Metyrapone abolishes spike–wave discharge seizures in genetic absence epilepsy rats from Strasbourg by reducing stress hormones

Stress is one of the most commonly reported triggers for seizures in patients with epilepsy, although the mechanisms that mediate this effect are not established. The clinical evidence supporting this is derived from patients' subjective experience of stress, and how this influences their own seizures. Animal models can be used to explore this phenomenon in controlled environments, free from subjective bias. Here, we used genetic absence epilepsy rats from Strasbourg (GAERS), a genetic rat model of absence epilepsy, to explore the influence of stress and stress hormones on spontaneous seizures.


| INTRODUCTION
Stress is the most frequently reported trigger for seizures in patients with epilepsy. A variety of different reports, usually acquired from patient questionnaires concerning seizure triggers, have documented that 20%-64% of patients believe stress to be the most common trigger for their seizures. [1][2][3][4][5][6] This does not appear to be limited to particular forms of epilepsy, being reported equally by both patients with focal and with generalized epilepsy. 1 Recently, there have been attempts to therapeutically exploit this by introducing behavioral interventions and stress-reduction strategies as a way of reducing seizure occurrence by limiting the triggers. 7 The biological mechanisms that underlie these apparent seizure-promoting effects of stress are unknown, but elevations in stress hormones are prime candidates to mediate this (reviewed in for example Maguire and Salpekar, 8 Gunn and Baram 9 ).
The hypothalamic-pituitary-adrenal (HPA) axis is the primary hormonal stress system in mammals. 10 Activation of this axis results in increased synthesis and sequential release of corticotrophin-releasing hormone (CRH), adrenocorticotropic hormone (ACTH), and cortisol (or the equivalent corticosterone in rodents). These hormones have each been implicated in the pathophysiology of seizures and epilepsy, primarily acting as powerful proconvulsant agents. [11][12][13][14][15] This suggests that a potential therapeutic strategy may revolve around inhibition of stress steroids, perhaps as a "precision medicine" approach for those patients who experience stress-induced seizures. Corticosterone is synthesized in the adrenal glands by conversion from 11-deoxycorticosterone (DOC), a reaction catalyzed by the enzyme 11-βhydroxylase. This hormone is the final end product of the HPA axis, and plays a variety of physiological roles, including in glucose metabolism, blood pressure, and inflammation, but is also important in psychological responses to stress. After release into the circulation in times of stress, it acts on steroid receptors in specific brain regions such as hippocampus to suppress the HPA axis in a negative feedback role. In certain disease states, such as major depressive disorder and Cushing syndrome, excessive production of cortisol is common, 16 which may contribute to disease severity and progression. Metyrapone, an inhibitor of 11-βhydroxylase, 14 is used clinically to prevent excessive cortisol synthesis in diseases associated with elevated cortisol levels such as Cushing syndrome, and has also been trialed as a therapy in depression. 17,18 To our knowledge, the effects of metyrapone in patients with epilepsy has not been investigated, but experimentally, the drug has been shown to reduce the incidence of acute seizures following hypoxia-ischemia injury, 19 reduce seizure-induced brain damage, 20,21 and reverse the early life stress-induced reduction in seizure threshold in the amygdala kindling model. 14 Together, this evidence suggests that metyrapone may be a useful therapy against epileptic seizures.
Genetic absence epilepsy rats from Strasbourg (GAERS) are a well-validated and widely used rat model of genetic generalized epilepsy. 22,23 Adult rats exhibit spike-wave discharge (SWD)-like events that morphologically appear similar to absence seizures in patients with childhood absence epilepsy, and their pharmacological profile matches clinically used medications. [24][25][26] In addition, they also exhibit marked anxiety-and depressionlike behaviors, [27][28][29] which are commonly observed comorbidities in patients with epilepsy. 30 Because of the regular incidence of seizures in GAERS, occurring up to once per minute, this model provides an appropriate epilepsy model to examine how environmental exposures influence spontaneous seizure activity. As such, here we investigated the interactions between stress, stress hormones, and seizures in GAERS.

