Abstract
Sudden unexpected death in epilepsy (SUDEP) has been linked to respiratory dysfunction, but the mechanisms underlying this association remain unclear. Here we found that both focal and generalized convulsive seizures (GCSs) in epilepsy patients caused a prolonged decrease in the hypercapnic ventilatory response (HCVR; a measure of respiratory CO2 chemoreception). We then studied Scn1aR1407X/+ (Dravet syndrome; DS) and Scn8aN1768D/+ (D/+) mice of both sexes, two models of SUDEP, and found that convulsive seizures caused a postictal decrease in ventilation and severely depressed the HCVR in a subset of animals. Those mice with severe postictal depression of the HCVR also exhibited transient postictal hypothermia. A combination of blunted HCVR and abnormal thermoregulation is known to occur with dysfunction of the serotonin (5-hydroxytryptamine; 5-HT) system in mice. Depleting 5-HT with para-chlorophenylalanine (PCPA) mimicked seizure-induced hypoventilation, partially occluded the postictal decrease in the HCVR, exacerbated hypothermia, and increased postictal mortality in DS mice. Conversely, pretreatment with the 5-HT agonist fenfluramine reduced postictal inhibition of the HCVR and hypothermia. These results are consistent with the previous observation that seizures cause transient impairment of serotonergic neuron function, which would be expected to inhibit the many aspects of respiratory control dependent on 5-HT, including baseline ventilation and the HCVR. These results provide a scientific rationale to investigate the interictal and/or postictal HCVR as noninvasive biomarkers for those at high risk of seizure-induced death, and to prevent SUDEP by enhancing postictal 5-HT tone.
SIGNIFICANCE STATEMENT There is increasing evidence that seizure-induced respiratory dysfunction contributes to the pathophysiology of sudden unexpected death in epilepsy (SUDEP). However, the cellular basis of this dysfunction has not been defined. Here, we show that seizures impair CO2 chemoreception in some epilepsy patients. In two mouse models of SUDEP we found that generalized convulsive seizures impaired CO2 chemoreception, and induced hypothermia, two effects reported with serotonergic neuron dysfunction. The defects in chemoreception and thermoregulation were exacerbated by chemical depletion of serotonin and reduced with fenfluramine, suggesting that seizure-induced respiratory dysfunction may be due to impairment of serotonin neuron function. These findings suggest that impaired chemoreception because of transient inhibition of serotonergic neurons may contribute to the pathophysiology of SUDEP.
Introduction
Sudden unexpected death in epilepsy (SUDEP) is the main cause of premature death in patients with epilepsy (Thurman et al., 2014). Most witnessed cases are preceded by a generalized convulsive seizure (GCS; Devinsky et al., 2016), and those that occur during epilepsy monitoring unit (EMU) surveillance together with data from animal models indicate that death is often because of postictal respiratory arrest (Ryvlin et al., 2013; Massey et al., 2014). However, the mechanisms that lead from seizures to fatal respiratory dysfunction remain largely unknown.
Under awake, resting conditions, depth and rate of breathing are primarily regulated by two sources of “tonic respiratory drive”: (1) wakefulness drive mediated by descending projections from the forebrain and (2) chemoreceptor drive mediated by CO2 chemoreceptors (Mahamed et al., 2001; Richerson and Boron, 2003; Teran et al., 2014). There is evidence that GCS can impair both of these sources of respiratory drive, which could explain why terminal apnea occurs after a GCS in SUDEP cases (Ryvlin et al., 2013). For example, patients have been observed to remain immobile in the prone position before death (Langan et al., 2000; Ryvlin et al., 2013), suggesting they are not awake. This lack of cortical activation in obtunded postictal patients would result in loss of wakefulness drive.
There is also evidence that CO2 chemoreception is impaired after seizures. Patients with Dravet syndrome (DS), an epileptic encephalopathy with a high incidence of SUDEP (Shmuely et al., 2016), can display prolonged hypoventilation after a GCS causing increased transcutaneous CO2 (tcCO2; Kim et al., 2018). This suggests that respiratory CO2 chemoreceptors are impaired after seizures. If true, this postictal dysfunction of CO2 chemoreception would result in loss of chemoreceptor drive. However, CO2 chemosensitivity has not been directly measured after seizures, such as with the hypercapnic ventilatory response (HCVR).
There are several groups of neurons that are candidates for central CO2 chemoreceptors (Teran et al., 2014). These include serotonergic neurons in the medulla and midbrain, which are highly sensitive to CO2 (Teran and Richerson, 2020). They would be expected to increase their firing rate in response to the increase in CO2 that occurs because of hypoventilation/apnea during the ictal and postictal state. Instead, it has previously been shown that a subset of serotonergic neurons are inhibited during seizures (Zhan et al., 2016). If it is common for serotonergic neurons to be inhibited during seizures, that could explain the postictal decrease in CO2 chemoreception described above.
Some serotonergic neurons are also critically important for thermoregulation. Genetic deletion (Hodges et al., 2008) or chemogenetic silencing (Ray et al., 2011) of serotonergic neurons in mice blunts CO2 chemoreception and also causes thermoregulatory failure. Similarly, targeted lesions of serotonergic neurons with diphtheria toxin (Cerpa et al., 2014) or inhibition of 5-hydroxytryptamine (5-HT, serotonin) synthesis with para-chlorophenylalanine (PCPA; Murray et al., 2015) impairs thermoregulation.
Serotonin has been implicated in the pathophysiology of SUDEP (Petrucci et al., 2020), but the mechanisms of this link remain unclear. Dysfunction of the 5-HT system increases mortality after seizures (Tecott et al., 1995; Buchanan et al., 2014), and activation of serotonergic neurons or receptors increases survival in SUDEP models (Tupal and Faingold, 2006, 2019; Faingold et al., 2011; Buchanan et al., 2014; Zeng et al., 2015; Zhang et al., 2016, 2018; Petrucci et al., 2020). Notably, recent randomized-controlled trials show that fenfluramine, whose mechanism of action includes several 5-HT agonist effects (Fuller et al., 1978; Baumann et al., 2014; Tupal and Faingold, 2021), decreases convulsive seizure frequency in children with DS (Lagae et al., 2019; Nabbout et al., 2020; Gogou and Cross, 2021). Here, we studied patients with epilepsy and two mouse models of SUDEP to directly measure whether GCS cause postictal impairment of CO2 chemoreception and thermoregulation as would be expected if there is postictal inhibition of a sufficient number of serotonergic neurons. Using PCPA and fenfluramine in DS mice we also obtained evidence that serotonergic mechanisms appear to be involved.
