted that, following cessation of treatment, both the percentage time in seizure (F1,11 = 9.92; p = 0.013) and seizure frequency (F1,11 =32.20; p < 0.001) increased compared to during the treatment period. However, these outcomes still remained significantly lower than control-treated rats. No significant differences were observed in seizure duration either during (F1,11 = 2.99; p = 0.117) or after (F1,11 = 2.95; p = 0.120) ESX treatment (Fig. 2D).

Chronic ethosuximide treatment alleviates behavioral disturbance in GAERS

At 7 weeks, the distance traveled in the open field was significantly reduced in GAERS compared to NEC rats (F3,22 = 25.61; p < 0.001; Fig. 3), but this measure was not significantly influenced by ESX treatment (F3,22 = 1.96; p = 0.178). At 34 weeks, when all GAERS are exhibiting seizures, a different profile was observed: the significant effect of strain on behavior was maintained (F3,22 = 4.57; p = 0.046), but we now also observed a significant effect of treatment (F3,22 = 12.08; p = 0.003), and a significant strain vs. treatment interaction (F3,22 = 5.125; p = 0.036), such that ESX treatment increased the extent of exploration of the open field but only in GAERS (Fig. 3C).

Figure 3.

Suppression of seizures is accompanied by improved behavioral outcome in GAERS. Representative traces of the path traveled by an NEC rat (A) and a GAERS (B) in the open field— the black square represents the trial start position. Note the reduced total distance and entry into the central area of the maze of the highly anxious GAERS. At 7 weeks, prior to the onset of seizures, GAERS displayed an anxiogenic phenotype compared with NEC rats, as evidenced by (C) significantly reduced distance travelled in the open field and (D) significantly reduced entries into the central area, and this phenotype was not influenced by ESX treatment. At 34 weeks, however, this anxiogenic phenotype was ameliorated in ESX-treated GAERS, compared to control-treated GAERS. *Indicates significant (p < 0.05) post hoc comparison with relevant NEC group. #Indicates significant (p < 0.05) post hoc comparison with control-treated GAERS. Data represent group mean + SEM. Sample sizes: GAERS treated with ESX – n = 5, with water – n = 6; NEC treated with ESX – n = 7, with water – n = 4.

We observed the same trends when assessing the number of times the animals ventured into the inner area of the open field. At 7 weeks, number of entries was significantly reduced in GAERS compared to NEC rats (F3,22 = 17.7; p < 0.001; Fig. 3D), but not affected by ESX (F3,22 = 0.02; p = 0.896). At 34 weeks, however, ESX treatment significantly influenced the number of entries into the central area (F3,22 = 8.95; p = 0.0078); but a nonsignificant effect of strain for this measure (F3,22 = 2.57; p = 0.126). It is important to note that we also observed a significant interaction between these variables at this age (F3,22 = 4.41; p = 0.05).

Ethosuximide differentially alters DNA methyltransferase gene expression

To investigate potential molecular mediators of the enduring effects of ESX, we examined the gene expression profiles of the DNA methyltransferase family of enzymes—DNMT1, DNMT3A, and DNMT3B—in the somatosensory cortex of treated and untreated rats (Fig. 4). Expression of these enzymes is strongly associated with gene transcriptional repression (Attwood et al., 2002), which we hypothesized as being related to the enduring effects of ESX. Using quantitative PCR (qPCR), we found that DNMT1 and DNMT3A mRNA expression were significantly elevated in GAERS that were chronically treated with ESX (Fig. 4A,B). Altered expression of these genes, specifically in treated epileptic animals, would suggest a possible role in suppressing epileptogenesis. ANOVA demonstrated increased DNMT1 expression in GAERS compared to NEC rats (F3,22 = 8.131; p = 0.011), and in ESX vs. water-treated rats (F3,22 = 17.99; p = 0.0006), but without an interaction (F3,22 = 1.568; p = 0.23). In the case of DNMT3A mRNA expression, significant strain (F3,22 = 6.199; p = 0.023) and treatment (F3,22 = 6.978; p = 0.017) effects were also observed, but this time a significant interaction between groups was observed (F3,21 = 5.271; p = 0.035), with ESX-treated GAERS showing significantly higher expression levels than the other three groups (p < 0.05).

Figure 4.

DNA methyltransferase (DNMT) mRNA expression is differentially altered in GAERS. Expression of DNMT1 (A) and DNMT3A (B) are elevated in chronically treated ESX-treated GAERS, whereas DNMT3B is elevated in GAERS irrespective of drug treatment (C). No gene expression changes were induced by acute treatment with ESX (D) *Indicates significant (p < 0.05) post hoc comparison with relevant NEC group. &Indicates significant (p < 0.05) post hoc comparison with relevant control-treated group. Data represent group mean + SEM. Sample sizes as for Fig. 3.

