Loss of cyclin‐dependent kinase‐like 5 results in susceptibility to audiogenic seizures in mice

Abstract

CDKL5 deficiency disorder (CDD) is a severe neurodevelopmental encephalopathy characterized by early‐onset, treatment‐resistant epilepsy. Mice lacking CDKL5 display several clinically relevant phenotypes, but spontaneous seizures are not consistently reported, and it is unknown if CDD model mice are susceptible to sensory stimulus‐triggered seizures, a well‐documented clinical feature of CDD. Here, we tested the hypothesis that CDKL5 deficiency confers susceptibility to audiogenic seizures (AGS). We exposed adult male Cdkl5 knockout, female heterozygous, and wildtype littermates (P80‐217) to audiogenic challenges and, in a separate cohort, monitored for spontaneous seizures. Audiogenic stimulation triggered severe, lethal (80%) seizures in Cdkl5 knockout mice. In contrast, heterozygous mice were largely resistant to audiogenic stimulus (92% survival). These findings establish susceptibility to AGS as a highly penetrant phenotype in a CDD mouse model. Furthermore, spontaneous seizures were detected in a subset of Cdkl5 knockout mice during chronic video‐EEG monitoring. AGS may provide a translationally relevant screen for investigating hyperexcitability and for evaluating potential therapeutics to prevent seizures in CDD.

Plain Language Summary

CDKL5 deficiency disorder (CDD) is a severe genetic condition causing early‐onset seizures. Mice with the same mutation are useful models but don't consistently have epilepsy. We tested if these mice in our lab are sensitive to sound‐triggered seizures. We discovered that male CDD mice are highly vulnerable to sound, which triggered severe seizures in most of them. Female CDD mice and normal mice were resistant. This is the first report of sound‐triggered seizures in a CDD model and provides a useful new method to study epilepsy in CDD and screen for antiseizure treatments.

Key points.

• Loss of CDKL5 results in susceptibility to audiogenic seizures in mice.

• Audiogenic stimulation provides a robust, noninvasive, and penetrant seizure induction method in a CDKL5 mouse model.

• Spontaneous generalized tonic–clonic seizures were detected in a subset of CDKL5 knockout mice during chronic video‐EEG recording.

1. INTRODUCTION

CDKL5 deficiency disorder (CDD) is a rare X‐linked disorder caused by mutations in the CDKL5 gene. Patients display global developmental delays, autism‐like features, and debilitating, early‐onset, and often drug‐resistant epilepsy. ¹ Current treatment focuses on symptom control although efforts are underway to develop gene therapies.

Clinical case reports of CDD patients frequently describe reflex seizures, triggered by external stimuli such as auditory, visual, tactile, or thermal cues.^{2, 3, 4} This suggests that sensory hyperexcitability and network vulnerability to external provocation may be a key component of the CDD epilepsy phenotype.

Many features of CDD are recapitulated in mice lacking Cdkl5 including impaired cognition, autism‐like behavior, and visual processing abnormalities.^{5, 6} There have been conflicting results, however, in the occurrence of spontaneous seizures. Studies using EEG have reported finding no seizures in knockout mice.^{5, 7, 8, 9} Other studies observed spontaneous epileptiform activity and seizures in constitutive knockouts^{10, 11} and in conditional knockouts. ¹²

While CDD mouse models show susceptibility to chemical convulsants, including kainic acid (KA) and PTZ,^{8, 13} and mechanically and thermally induced seizures in drosophila models, ¹⁴ the susceptibility of Cdkl5‐deficient mice to reflex seizures triggered by sensory stimuli remains largely unexplored. We hypothesized that CDKL5 deficiency confers a susceptibility to audiogenic seizures (AGS). Here, we evaluated knockout and heterozygous CDD mice and their wild‐type littermates to a standardized AGS protocol used to evoke seizures in mice lacking Ube3a. ¹⁵ We demonstrate that complete, but not partial, Cdkl5 deficiency results in susceptibility to seizures in most mice in this test. We also report spontaneous seizures in a pilot chronic EEG study in constitutive Cdkl5 knockout mice. Together, this establishes a useful model for investigating reflex seizures and seizure‐induced mortality that may have applications in understanding mechanisms of hyperexcitability and use in screening or evaluating therapeutic interventions in CDD.