Materials and Methods
Video-EEG and HCVR measurement in epilepsy patients
Eligible patients were 18 years or older with confirmed epilepsy who were undergoing video-EEG (vEEG) monitoring in the EMU at the University of Iowa Hospitals and Clinics from March 2015 to November 2019. Continuous time-synchronized video and EEG were recorded using a Nihon Kohden acquisition system. EEG electrodes were placed using the international 10–20 system with addition of T1 and T2 electrodes. Recordings of each seizure were reviewed by a board-certified epileptologist (R.K.S.). Patients with a history of cardiopulmonary disease, stroke or space-occupying brain lesion were excluded. Women were excluded if pregnant or if pregnancy could not be ruled out. The HCVR, defined here as the slope of the line describing the minute ventilation (VE) versus end-tidal CO2 (ETCO2) relationship (ΔVE/ΔETCO2) fit with a least-squares regression, was measured using a previously described modified hyperoxic rebreathing technique (Sainju et al., 2019). Only data points above the ventilatory recruitment threshold were included, as these reflect chemoreflex sensitivity (Jensen et al., 2005). Individual breaths that were significant outliers because of behavioral changes were excluded. Postictal HCVR testing was performed as soon as patients were able to cooperate with testing after focal seizures, and was delayed for at least 2 h postictal for GCS given a lack of human safety data for HCVR measurement performed immediately after GCS. Cardiorespiratory function was monitored for apnea with the use of a nasal pressure transducer (BiNaps, Salter Labs), oronasal thermistor (ThermiSense, Salter Labs) and chest and abdominal belts (Pro-Tech zRIP DuraBelt, Philips Respironics). All procedures involving patients were approved by the institutional review board at the University of Iowa. All patients provided written informed consent.
Mouse husbandry and genotyping
Breeding and genotyping of DS and D/+ mice have been previously described (Auerbach et al., 2013; Wagnon et al., 2015). A pair of Scn1aR1407X/+ heterozygous (DS) male mice on a C3HFeB/HeJ background were bred with wild-type (WT) C3HFeB/HeJ female mice obtained from Jackson Laboratory to establish a breeding colony. A pair of Scn8aN1768D/+ heterozygous (D/+) male mice on a C57BL/6J background (Wagnon et al., 2015) were obtained from the Meisler lab at the University of Michigan and used to establish a breeding colony. For both colonies, wild-type (WT) littermates were used as controls. Both DS and D/+ mice have a high incidence of spontaneous seizures and seizure-induced mortality (Wagnon et al., 2015; Kim et al., 2018). These animals exhibit a normal growth curve and behavior, until they experience a fatal seizure (unpublished observations). Animals of both sexes were used for all experiments. Mice were housed in a 12/12 h light/dark cycle (lights on 7 A.M. to 7 P.M.) in standard cages with food and water available ad libitum. All procedures and experiments involving mice were conducted with approval of the University of Iowa Institutional Animal Care and Use Committee, and in strict accordance with the recommendations of the ACP Guide for the Care and Use of Laboratory Animals (National Research Council, 2011).
Whole-animal plethysmography
Ventilation was measured using standard barometric whole-body plethysmography as previously reported (Hodges et al., 2008; Massey and Richerson, 2017; Kim et al., 2018). Briefly, a commercially available plethysmography chamber [Buxco, Data Sciences International (DSI)] was supplied with continuous gas flow (700 ml/min) from certified calibrated tanks containing prefilled gas mixtures (Praxair). Tidal volumes were calculated from measurements of chamber pressure, ambient barometric pressure (obtained from Iowa City Airport data), animal and chamber temperature and chamber humidity (Drorbaugh and Fenn, 1955). Inspiratory peaks were detected automatically with custom software and verified manually. A total of at least 60 s of data, with each segment lasting at least 6 s in duration, were selected for analysis. These segments were taken from periods of quiet wakefulness that did not contain behavioral artifacts (chewing, licking, sniffing, locomotion, etc.), because that allows measurement of homeostatic automatic breathing, which is dependent on chemoreceptor drive, and not breathing movements involved in many conscious behaviors that are not related to blood gas control (Arthurs et al., 2023). Breathing was analyzed after steady-state was reached during periods of gas exposure to normocapnia and hypercapnia.
To measure baseline HCVR, plethysmography recordings were obtained for >20 min from DS and WT mice in 0% CO2 (normocapnia) followed by 7- to 10-min exposures to increasing levels of inspired CO2 (3, 5, and 7%). In all plethysmography experiments, breathing was measured under hyperoxic (50% O2; variable CO2; balance N2) conditions to minimize input from peripheral chemoreceptors (Lahiri and DeLaney, 1975). VE was calculated as the product of respiratory frequency (fR) and tidal volume (VT) and normalized to animal weight. Measurements were then normalized to values at 0% CO2 for each genotype. Body temperature was measured with temperature probes (IPTT300; BMDS) inserted into the abdominal cavity at least 3 d before recordings. Unless otherwise noted, all experiments were performed at room temperature between 10 A.M. and 5 P.M. DS mice have a low likelihood of spontaneous seizures during the daytime (Teran et al., 2019). Mice were observed for at least 30 min before measuring baseline HCVR to ensure they did not have a seizure. If they did, the study was delayed, typically until the next day.
HCVR before and after heat-induced seizures
A preictal HCVR was obtained as described above, except that measurements were only made at 0% and 7% CO2. Hyperthermia was induced in DS and WT mice by placing the animal in a 10 L beaker and slowly increasing core body temperature with a heat lamp until a seizure occurred that qualified as Racine 5 (R5) in a modified Racine scale (Racine, 1972; bilateral myoclonus and rearing with loss of postural control) or a body temperature of 42.5°C was reached. In some cases, seizures progressed to GCS with or without tonic hindlimb extension. Since seizures that progressed to a GCS with tonic hindlimb extension were often fatal, an effort was made to prevent such progression to minimize animal deaths while achieving relative uniformity of seizure severity. This was done by minimizing exposure to the heat lamp and turning it off when body temperature reached ∼41°C, as determined by pilot experiments that showed body temperature in DS mice often increased too quickly after the 41°C point and thus caused seizure progression and death. As controls, WT littermates were heated to the same average temperature at which DS mice developed a R5 seizure (41.9 ± 0.5°C; data not shown), except the WT mice did not have a behavioral seizure. Following heat-induced seizures (DS mice) or hyperthermia (WT mice), animals were placed back inside the plethysmography chamber (in which the ambient temperature remained at room temperature) and an HCVR was obtained ∼10–15 min later.
HCVR and continuous body temperature monitoring
To measure changes in body temperature and metabolism induced by seizures, a separate cohort of DS mice was instrumented with abdominal telemetry temperature probes (Emitter G2; Starr Life Sciences Corp.) and allowed at least 3 d of recovery. Plethysmography was performed as described above with the chamber on a telemetry receiver (ER-4000; Starr Life Sciences Corp.). Chamber gas flow rate was decreased to 200 ml/min to allow changes in chamber O2 and CO2 to be more accurately measured (S-3A and CD-3A; AEI Technologies). Ventilation in 0% and 7% CO2 was measured before and after seizures were induced with hyperthermia. Oxygen consumption (VO2) was calculated by subtracting the outflow fraction of O2 from the inflow fraction of O2 and multiplying the difference by the chamber flow rate. Following HCVR measurements, animals were returned to their home cages on a telemetry receiver at room temperature (22–24°C). Cotton bedding was reduced to prevent rapid temperature recovery by nesting. Postictal core body temperature was recorded in 10 s epochs for an additional 8–24 h.