Effects of ESX on DNMT1 and DNMT3A expression were not simply a nonspecific drug response, since ESX did not change expression of these enzymes in NEC animals (Fig. 4A,B). In addition, ESX did not affect all methyltransferase genes in GAERS: DNMT3B expression demonstrated a different profile (Fig. 4C), with GAERS displaying significantly greater levels of gene expression (F3,22 = 112.0; p < 0.0001). In contrast to the other family members, DNMT3B mRNA expression was not significantly altered by ESX (F3,22 = 1.852; p = 0.191).

To determine whether the effects on DNMT gene expression were related to the antiepileptogenic effects of ESX, or to a nonspecific effect of the drug, we next treated separate adult male GAERS with an acute injection of ESX or vehicle at a dose known to suppress seizures (Tringham et al., 2012). We found no changes in gene expression of any of the DNMT family of genes between treated and untreated rats (DNMT1: t8 = 0.45; p = 0.66, DNMT3A: t8 = 0.96; p = 0.36, DNMT3B: t8 = 0.49; p = 0.64, Fig. 4D), ruling out acute exposure to ESX as a driving factor.


Here we demonstrate that chronic treatment with ESX induces a disease-modifying effect in the GAERS rat model of GGE. This extends upon previous research demonstrating an antiepileptogenic action of chronic ESX treatment in WAG/Rij rats (Blumenfeld et al., 2008; Russo et al., 2010, 2011; Sarkisova et al., 2010), a model of GGE with different genetic causation (Gauguier et al., 2004; Rudolf et al., 2004; Powell et al., 2009), suggesting that this effect is broadly applicable across absence epilepsy models. In addition, we showed that chronic suppression of seizures mitigates the anxiety-like behavioral phenotype in GAERS, an effect that presented itself only after the onset of seizures.

GAERS from our Melbourne colonies exhibit a highly anxious phenotype, which is present prior to the onset of the epilepsy (Jones et al., 2008). Our previous studies characterizing this phenotype employed GAERS/NEC rats originally obtained from Hull, United Kingdom, in 2002 (Jones et al., 2008, 2010; Bouilleret et al., 2009), and contrasted from a much earlier report on the original Strasbourg colony, which did not find an hyperanxious phenotype (Vergnes et al., 1991). It is important to note that we now replicate this phenotype using GAERS/NEC rats from our inbred colonies sourced from the founder colony from Strasbourg in 2007, providing broad validation of the association of this behavioral trait with different GAERS colonies. Exposure to a novel open field is a widely used measure of anxiety-like behavior that demonstrates predictive validity (Prut & Belzung, 2003), and this apparatus gives strong and reproducible data when testing GAERS. However, caution must be taken when using only a single test for behavioral outcomes, and future studies should use batteries of assays to comprehensively characterize anxiety-like phenotypes in this and other rat strains.

An intriguing aspect of the disease-modifying effects of ESX treatment observed here was the mitigation of the anxiety-like behavioral comorbidity. Because the GAERS strain was originally bred for the epilepsy phenotype, the emergence of the behavioral abnormality is likely to be linked in some way to the epilepsy, even though it appears to manifest before seizures begin. Clinically, this ontogeny has also been reported, with pediatric patients with GGE frequently experiencing anxiety disorders prior to the first recognized seizure (Jones et al., 2007). Here we show that the behavioral deficit is significantly attenuated by ESX treatment, but only in chronically epileptic animals, suggesting an indirect effect of the treatment on behavior that is dependent on the presence of seizures. This finding also agrees with a previous report describing amelioration of depression-like behavior during chronic ESX treatment in the WAG/Rij model (Sarkisova et al., 2010), and further enhances the postulate of bidirectionality between epilepsy and psychiatric comorbidities (Kanner, 2011). Here, we did not examine any measure of depression-like behaviors, so we are unable to comment on whether depression phenotypes are similarly improved by ESX in GAERS.

