Progressive cardiorespiratory dysfunction in Kv1.1 knockout mice may provide temporal biomarkers of pending sudden unexpected death epilepsy (SUDEP): the contribution of orexin

Abstract

Objective:

Immediately preceding Sudden Unexpected Death in Epilepsy (SUDEP), patients experienced a final GTC seizure, rapid ventilation, apnea, bradycardia, terminal apnea and asystole. Whether a progressive pathophysiology develops and increases risk of SUDEP remains unknown. Here, we determined (i) heart rate, respiration and blood oxygen saturation (SaO₂) in low risk and high risk KO mice, and (ii) whether blocking receptors for orexin, a cardiorespiratory neuromodulator, influences cardiorespiratory function mice or longevity in high risk KO.

Methods:

Heart rate and SaO₂ were determined noninvasively with ECGenie and pulse oximetry. Respiration was determined with non-invasive airway mechanics technology. The role of orexin was determined within subject following acute treatment with a dual orexin receptor antagonist (DORA, 100 mg/kg). The number of orexin neurons in the lateral hypothalamus was determined with immunohistochemistry.

Results:

Intermittent bradycardia was more prevalent in high risk KO mice, an effect that may be the result of increased parasympathetic drive. High risk KO mice had more orexin neurons in the lateral hypothalamus. Blocking orexin receptors differentially influenced heart rate in KO, but not WT mice. When DORA administration increased heart rate, it also decreased HRV, breathing frequency and/or hypopnea-apnea. Blocking orexin receptors prevented the methacholine (MCh)-induced increase in breathing frequency in KO mice and reduced MCh-induced seizures, via a direct or indirect mechanism. DORA improved oxygen saturation in KO mice with intermittent hypoxia. Daily administration of DORA to high risk KO mice increased longevity.

Significance:

High risk KO mice have a unique cardiorespiratory phenotype that is characterized by progressive changes in five interdependent endpoints. Blocking orexin receptors attenuates some of these endpoints and increases longevity, supporting the notion that windows of opportunity for intervention exist in this preclinical SUDEP model.

INTRODUCTION

Epilepsy is a chronic neurological disorder characterized by recurrent and unprovoked seizures and affects approximately 50 million people globally¹. Sudden Unexpected Death in Epilepsy (SUDEP) is one of the leading causes of mortality in epilepsy affecting 1:1000 people with epilepsy every year, with the risk increasing to 1:150 in those with refractory seizures². Risk factors for SUDEP include refractory seizures, severe generalized tonic clonic (GTC) seizures, cognitive impairments and cardiac arrhythmias^2–8. The multi-center MORTEMUS clinical study reported that immediately prior to SUDEP, patients experienced a final GTC seizure, rapid ventilation, apnea, bradycardia, terminal apnea and asystole⁴. All patients that succumbed to SUDEP experienced previous severe seizures earlier in their lives and the sequence of events following the seizure resulted in recovery. Whether pathophysiology progressively develops that overtime increases the probability of SUDEP remains unknown. Here, we tested the overall hypothesis that endpoints progressively change as risk for sudden death increases in a pre-clincial SUDEP model. Endpoints were selected based on their occurrence in SUDEP patients prior to death, specficially heart rate, breathing rate, rapid breathing induced apnea, hypopnea-apnea, and severe seizures. Identification of temporal biomakers would allow endpoints to be monitored and provide windows of opportunity for intervention in attempts to postpone or prevent SUDEP.

To test this hypothesis, the Kv1.1 knockout (KO) mice SUDEP model was used. KO mice model the loss of Kv1.1 protein function that is associated with human SUDEP^9,10. These mice have severe spontaneous recurrent seizures^11–14. Similar to humans, KO mice experience SUDEP risk factors including multiple GTC seizures, cognitive impairments and cardiac arrhythmias^12–15. KO mice are the only SUDEP model with a predictable lifespan and 100% mortality. The median age of sudden death is postnatal day (P) 46 and referred to as SD₅₀ (or when 50% of mice have died). The mean age of sudden death is P45 ± 1 day (Figure 1A; updated from¹²). With this model it is feasible to monitor endpoints at different ages relative to the probability of death.

Figure 1. Intermittent bradycardia increases as the probability of death increases.

(A) Survival curve of the KO colony. The median age of sudden death is postnatal day (P) 46 and is referred to as SD₅₀ (or when 50% of mice have died). The mean age of sudden death is P45 ± 1 day. (B) A representative ECG trace using non-invasive ECGenie recording techniques in non-anesthetized mice. P-Q-R-S-T components of the ECG waveform are indicated. (C) Scatterplot of the mean heart rate (HR) of ECG segments from all KO and age-matched WT mice in the first two columns. ECG segments at individual ages from P28-P51 KO mice are depicted. Bradycardia segments are highlighted in the shaded region. Intermittent bradycardia is more apparent in older KO mice (Brown-Forsythe test: F (17,1665) = 58.91, p < 0.0001). HR of WT littermates remained stable with age and were thus pooled (n = 359 WT segments and 712 KO segments from 4–5 mice/genotype/age). (D) HR binned by probability of sudden death (SD₂₀, SD₃₅ and SD₅₅). Mean HR of SD₃₅ and SD₅₅ KO mice were decreased when compared to WT (F (1,91) = 50.09, p < 0.0001), and within group across age (F (2,91) = 14.73, p < 0.0001; Two-way ANOVA with Sidak’s multiple comparison post hoc text). (E) Short-term heart rate variability (HRV) positively correlates with increased probability of death in KO mice as evidenced by increased rmSSD when compared between and within groups (F (1, 90) = 23.38, p < 0.001; Two-way ANOVA with Sidak’s post hoc test). (F) Bar graph depicting the change in heart rate following administration of atropine (n = 4–5). Red ‘+’ indicates differences within KO group across age, ⁺p < 0.05, ⁺⁺⁺ p < 0.001, ⁺⁺⁺⁺ p < 0.0001. Blue * indicates statistical difference from WT cohorts. * p < 0.05, ** p < 0.01, **** p < 0.0001. ECG, electrocardiogram; KO, Kv1.1 knockout; rmSSD, root mean square of the successive differences; SD_#, indicates # percent of the colony has experienced sudden death; WT, wild type.

