Neuronal plasticity contributes to postictal death

Abstract

Repeated generalized tonic-clonic seizures (GTCSs) are the most critical risk factor for sudden unexpected death in epilepsy (SUDEP). GTCSs can cause fatal apnea. We investigated neuronal plasticity mechanisms that precipitate postictal apnea and seizure-induced death. Repeated seizures worsened behavior, precipitated apnea, and enlarged active neuronal circuits, recruiting more neurons in such brainstem nuclei as periaqueductal gray (PAG) and dorsal raphe, indicative of brainstem plasticity. Seizure-activated neurons are more excitable and have enhanced AMPA-mediated excitatory transmission after a seizure. Global deletion of the GluA1 subunit of AMPA receptors abolishes postictal apnea and seizure-induced death. Treatment with a drug that blocks Ca²⁺-permeable AMPA receptors also renders mice apnea-free with five-fold better survival than untreated mice. Repeated seizures traffic the GluA1 subunit-containing AMPA receptors to synapses, and blocking this mechanism decreases the probability of postictal apnea and seizure-induced death.

Keywords: AMPA; Apnea; Brainstem plasticity; Epilepsy; GluR1 subunit; SUDEP; Seizures.

Fig. 1.. Repeated generalized seizures gradually worsen, leading to postictal apnea and death.

(A) Human SUDEP risk increases as a function of yearly GTCS frequency (data plotted from Hesdorfer et al. 2011, Epilepsia(Hesdorffer et al., 2011)). (B) We recorded EEG, breathing, and heartbeat simultaneously as mice experienced GTCSs every other day for 10 seizures. (C) Kaplan-Meier survival curve demonstrates decreased survival with repeated seizures. (D) Cumulative probability of postictal apnea and seizure-induced death increase with repeated GTCSs. Inset graph: Apnea correlates with death. (E) Representative bilateral EEG recording (black), breathing (blue), and heartbeat (red) of a first seizure with ictal apnea only. (F) Repeated seizure with postictal apnea and subsequent recovery from it. (G) Fatal repeated GTCS with terminal postictal apnea and gradual bradycardia. (H) Postictal apnea occurred more frequently as the number of GTCSs (columns) increased in individual C57BL/6 mice (rows, n = 17). Grey and black indicate postictal apnea duration (< or > 30 s), and white crosses indicate death day. (I) Behavioral seizures worsened with repeated GTCSs (columns) in individual C57BL/6 mice (rows). Grey and black indicate stage 5 or 6 seizures, and white crosses indicate death day. (J) Ratio of mice that had only stage 5 seizures (grey) vs at least one stage 6 seizure (black). (K) Latency to seizure onset (min) decreased with repetition. Data are mean ± SEM.

Fig. 2.. Seizures repeat and expand network, causing brainstem plasticity.

(A) 3D brain reconstruction with sequential 45 μm sections overlaid on top of each other as shown in c after four repeated GTCSs (the first stage 6 seizure occurred on day 4) vs (B) after a single stage 5 seizure in TRAP2 mice. (C) Schematic of the number and location of slices used for 3D reconstruction in a and b. CeA – central amygdala, MeA – medial amygdala, BLA – basolateral amygdala, PAG – periaqueductal gray, D. raphe – dorsal raphe, P.B. – parabrachial nucleus, K.F. – Kölliker-Fuse nucleus. (D) Repeated second seizure (green, c-Fos, or (E) ARC) recruits the same neurons as the first seizure (red, tdTomato) plus additional neurons, expanding the network. (F) Quantification demonstrates % of neurons active during the first seizure that were also active during the second seizure across the motor cortex. (G) Quantification of the number of cells per slice in the dorsal raphe and PAG nucleus in TRAP2 mice after repeated seizures (n = 5 mice) vs after the first seizure (n = 4 mice). Color gradation of symbols in the first column represents the number of repeated seizures that occurred before quantification (1^st stage 6 seizure on: day 1 (white), day 4 (light grey), day 6 (dark grey), day 9 (black)). (H) GTCSs expand the neuronal circuits active during subsequent GTCSs, activating the first initial neurons and additional ones. Data are mean ± SEM, *p < 0.05, **p< 0.01.

Fig. 3.. Seizure-activated neurons are more excitable than nonactive neurons and demonstrate enhanced AMPA transmission after a single seizure.