| Subjects
We performed a series of studies using GAERS and nonepileptic control (NEC) rats. 22 Male adult GAERS (n = 38) and NEC rats (n = 4) were bred and housed at the Department of Medicine, Royal Melbourne Hospital, University of Melbourne Biological Research Facility on a 12-h light/dark cycle (lights on at 6 a.m.), with food and water available ad libitum. All experiments were conducted in our laboratories in the Department of Medicine, Royal Melbourne Hospital, University of Melbourne, and approved by the Florey Neuroscience Institute Animal Ethics Committee (#14-027).

| Stress responsivity
First, we assessed the corticosterone stress response in GAERS and NEC rats (n = 4/group). We have previously

Key Points
• GAERS, a model of absence epilepsy, exhibit elevated corticosterone • Acute stress increases seizures in GAERS • Exogenous corticosterone similarly increases seizures • Inhibition of corticosterone synthesis with metyrapone prevents seizures described this procedure. 31,32 Briefly, at 11 a.m., rats were placed in Perspex Broome rodent restraint tubes (SDR Clinical Technology), with the tail exposed at one end. A small nick was made at the tail tip, and .1 mL blood was collected in a heparinized syringe and expunged (without the needle) into an Eppendorf tube and stored on ice. The restraint tube was then fastened and held in position for 30 min, at which time another blood sample was taken.
Blood was centrifuged at 3000 × g for 10 min at 4°C, and the plasma isolated and stored until assayed at −20°C.
To assess serum corticosterone levels, we used a radioimmunoassay (MP Biomedicals), performing the experiment as per the manufacturer's instructions. All samples were processed in duplicate.

| Electrode implantation surgery
GAERS underwent surgical implantation of cortical recording electrodes to facilitate visualization of electroencephalogram (EEG), as previously described. 29,33 Briefly, rats were anesthetized with isoflurane, placed in a stereotaxic frame, and the scalp shaven and skull exposed. Six burr holes were drilled through the skull, taking care not to breach the dura. These holes were located at ±2 mm anterior/posterior, ± 2 mm lateral to bregma, and ± 2 mm lateral, 2 mm anterior to lambda. Stainless steel screw electrodes (Plastics One, Bioscientific) were then gently screwed into these holes, and the electrodes connected to a pedestal (Plastics One). The posterior electrodes acted as the ground and reference electrodes. The pedestal was secured to the animal's head using dental cement (Henry Schein Halas), and the animals were allowed to recover for at least 1 week prior to any further experimentation. Animals were isolated after surgery to avoid experimental loss.

| Stress and seizure studies
Next, we investigated whether acute stressors influenced seizure incidence in the GAERS model. After animals had recovered from surgery, GAERS (n = 8) were connected to EEG recording cables (Plastics One) via the implanted pedestal. EEG was acquired using Profusion P3 software (Compumedics) unfiltered and digitized at 256 Hz. Animals were connected in their home cages for 1 week continuously to avoid excessive handling associated with connection of cables, and all recordings done in the home cage. After 48 h, the first stressor occurred: noise stress. Commencing at 11 a.m., a radio playing loud rap music was switched on in the home room, while the EEG of the rats was being recorded. This method of inducing stress was previously described as part of a multimodal stress paradigm, 34 which we adapted here as a unimodal stressor. The noise levels of this music ranged from 57 to 89 dB, averaging 82 dB, as measured using the Sound Meter 2.15 phone app. After 30 min, this was turned off, and the EEG continued to record. Only rats under experimentation were in the room at this time. Forty-eight hours later, the second stressor was initiated: cage tilt stress. 35 This also commenced at 11 a.m., and consisted of putting chocks of wood under the side of each cage to raise them to be angled at 45°. Thirty minutes later, the chocks were removed, and EEG continued to record. After completion of these studies, one clean EEG channel was exported and analyzed for seizures using automated seizure detection software (SpikeWave Finder). 36 We analyzed three 4-h sections of EEG from each animal from 10 a.m. to 2 p.m. on the day before stress (sham), the day of noise stress, and the day of cage tilt stress. Exported traces were coded before review, so the reviewer was blinded to the treatment occurring that day. Data were separated into 30-min epochs for presentation and analysis purposes.