We have previously reported that elevated breathing frequency and apnea are apparent in KO mice and worsen thru SD₉₀ (when 90% of mice have died)¹⁶. Here, we tested three hypotheses. Our first hypothesis tested whether intermittent bradycardia would be apparent in KO mice with higher probability of death (i.e. high risk). We report that high risk KO mice have intermittent bradycardia. The second hypothesis focused on a mechanism upstream of bradycardia, increased respiration, apnea and seizures. We have previously reported that there is an increase in the number of cells expressing the cardiorespiratory neuromodulator orexin in high risk KO mice¹⁷. Heart rate, apnea, breathing frequency and seizures were determined within subject during vehicle treatment, then again 48-hrs later following administration of a dual orexin receptor antagonist (DORA). In high risk KO mice with bradycardia, acute DORA treatment increased heart rate, reduced breathing frequency and the number of KO mice having apnea and seizures. Finally, we tested the third hypothesis that initiating daily treatment with a DORA on SD₅₀, i.e. a time when multiple endpoints have changed, would postpone SUDEP in KO mice. Daily DORA treatment did increase longevity in high risk KO mice.

MATERIALS AND METHODS

Animals:

Kcna1^+/− mice (on the C3HeB/FeJ background strain) were bred in a quiet room in the Animal Resource Facilities at Creighton University. Both male and female mice were used in this study. To increase rigor, WT and KO cohorts were age-matched and paired to experience the treatments and tests at the same time on the same days. All mice were on a 12 hour light/dark cycle and provided with food and water ad libitum. Experiments were conducted between zeitgeber time 03:00–09:00 (00:00 is lights on) to avoid potential confounding circadian variability. The genotype of the mice was determined with tail clips using Transnetyx, Inc (Cordova, TN, USA). All the experiments were carried out in accordance with the guidelines of the National Institutes of Health and approved by the Institutional Animal Care and Use Committee at Creighton University.

Experimental Designs:

Experiment Ia: In the first cohort, heart rate and heart rate variability (HRV) were determined in KO and WT littermates at P28–51. Data from all ages were analyzed or binned into three groups: P28–36, P37–43 and P44–51 when ~20%, ~35% and ~55% of KO mice have succumbed to sudden death (SD), respectively (i.e. SD₂₀, SD₃₅, SD₅₅ calculated from the colony survival curve, Figure 1A). Mice were euthanized after testing. Experiment Ib: In a separate cohort of SD₅₅ KO and age-matched WT mice, baseline ECG was recorded for 30 min. Fifteen minutes following administration of the parasympathetic acetylcholine muscarinic receptor antagonist, atropine sulphate (3mg/kg), heart rate was recorded for 30 min. Mice were euthanized after testing.

Experiment IIa: In a separate cohort of age-matched WT and KO mice, the number of orexin neurons in the lateral hypothalamus was determined using immunohistochemistry. Orexin cell number was determined in WT and KO mice at two SD₀ ages (prior to epilepsy onset on P12, and during the week of epilepsy onset on P21), SD₁₀ and SD₃₀. Experiment IIb: This within subject study was conducted in a separate cohort. KO and WT mice were injected with vehicle (VEH) on either SD₂₀ (P30–32), SD₃₅ (P38–40), or SD₅₅ (P44–49). Three different groups were used for testing at each age. KO mice were not selected a priori, those that survived to the respective age were used. Thirty minutes following injection, ECG was monitored for 30 min, then breathing frequency, hypopnea and apnea were determined. Forty-eight hours later, the dual orexin receptor antagonist (DORA) ACT-078573 (100 mg/kg, i.p.) was administered (donated by Actelion; Allschwil, Switzerland) (VEH was 50/50 DMSO/sterile saline). Thirty minutes after injection, ECG endpoints were recorded for 30 min, then respiratory endpoints were recorded. Based on the pharmacokinetics of DORA, cardiorespiratory endpoints were monitored during peak plasma concentration (30 min – 3 hrs)^18,19. Mice were euthanized after testing. Experiment IIc: In a separate cohort of SD₃₅ KO and age-matched WT mice, baseline ECG was recorded for 30 min, then again following administration of atropine sulphate (3mg/kg). The following day, DORA (100 mg/kg, i.p.) was administered acutely. After 30 min, baseline heart rate was recorded for 30 min, then again following administration of atropine sulphate (3 mg/kg).

Experiment III: To determine whether late-onset DORA treatment influenced longevity, DORA (200 mg/kg, i.p.) was administered daily in a separate cohort of KO mice starting at P45 until natural mortality. New survival curves for mice that had survived to P45 were generated and the median and mean ages of mortality were calculated.

Electrocardiography (ECG):

ECG was recorded non-invasively in non-anesthetized, conscious KO and WT littermates using the ECGenie recording platform (Mouse Specifics, Inc., Boston, MA, USA). Heart rate and HRV is dependent on the state of vigilance. Thus, all ECG recordings were conducted when mice were awake. Briefly, mice were removed from their home cage, placed on the ECGenie recording platform. ECG signals were recorded for 30 mins at a sampling rate of 2 kHz (Figure 1B). Segments were sampled equidistantly (approximately every minute) during the recording session. To determine the incidence of intermittent bradycardia or tachyarrhythmia, segments of at least 25–75 P-Q-R-S-T complexes were manually reviewed to ensure they were noise free; cardiac abnormalities including premature ventricular contraction, atrioventricular block, QT prolongation and sinus exit block have been reported in the KO mice²⁰. Mean heart rate was determined for each segment for each animal using the e-MOUSE software (Mouse Specifics, Inc., Boston, MA, USA) and signal fidelity was manually verified. The mean heart rate for WT was 789 ± 48 bpm (mean ± SD) with a range of 628–874 bpm. In humans, bradycardia is defined as heart rate less than the normal range (i.e. 60 bpm for humans). The normal range includes 90–95% of values distributed around the mean heart rate of control subjects. In the study herein, 95% of values fell within the range defined by twice the standard deviation of the mean for each WT control cohort. An ECG segment was operationally defined as ‘bradycardia’ if the rate was less than twice the standard deviation of the WT mean. Segments defined as bradycardia are highlighted by the pink box in Figure 1C. DORA treatment did not influence heart rate of non-treated WT (Figure 1C) or VEH-treated WT control cohorts (Figure 2C and 2D), supporting previous studies. DORA treatment was considered to significantly change KO heart rate if the following operational criteria were met: (i) KO heart rate during DORA significantly differed from the within subject VEH rate and (ii) the change in KO heart rate was at least twice the standard deviation of the average DORA-induced nonsignificant influence on WT heart rate (i.e. ≥ 30 bpm).