Seizure induction in Scn8aN1768D/+ mice
To evoke seizures in D/+ mice, a sonicator (F60 Sonic Dismembrator, Fisher Scientific) was used to produce an audiogenic stimulus directly adjacent to the plethysmography chamber. The stimulus was a broad spectrum noise with intensity of 120 dB that lasted up to 30 s or until the animal developed a behavioral seizure. Seizures were documented with a webcam (FL8910W; Foscam Digital Technologies) using commercially available software (Blue Iris 4; Foscam Digital Technologies).
5-HT depletion with PCPA
A cohort of DS mice instrumented with Emitter G2 telemetry temperature probes was exposed to 0% and 7% CO2 to obtain a baseline HCVR, after which they were randomly assigned to receive either PCPA or vehicle. PCPA methyl ester (C3635; Sigma) was delivered at 800 mg/kg intraperitoneally for four to five consecutive days. Volume-matched saline injections were given to littermate control mice. Mice were returned to their home cage after each injection and video recordings were made as described above to document seizures or death. After 4 d, HCVR measurements were made before and after heat-induced seizures. Because PCPA treatment increased postictal mortality rate in DS mice, treated and control mice were placed on a cooling block for 3–5 min to abort the seizure once it reached an R5 level (Racine, 1972). The average postictal core temperature nadir was not significantly different between vehicle-treated animals placed on a cooling block following a hyperthermic seizure (32.5 ± 1.1°C, n = 21) and those that were not cooled (33.4 ± 2.1°C, p = 0.146, n = 23).
Fenfluramine treatment
A cohort of DS mice instrumented with Emitter G2 telemetry temperature probes was exposed to 0% and 7% CO2 to obtain a baseline HCVR, after which they were randomly assigned to receive either fenfluramine or vehicle. Fenfluramine hydrochloride (F112; Sigma) was dissolved in saline and delivered at 20 mg/kg intraperitoneally ∼45 min later an R5 seizure was induced with hyperthermia. The 20 mg/kg dose of fenfluramine was based on recent findings of Tupal and Faingold (2019). Volume-matched saline injections were given to control DS mice. The HCVR was repeated 10–15 min after seizures in fenfluramine-treated and control DS mice.
Effect of temperature on measurement of VE and HCVR
To determine whether changes in body temperature were responsible for changes we observed in the HCVR after seizures, a cohort of DS mice instrumented with Emitter G2 telemetry temperature probes were treated with PCPA for four consecutive days as described above. On day 4, each animal was placed inside a prechilled glass beaker on a telemetry receiver to lower core body temperature to 32°C, after which an HCVR was obtained as above. Body temperature was then increased to 37°C using a heat lamp, and the HCVR was repeated. None of these mice developed seizures.
Experimental design and statistical analysis
All animal data were collected and analyzed off-line using custom-written software (MATLAB; The MathWorks). Unless otherwise indicated, statistical significance in our plethysmography experiments involving multiple gas compositions was determined using a two-way repeated-measures ANOVA followed by the Holm–Sidak multiple comparison test, since our experiments manipulated two different variables: inspired CO2 levels and animal genotype. Wilcoxon signed-ranks test was used for nonparametric comparisons for HCVR measurements before and after seizures in epilepsy patients, metabolic changes (ΔVE/VO2) in response to hypercapnia in DS mice, and ventilatory measurements in DS and D/+ mice. When data followed a normal distribution, multiple paired t tests with correction for multiple comparisons using the Holm–Sidak method were used to compare ventilatory measurements between DS mice and WT littermates before and after hyperthermia. All mice in each experimental group were induced to have R5 level seizures that were similar in severity and duration. For some groups seizures were video recorded for later determination of seizure duration. Mann–Whitney tests were used to compare the average difference in body temperature nadir (ΔBT) between HCVR severity groups (mild vs severe). A Pearson correlation coefficient was computed to assess the relationship between HCVR (% interictal) and body temperature nadir. Unpaired t tests with correction for multiple comparisons using the Holm–Sidak method were used to compare HCVR and body temperature before and after treatments (PCPA vs vehicle). Fisher's exact test (two-tailed) was used to compare incidence of seizure-induced respiratory arrest (S-IRA) and death between two treatment groups. Statistical significance was set at p < 0.05. When data are presented as X ± Y, X is the group mean and Y is the standard deviation. All error bars on graphs represent the SEM. Statistical comparisons and data graphics were generated using Prism 9 (GraphPad Software).
Data availability
The data that support the findings of this study are available from the corresponding author, on reasonable request.
Results
Seizures caused prolonged blunting of CO2 chemoreception in epilepsy patients
We previously reported that central CO2 chemoreception in epilepsy patients, as measured by hyperoxic (50% O2) HCVR testing during an interictal period, correlated negatively with the duration and magnitude of postictal elevation of tcCO2 after GCS (Sainju et al., 2019). We now report interictal and postictal HCVR measurements in a subset of patients from the same cohort. A total of 123 patients with epilepsy were enrolled in the study, of which 83 (44 women) were eligible for HCVR measurement. All 83 had an interictal HCVR measured at bedside in the EMU. Postictal HCVR data were available in seven patients (6 women, mean age 32 ± 6.5 years) after nine seizures (Table 1). All seizures with postictal HCVR measurements were of focal onset on EEG (three right temporal, two left temporal, two left frontal, one left frontotemporal, and one nonlocalized/nonlateralized), of which four progressed to GCS and five did not. Postictal HCVR measurement was significantly delayed after GCS compared with focal seizures (229 ± 86 vs 68 ± 19.2 min, p = 0.004). Eight of nine postictal measurements (all four GCS and four of 5 focal) showed a decrease in HCVR compared with interictal (ΔVE/ΔETCO2 = 2.585 ± 1.5 vs 1.359 ± 0.731 l/min/mmHg; interictal vs postictal, p = 0.0391) and an increase for one focal seizure (Fig. 1). None of the patients displayed postconvulsive central apnea (PCCA). Baseline and postictal HCVR measurements for each patient are listed in Table 2.
Clinical characteristics of patients
Baseline and postictal HCVR
The HCVR was blunted postictally in epilepsy patients. A, Derivation of HCVR (ΔVE/ΔETCO2) in a single subject before (blue triangles) and ∼1.5 h after (red triangles) a focal seizure. Each data point represents a single breath, plotted sequentially as ETCO2 gradually rose with rebreathing. Values of ETCO2 were rounded to the nearest integer. Lines are the least squares fit of the data. B, Derivation of HCVR in a different subject before (blue circles) and ∼3 h after (red circles) a GCS. C, HCVR before (blue) and after (red) focal seizures (triangles, n = 5 seizures in 4 subjects) or GCS (circles, n = 4 seizures in 3 subjects). The postictal HCVR was significantly depressed compared with interictal. Error bars represent mean ± SEM (Wilcoxon matched-pairs signed-rank test, *p < 0.05).