| Pharmacological studies
We next conducted a series of acute pharmacological challenge studies (n = 7-9 animals per study), all of which followed the same general protocol, as described previously. 37 GAERS were connected to EEG cables for at least 60 min prior to injection of drug, to record a stable baseline of seizure occurrence. Animals were then injected with drug, and the EEG recording persisted for a further 2 h. Separate cohorts of rats were used for each drug study, and drug doses within each study were delivered in a randomized fashion, with at least 48 h separating treatment days. Studies were always done in the daytime, with injections occurring between 10 a.m. and 1 p.m. The drugs used were: corticosterone, purchased from Wako Pure Chemical Industries (0-30 mg/kg dissolved in 50% ethanol and saline); metyrapone, purchased from Sigma-Aldrich (0-100 mg/kg dissolved in 20% dimethylsulfoxide [DMSO] and saline), and DOC purchased from Sigma-Aldrich (0-30 mg/kg dissolved in βcyclodextrin in 80% Na 2 HPO 4 and 20% DMSO). All drugs were delivered intraperitoneally. Seizures were analyzed as described above, with the reviewer blinded to the treatment each animal received until the analyses were complete.

| Statistical analysis
Corticosterone response to stress was assessed using two-way analysis of variance (ANOVA) with repeated measures. When assessing effects of stress intervention or drug over time, we used two-way ANOVA with repeated measures. Effects of drug on seizure incidence, average seizure duration, and percentage of time spent in seizure following injection were compared between drug doses using one-way ANOVA with repeated measures. Effects of drug on the latency to seizure, and on the baseline seizure incidence across time, were conducted using a paired Student t-test. Bonferroni post hoc test was incorporated where appropriate. Data are presented as mean ± SEM, with statistical significance considered when p < .05. All statistical analyses were conducted using GraphPad Prism (v9).

| GAERS exhibit elevated corticosterone levels
We first tested the HPA axis response to acute stress in GAERS and NEC rats. Rats received 30-min restraint stress, and the effect on blood corticosterone levels was measured. As anticipated, restraint stress significantly elevated corticosterone levels (F = 11.09, p = .005; Figure 1). In addition, we found that GAERS showed elevated levels of corticosterone compared to NEC, both at baseline and after stress (F = 5.96, p = .028), but there was no significant interaction identified between strain and stress (p = .7). Therefore, GAERS show elevated corticosterone levels at baseline, but relatively normal HPA axis responsivity to this stressor.

| Acute stress promotes seizures in GAERS
After establishing marked elevations in baseline stress hormones in GAERS, we next sought to determine whether acute stress could influence seizure incidence. We applied two sequential stressors to GAERS: noise stress and cage tilt stress. We found that both stressors significantly elevated seizures. Specifically, we found a significant time × stress interaction following noise stress (F = 2.268, p = .023; Figure 2A), such that seizures increased during the 30-min stressor (p = .13), and in the subsequent 30-min time block (p < .001). Similarly, the total time spent in seizure was significantly increased in these two time slots (p < .05; Figure 2B) before returning to baseline levels. However, the average duration of seizure events was not influenced throughout the stress period (F = .05, p = .82; Figure 2C). Similar effects were observed as a consequence of cage tilt stress, with significant interactions between stress and time for seizure incidence (F = 3.474, p = .001; Figure 2D), and time in seizure (F = 2.835, p = .005; Figure 2E). However, in this case, we saw bimodal effects of stress; during the stress period, seizures were significantly elevated (p = .018), whereas 60-90 min after the cessation of stress, seizure incidence significantly reduced (p < .05). Again, we saw no effect of stress on seizure duration (F = .07, p = .90; Figure 2F).

| Exogenous corticosterone increases seizures
One potential mediator for the effects of stress on seizures in GAERS is stress-induced release of corticosterone.
To determine the effect of raised stress hormone levels, we treated GAERS with exogenous corticosterone. We found that injection of corticosterone dose-dependently increased seizures (F = 4.384, p = .047; Figure 3A), with a dose of 30 mg/kg reaching significance after post hoc testing. Corticosterone administration also significantly increased the time spent in seizure (F = 5.854, p = .0102; Figure 3B), but no effect on seizure duration was observed (F = .938, p = .42; Figure 3C).