Figure 2. Blocking orexin receptors differentially influences heart rate (HR) in KO mice.

(A) A representative brightfield photomicrograph depicting orexin immunopositive neurons in the lateral hypothalamus at 20X magnification. Scale bar = 100 μm. (B) The mean number (±SEM) of orexin-immunopositive cells in the LH at two SD₀ ages (the first age was prior to epilepsy onset at P12 and the second age was during the week of epilepsy onset at P21), SD₁₀ and SD₃₀. The number of cells were normalized to SD₃₀ wild type (WT) mice. (C-D) Box-whisker plot depicting the median HR, 1^st and 3^rd quartiles and the minimum and maximum values for each subject following vehicle (VEH, red bars), then following acute DORA administration (diagonal or white bars) at (C) SD₂₀ and (D) SD_35–55. Blue line indicates the bradycardia threshold for each WT VEH cohort. DORA either reduced HR (blue shaded region, downward diagonal lines), increased HR (green shaded region, upward diagonal lines), or did not change HR (unshaded regions, white bars) in KO mice (Paired t tests). Arrowheads denote whether DORA increased (upward) or decreased (downward) within subject bradycardia incidence. Horizontal arrows indicate no bradycardia was detected during VEH or DORA. (E) For individual KO mice, the effect of DORA on HR is negatively correlated with the effect of DORA on bradycardia (Pearson r, p < 0.0001). (F) For individual KO mice, the effect of DORA is negatively correlated with the degree of difference from WT during VEH (Pearson r, p < 0.01). (G) HRV (rmSSD) is higher in SD_35–55 KO VEH when compared with WT VEH (F (2, 65) = 3.95, p < 0.05) and SD₂₀ KO cohorts (F (2, 65) = 10.73, p < 0.005, Two-way ANOVA with Sidak’s test). (H) The effect of atropine on HR before and after pretreatment with DORA in WT and KO cohorts. ‘+’ indicates differences within KO group across age. Blue * indicates KO group differs from WT. Red * indicates DORA KO group differs from KO VEH controls, * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. And ⁺ p < 0.05, ⁺⁺ p < 0.01, ⁺⁺⁺ p < 0.001, ⁺⁺⁺⁺ p < 0.0001. DORA, dual orexin receptor antagonist; HR, heart rate; HRV, heart rate variability; KO, Kv1.1 knock out; LH, lateral hypothalamus; rmSSD, root mean square of the successive differences; SD_#, indicates # percent of the colony has experienced sudden death; WT, wild type.

Orexin immunohistochemistry and stereology:

Mice were quickly anesthetized with isoflurane, decapitated and brains were quickly removed and frozen in methyl-butane on dry ice. Six LH-containing sections from −1.70mm Bregma thru −2.06mm Bregma were selected using unbiased systematic random sampling and immunolabeled for orexin. Sections were fixed in 4% paraformaldehyde, washed with 0.01 M PBS/0.3% Triton X-100, pH 7.4 (PBS-T) and blocked with 10% normal goat serum/PBS-T. Sections were incubated with a rabbit anti-orexin antibody that detects both orexin peptides A and B (1:1000, Peninsula Laboratories, San Carlos, CA, USA) overnight at room temperature. Sections were incubated with Alexa Fluor-594 conjugated goat anti-rabbit IgG (1:600, Invitrogen, Grand Island, NY, USA) for 3 hours. The number of orexin-positive neurons were counted using unbiased stereology^17,21. The total number of orexin neurons were normalized to WT mice (SD₃₀ age-matched) and expressed as the mean ± SE.

Respiratory measurements:

For Experiment II, following the ECG recording non-anesthetized, conscious mice were acclimated to a non-invasive airway mechanics chamber (NAM; Buxco Finepointe/DSI Inc., Minneapolis, MN, USA). This was approximately one hour after VEH or DORA administration. Mice were exposed to aerosolized PBS for 1 min and baseline responses were recorded for 3 min. Data were acquired with Buxco FinePointe acquisition software V2.3.1.9. Breathing frequency was determined. To account for varying body weights, frequency was normalized to individual animals’ weights. The duration of two respiratory cycles was calculated for each individual animal at baseline using the equation: 2*(frequency of breaths per min / 60 sec). An apnea event was defined as the decrease of airflow ≥ 90% for a period ≥ two complete respiratory cycles. Hypopnea was defined as the reduction of airflow ≥ 30% for a period ≥ two complete respiratory cycles (i.e. shallow breathing). Rapid breathing-induced apnea was determined. Frequency (breaths min⁻¹ gram⁻¹) 10 min prior to and following each apnea event was calculated. Rapid breathing was defined as frequency that was greater than a 50% increase from the subject’s mean frequency (mean increase was 68 ± 1.4%).

Seizure monitoring:

Methylcholine (MCh) induces seizures in KO mice¹⁶. Following PBS baseline recording, mice were exposed to aerosolized (MCh at 6, 12, 24 and 48 mg/ml). Each consecutive MCh concentration was administered for 1 min and responses were continuously recorded for the subsequent 3 min as we have previously described¹⁶. Seizure monitoring occurred during the entire course of the MCh experiments (4 min at PBS and each MCh concentration = 40 mins). The limited space of the NAM chamber prevents electroencephalography (EEG) recordings. Severe seizures were video-monitored, manually scored, and video-verified using a modified Racine scale (0-normal; 1-immobility; 2-head bobbing; 3-forelimb/hindlimb clonus and tail extension; 4- generalized clonus with forelimb tonic extension; 5-severe tonic-clonic seizures with hindlimb tonic extension) as previously described^14,16. Severe behavioral seizures (Types 3–5) and at least 10 sec in duration were included in the analyses.