The HCVR was inhibited after heat-induced seizures in Scn1aR1407X/+ mice
To determine whether DS mice have impaired interictal chemoreception, we challenged WT and DS mice to an increase in ambient CO2. Using whole-body plethysmography, 21- to 25-d-old WT (n = 12) and DS (n = 14) mice were exposed to 0%, 3%, 5%, and 7% CO2. This age range was chosen to study baseline breathing as our DS mice have a high incidence of seizure-induced death centered in this period (Teran et al., 2019). DS mice exhibited increased VE in response to graded increases in CO2 that was not significantly different from that of WT littermates (p = 0.0923; Fig. 2A). fR responses to hypercapnia were similar between genotypes (p = 0.7796; data not shown). DS mice displayed lower VT responses compared with WT mice (F(1,24) = 11.33, p = 0.0026), only at 7% CO2 (171 ± 29% of baseline in DS vs 204 ± 24% in WT; data not shown). This is different from what was reported for a Dravet syndrome mouse model with a different mutation (A1783V) expressed selectively in inhibitory neurons, which had a more severe phenotype with 100% lethality by P25 (Kuo et al., 2019). In those mice there was a decrease in fR but no change in VT. However, more consistent with our data, A1783V conditional mutant mice had only a minimal decrease in VE, which is the parameter that is relevant for control of systemic partial pressure of CO2 (Pco2). Additionally, we did not observe prolonged apnea or ataxic breathing in DS mice under baseline conditions (data not shown).
The interictal HCVR was normal in Scn1aR1407X/+ mice but was depressed after heat-induced seizures. A, Minute ventilation (VE) in response to graded increases in inhaled/ambient CO2. Scn1aR1407X/+ (DS) mice (n = 14) showed no differences in VE compared with WT littermates (n = 12). Two-way repeated measures ANOVA, with genotype and conditions as factors. Values are normalized to measurement at 0% CO2 for both groups of mice. B, VE, at 0% and 7% CO2 before and after heat-induced seizures in DS mice (n = 14) and equivalent hyperthermia in WT (n = 12) littermates. C, Data from (B displayed as HCVR (defined as the change in VE in response to increase in CO2 from 0% to 7%; ΔVE/ΔCO2) in DS mice and WT littermates before (preheat) and after (postheat) hyperthermia. DS mice displayed a significantly smaller HCVR after heat-induced seizures. There was no change in the HCVR in WT mice after hyperthermia to the same temperature (without a seizure). Paired t tests were used for B, C. **p < 0.01, ns - not significant.
We then measured ventilation at 0% and 7% CO2 and calculated the HCVR in DS mice before and after heat-induced seizures. All DS mice had R5 level seizures that were behaviorally comparable in severity and duration. The preictal VE and HCVR (change in VE in response to change in CO2, or ΔVE/ΔCO2) in DS mice (n = 14) were not different from WT littermates (n = 12; Fig. 2B,C). In contrast, the HCVR in DS mice following a heat-induced GCS was 43% lower than baseline (2.815 ± 1.28 vs 4.945 ± 2.214 ml/min/7% CO2, p = 0.001434), primarily driven by a 64% decrease in the change of fR in response to 7% CO2 (45 ± 35 vs 126 ± 48 breaths/min, p < 0.0001; data not shown). After heating, WT mice displayed a trend toward an increase in fR in response to 7% CO2 compared with before heating (145 ± 35 vs 124 ± 35 breaths/min, p = 0.054; data not shown), but no difference in HCVR (5.001 ± 1.595 vs 5.579 ± 1.182 ml/min/7% CO2, p = 0.4498; Fig. 2C). There were no differences in VT after heating in either genotype (ΔVT, 11.822 ± 1.6 vs 10.559 ± 3.4 µl/g/7% CO2 in WT, p = 0.222 and 10.423 ± 4 vs 9.331 ± 3.7 in DS, p = 0.379; data not shown). There were no sex-dependent differences observed.
DS mice had transient postictal impairment of thermoregulation that was proportional to blunting of the HCVR
A postictal decrease in the HCVR could be because of inhibition of serotonin neurons (Zhan et al., 2016). If so, a deficit in thermoregulation would also be expected to occur (Hodges et al., 2008; Ray et al., 2011). We measured the HCVR before and after heat-induced seizures (all R5 level) in a cohort of postnatal day (P)22–P34 DS mice instrumented with Emitter G2 abdominal telemetry temperature probes (n = 23). This age range was used to allow more animals to be eligible for study than in the experiments above, while still including the age range at which death was common. Mice in which there was a large decrease in the HCVR (postictal <50% of preictal ΔVE/ΔCO2, n = 11 of 23; Fig. 3A,B) displayed a larger drop in body temperature in the first hour postictally (−5.1 ± 1.3 vs −2.6 ± 2.7°C, p = 0.0317; Fig. 3E) than those in which there was a mild or no decrease in HCVR (postictal >50% of preictal ΔVE/ΔCO2, n = 12 of 23; Fig. 3C,D). Taken together, the HCVR of the two groups combined (mild and severe) was decreased postictally by 45% (1.989 ± 0.993 vs 3.642 ± 0.806 ml/min/7% CO2) because of decreases in both fR and VT. There was also a significant postictal decrease in VE at 0% CO2 (0.857 ± 0.302 vs 1.597 ± 0.498 ml/min, p < 0.0001). A decrease in body temperature did not occur in WT animals (n = 7) after raising their body temperature to an equivalent level (seizures also did not occur; Fig. 3E). A Pearson correlation coefficient was computed to assess the relationship between the change in the HCVR induced by seizures and the BT nadir (°C), and there was a significant correlation between the two variables (R2 = 0.3099, n = 23, p = 0.0058; Fig. 3F). There was nothing else that distinguished the mild versus severe blunting groups other than the change in HCVR and body temperature, such as age, sex, weight, seizure duration or seizure severity.
Thermoregulation and the HCVR were both impaired after seizures in Scn1aR1407X/+ mice. A, Breathing traces of a DS mouse with severe blunting of the HCVR after a seizure. Vertical scale bar represents one arbitrary unit of plethysmography chamber pressure, and horizontal bar represents 2 s. B, Summary of data from DS mice with severe postictal blunting of the HCVR (n = 11). C, Breathing traces of a DS mouse without blunting of the HCVR after a seizure. D, Summary of data from mice with mild postictal blunting of the HCVR (n = 12). E, DS mice with the greatest postictal decrease in the HCVR had a larger drop in body temperature (ΔBT, red, n = 11) in the first hour postictally (shaded box) than those mice in which there was a mild decrease in the HCVR (blue, n = 12; Mann–Whitney test, *p = 0.0317). For comparison, body temperature is shown for seven WT mice that were heated to the same temperature but did not have a seizure. Solid lines are the mean body temperature and the shaded regions are the SE. Inset shows individual values and mean drop in body temperature (ΔBT ± SEM) for the shaded period compared with baseline temperatures for the two groups of DS mice. F, There was a significant correlation between the postictal decrease in HCVR (ΔVE/ΔCO2 as % of preictal) and body temperature nadir. Red and blue symbols were selected based on whether postictal HCVR was > or <50% of the preictal value (R2 = 0.3099, n = 23, **p < 0.01).
