A Clue to Seizure-Induced SUDEP Risk: Loss of Brainstem Serotonergic Control of Cardiorespiratory Function

Impaired Serotonergic Brainstem Function during and after Seizures

Zhan Q, Buchanan GF, Motelow JE, Andrews J, Vitkovskiy P, Chen WC, Serout F, Gummadavelli A, Kundishora A, Furman M, Li W, Bo X, Richerson GB, Blumenfeld H. J Neurosci 2016;36:2711–2722.

Impaired breathing, cardiac function, and arousal during and after seizures are important causes of morbidity and mortality. Previous work suggests that these changes are associated with depressed brainstem function in the ictal and post-ictal periods. Lower brainstem serotonergic systems are postulated to play an important role in cardiorespiratory changes during and after seizures, whereas upper brainstem serotonergic and other systems regulate arousal. However, direct demonstration of seizure-associated neuronal activity changes in brainstem serotonergic regions has been lacking. Here, we performed multiunit and single-unit recordings from medullary raphe and midbrain dorsal raphe nuclei in an established rat seizure model while measuring changes in breathing rate and depth as well as heart rate. Serotonergic neurons were identified by immunohistochemistry. Respiratory rate, tidal volume, and minute ventilation were all significantly decreased during and after seizures in this model. We found that population firing of neurons in the medullary and midbrain raphe on multiunit recordings was significantly decreased during the ictal and post-ictal periods. Single-unit recordings from identified serotonergic neurons in the medullary raphe revealed highly consistently decreased firing during and after seizures. In contrast, firing of midbrain raphe serotonergic neurons was more variable, with a mixture of increases and decreases. The markedly suppressed firing of medullary serotonergic neurons supports their possible role in simultaneously impaired cardiorespiratory function in seizures. Decreased arousal likely arises from depressed population activity of several neuronal pools in the upper brainstem and forebrain. These findings have important implications for preventing morbidity and mortality in people living with epilepsy.

Commentary

A subset of temporal lobe epilepsy (TLE) patients present comorbid conditions associated with a dysfunction of the serotonin axis, including depression, sleep apnea, and sudden unexpected/unexplained death in epilepsy (SUDEP). To date, the mechanism underlying these associations has been unclear, but potential culprits have been proposed: decreased serotonin levels, impaired 5-HT receptor function, or a decrease in serotonin neuron population or neuronal firing. Zhan et al. directly demonstrated that postictal generalized EEG depression is associated with decreased 5-HT neuronal firing in the lower brainstem and subsequent suppressed breathing, a major risk factor associated with SUDEP. Over 30% of patients with TLE have seizures that are refractory to commonly used antiseizure drugs (ASDs), and 9 in 1,000 of these intractable patients will die yearly from SUDEP (1). The relatively high risk of mortality presents a dire need for the 1) identification of reversible mechanisms contributing to seizure-induced SUDEP risk in animal models of epilepsy and, more importantly, 2) development of novel therapies (or repurposing old standbys) to prevent SUDEP fatalities. Zhan et al. have presented a seizure-induced mechanism responsible for cardiorespiratory arrest for which there are clinically available treatments (i.e., SSRIs).

Both human and animal data demonstrate that serotonin plays a role in epilepsy. Elevated extracellular serotonin (5-HT) levels lower thresholds for both focal and generalized seizures and inhibition of brain 5-HT decreases audiogenically, chemically and electrically evoked seizures (2). Several ASDs including carbamazepine, phenytoin, valproic acid, lamotrigine, and zonisamide have been shown to increase endogenous extracellular 5-HT concentration as part of their anti-seizure mechanism. Transgenic mice lacking 5-HT_1A or 5-HT_2C receptors demonstrate increased seizure activity or lower seizure threshold (1). The recreational drug MDMA has specific effects on the serotonin system, first causing massive release of 5-HT and then damage to serotonergic synaptic terminals, which can contribute to seizure generation (3). Human epilepsy brains demonstrate decreased 5-HT_1A receptor binding (4).