Oximetry:

Arterial blood oxygen saturation was measured noninvasively using MouseOx Plus infrared pulse oximetry (Starr Life Sciences Corp, Oakmont, PA, USA) as we have described previously¹⁶. Following removal of the hair around the neck, a CollarClip Sensor (Starr Life Sciences Corp) was fitted in SD₃₅ KO mice and age-matched WT controls. Eight parameters, including oxygen saturation (SaO₂), pulse rate and breathing rate, were acquired at 5 Hz and mice were recorded for 1 hr (resulting in over 11,000 data points). Signal fidelity is reliable when all eight parameters are measured as error free, thus only error-free measurements were included in the analyses. During each recording session, state of vigilance and seizure onset, duration and severity were recorded by a trained technician. Data points during wake and seizures were analyzed. Data points below 90% SaO₂ were demarcated as hypoxic.

Statistics:

Data are presented as the mean ± S.E. Statistical significance was determined with Prism6 software (Graphpad Software Inc., La Jolla, CA, U.S.A.) using the following tests: two-way ANOVA with Sidak’s multiple comparison post hoc test, unpaired t-tests, paired t-tests, Brown-Forsythe test, Chi-Square test, two-way ANOVA with Sidak’s multiple comparison post hoc text, Fisher’s exact test, one-way ANOVA with Dunnett’s post hoc test, two-way ANOVA with repeated measures and Gehan-Breslow-Wilcoxon test. The statistical test used for each endpoint is detailed in the corresponding Figure Legend.

RESULTS

Experiment I

Intermittent bradycardia is more prominent in KO mice that have a higher probability of death.

Previously Glasscock et al. reported bradycardia events and increased heart rate variability in KO mice on a Tac:N:NIHS-BC background aged 1 – 3 months^13,22. On this strain background 60% die by P70, but it is unclear whether all KO mice have early mortality²². Here, we used KO mice on the C3H/FeJ background, which have 100% mortality by P78 (Figure 1A)¹². We found that KO mice aged P28–51 also exhibited intermittent bradycardia and overall had lower heart rates compared to WT (789 ± 3 bpm vs. 697 ± 3 bpm, p < 0.0001, n = 359 WT ECG segments from 14 mice; 712 KO ECG segments from 18 mice)(Figure 1C, bradycardia segments are highlighted, defined in the Methods).

When data were analyzed based on age it was apparent that heart rates decreased and the incidence of bradycardia increased in KO mice at older ages with a higher probability of death (Figure 1C). The heart rate of WT littermates remained stable across segments, mice and ages, thus data were pooled. Mean heart rate for each subject was determined and data from all ages were binned into three groups based on the approximate probability of sudden death (~SD₂₀, ~SD₃₅, ~SD₅₅) (Figure 1D). Heart rates of younger SD₂₀ KO mice resembled those of age-matched WT littermates. In contrast, the mean heart rate of high-risk SD₅₅ KO mice dropped by more than 15% when compared within genotype and between age-matched WT littermates. Short-term HRV was higher in SD₃₅ and SD₅₅ KO mice, as evidenced by the increase in rmSSD (Figure 1E).

Parasympathetic drive is increased in SD₅₅ KO mice.

Increased HRV is associated with increased parasympathetic tone. In a new cohort of mice, parasympathetic tone was determined in SD₅₅ KO mice. Heart rate was measured at baseline, then again following blockade of parasympathetic input with the muscarinic receptor antagonist atropine. Atropine increased heart rate significantly more in KO mice with intermittent bradycardia compared with WT (Figure 1F). Collectively, these data suggest that intermittent bradycardia is detectable in KO mice at SD₅₅ and may be driven in part by an increase in parasympathetic drive.

Experiment II

The number of orexin neurons in KO LH is increased by SD₃₀.

Orexin neurons contribute to the regulation of cardiorespiratory centers and can be upstream of parasympathetic drive.^23–39 We previously reported that the number of orexin-immunopositive neurons is increased in KO mice (pooled SD_27–65) when compared with WT¹⁷. Here, we determined whether this difference was present at ages during the development of epilepsy. At younger ages, when the probability of death for KO mice is 0 (SD₀ ages are P12, prior to epilepsy onset, and P21, during the week of epilepsy onset), the number of orexin-positive neurons in KO LH resembled WT (Figure 2A, 2B). Orexin-labeled cells increased in both SD₁₀ KO mice and age-matched WT mice by similar degrees. However, by SD₃₀ the number of orexin-positive neurons continued to increase in KO mice, but not WT controls. Thus, there is a rise in the orexinergic system at the same time that we detect cardiorespiratory abnormalities (Experiment I and¹⁵).

DORA differentially influences heart rate in KO mice, but not in WT mice.

Next, we tested the hypothesis that the augmented levels of orexin in KO mice contribute to the intermittent bradycardia. The mean heart rate for individual KO mice following VEH and DORA is depicted at SD₂₀ (Figure 2C) and at SD_35–55 (Figure 2D; there were no differences between SD₃₅ and SD_55, so data were combined). Intermittent bradycardia was defined as having multiple ECG segments that are less than twice the standard deviation of age-matched WT VEH (indicated by the blue lines in Figures 2C and 2D). Similar to data in Experiment I, approximately half of KO mice experienced intermittent bradycardia (12 of 25 mice), with the number of mice that experienced intermittent bradycardia being infrequent at SD₂₀ (1 of 7) and more common at SD_35–55 (11 of 18) (p < 0.05, SD_35–55 vs. age-matched WT VEH, Fisher’s exact test).

When considering the overall mean heart rate, DORA treatment did not influence control WT heart rates (Δ in heart rate 0.63 ± 2.5%, p > 0.05, n = 8 mice, data not shown) supporting previous studies^40–42. In contrast, DORA changed heart rate in a majority of KO mice. Within subject analyses identified three responder groups based on whether DORA increased, decreased or did not change heart rate (plotted in green, blue or white regions, respectively). In the KO mice with intermittent bradycardia (1 SD₂₀ and 11 SD_35–55), acute DORA treatment reduced the incidence of bradycardia within subject by 74 ± 11% (range 15% - 100% reduction, p < 0.01, Wilcoxon matched-pairs test) in 11 of 12 mice, with 7 mice having no bradycardia after treatment (mice in which DORA decreased bradycardia are indicated by downward arrows in Figures 2C and 2D). DORA treatment also reduced the within subject coefficient of variation (p < 0.01, Wilcoxon matched-pairs test). Interestingly, in 4 of these subjects in which bradycardia was reduced, DORA did not influence their mean heart rates (Figure 2D, white region). In contrast, 13 KO mice did not have intermittent bradycardia during VEH. In half of these subjects, acute DORA treatment increased the within subject incidence of bradycardia by 55 ± 9% (range 24% - 85% of segments in 6 of 13 KO mice; 4 of 6 SD₂₀ and 2 of 7 SD_35–55 mice, indicated by upward arrows). In the remaining mice, DORA had no influence (indicated by horizontal arrows). Collectively, these data suggest that intermittent bradycardia (i) can be reduced in KO mice by acute DORA treatment, and (ii) may be promoted by DORA treatment in those with heart rates in the normal range. As depicted in Figure 2E, when DORA increased heart rates, the incidence of bradycardia decreased; when DORA reduced heart rates, the incidence of bradycardia increased (p < 0.0001, Pearson’s r correlation).