| Inhibition of corticosterone synthesis abolishes seizures
After establishing that stress and exogenous corticosterone administration elevate seizures, and noting the elevated levels of baseline endogenous corticosterone in GAERS, we next investigated whether inhibition of corticosterone synthesis with metyrapone has the opposite effect. Injection of metyrapone led to a dramatic dose-dependent reduction in seizure incidence (F = 26.24, p < .0001; Figure 4A) and  Figure 4B). We also found a significant reduction in seizure duration after 100 mg/kg metyrapone (F = 5.351, p = .0185; Figure 4C). To further examine the efficiency of high-dose metyrapone to inhibit seizures, we calculated the latency to recording the first event after injection, finding a highly significant increase in this measure, compared to vehicle (t = 8.0, p < .0001; Figure 4D). We also tracked the reemergence of seizures after metyrapone injection. Seizures started to emerge approximately 40 min after metyrapone injection, but stayed suppressed for the entire recording period of 2 h, significantly different from vehicle treatment (F = 45.66, p < .0001; Figure 4E). Finally, we assessed whether the effects of 100 mg/kg metyrapone persisted until the next testing day. No statistical differences were identified when comparing the baseline seizure incidence occurring immediately prior to metyrapone injection to the baseline seizure incidence of the next test day (baseline seizures premetyrapone = 40.3 ± 3.5 per hour; baseline seizures next test day = 41.7 ± 5.2 per hour; t = .3, p = .80). This suggests that the effects of drug had dissipated over this period.

| DOC, a corticosterone precursor, increases seizures
Finally, we investigated whether a precursor of corticosterone, the neurosteroid DOC, might influence seizures. Although injection of the DOC, which is the direct precursor of corticosterone, did not significantly influence seizure occurrence (F = 1.423, p = .275; Figure 5A) we did observe a slight, but significant, elevation in time spent in seizure (F = 3.56, p = .039; Figure 5B), but no post hoc significance was observed. Seizure duration was not impacted by DOC treatment (F = 2.496, p = .113; Figure 5C).