The postictal decrease in baseline ventilation and impaired HCVR was not because of a decrease in metabolic rate
Ventilation is normally coupled to metabolic rate (Davis et al., 1980; Newstead, 1987; Mortola and Frappell, 2000; Richerson and Boron, 2003; Mortola and Maskrey, 2011), and a decrease in baseline ventilation and a blunted HCVR could either be because of impaired chemoreception or a decrease in metabolism. To determine whether the respiratory changes could be explained by changes in metabolism, we measured VO2 during a subset of plethysmography experiments and calculated VE/VO2 (Fig. 4). In the subset of DS mice with severe blunting of the HCVR postictally (n = 11 of 23), VE/VO2 in normocapnia (0% CO2) was 54% lower following heat-induced seizures compared with preictal (8.336 ± 4.915 vs 17.927 ± 5.707, p = 0.001952; Fig. 4A). The change of VE/VO2 in response to an increase in CO2 from 0% to 7% was decreased by 47% following heat-induced seizures (p = 0.0043; Fig. 4B). In contrast, VO2 did not change significantly postictally in normocapnia (0.133 ± 0.094 vs 0.099 ± 0.026 ml O2/min, p = 0.6377; Fig. 4C), indicating that there was a decrease in baseline ventilation relative to metabolic rate (i.e., hypoventilation). Thus, the decrease in VE was largely because of a decrease in chemoreception. VO2 was significantly lower postictally in 7% CO2 (0.039 ± 0.036 vs 0.066 ± 0.027 ml O2/min, p = 0.005851; Fig. 4C).
The effect of seizures on the HCVR was not because of a change in metabolism. A, Summary of VE/VO2 data from DS mice shown in Figure 3B with severe blunting of the HCVR (n = 11). B, Data from A expressed as ΔVE/VO2 in response to increase in CO2 from 0% to 7% in DS mice before (preictal) and after (postictal) heat-induced seizures. ΔVE/VO2 was significantly reduced in DS mice following heat-induced seizures consistent with blunting of chemoreception rather than a decrease in metabolism. C, Summary of VO2 data from DS mice in A. D, Summary of VE/VO2 data from DS mice shown in Figure 3D with mild blunting of the HCVR (n = 11). One mouse was excluded because of incomplete collection of VO2 data. E, Data from D expressed as ΔVE/VO2 in response to 7% CO2, which was not significantly reduced following heat-induced seizures. F, Summary of VO2 data from DS mice in D. Statistical analyses for A, D as well as C, F were done with Wilcoxon matched-pairs signed-rank tests with correction for multiple comparisons using the Holm–Sidak method. Paired t tests were used for B, E. **p < 0.01.
In the subset of DS mice with mild blunting of the HCVR postictally (n = 11 of 23; one mouse was excluded because of incomplete collection of VO2 data.), VE/VO2 in normocapnia was 57% lower following heat-induced seizures compared with preictal (7.856 ± 4.869 vs 18.193 ± 4.975, p = 0.001952; Fig. 4D). There was no effect of seizures on the change in VE/VO2 induced by 7% CO2 in the mild group (p = 0.7647; Fig. 4E) and VO2 did not change significantly postictally in normocapnia (0.174 ± 0.146 vs 0.084 ± 0.015 ml O2/min, p = 0.1231; Fig. 4F). However, postictal VO2 was significantly lower in 7% CO2 (0.046 ± 0.022 vs 0.057 ± 0.020 ml O2/min, p = 0.001952; Fig. 4F).
Postictal hypothermia and impairment of the HCVR were not because of preictal hyperthermia
To determine whether impairment of the HCVR and thermoregulation after heat-induced seizures in DS mice is also seen in other mouse seizure models, and to rule out the possibility that the decrease in body temperature was a response to the hyperthermia used to induce seizures, we used D/+ mice for their susceptibility to nonfatal GCS in response to an audiogenic stimulus (Wengert et al., 2021). We measured the HCVR in P30–P37 D/+ mice (n = 9) instrumented with Emitter G2 abdominal temperature telemetry probes before and after a series of evoked GCS. We used an older age for D/+ mice than DS mice as they are significantly smaller in size, and thus could not be instrumented with E-mitters for temperature monitoring until they were older. Because seizures in D/+ mice are relatively short compared with those in DS mice, we evoked a GCS every 5 min for a total of four seizures. All seizures qualified as R5 on a modified Racine scale (Racine, 1972), and consisted of a short period of wild-running and loss of postural tone, immediately followed by a pronounced tonic phase associated with hindlimb extension and S-IRA that lasted ∼5 s (Fig. 5A). None of the seizures were fatal. There was a significant decrease in ventilation in 7% CO2 postictally (VE = 5.633 ± 0.871 vs 8.154 ± 1.753 ml/min postictal vs preictal, p = 0.007797) but not in normocapnia (VE = 2.088 ± 0.577 vs 2.199 ± 0.956, p = 0.9102) in D/+ mice (n = 9; Fig. 5B). The HCVR was decreased postictally by 40% compared with preictal (ΔVE/ΔCO2 = 3.545 ± 0.738 vs 5.955 ± 1.059 ml/min/7% CO2, p < 0.0001; Fig. 5C). Interestingly, core body temperature dropped slightly after each evoked seizure, and ∼30 min after the last evoked GCS, core body temperature was 1.3 ± 0.6°C lower than preictal (Fig. 5D). Mice were returned to their home cages on telemetry receivers and their core body temperature returned to preictal levels ∼1.5 h after the last evoked GCS (Fig. 5D).
Nonthermogenic seizures also caused impairment of the HCVR and thermoregulation in Scn8aN1768D/+ and Scn1aR1407X/+ mice. A, Representative plethysmography trace of a seizure evoked (blue arrow) in a D/1 mouse with an audiogenic stimulus (blue dotted line) leading to apnea. B, VE at 0% and 7% CO2 before and after four evoked GCS in D/+ mice (n = 9; Wilcoxon matched-pairs signed-rank test, **p < 0.01). C, Data from B expressed as HCVR (VE/ΔCO2) in D/+ mice before and after four evoked GCS. The HCVR was significantly reduced in D/+ mice following audiogenic seizures (paired t test, ****p < 0.0001). D, D/+ mice displayed a transient drop in body temperature following four stimulus-evoked GCS (black dotted lines). Solid line is the mean body temperature and the shaded regions are the SE (n = 8). E, Representative trace showing a transient drop in body temperature occurred following a spontaneous nonfatal seizure in a female P33 DS mouse. Animal was instrumented with Emitter G2 temperature telemetry probe and placed in home cage for continuous surveillance. The gray bar shows the time of a 50-s-long spontaneous R5 seizure. Note that activity monitor shows that the drop in temperature was not because of the animal sleeping. The inset in the top right shows 12 h of recording including the 4 h from the main figure.
To confirm the conclusion that postictal hypothermia was not caused by preictal hyperthermia we monitored video and body temperature of DS mice implanted with Emitter G2 telemetry probes. Two mice (of n = 13) had spontaneous nonfatal R5 level seizures with hindlimb extension and subsequently had a transient decrease in their body temperature by 0.85°C and 1.35°C (representative trace of a P33 DS mouse shown in Fig. 5E).