The effect of DORA on the mean heart rate of individual subjects negatively correlated with the degree of difference between WT and KO mice during VEH treatment (i.e. if KO heart rate was lower than WT, DORA increased heart rate; but if KO heart rate was higher than WT, then DORA decreased heart rate; p < 0.01, Figure 2F). HRV of SD3_50–55 VEH-treated KO mice was higher than WT, supporting data from Experiment I. DORA treatment did not influence WT HRV (p > 0.05, n=8, data not shown), but did reduce HRV of KO mice to values that did not differ from KO VEH or age-matched WT VEH controls (Figure 2G).

DORA facilitates atropine-induced effects on heart rate in SD₅₅ KO mice.

In a new cohort of WT and SD₅₅ KO mice, we measured heart rate at baseline and in the presence of atropine. Twenty-four hours later DORA was administered and heart rates during baseline and atropine were determined. DORA treatment increased the atropine effects on heart rate in KO mice, not WT mice (Figure 2H). These data suggest that inhibiting orexin receptors facilitates parasympathetic blockade via atropine in KO mice.

DORA differentially influences hypopnea and apnea in KO mice, but not in WT mice.

We previously found that KO mice exhibit respiratory abnormalities including increased hypopnea and apnea in high risk KO mice¹⁵. Orexin neurons also regulate central respiratory centers^{23,24,33–39,25–32}, therefore, we determined the effect of DORA treatment on KO respiration. The number of subjects that experienced hypopnea and apnea increased with age, affecting 75–100% of SD₅₅ KO mice (Figure 3A), supporting previous findings¹⁶. Breathing frequency surrounding each apnea event was calculated to determine whether rapid breathing-induced apnea occurred in KO mice (resembling the atypical breathing response observed in epilepsy patients prior to SUDEP). Thirty-three percent of low risk SD₂₀ mice and 64% of high risk SD₅₅ mice experienced rapid-breathing induced apnea.

Figure 3. DORA treatment reduces apnea in a majority of SD_35–55 KO mice.

(A) Bar graphs depicting the percent of subjects that experienced hypopnea (left) and apnea (right) following acute VEH (red diamond bars), then acute DORA treatment (red header bond bars) (n = 5 SD₂₀, 15 SD_35–55). (B) Percent of subjects per responder group (increased in green with upward diagonal lines, decreased in blue with downward diagonal lines and no change in white). (C) Scatterplot depicting the degree of DORA-induced reduction of hypopnea (Paired t test: p < 0.0001) and apnea (Paired t test: p < 0.0001). The degree of reduction was not dependent on age. DORA, dual orexin receptor antagonist; SD_#, indicates # percent of the colony has experienced sudden death; VEH, vehicle.

DORA treatment did not influence the low frequency of hypopnea and apnea in WT mice (p > 0.05, n=8; data not shown), supporting previous studies^40,41, or the incidence in SD₂₀ KO mice (Figure 3A). In contrast, acute DORA treatment decreased the number of SD₃₅ and SD₅₅ KO mice exhibiting hypopnea and apnea (Figure 3A). The direction of the actual effects of the DORA varied. KO data were stratified into three responder groups based on whether acute DORA treatment increased, decrease or did not change hypopnea and apnea (Figure 3B). The proportion of mice in which DORA decreased hypopnea and apnea was greater in SD_35–55 mice. In mice in which DORA reduced frequency (i.e. SD₂₀ and SD_35–55), the degree of reduction was similar for both hypopnea and apnea (50–100% reduction) (Figure 3C).

When DORA significantly increased heart rate, breathing frequency and/or hypopnea-apnea was also reduced.

The bidirectional influence of DORA on both heart rate and hypopnea-apnea prompted additional analyses to determine whether the drug’s effect caused the same directional changes for both endpoints. In 75% of KO mice with DORA-induced increase in heart rate, DORA decreased hypopnea-apnea. As depicted Figure 4A, breathing frequency was increased in KO VEH cohorts, supporting previous findings¹⁶. Acute DORA treatment reduced frequency in KO mice (Figure 4A). In 100% of KO mice with DORA-induced increase in heart rate, DORA decreased breathing rate. These data suggest that when orexin receptor blockade increases heart rate, it also reduces breathing frequency and/or hypopnea-apnea.

Figure 4. DORA reduces breathing frequency at all ages.

(A) Data are normalized to age-matched WT littermates (blue solid bar). Black * indicates data differ from WT VEH, * p < 0.05 and ** p < 0.01. Acute DORA treatment (red header bond bars) reduced KO breathing rates (paired t test, red ** p < 0.01.). (B) Within subject breathing frequency during VEH and DORA. DORA reduced breathing frequency during MCh at SD₂₀ (F(1,20) = 71.8, p < 0.0001), at SD₃₅ (F(1,40) = 11.3, p < 0.005, data not shown) and (C) at SD₅₅ (F(1,40) = 23.7, p < 0.0001; Two-way ANOVA with Sidak’s post hoc test), * p < 0.05, ** p < 0.01, *** p < 0.001. DORA, dual orexin receptor antagonist; KO, Kv1.1 knockout; MCh, methacholine; SD_#, indicates # percent of the colony has experienced sudden death; VEH, vehicle; WT, wild type.

DORA reduced the MCh-induced increase in breathing frequency and seizures.