2. METHODOLOGY

The mouse model of CDD used was the B6.129(FVB)‐Cdkl5 ^tm1.1Joez /J (Jackson Laboratory, strain #021967, ⁵ ). Breeding scheme is as recently described. ¹⁶ A total of 54 mice were used in this study (Figure 1A).

FIGURE 1.

Study workflow and behavioral seizure phenotyping in Cdkl5 deficient mice. (A) Workflow schematic of the study. A total of 54 mice were used for this study. Thirty‐seven naive mice were used for audiogenic stimulus over three trials (Knockout n = 10, Heterozygous n = 12, Wildtype n = 15). Seventeen mice were implanted with continuous EEG and video monitoring. Following a 7‐day recovery period, five Knockout mice underwent two audiogenic stimulus trials. On the third trial, five mice were given I.P KA: two mice also received audiogenic stimulus and three mice served as subthreshold KA controls. Of the remaining 12 mice (six Knockout, six Wildtype), mice were left under video monitoring. (B) Behavioral seizure phenotyping. In response to audiogenic stimulus, Cdkl5 deficient mice exhibit wild running (pink), a clonic phase involving myoclonic jerking of the hindlimbs (salmon), a tonic phase characterized by hind limb extension (crimson) and the terminal phase characterized by atonia of muscles (gray). All terminal seizures presented in this order.

2.1. Audiogenic‐stimulus exposure and seizure assessment

The protocol for inducing AGS was adapted from Campbell et al. ¹⁵ All mice were habituated for at least 30 min prior to testing. Mice were placed in a polycarbonate test cage (365 × 207 × 140 mm). The acoustic stimulus was applied by scratching a metal rod across the metal cage lid for 30 s or until the animal had a behavioral seizure. Sound intensity was maintained between the range of 100–110 dB. Testing was repeated over three trials spaced at least 24 h apart (Figure 1A). AGS procedures were carried out by a male experimenter blinded to genotype.

Responses to AGS were recorded by camera (Logitech model C920 HD) or iPhone 15 PRO Max. Wild running phase characterized by rapid, erratic running within the cage, a clonic phase distinguished by myoclonic jerking of the hindlimbs, and a tonic phase characterized by hindlimb extension and muscular rigidity. The end of the tonic phase was evident at the onset of bodily atonia ¹⁷ (Figure 1B).

2.2. Surgery, KA treatment and EEG acquisition

Briefly, mice (weight: 25–30 g; age 10–14 weeks) were anesthetized with isoflurane (5% induction, 2% maintenance) and placed in a mouse‐adapted stereotaxic frame. An EEG transmitter unit (model HD‐X02; DSI) was implanted under the skin in a subcutaneous pocket along the dorsal flank of the mouse. The EEG signal was recorded from skull‐fixed screws over the dorsal hippocampus (AP: −2.5, ML: +2.0) and V1 (AP: −3.6, ML: −2.3) regions with the reference electrodes positioned over the prefrontal cortex. Postsurgery animals were housed individually and allowed 1 week to recover before initiating continuous video‐EEG recording for a minimum of 7 days. EEG signals were recorded at a sampling rate of 500 Hz and acquired with Ponemah (v6.51, DSI) and exported as EDF files for analysis. Cdkl5 knockout mice received an I.P. dose of KA (10 mg/kg).

2.3. EEG analysis

Continuous electroencephalography (EEG) recordings were acquired in European Data Format (EDF) and processed using MNE‐Python (version 1.10.2). ¹⁸ Signals were filtered using a 0.5–80 Hz finite impulse response (FIR) band‐pass filter implemented with zero‐phase forward–reverse convolution to suppress slow drift and high‐frequency noise. Filter design used a Hamming window.

Time‐frequency decomposition was performed using a short‐time Fourier transform (STFT) with the following parameters: NFFT = 512, overlap = 50%, frequency range visualized = 0–80 Hz, and power scaling = 10 × log₁₀ (μV²/Hz). Seizures were manually detected and identified by electrographic measures of sudden onset of high‐amplitude and rhythmic discharges lasting longer than 5 s followed by a postictal suppression and accompanied by behavioral correlates on synchronized video. EEG signals were visualized using EDFbrowser (v2.13) for video correlation. Events were then visualized using a custom‐written raster plot script in R across the final 28 h of recording to capture all events, showing the onset of a seizure event.