| DISCUSSION
Here we investigated the influence of stress on seizures in a well-validated rodent model of absence epilepsy, the GAERS model. 23 We found that GAERS exhibit heightened HPA axis function, indicated by elevated circulating baseline corticosterone levels, and it is possible that this is related to the chronically elevated anxiety levels observed Hashed areas indicate the periods of stress. *p < .05, ***p < .001 indicate significant differences at the indicated time point between stress and sham conditions. n = 8/group. in this strain. 27,29,36,38 We also show that elevated corticosterone is related to seizure expression; exogenous administration modestly increases seizure incidence, whereas inhibition of corticosterone synthesis by metyrapone completely suppresses seizures. None of the rats had seizures in the first 20 min after metyrapone injection, in contrast to vehicle-treated rats, which started having seizures again almost immediately. We therefore present this as a model that can be used to explore the interactions between stress, stress hormones, and seizures with relevance to patient populations, and promote metyrapone as a potential new therapeutic agent to be used in the treatment of seizures.
There is substantial literature documenting the increased prevalence of seizures following stress in humans. [1][2][3][4][5][6]39 For example, one study included 71 subjects with epilepsy who completed prospective diaries each evening, reporting the perceived likelihood of experiencing a seizure the next day. They also recorded all seizures, and possible triggers, such as stress levels, sleep, alcohol use, and anxiety. The authors found that stress and anxiety measures were significantly associated with an increased risk of experiencing a seizure the next day. These studies are somewhat limited, due to the subjective nature of stress perception, and because of ethical challenges associated with inducing stress, it is difficult to demonstrate causation. However, these clinical studies are supported by a large literature in animals, studies that can overcome these limitations. The findings from animal studies are quite varied, with several studies showing increased seizure risk/severity, 40,41 or alternatively antiseizure effects of different stressors. [42][43][44] Often, these studies use acute models of convulsive seizures, as opposed to chronic models of epilepsy; acute models have the advantage of being quick and well controlled, but do not accurately portray the structural/functional abnormalities of an epileptic brain. In addition, little is known about how stress influences absence seizures, although clinically, there do not appear to be marked differences in effects of stress on different epilepsy types. 1,5 One relevant study looked at seizures in the WAG/Rij model of absence epilepsy, and showed that following a foot shock, seizures are initially suppressed and then increased. 45 The effects on seizures were magnified upon repeated exposure to the foot shock, suggesting a delayed, proconvulsant effect of the stress. Another study examined the incidence of SWD in mice harboring a mutation in the Scn8a sodium channel gene. Restraint stress for 20 min increased SWD seizures immediately following the stressor, and peaked 1 h after stress. 40 These published reports agree with the current study. Here, we chose to use two different complementary stressors-tilting the home cages to a 45° angle 35 and playing loud rap music 34 -to gauge whether these influenced seizure incidence (restraint stress was not possible for this outcome, because the restraint tube prevented connection of the EEG cable). If we identified similar and consistent effects of these stressors, we could conclude that such effects are globally relevant to stress, but if discrepancies were observed, then any unique effects of one stressor may be idiosyncratic to F I G U R E 3 Corticosterone increases seizures in genetic absence epilepsy rats from Strasbourg. (A) Seizures were dosedependently increased by exogenous corticosterone injection.
(B, C) The percentage of time spent in seizure was also increased by corticosterone (B), but no effect was observed on seizure duration (C), compared to vehicle control. *p < .05, **p < .01 indicate significant difference compared to vehicle treatment. n = 9/group. that particular stressor. Both stressors elevate seizure incidence, effects that persisted for at least the duration of the stressor. Interestingly, following conclusion of the cage tilt stress, we observed a pronounced seizure-suppressive effect that lasted for ~90 min. This was not expected, was not observed with the noise stress, and indicates that the effects of stress are not constrained to the period of the stress itself. Further comparisons between these stressors could be made, for example, mapping the time course and magnitude of the corticosterone response to each stressor, to then relate to changes in seizure incidence. This was not feasible in our study, because the handling associated with blood sampling would have interfered with natural occurrence of seizures. However, it should be noted that a previous study showed similar corticosterone profiles in mice following restraint and noise stressors. 34 It is possible that the mechanisms of how stressors might promote seizures involve excessive release of stress hormones. These hormones, including those involved in the HPA axis circuit, such as CRH, ACTH, and corticosterone, have been implicated in seizures and epilepsy, primarily due to their potent actions in hippocampus. 46 In our study, exogenous corticosterone increases seizure incidence in GAERS, suggesting that this might be a mediator of the effects of stress on seizures. This agrees with a previous study whereby epileptic mice exhibited more epileptiform bursts during chronic exposure to corticosterone. 47 Corticosterone acts at two membrane-bound steroid receptors: glucocorticoid and mineralocorticoid receptors. 