The parasympathomimetic MCh is muscarinic cholinergic receptor agonist used to test airway hypersensitivity by inducing bronchoconstriction. We previously reported that when challenged with nebulized MCh, KO mice have an atypical response in which breathing rates increase, whereas WT respond to MCh by decreasing their breathing frequency¹⁵. Responses to MCh were determined during VEH, then again 48 hrs later following acute DORA administration. DORA significantly reduced breathing frequency during MCh at all ages (Figure 4, B). DORA did not influence breathing frequency at the higher MCh concentrations (24 and 48 mg/ml) in SD₅₅ KO mice. These data suggest that acute blockade of orexin receptors in KO mice is able to attenuate their atypical increased breathing frequency response to the MCh challenge.

Seizures are triggered clinically with hyperventilation and we have reported that MCh not only increases breathing rate, but also triggers seizures in KO mice¹⁶. Here, the evoked seizures did lead to mortality in one subject during VEH, supporting previously reported data¹⁶. This subject was removed from the study due to the lack of DORA treatment data. The remaining mice were examined to determine whether acute DORA treatment attenuated the MCh-triggered severe seizures. Population data indicate that the percentage of KO mice that experienced severe seizures increased from 67% to 100% with age (Figure 5A). Forty-eight hours later, the same mice received acute DORA administration and the percent of KO that experienced seizures dropped by half. Figure 5B depicts the mean cumulative severe seizure incidence with increasing MCh concentrations during VEH and DORA. The total number of severe seizures (during PBS and all MCh concentrations) were combined (Figure 5C). Acute DORA treatment significantly reduced the total number of seizures at all ages and 50% of SD₅₅ KO mice did not experience a seizure. These data suggest that blocking orexin receptors either directly or indirectly attenuates MCh-induced seizures.

Figure 5. Acute DORA treatment protects against MCh-induced seizures.

Data for SD₂₀ are depicted in the top row, SD₃₅ in the middle row and SD₅₅ in the bottom row. (A) The cumulative percentage of VEH-treated KO mice that experienced severe seizures increased from 67% to 100% with age (red diamond bars). DORA treatment reduced the overall number of KO mice susceptible to seizures at each age (red header bond bars). (B) Seizures increase with higher MCh concentrations at SD₂₀ (F (4, 20) = 3.5, p < 0.05), at SD₃₅ (F (4,36) = 10.9, p < 0.0001) and at SD₅₅ (F (4, 36) = 13.8, p < 0.0001; Two-way ANOVA with repeated measures). Acute treatment with DORA reduced MCh-seizures at SD₂₀ (F (1, 5) = 4.9, p < 0.05), at SD₃₅ (F (1, 9) = 6.6, p < 0.05) and at SD₅₅ (F (1, 9) = 7.4, p < 0.05; Two-way ANOVA with repeated measures). (C) Total seizures during the MCh challenge was determined per subject. Within subject analyses indicates acute DORA treatment reduced the number of seizures at SD₃₅ and SD₅₅ (paired t tests). Blocking orexin with DORA completely prevented seizures in 25% of SD₅₅ KO mice. DORA, dual orexin receptor antagonist; KO, Kv1.1 knockout; MCh, methacholine; SD_#, indicates # percent of the colony has experienced sudden death; VEH, vehicle.

DORA attenuates intermittent hypoxia.

It is well established that severe seizures and apnea can promote hypoxia, and that hypoxia can trigger bradycardia. We have previously reported high risk KO mice have chronic intermittent hypoxia¹⁶. Here, blood oxygen saturation, pulse rates and breathing rates were simultaneously recorded using pulse oximetry in a new cohort of SD₃₅ KO and age-matched WT mice. While SaO₂ was relatively stable in WT mice, it was more variable (i.e. has a greater range over time and dipped below 90%, or hypoxia, often) in KO mice as depicted in Figure 6A, supporting previous reports¹⁶. In KO mice, stable SaO₂ was detected during eupnea (Figure 6B), and during normal and reduced pulse rates (Figure 6C). Hypoxia occurred during seizures (Figure 6D), and prior to a reduction in pulse rate (Figure 6E). However, hypoxia was also evident during seizure-free periods with normal breathing rates (Figure 6F). To determine blood gas stability over time, the hypoxia fraction (the percent of hypoxic measurements) and the coefficient of variation were determined for each subject, as previously reported¹⁶. KO mice with intermittent hypoxia had a greater hypoxia fraction compared with WT. Acute DORA treatment reduced the hypoxia fraction and coefficient of variation in these KO mice (Figure 6G).

Figure 6. Blocking orexin receptors reduces the hypoxia fraction and increases longevity.

(A) Representative SaO₂ traces from a KO (blue line) and WT (black line) mouse. The KO trace depicts greater SaO₂ fluctuation and two hypoxic events (i.e. when the trace dips below 90% saturation into the blue shaded area). (B-F) SaO₂ traces from KO mice. (B) Stable SaO₂ during eupnea (green line), (C) normal pulse rate and low pulse rate (red lines). (D-F) Hypoxic events occurring (D) during seizures, (E) preceding low pulse rate (red line) and (F) during eupnea (green line). (G) Blood gas stability was determined by calculating the hypoxia fraction (percentage of hypoxic measurements) and the coefficient of variation (COV). Acute DORA treatment reduced the hypoxia fraction and the COV in KO mice (red header bond bars). Data are expressed as mean ± SEM. N = 789 measurements from 3 KO mice, 1176 measurements from 3 KO DORA mice, 10306 measurements from 4 WT mice. Red ‘+’ indicates statistical difference from KO, + p < 0.05, ++ p < 0.01; black ‘*’ indicates significantly differs from WT, * p < 0.05, ** p < 0.01. (H) Survival curves of three KO cohorts. The survival curve of the colony is provided for reference (grey dashed line). The control curve that included only KO mice ≥ P45. Late-onset DORA treatment shifted the median age of death from P51 (KO (P45+)) to P59 (KO DORA; p = 0.050, Gehan-Breslow-Wilcoxon test) and increased the mean age of mortality (SD₅₀) from P52.7 ± 0.7 days to P59.2 ± 3.2 days (p < 0.05). N = 6 KO DORA and 87 KO mice ≥ P45. DORA, dual orexin receptor antagonist; KO, Kv1.1 knockout; SaO₂, blood oxygen saturation; SD_#, indicates # percent of the colony has experienced sudden death; VEH, vehicle; WT, wild type.

Experiment III:

Late-onset treatment with DORA increases longevity.