2.4. Statistics

Survival was defined as the absence of seizure‐induced death within each trial. For animals that did not die during a given trial, survival time was right censored at the trial number reached (maximum = 4). Differences in survival across genotypes were assessed using the log‐rank (Mantel‐Cox) test. Statistical significance was evaluated at α = 0.05. When global significance was detected, pairwise post hoc comparisons were performed with Benjamini–Hochberg correction for multiple testing. Survival estimates at the final challenge (Trial 4) were summarized to extract median survival probabilities and corresponding 95% confidence intervals.

3. RESULTS

3.1. CDD knockout mice are susceptible to AGS

To explore if CDD mice are susceptible to sound‐induced seizures, animals were exposed to an audiogenic stimulus through mechanical abrasion (Video S1), as described. ¹⁵ We compared survival across genotypes using a Kaplan–Meier analysis. Survival probability differed markedly between groups (log‐rank test: χ² = 26.1, p = 2.0 × 10⁻⁶). Wildtype mice exhibited complete survival across all three trials (100%, 95% CI: 100%–100%), whereas heterozygous mice displayed only a modest reduction in survival (92%, 95% CI: 77%–100%). In contrast, knockout mice showed a profound survival deficit, with only 20% of animals surviving the final challenge (95% CI: 5.8%–69%) (Figure 2A).

FIGURE 2.

Survival and behavioral seizure responses to repeated audiogenic stimulation in Cdkl5 deficient mice. (A) Kaplan–Meier survival curve across three audiogenic stimulus trials for Cdkl5 knockout males (n = 10), Cdkl5 heterozygous females (n = 12), and wildtype controls (n = 15). Across the three trials, 80% of knockout mice (8/10) died, one heterozygous mouse (8%) died during the first trial, and no deaths occurred in wildtype mice. (B) Audiogenic seizure assessment in naïve Cdkl5 knockout (n = 8) and heterozygous mice (n = 1) which died during the audiogenic stimulus trials. Stacked bars show mean durations of seizure phases following the sound stimulus: Latency to wild running, wild running, clonic phase, and tonic hindlimb extension leading to atonia. (C) Boxplots of wild‐running duration across Trials 1–3. Incidence and duration of wild running were quantified from recorded videos for each genotype and trial. Knockout mice exhibited significantly longer wild‐running durations than wildtype mice in each trial (Trial 1: p = 0.00916; Trial 2: p = 0.0445; Trial 3: p = 0.00337; Kruskal–Wallis with Dunn's post hoc test and Benjamini–Hochberg correction). Heterozygous values did not differ significantly from wildtype.

Pairwise comparisons confirmed that knockout mice had significantly lower survival than both wildtype (p = 4 × 10⁻⁵, Benjamini–Hochberg corrected) and heterozygous mice (p = 0.0012). There were no significant differences between wildtype and heterozygous mice (p = 0.264). These findings demonstrate that loss of the gene dramatically increases vulnerability to seizure‐induced death during repeated audiogenic stimulation, whereas partial loss is largely protective (Figure 2A).

Prior to the onset of seizure events, Cdkl5 mice displayed wild running for seconds at a time (m = 4.22, SD = 2.20). Seizure onset was observed as clonic before tonic hindlimb extension, which resulted in movement cessation (mean duration of clonic phase m = 3.89, SD = 0.99, mean duration of tonic phase m = 10.33, SD = 2.94) (Figure 2B, Video S1).

Knockout mice displayed wild running, revealing a significant disparity when compared to wildtypes across all trials. (Trial 1 KO vs. WT p = 0.00916, Trial 2 KO vs. WT p = 0.0445, Trail 3 KO vs. WT p = 0.00337, Kruskal–Wallis plus Dunn post hoc and Benjamini–Hochberg for multiple corrections), and heterozygous differed significantly by Trial 3 (p = 0.00442), (Figure 2C.) Wildtypes and heterozygous mice exhibited wild running, but to a lesser extent. All except one knockout mouse ran (9/10), four heterozygous mice ran (4/12), and only two wildtype mice ran (2/15).