(A) A fluorescent/DIC image shows bright tagged neurons in layer 2/3 of the motor cortex in an acute brain slice of a TetTag mouse after a single seizure. (B) An example of motor cortical layer 2/3 pyramidal neuron filled with biocytin during the recording. All recorded neurons were identified post-hoc. (c) The rest membrane potentials, (D) membrane time constant, (E) membrane resistance were similar in tagged (green) and untagged (black) neurons. (F) Action potential (A.P.) threshold was lower in tagged (green) than untagged (black) neurons. (G) A.P. amplitudes, (H) A.P. wide were similar in tagged (green) and untagged (black) neurons. (I) Similar membrane resistance but lower rheobase current was in tagged (green) than untagged (black) neurons. Dashed lines show the mean values. (J,K) The representative traces illustrate action potentials evoked at different current injections. (L) Frequency-current (F-I) plot illustrates a higher frequency of action potentials evoked by each current injection in tagged neurons. (M) Instantaneous frequency vs action potential sequence based on analysis of A.P.s evoked by current injection, examples are shown are in j and k. (N) Cumulative ratio (%) of sEPSC inter-event intervals in tagged (green) and untagged neurons (black). (O) Average sEPSC traces from a tagged (green) and an untagged surrounding neuron (black). (P) Cumulative ratio (%) of sEPSC amplitudes in tagged (green) and untagged neurons (black). (Q) The frequency distribution of sEPSCs for 16 untagged neurons fitted with Gaussian distribution (black line). (R) The frequency distribution of sEPSCs for 13 tagged neurons with the sum of two Gaussian distributions (green curve) and two separated single Gaussian distributions (red curves). (S) Rise time was similar in tagged and untagged neurons. (T) Weighted decay tau was similar in tagged and untagged neurons. Data are mean ± SEM (SD: c-h), n = 11 per group, *p < 0.05, **p< 0.01, ****p< 0.0001.

Fig. 4.. Mice without the GluA1 subunit of AMPA receptor do not develop postictal apneas, have less severe seizures, and survive.

(A,B) Representative EEG (black), breathing (blue), and heartbeat (red) recordings in a GluA1 K.O. mouse without postictal apnea and a W.T. littermate with fatal postictal apnea and gradual bradycardia until death. (C) Postictal apneas appeared more frequently in W.T. compared to GluA1 K.O. mice with repeated seizures (crossed white background indicates headset fell out or breathing recording not saved). (D) Seizures in W.T. mice became progressively longer compared to K.O. mice, in which they stayed of the same duration (each point represents a single mouse). (E) Seizure behavior became more severe with repetition in W.T. than in GluA1 K.O. mice. (F) W.T. mice had more stage 6 (black) seizures than GluA1 K.O.s. (G) Kaplan-Meyer survival curve demonstrates more GluA1 K.O.s survived than W.T.s. Data are mean ± SEM, *p < 0.05, ****p< 0.0001.

Fig. 5.. GluA1 subunit of AMPA receptors sustains repeated GTCSs.

(A) Heat maps illustrate the severity of behavioral seizure scores in GluA1 K.O. vs W.T. mice (red indicates death; black indicates more severe seizures; rows signify individual animals (10 animals were randomly chosen to represent each group)). (B) All W.T. mice were kindled, whereas 60% of GluA1 K.O. mice remained unkindled despite stimulation. (C) Threshold to induce seizure was higher in GluA1 K.O. than in W.T. mice. (D) Behavioral seizure scores were lower in GluA1 K.O. than W.T. mice (median and error, 95% CI). (E) Seizure duration (after-discharge, s) was shorter in GluA1 K.O. than in W.T. mice. Data are mean ± SEM, * p < 0.05.

Fig. 6.. Blockade of calcium-permeable AMPA receptors prevents postictal apnea and seizure-induced death.

(A) IEM-1460 blocks calcium-permeable AMPARs. (B) Cumulative ratio (%) of sEPSC inter-event intervals decreased after IEM (green) application (before is black) only in activated neurons. (C) Cumulative ratio (%) of sEPSC amplitudes decreased after IEM (green) application (before is black) only in activated neurons. (D) IEM-1460 or saline were injected i.p. 15 min before each PTZ injection every other day for a total of 10 seizures. (E,F) IEM application (grey) did not affect sEPSC inter-event interval frequency (e) or amplitude (f) in nonactive neurons. (G) Postictal apnea did not occur in IEM-treated mice. (Mice #3&4 died in a home cage without a seizure in between assigned PTZ injection days). (H) Behavioral seizures worsened with repetition in saline-injected mice but not in IEM treated. (I) Overall seizure duration in saline-injected mice was longer. (J) Kaplan-Meyer survival curve demonstrates more IEM-treated mice survived than saline injected. (K) We propose repeated seizures traffic the GluA1 subunit-containing AMPA receptors to synapses, and blocking this mechanism leads to better survival.