Brainstem serotonergic neurons are highly specialized. The midbrain dorsal raphe 5-HT neurons project rostrally to modulate arousal. The serotoneric neurons of the medullary raphe project caudally to modulate cardiorespiratory function (5, 6). Therefore, Zhan et al. conducted single and multiunit recordings of neurons of the medullary and midbrain raphe in anesthetized rats to determine changes in firing during ictal and postical epochs that may contribute to altered respiration and arousal, respectively. Serotonin neurons were identified using immunohistochemistry. Labelled 5-HT neurons were located in the dorsal, ventral, and caudal subnuclei of the dorsal raphe after juxtacellular recordings were completed. Simultaneous EKG and breathing plethysmography measurements were conducted to monitor correlated changes in respiratory and cardiac function during and after evoked seizures via hippocampal electrical stimulation. Mean minute ventilation, respiratory rate, tidal volume, and heart rate (expressed as percent change relative to 30-second preictal baseline) all decreased during and after seizures. Mean multiunit activity from the medullary and midbrain raphe and single unit activity from medullary 5-HT neurons decreased in the ictal and postictal periods relative to preictal periods. However, seizures did not significantly affect 5-HT single unit activity in the midbrain dorsal raphe.

Raw traces of local field potentials, multiunit activity, and respiration demonstrated a clear decrease in multiunit activity during seizures in the medullary raphe nucleus and midbrain dorsal raphe correlated with decreased breathing, which remains decreased during a refractory postictal period (with EEG depression recorded in local field potentials), whereupon, within 10 seconds, normal breathing returns. The same pattern is true for multiunit activity in the dorsal raphe and medullary raphe. However, following spike sorting of single unit recordings, a decreased rate of firing is noted for serotonergic neurons in the medulla, and changes in firing of midbrain serotonin neurons are mixed. Caution is required in interpreting single unit activity from identified serotonergic neurons with multiunit recordings when sampled from a wide population of neurons. The authors focused on the regional differences due to specific serotonergic cardiorespiratory functions associated with medullary control centers. The output of the midbrain raphe would more likely be associated with arousal, and nonserotoneric neurons included in the population recorded may also be responsible for multiunit firing rate changes. Arousal was not analyzed in this study nor was the severity of the evoked seizures, both of which could have contributed to or been affected by different degrees of postictal suppression.

Zhan and colleagues do not address the lack of difference in serotonin neuron firing rates or impairments in functional sequelae (i.e., respiration or heart rate) between ictal epochs relative to postictal epochs. Previous work by this group quantified postictal EEG suppression and cardiorespiratory arrest as contributing to SUDEP (7) and their monitored SUDEP cases in the MORTEMUS study invariably displayed seizure-induced postictal generalized EEG suppression (PGES) associated with depression of cardiorespiratory function (apnea/hypoventilation) and, finally, asystole (8). This suggests that PGES may be a cause or a surrogate marker of the primary cause of cardio-respiratory dysfunction. An alternative but related explanation is that the hypoventilation and apnea that occur during the seizure contribute to electrocerebral shutdown (9).

The rat model of stimulus-evoked seizures does not investigate SUDEP, but that is not necessary to investigate mechanisms of seizure-induced brainstem impairment. The authors used a well-controlled rat seizure model, electrical stimulation-induced seizures in the hippocampus, to access serotonergic neurons of brainstem modulatory regions that have been hypothesized to contribute to a state of cardiac, respiratory, and arousal system vulnerability after a seizure. It would be worthwhile to investigate whether epileptic brains from spontaneously seizing animals—rather than normal animals—had impairments in serotonergic neuronal firing in the medulla and midbrain raphe associated with cardiovascular impairment. At baseline or interictally, the epileptic animal may have a lowered brainstem 5-HT firing rate and already impaired cardiovascular function, suggesting that a stimulus to the hippocampus similar to that delivered in these normal animals would create the perfect storm necessary to cause cardiorespiratory arrest.

No attempt was made to reverse or prevent the very clear seizure-induced decrease in serotonergic firing in the medullary raphe or secondary effects on breathing and heart rate. The required use of ketamine and other anesthetics in these experiments modulates 5-HT neuronal activity, complicating the design of targeted therapy experiments. However, pretreatment with an SSRI has previously reduced the frequency of respiratory arrest in DBA/2 mice (10). A welcome future direction for this research would be to determine whether therapeutics that increase 5-HT levels (i.e., 5-HT receptor agonists or SSRIs) might affect seizure-induced PGES, serotonergic neuronal firing in the medullary raphe, or cardiorespiratory measures in multiple seizure and epilepsy models.

Knowledge about the normal function of the 5-HT system in the medulla raphe and the effect of evoked seizures on that discrete serotonergic neuronal population provides insight into breathing impairment after acute seizures, enabling a better understanding of the cause of death in SUDEP. These findings may lead to better ways to identify at-risk patients and treat them to prevent SUDEP, but more research is required in epileptic models. As has been the history with ASD development, the normal brain is likely to be less clinically predictive, and prognostic indicators of SUDEP will need to be validated in spontaneously seizing animal models.