Depletion of serotonin with PCPA decreased the interictal HCVR and core body temperature of DS mice
To determine whether 5-HT depletion replicates the observed postictal defects in CO2 chemoreception and thermoregulation in DS mice, we used the tryptophan hydroxylase inhibitor PCPA, which reduces brain 5-HT to <12% of baseline levels after three to four daily treatments (Murray et al., 2015). DS mice (P29–P35) were exposed to 0% and 7% CO2 during plethysmography to obtain a baseline HCVR, after which they were treated with either PCPA or vehicle once per day for four consecutive days. Older DS mice were used for this experiment as the HCVR is not fully developed at younger ages. Preliminary experiments in P21–P25 DS mice showed that the HCVR did not plateau until ∼P29, which was consistent with previous data obtained in a different WT mouse strain showing an age dependence of the HCVR over a similar range (Cerpa et al., 2017). Mice were housed in home cages with plenty of bedding to prevent hypothermia, which is known to cause insomnia and might have increased seizures (Murray et al., 2015). On day four, a preictal HCVR was obtained approximately 1 h following the final injection, a time at which body temperature is decreased after PCPA injection (Murray et al., 2015). In DS mice treated with vehicle (n = 21), there was no difference in the HCVR at 4 d compared with baseline (ΔVE/ΔCO2 = 3.711 ± 0.931 vs 3.761 ± 0.935 ml/min/7% CO2, p = 0.8624; Fig. 6A). In contrast, the HCVR in PCPA-treated mice (n = 21) was significantly reduced compared with baseline (2.509 ± 0.798 vs 3.643 ± 0.629 ml/min/7%CO2, p = 0.000118; Fig. 6A). Core body temperature was also significantly reduced after PCPA treatment compared with vehicle (34.6 ± 1.3 vs 37.2 ± 0.6°C, p < 0.00001) after mice were placed in a cage without bedding at an ambient temperature of 22–24°C (Fig. 6B), as reported previously (Murray et al., 2015).
Serotonin depletion in DS mice impaired the interictal HCVR and thermoregulation, partially occluded postictal impairment of the HCVR, and increased seizure-induced mortality. A, The HCVR (ΔVE/ΔCO2) was significantly lower in DS mice after treatment with PCPA for 4 d relative to baseline (paired t test). B, Core body temperature was significantly lower in DS mice treated with PCPA for 4 d (unpaired t test, n = 21 per group). C, ΔVE in response to 7% CO2 at baseline, after 4 d of treatment with either vehicle (n = 16) or PCPA (n = 14), and after a heat-induced seizure. HCVR was significantly lower after 4 d of PCPA treatment compared with vehicle (unpaired t test). In contrast, postictal HCVR was not significantly different between treatment groups. Data plotted in panel C is the same as in panel A except that mice that died during seizures were excluded. D, Difference in HCVR after heat-induced seizures between treatment groups (post-HCVR – pre-HCVR). PCPA-treated DS mice displayed a smaller reduction in postictal HCVR compared with the vehicle group, indicating PCPA partially occluded the postictal HCVR depression observed in untreated animals (unpaired t test). E, Core body temperature in DS mice treated with either PCPA (n = 10, green) or vehicle (n = 8, black) up to 8 h following a heat induced seizure (black arrow). PCPA-treated DS mice displayed thermoregulatory dysfunction following heat-induced seizures that was more severe than vehicle-treated mice. All of these animals survived. F, DS mice treated with PCPA had a higher incidence of seizure-induced mortality (n = 21 out of 48) compared with those treated with vehicle (n = 8 out of 41). Two mice in the PCPA-treated group died after spontaneous seizures before induction of a seizure by hyperthermia. Fisher's exact test. *p < 0.05, **p < 0.01, ****p < 0.0001.
Serotonin depletion partially occluded postictal impairment of the HCVR and worsened postictal hypothermia
After obtaining an HCVR ∼1 h following the final treatment of DS mice with PCPA or vehicle, a seizure was induced with hyperthermia. The average temperature at which DS mice developed a convulsive seizure was not significantly different between treatment groups (39.0 ± 0.9°C vs 39.4 ± 0.7°C after PCPA vs vehicle, p = 0.2268). The duration of most R5 seizures was relatively short (typically <2 min) and was equivalent between PCPA and vehicle-treated animals. Three of 21 DS mice in the vehicle group died from S-IRA and two of 21 had to be excluded after developing status epilepticus. In contrast, seven of 21 DS mice in the PCPA treatment group died from S-IRA. In animals in which the heat-induced seizure was not fatal, a postictal HCVR was obtained. The HCVR in vehicle-treated DS mice (n = 16 of 21) following a heat-induced GCS was 54% smaller than preictal (ΔVE/ΔCO2 = 1.757 ± 0.654 vs 3.788 ± 0.974 ml/min/7% CO2, p < 0.0001; Fig. 6C). PCPA-treated animals (n = 14 of 21) displayed a postictal HCVR that was 48% smaller than preictal (1.306 ± 0.441 vs 2.524 ± 0.788, p = 0.00027). The preictal HCVR was significantly smaller in PCPA-treated DS mice compared with that in the vehicle group (p = 0.001784). The postictal HCVR was also smaller in PCPA-treated animals compared with vehicle, but this was not statistically significant (p = 0.07436). However, PCPA-treated DS mice displayed a smaller reduction in postictal HCVR compared with vehicle group, calculated as post-HCVR – pre-HCVR, indicating PCPA partially occluded the postictal HCVR depression observed in vehicle-treated animals (p = 0.043; Fig. 6D). There was a significantly larger postictal drop in body temperature in PCPA-treated DS mice that persisted >8 h (28.9 ± 4.6°C with PCPA vs 36.9 ± 0.3°C with vehicle at 8 h; p = 0.00041; Fig. 6E). Hypothermia lasted for >12 h in five of 14 mice. All mice survived and recovered from hypothermia provided they were treated with saline to maintain hydration (0.1 ml/g, i.p., q 12 h). Seizure burden, as determined by duration of total seizure activity classified as Racine scale 5 and above, did not correlate with severity of postictal HCVR depression calculated as a percentage of preictal HCVR in PCPA (R2 = 0.1359, n = 5, p = 0.5415) or vehicle-treated (R2 = 0.1785, n = 7, p = 0.3450) mice (data not shown).
As above, we observed a higher incidence of seizure-induced mortality in DS mice treated with PCPA following heat-induced seizures (fatality rate 33% after PCPA vs 14% after vehicle). To further investigate this, additional cohorts of DS mice (P22–P35) were treated with either PCPA or vehicle for four or five consecutive days and their susceptibility to seizure-induced mortality was assessed. Overall, 20% of DS mice treated with vehicle (n = 8 out of 41) died following a heat-induced seizure. Treatment with PCPA significantly increased mortality to 44% (n = 21 of 48, p = 0.0227; Fig. 6F). Within the PCPA-treated group, younger DS mice (P22–P25) treated for 5 d had a 64% fatality rate because of seizures (n = 7 out of 11) compared with 29% in vehicle-treated littermates (n = 2 out of 7). Older DS mice (P29–P35) treated for 5 d with PCPA had a 44% fatality rate (n = 7 out of 16) compared with 23% with vehicle (n = 3 out of 13). Two of the seven deaths from the P22–P25 age cohort were because of a spontaneous seizure after four consecutive days of treatment with PCPA.
