Abstracts

Rationale: Sudden unexpected death in epilepsy (SUDEP) is the leading cause of death in refractory epilepsy patients. Human and animal models suggest that seizure-related respiratory arrest may instigate the cascade that produces cardiorespiratory arrest and death. Recent studies show that electrical stimulation of the amygdala can produce apneas, and seizure spread to the amygdala coincides with onset of ictal apneas; however, these results can be inconsistent. This work aimed to determine if an extended amygdalar structure, the bed nucleus of the stria terminalis (BNST) is involved in seizure-related death in the DBA/1J audiogenic mouse model of SUDEP. The BNST has projections to many brainstem respiratory-related nuclei and dense interconnections to other extended amygdalar nuclei such as the central amygdala (CeA), making it a potential intermediary in seizure-related death.

Methods: We used DBA/1J audiogenic seizure (AGS) mice, a characterized model with seizures in response to audio tones. Animals were “primed” to have audiogenic seizures at P21 by daily exposure to an audio tone ( >90dB), repeated for 3-7 days or until animals had 3 full tonic-clonic seizures. Animals were ventilated following full seizures, which are associated with respiratory arrest and death.

After priming, bilateral, stereotaxic injections into the BNST were used to introduce either a control virus (AAV-DJ-CMV-eGFP) or a virus (AAV-DJ-CMV-eGFP-TeNT) to express tetanus neurotoxin (TeNT) under a general neuronal promoter to diminish presynaptic output of BNST neurons. After 3 weeks, whole-body plethysmography was conducted. Animals were re-tested for AGS and mortality associated with full seizures was recorded. Injection accuracy and viral efficiency were evaluated by immunohistochemistry (IHC) upon death. A subset of virally-injected animals was used for electrophysiologic experiments, recording neurons in brainstem targets of the BNST– the parabrachial nucleus and periaqueductal gray.

A subset of animals did not undergo stereotaxic injections. After five days of audio exposure they were re-tested for AGS and ventilated. IHC was conducted to measure neuronal activation with the immediate early gene marker c-Fos in the BNST, amygdala, and brainstem.

Results: Strong BNST c-Fos activation occurred only following convulsive seizures (n=14, p< 0.001). The PAG and inferior colliculus expectingly showed strong activation by convulsive and wild-running seizures. There was no activation in the amygdala following seizures.

Animals with good bilateral targeting of virally expressed TeNT in the BNST had a significant increase in survival from tonic-clonic seizures compared to control virus (42% vs 8%, n=12, 25, p=0.02). Experimental mice showed no change in baseline breathing parameters or hypercapnic response compared to control mice.

Conclusions: We show that a deep brain region, the BNST, a component part of the extended amygdala, is activated only by seizures that are associated with respiratory arrest. Its output may be critical for mediation of seizure-related respiratory arrest and death in a mouse model of SUDEP.

Funding: Please list any funding that was received in support of this abstract.: American Epilepsy Society Junior Investigator Award; Center for SUDEP Research Pilot Award.