48 The latter is a high-affinity receptor, occupied at resting levels of corticosterone, whereas the glucocorticoid receptor is only activated in times of elevated corticosterone levels, such as during stress. Typically, receptor activation results in translocation to the nucleus, and initiation of gene transcription, but nongenomic actions of corticosterone have also been identified. For example, corticosterone enhances the frequency of postsynaptic potentials and glutamate release, which seems to be mediated by a membrane-bound mineralocorticoid receptor. 49 Given the electrophysiological nature of seizures, and the abrupt effects of stress on seizures, it is plausible that the corticosterone is acting through this membrane-bound receptor to increase seizure occurrence, although further experiments would be required to demonstrate this. At this stage, we cannot rule out a genomic role for the effects of corticosterone, which seem particularly pertinent to the delayed suppression of seizures by cage tilt stress occurring 1 h after stress cessation. In addition, most literature has focused on glucocorticoid actions in limbic structures, such as hippocampus. Absence seizures in GAERS oscillate between thalamus and primary somatosensory cortex, so it is possible that these structures may mediate When comparing effects on latency to first seizure after injection, 100 mg/kg metyrapone significantly delayed this measure, compared to vehicle. (E) When looking at the time course of events after injection, 100 mg/kg metyrapone completely abolished seizures in all rats for >30 min after injection, and seizure incidence remained low for the duration of the recording session. *p < .05, **p < .01, ***p < .001, ****p < .0001 indicate significant differences compared to vehicle treatment. n = 8/group. the effects of corticosterone. However, no study to date examines expression of steroid receptors in these regions in GAERS and NEC rats, or their functionality. One limitation of our approach using systemically delivered drugs is that the site of action within the brain cannot be identified; future studies could identify this using intracerebral infusions of drug to cortical and thalamic regions.
We also found an elevation in baseline corticosterone levels in GAERS, and this agrees with one previous study examining another model of absence epilepsy: the WAG/Rij model. 45 It is intriguing to consider whether a bidirectional relationship exists, such that increases in stress hormones enhance seizures and seizures themselves increase stress hormone levels. We have previously shown that kindled seizures elicit a pronounced corticosterone response, 14,32 but we are not aware of any studies that examine the acute endocrinological consequences of absence seizures. If such a relationship exists, interventions geared to reducing seizures, such as antiseizure drugs or neuromodulatory techniques, might reduce corticosterone levels. Motivated by the finding of elevated corticosterone in GAERS, we tested whether metyrapone, an inhibitor of 11-βhydroxylase that catalyzes the production of corticosterone, would be able to reduce seizures. Metyrapone markedly and dose-dependently suppressed seizures in GAERS, completely abolishing seizures at the highest dose for >20 min. Given its established safety record, this property promotes metyrapone as a strong candidate for repurposing as a new antiseizure compound. Further experiments are required to fully characterize the effects of metyrapone, for example, determining the duration of action (although we found this does not extend to 48 h), effects of repeated testing/tolerance, and examination in different models of epilepsy, such as acquired epilepsy models, to ascertain its effects on other seizure types. These would be required studies prior to any attempts at clinical translation.
To our knowledge, the only pharmacological target of metyrapone is 11-βhydroxylase. Inhibition of this enzyme results in suppression of synthesis of corticosterone, 14 and given the effects of stress, and of exogenous corticosterone (both of which increased seizures), it is attractive to speculate that the effects of metyrapone are driven by reducing corticosterone levels. However, another consequence of inhibition of 11-βhydroxylase is the increased levels of corticosterone precursors, including neurosteroids such as the immediate precursor DOC. Neurosteroids, in particular tetrahydrodeoxycorticosterone, 50 are powerful seizure suppressants, so it is feasible that the mechanism by which metyrapone reduces seizures is actually related to an increase in the neurosteroid precursors of corticosterone. To address this, we exogenously administered DOC to GAERS, and found that this increased seizures. This effectively rules this out as a mechanism by which metyrapone reduces seizures, because one would predict that if that F I G U R E 5 Treatment with deoxycorticosterone (DOC) does not reduce seizures in genetic absence epilepsy rats from Strasbourg. (A, B) Seizure incidence was not impacted by DOC treatment (A), compared to vehicle, although the percentage of time spent in seizure was significantly increased (B), compared to vehicle. (C) Seizure duration was not impacted by DOC administration. #p < .05 indicates effect of treatment. n = 7/group. were the case, DOC would similarly reduce seizures in GAERS. It is interesting to speculate how DOC increases seizures in GAERS, because neurosteroids powerfully suppress convulsive seizures, 50 which is in contrast to the observed elevation in absence seizures seen here. Neurosteroids act to modulate γ-aminobutyric acid type A (GABA A ) receptors, 50 and it is appreciated that enhancement of GABA neurotransmission increases SWD seizures in GAERS, 23 pointing to a likely mechanism of DOC action.
To summarize, here we show that GAERS exhibit elevations in stress hormone levels, and that stress and exogenous corticosterone administration result in increased SWD seizures. Inhibition of corticosterone synthesis by metyrapone abolishes the occurrence of seizures, promoting this compound as a novel antiabsence compound ready for clinical trials.

AUTHOR CONTRIBUTIONS
Gabi Dezsi, Ezgi Ozturk, Georgia Harris, and Cornelius Paul conducted the experiments and analyzed the data. Terence J. O'Brien designed the study and edited the paper. Nigel C. Jones conceptualized the study, acquired funding, analyzed the data, and wrote the paper.