Changes in body temperature are known to alter the ventilatory response to O2 and CO2 (Maskrey, 1990). To verify that the observed decrease in the HCVR after seizures and after treatment with PCPA was not an indirect effect of the decrease in core body temperature, a cohort of DS mice instrumented with Emitter G2 telemetry temperature probes was treated with PCPA for 4 d as described above (n = 4). On day 4, core body temperature and the HCVR were measured approximately 1 h after the last PCPA treatment. The HCVR was obtained again in the same animals after cooling (∼32°C), and then again after heating (∼37°C) their core body temperature (actual body temperature was 32.4 ± 0.5°C, then 36.9 ± 0.3°C). As expected (Mortola and Frappell, 2000; Mortola and Maskrey, 2011), decrease in body temperature increased VE (Fig. 7A), the opposite of the effect of seizures and PCPA, and the HCVR in PCPA-treated DS mice did not change significantly by altering body temperature (ΔVE/ΔCO2 = 1.668 ± 1.142 vs 1.826 ± 0.692 ml/min/7% CO2 at 32°C vs 37°C, n = 4, p = 0.625; Fig. 7B).
Hypothermia did not directly cause a decrease in HCVR in PCPA-treated DS mice. Body temperature was altered in PCPA-treated DS mice (P34, n = 4) by ambient cooling or heating. The same mice were used for both conditions. A, Minute ventilation (VE) in 0% or 7% CO2 was changed by cold or warm conditions, but there was no change in B the HCVR (ΔVE/ΔCO2) Wilcoxon matched-pairs signed-rank test, p = 0.6250. Mean ± SEM are plotted for each temperature challenge.
Fenfluramine reduced postictal dysfunction of ventilation and thermoregulation in DS mice
DS mice (P27–P36) with Emitter G2 abdominal telemetry temperature probes were used to measure a baseline HCVR, after which they were treated with either fenfluramine (20 mg/kg, i.p.) or saline ∼45 min before seizure induction. DS mice were used at this age range to allow comparison to the PCPA experiment. The average temperature at which DS mice developed a convulsive seizure was not significantly different between treatment groups (37.1 ± 1.2°C after fenfluramine vs 37.9 ± 1°C after vehicle, p = 0.0873). All animals were placed on a cooling block for 3–5 min to terminate seizures. Similar to the findings above, DS mice treated with vehicle had a significant decrease in ventilation in normocapnia postictally (VE = 0.836 ± 0.379 vs 1.528 ± 0.528 ml/min postictal vs preictal, p = 0.0040; n = 8). In contrast, there was no difference in ventilation in normocapnia postictally in fenfluramine-treated animals (VE = 1.648 ± 0.933 vs 1.573 ± 0.395 ml/min postictal vs preictal, p = 0.7576; n = 10; Fig. 8A). The HCVR in vehicle-treated DS mice following a heat-induced seizure was 49% lower than preictal (ΔVE = 2.067 ± 0.867 vs 4.027 ± 1.865 ml/min/7% CO2 postictal vs preictal, p = 0.01351), whereas fenfluramine-treated animals displayed a postictal HCVR that was 29% lower than preictal (ΔVE/ΔCO2 = 2.748 ± 0.352 vs 3.867 ± 1.323 ml/min/7% CO2, p = 0.0241), which was significantly higher than the vehicle group (p = 0.03699; Fig. 8B). There was also a significantly smaller postictal drop in body temperature in fenfluramine-treated DS mice compared with control animals (−2.2 ± 1.1°C with fenfluramine vs −5.6 ± 1.1°C with vehicle, p = 0.0003; Fig. 8C).
Fenfluramine reduced seizure-induced respiratory dysfunction and hypothermia. A, Minute ventilation (VE) at 0% and 7% CO2 before (pre) and after (post) hyperthermia in DS mice treated with fenfluramine (Fen, 20 mg/kg, n = 10) or saline (Ctl, n = 8) 45 min before hyperthermia. Fenfluramine-treated DS mice displayed less respiratory depression after heat-induced seizures at normocapnia and hypercapnia compared with saline-treated animals (paired t tests). B, Data from A displayed as HCVR (ΔVE/ΔCO2) in response to increase in inhaled CO2 from 0% to 7% in DS mice treated with either fenfluramine or saline before (preictal) and after (postictal) heat-induced seizures. The HCVR was significantly more depressed in control DS mice following heat-induced seizures compared with fenfluramine-treated animals (unpaired t test). C, DS mice treated with fenfluramine (n = 8, pink) had a smaller drop in body temperature (ΔBT) in the first hour postictally (shaded box) than saline-treated mice (n = 7, black; Mann–Whitney test). Solid lines are the mean body temperature and the shaded regions are the SE. Inset shows individual values and mean drop in body temperature (ΔBT ± SEM) for the shaded period compared with baseline temperatures for the same groups of animals. *p < 0.05, **p < 0.01, ***p < 0.001.
Discussion
It is now generally accepted that a large subset of SUDEP cases are due to respiratory arrest after generalized seizures (Ryvlin et al., 2013; Teran et al., 2022), making it important to understand what makes breathing unstable during the postictal period. It is not clear how seizures cause respiratory dysfunction, but it has been hypothesized that inhibition of serotonergic neurons leads to impaired CO2 chemoreception and this contributes to the pathophysiology of SUDEP (Massey et al., 2014; Kim et al., 2018; Sainju et al., 2019). It has previously been shown that seizures inhibit a subset of serotonergic neurons in rats (Zhan et al., 2016), but it was not clear whether that subset was involved in control of breathing. Here, we provide the first direct evidence that seizures cause a prolonged postictal decrease in baseline ventilation, central CO2 chemoreception and thermoregulation, three changes expected to occur if serotonergic neurons involved in breathing are inhibited (Hodges et al., 2008; Ray et al., 2011). We also provide evidence that this may be because of inhibition of 5-HT neurons by seizures.
In MORTEMUS, breathing was present in all subjects after convulsions ended and continued for 1–2 min before apneas began (Ryvlin et al., 2013). Various studies have examined postictal features of nonfatal seizures such as postictal generalized EEG suppression (PGES) and PCCA, relating these findings to the risk of SUDEP. However, PGES occurs immediately on termination of seizures and has been reported to have an average duration of only ∼1.5 min for patients who later died of SUDEP (Lhatoo et al., 2010). PCCA is also short in duration (mean of 12 s) and often occurs immediately after seizures end (Vilella et al., 2019). Thus, PGES and PCCA may be biomarkers for SUDEP, but their timing is not what is expected for the cause of death. A pathophysiological cascade of processes that is causative for SUDEP is likely to be initiated during or shortly after seizures, but events that are critical for death must continue for much longer after the end of a seizure to be fatal. Thus, changes that cause SUDEP are likely to last at least tens of minutes to explain the findings seen in MORTEMUS.