3.2. Telemetry implantation affects seizure threshold

To electrographically characterize AGS, we implanted knockout mice (P65‐79) with telemetry devices and exposed mice to AGS challenge 7 days postsurgical‐recovery. However, these implanted mice (n = 5) failed to exhibit wild running or seizures during two audiogenic trials. We hypothesized that the surgery and recovery period raised the seizure threshold, attenuating the phenotype. To test this, implanted mice were given a subconvulsive dose of KA (10 mg/kg, i.p.), which does not induce overt seizures in this model. ⁸ Following this, a third audiogenic challenge successfully induced lethal seizures in two of five mice, which were captured by EEG (Figure 3A). The three control mice that received KA displayed no behavioral seizure with the audiogenic stimulus, confirming the interaction of subthreshold network hyperexcitability and auditory stimulus was required to trigger the lethal event.

FIGURE 3.

Audiogenic seizure susceptibility and electrophysiological signatures in implanted mice. (A) Implanted mouse undergoing an audiogenic stimulus challenge. The spectrogram (0–80 Hz) and raw waveform trace show behavioral seizure progression in a knockout mouse, including running, spasms, and hind‐limb extension following the onset of the audiogenic stimulus. (B) Raster plot of continuous EEG monitoring in implanted mice across the final 28 h of recording. Seizure events were detected only in knockout animals, with no epileptic episodes in wildtype controls. (C) Representative electrophysiological traces comparing genotypes. Wildtype mice exhibit stable EEG activity with no epileptiform discharges, whereas knockout animals show distinct seizure events characterized by high‐amplitude ictal activity. Spontaneous seizure data derive from a small cohort intended to demonstrate occurrence rather than estimate incidence.

3.3. Spontaneous seizures are detected in a subset of Cdkl5 knockout mice

A separate pilot cohort of telemetry‐implanted adult (P62‐101) Cdkl5 knockout male mice underwent continuous video‐EEG recordings. Several spontaneous bilateral convulsive seizures were detected in two Cdkl5 knockout males (2/6) while wildtype littermate controls exhibited no events (0/6) (Figure 3B,C). The seizures in Cdkl5 knockout mice displayed clustering over a 24‐h period and both mice that displayed spontaneous seizures died during monitoring (Figure 3B). One mouse had a total of nine seizures with a mean duration of 34.61 s (SD = 12.013), while the second mouse had a total of seven seizures with a mean duration of 25.46 s (SD = 5.727) (Table S1). Seizures followed a tonic–clonic behavioral phenotype which corresponded and evolved as the electrographic event progressed. Seizures followed a consistent behavioral phenotype across all events presenting as generalized tonic–clonic, where the tonic phase consisted of tail stiffness and posture immobility followed by a clonic phase consisting of bilateral synchronous jerks and repetitive limb movements, which corresponded and evolved as electrographic events progressed (Video S2). In the case of nonlethal seizures, mice would demonstrate a flaccid posture postictally followed by a gradual recovery. In the case of SUDEP, the events were followed by a rigid hind‐limb extension, rather than flaccid posture, and a lack of recovery.

4. DISCUSSION

CDD is a severe neurodevelopmental encephalopathy characterized by early‐onset, intractable epilepsy. While mouse models of CDD recapitulate many clinical features, their susceptibility to evoked and spontaneous seizures has remained inconsistent and poorly defined.^{5, 7, 8, 9, 10, 11, 12} The principal finding of our study is that Cdkl5 knockout mice exhibit a highly penetrant and severe susceptibility to AGS, resulting in lethality. Also, similarly to previous reports,^{10, 11} we further demonstrate that a subset of knockout mice experience spontaneous tonic–clonic seizures, which also proved fatal, although the limited cohort size precludes conclusions regarding incidence or penetrance.

Our study introduces a reliable, noninvasive, and nonpharmacological method for inducing seizures in a CDD mouse model. The use of audiogenic stimulation circumvents the issues associated with excitotoxic agents like KA, which can cause extensive neuronal damage and confound subsequent molecular and electrophysiological analyses. This paradigm provides a reproducible and clearly defined seizure endpoint, making it a potential tool for preclinical drug screening. This approach allows for testing therapeutic candidates against triggered seizures while simultaneously monitoring their impact on spontaneous seizure burden in the same subjects, offering a robust dual assessment of efficacy. This is crucial for modeling drug resistance, a hallmark of CDD. The translational relevance is underscored by the documented cases of auditory‐triggered seizures in CDD patients, ⁴ indicating that this model addresses a clinically significant disease mechanism.