Collectively, the data generated herein indicate a unique pathophysiological phenotype develops in KO mice by SD_35–55 (intermittent bradycardia, increased hypopnea-apnea, breathing frequency and seizures and chronic intermittent hypoxia); thus, a window of opportunity for intervention may exist. In addition, data indicate that acute DORA treatment attenuates several components of this phenotype in a majority of SD_35–55 KO mice. Thus, we determined whether daily treatment of DORA would postpone or prevent SUDEP. Treatment was initiated at a late timepoint, in which KO mice have an imminent probability of sudden death. The mean age of mortality of our entire KO colony to date is P44 ± 1 day (n=169). In a new cohort of surviving SD₅₀ KO mice, DORA (200 mg/kg, i.p.) was administered daily beginning on P45 (SD₅₀) until natural mortality (Figure 6H). The survival curve was compared against a control curve that included only KO mice ≥ P45. Late-onset DORA treatment shifted the median age of death for KO (P45+) mice from P51 to P59 (p = 0.050) and increased the mean age of mortality by 12% (SD₅₀ for KO ≥ P45 = P52.7 ± 0.7 days; the SD₅₀ for KO DORA = 59.2 ± 3.2 days, p < 0.05).

DISCUSSION

Overall, this study details the cardiorespiratory pathophysiological changes that occur as KO mice age and SUDEP probability increases. Our results suggest that there is a concurrent increase of the number of orexin neurons in high risk KO mice, and that antagonizing orexin receptors attenuates cardiorespiratory abnormalities and postpones death. Specifically, there are nine novel findings reported in this study. (i) KO mice experience a progressive cardiorespiratory pathophysiology that worsens as the probability of death increases. (ii) Intermittent bradycardia is more prevalent in high risk KO mice, and this is partially mediated by elevated parasympathetic drive. (iii) Blocking orexin receptors differentially influences heart rate in KO, not WT mice. (iv) DORA treatment facilitates the atropine-induced changes in heart rate in SD₅₅ KO mice. (v) When DORA administration increases heart rate, it also decreases HRV, breathing frequency and/or hypopnea-apnea. (vi) Blocking orexin receptors prevents the MCh-induced increase in breathing frequency in KO mice and (vii) either indirectly or directly reduces MCh-induced seizures. (viii) DORA improves oxygen saturation in KO mice with intermittent hypoxia. (ix) Initiating daily DORA treatment in high risk KO mice increases longevity.

Clinical studies indicate that frequent bradycardia incidence is a risk factor in sudden death⁴³. Within the field of epilepsy and SUDEP, ictal and postictal cardiac arrhythmias, including bradyarrhythmias, have been proposed to mechanistically contribute to SUDEP^44,45. Preclinical studies report that cardiac abnormalities, particularly bradyarrhythmias, exist in SUDEP models^12,13,20. However, it is unknown whether the occurrence or incidence is progressive and thus a potential contributor to susceptibility to SUDEP. One case report described a patient with epilepsy and normal heart rate (several months prior to SUDEP), began experiencing ‘bradycardia before SUDEP’; however, it is unclear whether the bradycardia was detected hours, days, weeks or months before SUDEP⁴⁶. In the current study, we have begun to investigate this relationship with a mouse model of SUDEP in which seizures and death occur in a reasonable temporal sequence in 100% of mice allowing the assignment of probability of death based on age. We report that intermittent bradycardia is more common in high risk KO mice. Similarly, we found that the number of KO mice that experienced apnea increased with risk, supporting previous studies¹⁶. Increased apnea is also associated with early mortality outside the field of epilepsy (particularly in obstructive sleep apnea) and within the field of epilepsy^{4,16,43,47,48}. If bradycardia or apnea represent a temporal biomarker of sudden death in KO mice then the biomarker should become apparent in mice before they die, regardless of their age. KO mortality ranges from P24 thru P78, a relatively tight time frame, but one that still offers a fair amount of variation. This may be the reason why intermittent bradycardia was not only apparent at SD₅₅, but also occurred in one SD₂₀ KO and multiple SD₃₅ KO mice. However, all subjects in this study were sacrificed after the experiment so this is speculative. Current follow-up studies are sampling endpoints throughout life until natural mortality to determine how many days/weeks prior to death endpoints change in KO mice.

In addition to bradycardia and apnea, the ‘rapid breathing-induced apnea’ may be another temporal biomarker. In the MORTEMUS study, the final GTC seizure-induced rapid breathing – a response that should trigger blood gas stability, recovery and survival – triggered apnea, bradycardia, terminal apnea and asystole⁴. Rapid breathing-induced apnea is a pathological respiratory response attributed in part to blood gas instability and occurs in people with central idiopathic apnea, a condition also associated with early mortality^49,50. Preclinical SUDEP studies report that after a severe seizure, a fraction of KO mice experience rapid breathing-induced apnea, however it is not known whether these mice were at low or high risk for sudden death^16,20. Here, we report that this response phenotype was more common in high risk SD₅₅ KO mice.

The response phenotype of rapid breathing-induced apnea further suggests problems maintaining blood gas stability. High risk KO mice have blood gas instability as evident by chronic intermittent hypoxia¹⁶. It is well established that hypoxia can occur during severe seizures, apnea, and can trigger bradycardia^51–55. Further, intermittent hypoxia was noted to also occur in high risk KO mice during seizure-free periods with eupnea and normal heart rates, indicating the chemoresponses necessary for maintaining blood gas stability may be fundamentally dysregulated.

Chronic intermittent hypoxia is associated with increased central chemosensitivity and probability of death in patients with heart failure^56–58. Central chemoreception promotes responses to repeated hypoxia and/or hypercapnia challenges, including increasing ventilation^34–36. There are multiple central sites with chemoreceptive neurons and orexin neurons in the lateral hypothalamus represents one such site. The number of orexin neurons progressively increases KO mice (Figure 2B)^15,16 and orexin levels are increased in patients with respiratory failure, a condition associated with blood gas instability and early mortality^16,17,59,60. Orexin neurons project onto numerous respiratory nuclei including rostral ventrolateral medulla, medullary raphe, locus coeruleus and nucleus tractus solitarius^{24,32–34,61} and act on orexin G-protein coupled receptors 1 and 2. Orexin neurons increase breathing rate in response to hypercapnia and hypoxia^{25–27,34,36–39,62} and administration of DORA reduces this response^28,29. Extracellular multielectrode recording data indicate that the chemosensitivity of putative orexin neurons is increased in high risk KO mice (Warren et al., unpublished observations). An increase in the number of orexin neurons (or potential increased chemosensitivity of each orexin neuron) may exacerbate the chemoresponse and promote pathophysiological responses in KO mice. For example, challenges (such as with MCh, reported herein, or a seizure) that cause fluctuations in blood gasses may trigger an orexin-receptor-mediated increase in breathing frequency; which could trigger apnea leading to hypoxia, which may trigger bradycardia.