An intriguing question posed by the effects of ESX treatment relates to the mechanism by which it induces disease-modifying effects. ESX is widely believed to act as an antagonist at low threshold T-type calcium channels, and there is evidence from pilocarpine-induced status epilepticus studies that T-type calcium channels play a role in limbic-acquired epileptogenesis (Becker et al., 2008), suggesting involvement of these ion channels. However, recent studies suggest that ESX may not be working as a pure antagonist at T-type channels (Goren & Onat, 2007), so the role of T-type channels in these effects is not clear. It is also feasible that the effect of ESX represents the inhibition of a “kindling-like” phenomenon, whereby a cycle of seizures begetting seizures is interrupted by the ESX treatment. Supporting evidence comes from observations that chronic treatment with other antiabsence drugs (e.g., levetiracetam and zonisamide) induce similar sustained effects on epilepsy development in WAG/Rij rats (Russo et al., 2011). However, chronic treatment with carbamazepine, which is recognized to acutely increase seizures in models of absence epilepsy (Liu et al., 2006), did not induce the anticipated sustained increase in seizure activity (Russo et al., 2011).

It is recognized that many changes in gene expression occur in GAERS over the course of disease development (Jones et al., 2011), potentially representing causal or contributory drivers of the epileptic and behavioral phenotypes. Epigenetic mechanisms represent methods of controlling gene expression, and they are typically mediated by changes in the structure of chromatin and other DNA binding proteins that modify the accessibility of transcription factors to their binding domains (Borrelli et al., 2008). Here we demonstrate that chronic ESX treatment in GAERS results in alterations in the expression levels of the DNMT enzymes that catalyze DNA methylation, the most widely studied epigenetic modification. Indeed, such alterations in DNA methylation enzyme expression may also be relevant to other types of epilepsy, since a recent report identified increases in DNMT1 and DNMT3a in brain samples from human patients with temporal lobe epilepsy (Zhu et al., 2012). The expression changes identified here would be expected to modify the DNA methylation landscape and influence expression of many downstream genes, some of which may be relevant to the long-term disease-modifying effects of ESX. The attraction of this explanation lies in the prevailing assumption that DNA methylation is a stable and enduring mark, and therefore able to continue to mediate an ongoing gene expression profile long after the intervention has ceased (in this case, ESX treatment). DNMT1 acts as the maintenance methyltransferase during DNA replication, adding methyl groups onto the new DNA strand in appropriate positions (Leonhardt et al., 1992). In addition to its role in cell division, it is highly expressed in postmitotic cells (Veldic et al., 2004), suggesting it also plays a role in regulating DNA methylation patterns in mature neurons, which may be relevant to neurologic function and dysfunction. DNMT3A and DNMT3B are de novo methyltransferases, acting during development to dramatically alter the DNA methylation landscape (Okano et al., 1998), and DNMT3A has also been shown to affect neuronal function in adulthood (LaPlant et al., 2010). We found a significant gene versus drug interaction for DNMT3A expression, such that this enzyme was upregulated in GAERS only if the animals received ESX treatment. A similar trend, but without a statistical interaction, was observed for DNMT1. These effects did not appear to be acute effects of ESX, since a single drug injection did not affect DNMT expression. This is intriguing, and it suggests that these changes may act to alter DNA methylation patterns to mediate the effects of ESX on disease development and severity. It may also explain why the ESX had no effect on anxiety-like behavior in the NEC rats. Elevated levels of DNMT3B in both treated and untreated GAERS compared to NEC rats indicates that, although this enzyme may play roles in any DNA methylation differences between the strains, it is unlikely to be a mediator of any long-term beneficial effects of ESX. Further studies are required to demonstrate how ESX treatment might influence DNMT expression in GAERS, whether the changes in DNMT levels are relevant to the effects of ESX, and by extension whether this biologic process can be targeted to induce disease-modifying effects in epilepsy (Qureshi & Mehler, 2010).

Another point of interest raised from this study is whether there is a critical developmental window where treatment must be initiated for effects on disease progression to be realized. The ESX treatment in this study, and the previous studies in WAG/Rij rats, was started in young rats (3 weeks of age) prior to the onset of spontaneous recurrent seizures. This is relevant to any potential translation of this research to humans, where in most circumstances patients present to the clinic after their first seizure. An important next step will be to examine whether chronic ESX treatment initiated after the onset of epilepsy also produces disease-modifying and anxiolytic effects observed in this study. Overall, the current work supports an increasing literature base suggesting that targeting epileptogenic pathways is a feasible therapeutic strategy, and perhaps brings closer the prospect of identifying the holy grail of epilepsy research: a cure.


This work was supported by an NHMRC project grant to NJ (#566544) and an NHMRC CDA Fellowship to NJ (#628466), and by NIH R01 NS049307 and R01 NS066974 to HB.


None of the authors has any conflict of interest, financial or otherwise, to disclose. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.