Seizure events proved to be highly lethal in Cdkl5 knockout mice, with both sound‐induced and spontaneous seizures resulting in fatalities. Electrophysiological profiles of both seizure types appeared different, possibly indicating distinct neural circuits being affected. Although, the seizure clustering and sudden death may reflect disruptions in cortical and subcortical circuits, particularly within brainstem and autonomic centers. ¹⁹ This vulnerability is not unique to CDD; similar phenotypes are seen in other monogenic epilepsy models including Kcna1 KO, Scn1a KO, Cacna1a ^S218L and Ryr2 ^R176Q and Scn8a ^N1768D mice, which also exhibit spontaneous seizures and increased susceptibility to audiogenic induction.^{20, 21, 22, 23} This commonality suggests the disruption of a final common pathway within brainstem circuitry, highlighting the crucial role of subcortical networks in seizure propagation and lethality. Alternatively, or in addition, deficiency in CDKL5 expression in cardiac tissue, which has been linked to prolonged QT intervals and structural abnormalities in both CDD mice and patients,^{24, 25} may represent a significant contributor to mortality in the model. This convergence of neural and cardiac risk factors is observed in other monogenic models with audiogenic susceptibility,^{15, 26, 27} pointing to a multifaceted pathophysiology underlying SUDEP.

The present study has some limitations. First, the absence of simultaneous cardiorespiratory monitoring means we cannot conclusively determine whether cardiac arrhythmia, central apnoea, or both were the primary cause of death. Secondly, while sufficient for hypothesis generation and to confirm the presence of spontaneous seizures and SUDEP, our chronic EEG cohort was limited in size. The mice that displayed spontaneous seizures were from the same breeding pair, and although two further implanted knockout mice shared this lineage and did not display seizure behavior, we cannot rule out shared background genetic factors contributing to this phenotype. Although spontaneous seizures were not investigated in nonimplanted mice, it appeared that there may have been an interaction between electrode implantation and seizure expression in this model, which may have affected the reduced seizure propagation in the AGS cohort and led to a possible increased seizure propagation in the chronic cohort. Finally, this study was conducted in adult mice; investigating susceptibility at younger ages is warranted to fully understand the developmental trajectory of neural hyperexcitability in CDD. Altogether, these findings collectively establish the Cdkl5 knockout mouse as a robust platform for studying reflex epilepsy, spontaneous seizures, and seizure‐induced mortality.

AUTHOR CONTRIBUTIONS

Original conception of the work: O.M., J.H., and S.E. Contribution and design of the experiment: O.M., J.H., S.E., B.E‐M., and D.C.H. Collection of data: J.H., S.E., and O.M. Analysis and interpretation: J.H., S.E., and O.M. Drafting the article or revising it for important intellectual content: J.H., D.C.H., and O.M. All authors approved the final version of the manuscript.

FUNDING INFORMATION

This research was funded by Research Ireland (grant number 22/PATH‐S/10668 to O.M. and 16/RC/3948 and 21/RC/10294_P2, to D.C.H.). The article processing charge (APC) was funded by Research Ireland (grant number 22/PATH‐S/10668).

CONFLICT OF INTEREST STATEMENT

None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal's position on issues involved in ethics publication and affirm that this report is consistent with those guidelines.

ETHICS STATEMENT

All procedures involving animals were performed in compliance with EU Directive 2010/63/EU on the protection of animals used for scientific purposes and procedures were approved by RCSI University of Medicine and Health Sciences' Research Ethics Committee (REC 1587), and under license from the Ireland Health Products Regulatory Authority (AE19127/P087). Mice were housed at controlled conditions including a room temperature (20°C–25°C) and humidity (40%–60%) on a 12 h dark–light cycle with ad libitum access to water and food.

Supporting information

Table S1.

Video S1.

Video S2.

DATA AVAILABILITY STATEMENT

Data are included in the manuscript and Supporting Information are available under request.