Many studies of peri-ictal respiratory function in epilepsy patients and animal models of SUDEP focus on measurement of apnea and respiratory rate (Faingold et al., 2010; Kuo et al., 2019; Wenker et al., 2021). However, there are other breathing abnormalities that can also be severely pathologic. Seizures appear to frequently elicit a form of postictal respiratory dysfunction short of frank apnea, where ventilation is decompensated for a prolonged period and fails to meet metabolic demands (Orringer et al., 1977). For example, a DS patient previously described who later died of SUDEP had an elevated tcCO2 that persisted for up to 3 h after the last seizure (Kim et al., 2018). However, fR remained near normal and apneas were infrequent and short, emphasizing that fR or apneas alone give an incomplete assessment of breathing, and that measuring VE is a more accurate way to assess gas exchange. The elevated tcCO2 in that patient indicated significant hypoventilation, which parallels what we observed here in DS mice. Similarly, none of the epilepsy patients reported here had PCCA, but had a decreased HCVR for approximately 4 h following GCS. Thus, the effects of seizures on tonic chemoreceptor drive (Fink, 1961; Richerson and Boron, 2003; Dubois et al., 2016) and CO2 chemoreception extend for several hours, although the seizures were remarkably short in comparison.
Hypoventilation is defined as a decrease in ventilation out of proportion to metabolic demand, and by definition leads to an increase in Pco2 (Randerath et al., 2017). We found that postictal ventilation in normocapnia was decreased to 54% of baseline in DS mice. We also quantified metabolic rate in DS mice by measuring VO2 and found it was increased postictally in normocapnia, possibly because of preceding convulsive activity (Schoknecht et al., 2017). The decreased VE/VO2 observed therefore indicates impaired ability to match VE with metabolic rate, and would lead to increased Pco2. GCS are characterized by vigorous muscular exercise, acute metabolic acidosis secondary to lactic acid accumulation (Orringer et al., 1977), increased metabolic activity in the brain, and hypoxia. Each of these effects would be expected to increase ventilation after a GCS, and yet we observed a decrease. These findings are consistent with postictal respiratory depression primarily being caused by a defect in CO2 chemoreception and decreased tonic chemoreceptor drive, which could result from seizure-induced inhibition of serotonergic neuron firing (Zhan et al., 2016).
Previous studies have shown that selective elimination or silencing of serotonergic neurons leads to blunting of CO2 chemoreception and failure to maintain core body temperature when exposed to cold, mainly because of a decrease in shivering and hypometabolism of brown adipose tissue (Hodges et al., 2008; Ray et al., 2011). Similar to effects of 5-HT system inhibition, heat-induced seizures in DS mice and audiogenic seizures in D/+ mice induced postictal impairment of CO2 chemoreception and thermoregulation.
We considered the possibility that the effect of seizures on body temperature in DS mice induced an artifact in measuring the HCVR. However, altering body temperature in PCPA-treated DS mice from 32–37°C decreased ventilation at both 0% and 7% CO2, but did not change ΔVE/ΔCO2 significantly. At the time postictal measurements were made in DS mice following hyperthermia, body temperature was dropping. As shown in Figure 7, this would have led to an artifactual increase in VE at 7% CO2, which would overestimate the HCVR after seizures. Thus, the resulting error would underestimate the blunting of the HCVR. We also observed blunting of the HCVR and prolonged hypothermia in D/+ mice after inducing seizures with an audiogenic stimulus, and body temperature decreased after spontaneous seizures in DS mice. Collectively, these results indicate that the decrease in HCVR and impaired thermoregulation are not because of the hyperthermia used to induce seizures in DS mice, and the impairment of these homeostatic functions occurs in at least two mouse models of generalized seizures. In adult human patients a drop in body temperature would not be expected given that the large heat capacity and small surface-to-volume ratio of the human body would necessitate a long time for core body temperature to drop.
We further evaluated whether seizures in DS mice caused inhibition of the 5-HT system by using subacute depletion of 5-HT with PCPA. This impaired the interictal HCVR and thermoregulation in DS mice, exacerbated seizure-induced hypoventilation and hypothermia, and partially occluded the decrease in the HCVR induced by seizures. These findings support the hypothesis that seizures induce prolonged postictal impairment of serotonergic neuron function. They also provide additional evidence that blunting of chemoreception is not an indirect effect of hypothermia, since seizures induced after PCPA treatment caused a smaller additional decrease in the HCVR than after vehicle treatment, despite a much more severe drop in body temperature. Instead, impaired chemoreception and thermoregulation are independent effects, both likely because of 5-HT system dysfunction.
The present study also found that fenfluramine administration significantly reduced postictal ventilatory depression, and rescued the HCVR and thermoregulatory impairment induced by seizures. Fenfluramine enhances the serotonin system in multiple ways, including augmenting carrier-mediated release of 5-HT and preventing its reuptake (Fuller et al., 1978; Baumann et al., 2014), and agonist effects on 5-HT receptors (Tupal and Faingold, 2021). It prevents S-IRA in DBA/1 mice (Tupal and Faingold, 2019) and was recently approved to treat seizures in DS patients. There may also be a reduced incidence of SUDEP in DS patients taking fenfluramine (Cross et al., 2021).
Focal and generalized seizures both impair breathing and the HCVR, whereas monitored and witnessed SUDEP almost always occurs after generalized seizures (Ryvlin et al., 2019). The difference may lie in the need for a depressed level of consciousness, which may be required for SUDEP. 5-HT is necessary for arousal to hypercapnia (Buchanan and Richerson, 2010; Smith et al., 2018), which can also be impaired after seizures, and usually more severely after generalized convulsive seizures.
The results presented here support a hypothesis that seizures decrease serotonergic drive to the respiratory network. This would lead to withdrawal of multiple stimulatory effects on breathing in addition to reducing the HCVR, including a decrease in tonic chemoreceptor drive (Mahamed et al., 2001; Teran et al., 2014), and decreased excitability of respiratory neurons (Richter et al., 2003; Ptak et al., 2009) including respiratory motor neurons (Holtman et al., 1986; Talley et al., 1997). A decrease in 5-HT tone would also impair autoresuscitation (a.k.a. gasping), which is dependent on 5-HT (Tryba et al., 2006; Cummings, 2021). Data from MORTEMUS indicated that none of the subjects displayed robust gasping, which should have occurred in response to asphyxia. The results presented here raise the possibility that postictal inhibition of the HCVR may be because of impaired function of the 5-HT system and may be a noninvasive biomarker of SUDEP risk (Kim et al., 2018; Sainju et al., 2019).
Footnotes
We thank Dr. Miriam Meisler (University of Michigan) for providing the B6.Scn8aN1768D/+ mice, Xiuqiong Zhou for mouse husbandry, and Lori Smith for technical contributions. We also thank Deidre N. Dragon, Harold Winnicke, Deanne Tadlock, and Tammy Bryant along with all other epilepsy monitoring unit (EMU) staff members for their contribution to the patient study. NINDSF31 – NS110333 (FAT), NIH/NINDS – R01 NS113764 (GBR & BKG), and NIH/NINDS – U01 NS090414 (GBR).
The authors declare no competing financial interests.
- Correspondence should be addressed to Frida A. Teran at frida-teran{at}uiowa.edu