Preclinical and clinical data indicate that orexin receptor antagonists do not influence heart rate, apnea, or SaO₂ in control cohorts^40–42. DORA treatment did not influence baseline heart rate in WT controls. In contrast, acute blockade of orexin receptors differentially influenced heart rate in 80% of KO mice. The bidirectional effect of DORA is consistent with the dual influence of orexin itself. While microinjection of orexin A in the rostral ventral medulla increases heart rate, microinjection in the nucleus ambiguous, caudal dorsolateral and medial subnuclei of the solitary tract elicits bradycardia^30,31,63. Further, studies have shown that optogenetic stimulation of orexinergic neurons in rats evoke about 60% excitatory and 40% inhibitory responses in the cardiac vagal neurons of the brainstem²³. Here, we found blocking orexin receptors exerted three responses (discussed with cautious speculation). (i) DORA treatment did not significantly influence basal heart rate of WT VEH mice, an effect observed in half of the KO mice with WT-like heart rates and no bradycardia. This may indicate these subjects are temporally farthest from sudden death. (ii) In the other half of the KO mice with WT-like heart rates and no bradycardia, DORA reduced heart rate and promoted bradycardia. This may reflect a potentially early biomarker of dysregulation wherein heart rate appears normal, but orexin is contributing to increasing it to a ‘normal’ rate. (iii) Inhibiting orexin receptors in KO mice with intermittent bradycardia reduced within subject incidence and ‘prevented’ bradycardia in 7 subjects. In this cohort, the influence of orexin on parasympathetic tone may contribute to pathological intermittent bradycardia. Future studies are required to better understand these nuances.

Due to the interdependency of these biomarkers (seizures, breathing frequency, apnea, chronic intermittent hypoxia and intermittent bradycardia), it is unclear whether the DORA effects are direct, indirect or multi-mechanistic. Reducing breathing frequency may indirectly improve other endpoints. If increased breathing frequency triggered a portion of the apneic events, then the DORA-induced reduction in apnea was indirect. Similarly, hyperventilation (increased tidal volume and frequency) is known to trigger seizures, thus the DORA-induced reduction in seizures may be the result of lower breathing frequency. However, previous reports have demonstrated that administration of orexin can promote seizures and blocking orexin receptors reduces seizures^17,64–66; thus the dynamic relationship between orexin and seizures remains unclear.

An additional confounding physiological variable that orexin neurons influence, and was not studied herein, is sleep. Orexin neurons project to several nuclei within the ascending reticular activating system to promote arousal from sleep. We have previously reported that orexin contributes to sleep deficiency in KO mice^17,67. Sleep deficiency is upstream of promoting cardiac arrhythmias, apnea, blood gas instability and exacerbating seizures^{16,17,48,67–69}. DORA treatment is FDA-approved to treat sleep disorders, and pre-clinical data report daily DORA treatment improves sleep disorders associated with epilepsy¹⁶, and DORA treatment improves sleep architecture in KO mice. It is unlikely that improvement in sleep or sleep architecture is responsible for improving the endpoints herein because testing occurred immediately after DORA administration while subjects were awake. Future studies will be required to determine whether promoting sleep a priori postpones or prevents onset of one or more of these temporal biomarkers and subsequent SUDEP. Here, we report that chronic, daily DORA treatment in high risk KO mice, which improves sleep architecture¹⁶, did increase longevity, however did not prevent sudden death. The modest increase in longevity is encouraging as treatment began in mice with high probability of being at the height of disease progression and near sudden death. Further studies are needed to determine whether beginning treatment earlier would be even more beneficial.

In conclusion, high risk KO mice have a unique phenotype that is characterized by progressive changes in five interdependent endpoints: severe seizures, chronic intermittent hypoxia, apnea, rapid breathing-induced apnea, and bradycardia, the same pathophysiology reported in patients immediately prior to succumbing to SUDEP^4,16,67. Uncontrolled severe seizures represent the primary risk factor for SUDEP, however it is not known whether high risk patients have chronic intermittent hypoxia and/or increased apnea. Our results further indicate that orexin receptor activation contributes to this cardiorespiratory pathophysiology in a complex manner. Whether effects are direct, indirect and require changes in orexin protein levels in the LH and/or receptor expression at target nuclei will be determined in future studies. Collectively, these data support the notion that biomarkers may hold temporal significance regarding proximity to death, and windows of opportunity for intervention exist in this preclinical SUDEP model.

Key point box.

• Epileptic Kv1.1 knockout (KO) mice that are at high risk for sudden death have a unique cardiorespiratory phenotype that is characterized by progressive changes in five interdependent endpoints: severe seizures, chronic intermittent hypoxia, apnea, rapid breathing-induced apnea, and bradycardia, a similar pathophysiology experienced by people with epilepsy immediately prior to succumbing to SUDEP.

• The intermittent bradycardia in KO mice is partially mediated by elevated parasympathetic tone.

• Acute treatment with a dual orexin receptor antagonist (DORA) differentially influences heart rate in KO, but not WT mice: the effect of DORA administration on individual subjects’ heart rate negatively correlates with the degree of difference between WT and KO baseline heart rate during vehicle treatment (i.e. if KO heart rate was lower than WT, DORA increases heart rate; but if KO heart rate was higher than WT, then DORA decreases heart rate).

• When acute DORA treatment increases heart rate in KO mice with bradycardia, it also decreases heart rate variability (HRV), breathing frequency and/or hypopnea-apnea.

• Acute DORA treatment improves oxygen saturation in KO mice with intermittent hypoxia and attenuates the methacholine (MCh)-induced increase in breathing frequency in KO mice and either indirectly or directly reduces MCh-induced seizures.

• Initiating daily DORA treatment in high risk KO mice increases longevity.